Share
array:24 [ "pii" => "S2173510723001404" "issn" => "21735107" "doi" => "10.1016/j.rxeng.2023.08.002" "estado" => "S300" "fechaPublicacion" => "2023-11-01" "aid" => "1528" "copyright" => "SERAM" "copyrightAnyo" => "2023" "documento" => "article" "crossmark" => 1 "subdocumento" => "fla" "cita" => "Radiologia. 2023;65:519-30" "abierto" => array:3 [ "ES" => false "ES2" => false "LATM" => false ] "gratuito" => false "lecturas" => array:1 [ "total" => 0 ] "Traduccion" => array:1 [ "es" => array:19 [ "pii" => "S0033833823001637" "issn" => "00338338" "doi" => "10.1016/j.rx.2023.08.001" "estado" => "S300" "fechaPublicacion" => "2023-11-01" "aid" => "1528" "copyright" => "SERAM" "documento" => "article" "crossmark" => 1 "subdocumento" => "fla" "cita" => "Radiologia. 2023;65:519-30" "abierto" => array:3 [ "ES" => false "ES2" => false "LATM" => false ] "gratuito" => false "lecturas" => array:1 [ "total" => 0 ] "es" => array:14 [ "idiomaDefecto" => true "cabecera" => "<span class="elsevierStyleTextfn">Original</span>" "titulo" => "Clasificadores de aprendizaje supervisado no lineales basados en radiómica de la TC cerebral sin contraste para predecir el pronóstico funcional en pacientes con hematoma intracerebral espontáneo" "tienePdf" => "es" "tieneTextoCompleto" => "es" "tieneResumen" => array:3 [ 0 => "es" 1 => "es" 2 => "en" ] "paginas" => array:1 [ 0 => array:2 [ "paginaInicial" => "519" "paginaFinal" => "530" ] ] "titulosAlternativos" => array:1 [ "en" => array:1 [ "titulo" => "Radiomic-based nonlinear supervised learning classifiers on non-contrast CT to predict functional prognosis in patients with spontaneous intracerebral hematoma" ] ] "contieneResumen" => array:2 [ "es" => true "en" => true ] "contieneTextoCompleto" => array:1 [ "es" => true ] "contienePdf" => array:1 [ "es" => true ] "resumenGrafico" => array:2 [ "original" => 1 "multimedia" => array:5 [ "identificador" => "fig0025" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => false "mostrarDisplay" => true "figura" => array:1 [ 0 => array:4 [ "imagen" => "fx1.jpeg" "Alto" => 754 "Ancho" => 1333 "Tamanyo" => 212256 ] ] ] ] "autores" => array:1 [ 0 => array:2 [ "autoresLista" => "E. Serrano, J. Moreno, L. Llull, A. Rodríguez, C. Zwanzger, S. Amaro, L. Oleaga, A. López-Rueda" "autores" => array:8 [ 0 => array:2 [ "nombre" => "E." "apellidos" => "Serrano" ] 1 => array:2 [ "nombre" => "J." "apellidos" => "Moreno" ] 2 => array:2 [ "nombre" => "L." "apellidos" => "Llull" ] 3 => array:2 [ "nombre" => "A." "apellidos" => "Rodríguez" ] 4 => array:2 [ "nombre" => "C." "apellidos" => "Zwanzger" ] 5 => array:2 [ "nombre" => "S." "apellidos" => "Amaro" ] 6 => array:2 [ "nombre" => "L." "apellidos" => "Oleaga" ] 7 => array:2 [ "nombre" => "A." "apellidos" => "López-Rueda" ] ] ] ] "resumen" => array:1 [ 0 => array:3 [ "titulo" => "Graphical abstract" "clase" => "graphical" "resumen" => "<span id="abst0005" class="elsevierStyleSection elsevierViewall"><p id="spar0005" class="elsevierStyleSimplePara elsevierViewall"><elsevierMultimedia ident="fig0025"></elsevierMultimedia></p></span>" ] ] ] "idiomaDefecto" => "es" "Traduccion" => array:1 [ "en" => array:9 [ "pii" => "S2173510723001404" "doi" => "10.1016/j.rxeng.2023.08.002" "estado" => "S300" "subdocumento" => "" "abierto" => array:3 [ "ES" => false "ES2" => false "LATM" => false ] "gratuito" => false "lecturas" => array:1 [ "total" => 0 ] "idiomaDefecto" => "en" "EPUB" => "https://multimedia.elsevier.es/PublicationsMultimediaV1/item/epub/S2173510723001404?idApp=UINPBA00004N" ] ] "EPUB" => "https://multimedia.elsevier.es/PublicationsMultimediaV1/item/epub/S0033833823001637?idApp=UINPBA00004N" "url" => "/00338338/0000006500000006/v1_202311020932/S0033833823001637/v1_202311020932/es/main.assets" ] ] "itemSiguiente" => array:19 [ "pii" => "S2173510723000691" "issn" => "21735107" "doi" => "10.1016/j.rxeng.2023.02.001" "estado" => "S300" "fechaPublicacion" => "2023-11-01" "aid" => "1477" "copyright" => "SERAM" "documento" => "article" "crossmark" => 1 "subdocumento" => "fla" "cita" => "Radiologia. 2023;65:531-45" "abierto" => array:3 [ "ES" => false "ES2" => false "LATM" => false ] "gratuito" => false "lecturas" => array:1 [ "total" => 0 ] "en" => array:13 [ "idiomaDefecto" => true "cabecera" => "<span class="elsevierStyleTextfn">Update in Radiology</span>" "titulo" => "Placenta accreta spectrum in early and late pregnancy from an imaging perspective. A scoping review" "tienePdf" => "en" "tieneTextoCompleto" => "en" "tieneResumen" => array:2 [ 0 => "en" 1 => "es" ] "paginas" => array:1 [ 0 => array:2 [ "paginaInicial" => "531" "paginaFinal" => "545" ] ] "titulosAlternativos" => array:1 [ "es" => array:1 [ "titulo" => "El espectro de placenta acreta en la etapa temprana y final del embarazo. Un repaso a través de la imagen" ] ] "contieneResumen" => array:2 [ "en" => true "es" => true ] "contieneTextoCompleto" => array:1 [ "en" => true ] "contienePdf" => array:1 [ "en" => true ] "resumenGrafico" => array:2 [ "original" => 0 "multimedia" => array:8 [ "identificador" => "fig0030" "etiqueta" => "Figure 6" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => true "mostrarDisplay" => false "figura" => array:1 [ 0 => array:4 [ "imagen" => "gr6.jpeg" "Alto" => 3404 "Ancho" => 2675 "Tamanyo" => 683039 ] ] "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0030" "detalle" => "Figure " "rol" => "short" ] ] "descripcion" => array:1 [ "en" => "<p id="spar0030" class="elsevierStyleSimplePara elsevierViewall"><span class="elsevierStyleBold">MRI features</span>. A 31-year-old woman at 24 weeks with placenta percreta. <span class="elsevierStyleBold">A</span>–<span class="elsevierStyleBold">C.</span> Placenta previa with heterogeneous signal in T2-weighted image is appreciated; T2-dark bands (narrow vertical arrow) are more conspicuous in the SSFSE sequence (A) compared to SSFP sequence (B). Subplacental hypervascularity (wide vertical arrow in A), placental bulge (wide horizontal arrow in A and C), normal T2-hypointense placental–myometrial interface (narrow horizontal arrow in A), myometrial thinning (narrow horizontal arrow in C), and placental tissue protruding into the bladder lumen (B) are all shown. <span class="elsevierStyleBold">D, E. Extrauterine extension.</span> Large flow voids in the bladder wall (bladder vessel sign, small arrows in D) imply bladder wall invasion, and parametrial vessel sign (large arrows in E) suggests parametrial invasion. <span class="elsevierStyleBold">F, G.</span> DWI clearly outlines the border between the placenta and myometrium, as placenta shows a very high signal intensity in diffusion imaging. Compare areas with normal myometrial thickness (F) to areas of significant myometrial thinning (G). Placenta (p); bladder (B); cervix (Cx).</p>" ] ] ] "autores" => array:1 [ 0 => array:2 [ "autoresLista" => "B. Moradi, J. Azadbakht, S. Sarmadi, M. Gity, E. Shirali, M. Azadbakht" "autores" => array:6 [ 0 => array:2 [ "nombre" => "B." "apellidos" => "Moradi" ] 1 => array:2 [ "nombre" => "J." "apellidos" => "Azadbakht" ] 2 => array:2 [ "nombre" => "S." "apellidos" => "Sarmadi" ] 3 => array:2 [ "nombre" => "M." "apellidos" => "Gity" ] 4 => array:2 [ "nombre" => "E." "apellidos" => "Shirali" ] 5 => array:2 [ "nombre" => "M." "apellidos" => "Azadbakht" ] ] ] ] ] "idiomaDefecto" => "en" "Traduccion" => array:1 [ "es" => array:9 [ "pii" => "S003383382300053X" "doi" => "10.1016/j.rx.2023.02.005" "estado" => "S300" "subdocumento" => "" "abierto" => array:3 [ "ES" => false "ES2" => false "LATM" => false ] "gratuito" => false "lecturas" => array:1 [ "total" => 0 ] "idiomaDefecto" => "es" "EPUB" => "https://multimedia.elsevier.es/PublicationsMultimediaV1/item/epub/S003383382300053X?idApp=UINPBA00004N" ] ] "EPUB" => "https://multimedia.elsevier.es/PublicationsMultimediaV1/item/epub/S2173510723000691?idApp=UINPBA00004N" "url" => "/21735107/0000006500000006/v2_202401250617/S2173510723000691/v2_202401250617/en/main.assets" ] "itemAnterior" => array:19 [ "pii" => "S2173510723001350" "issn" => "21735107" "doi" => "10.1016/j.rxeng.2022.11.007" "estado" => "S300" "fechaPublicacion" => "2023-11-01" "aid" => "1469" "copyright" => "SERAM" "documento" => "article" "crossmark" => 1 "subdocumento" => "fla" "cita" => "Radiologia. 2023;65:509-18" "abierto" => array:3 [ "ES" => false "ES2" => false "LATM" => false ] "gratuito" => false "lecturas" => array:1 [ "total" => 0 ] "en" => array:13 [ "idiomaDefecto" => true "cabecera" => "<span class="elsevierStyleTextfn">Original articles</span>" "titulo" => "Performance in prognostic capacity and efficiency of the Thoracic Care Suite GE AI tool applied to chest radiography of patients with COVID-19 pneumonia" "tienePdf" => "en" "tieneTextoCompleto" => "en" "tieneResumen" => array:2 [ 0 => "en" 1 => "es" ] "paginas" => array:1 [ 0 => array:2 [ "paginaInicial" => "509" "paginaFinal" => "518" ] ] "titulosAlternativos" => array:1 [ "es" => array:1 [ "titulo" => "Eficacia de la capacidad y la eficiencia pronósticas de la herramienta de inteligencia artificial Thoracic Care Suite de GE aplicada a la radiografía torácica de pacientes con neumonía COVID-19" ] ] "contieneResumen" => array:2 [ "en" => true "es" => true ] "contieneTextoCompleto" => array:1 [ "en" => true ] "contienePdf" => array:1 [ "en" => true ] "resumenGrafico" => array:2 [ "original" => 0 "multimedia" => array:8 [ "identificador" => "fig0015" "etiqueta" => "Figure 3" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => true "mostrarDisplay" => false "figura" => array:1 [ 0 => array:4 [ "imagen" => "gr3.jpeg" "Alto" => 2568 "Ancho" => 2925 "Tamanyo" => 318699 ] ] "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0125" "detalle" => "Figure " "rol" => "short" ] ] "descripcion" => array:1 [ "en" => "<p id="spar0080" class="elsevierStyleSimplePara elsevierViewall">Diagram of the included and excluded patients.</p> <p id="spar0085" class="elsevierStyleSimplePara elsevierViewall">CXR: chest X-ray; ED: emergency department; ICU: intensive care unit.</p>" ] ] ] "autores" => array:1 [ 0 => array:2 [ "autoresLista" => "J.M. Plasencia-Martínez, R. Pérez-Costa, M. Ballesta-Ruiz, J.M. García-Santos" "autores" => array:4 [ 0 => array:2 [ "nombre" => "J.M." "apellidos" => "Plasencia-Martínez" ] 1 => array:2 [ "nombre" => "R." "apellidos" => "Pérez-Costa" ] 2 => array:2 [ "nombre" => "M." "apellidos" => "Ballesta-Ruiz" ] 3 => array:2 [ "nombre" => "J.M." "apellidos" => "García-Santos" ] ] ] ] ] "idiomaDefecto" => "en" "Traduccion" => array:1 [ "es" => array:9 [ "pii" => "S0033833823000279" "doi" => "10.1016/j.rx.2022.11.012" "estado" => "S300" "subdocumento" => "" "abierto" => array:3 [ "ES" => false "ES2" => false "LATM" => false ] "gratuito" => false "lecturas" => array:1 [ "total" => 0 ] "idiomaDefecto" => "es" "EPUB" => "https://multimedia.elsevier.es/PublicationsMultimediaV1/item/epub/S0033833823000279?idApp=UINPBA00004N" ] ] "EPUB" => "https://multimedia.elsevier.es/PublicationsMultimediaV1/item/epub/S2173510723001350?idApp=UINPBA00004N" "url" => "/21735107/0000006500000006/v2_202401250617/S2173510723001350/v2_202401250617/en/main.assets" ] "en" => array:19 [ "idiomaDefecto" => true "cabecera" => "<span class="elsevierStyleTextfn">Original articles</span>" "titulo" => "Radiomic-based nonlinear supervised learning classifiers on non-contrast CT to predict functional prognosis in patients with spontaneous intracerebral hematoma" "tieneTextoCompleto" => true "paginas" => array:1 [ 0 => array:2 [ "paginaInicial" => "519" "paginaFinal" => "530" ] ] "autores" => array:1 [ 0 => array:4 [ "autoresLista" => "E. Serrano, J. Moreno, L. Llull, A. Rodríguez, C. Zwanzger, S. Amaro, L. Oleaga, A. López-Rueda" "autores" => array:8 [ 0 => array:3 [ "nombre" => "E." "apellidos" => "Serrano" "referencia" => array:1 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">a</span>" "identificador" => "aff0005" ] ] ] 1 => array:3 [ "nombre" => "J." "apellidos" => "Moreno" "referencia" => array:1 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">b</span>" "identificador" => "aff0010" ] ] ] 2 => array:3 [ "nombre" => "L." "apellidos" => "Llull" "referencia" => array:1 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">c</span>" "identificador" => "aff0015" ] ] ] 3 => array:3 [ "nombre" => "A." "apellidos" => "Rodríguez" "referencia" => array:1 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">c</span>" "identificador" => "aff0015" ] ] ] 4 => array:3 [ "nombre" => "C." "apellidos" => "Zwanzger" "referencia" => array:1 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">d</span>" "identificador" => "aff0020" ] ] ] 5 => array:3 [ "nombre" => "S." "apellidos" => "Amaro" "referencia" => array:1 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">c</span>" "identificador" => "aff0015" ] ] ] 6 => array:3 [ "nombre" => "L." "apellidos" => "Oleaga" "referencia" => array:1 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">e</span>" "identificador" => "aff0025" ] ] ] 7 => array:4 [ "nombre" => "A." "apellidos" => "López-Rueda" "email" => array:1 [ 0 => "alrueda81@hotmail.com" ] "referencia" => array:3 [ 0 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">e</span>" "identificador" => "aff0025" ] 1 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">f</span>" "identificador" => "aff0030" ] 2 => array:2 [ "etiqueta" => "<span class="elsevierStyleSup">*</span>" "identificador" => "cor0005" ] ] ] ] "afiliaciones" => array:6 [ 0 => array:3 [ "entidad" => "Departamento Radiología, Hospital Universitario Bellvitge, Hospitalet de Llobregat, Barcelona, Spain" "etiqueta" => "a" "identificador" => "aff0005" ] 1 => array:3 [ "entidad" => "Clínica Iribas-IRM, Asunción, Paraguay" "etiqueta" => "b" "identificador" => "aff0010" ] 2 => array:3 [ "entidad" => "Departamento de Neurología, Hospital Clínic, Barcelona, Spain" "etiqueta" => "c" "identificador" => "aff0015" ] 3 => array:3 [ "entidad" => "Departamento Radiología, Hospital del Mar, Barcelona, Spain" "etiqueta" => "d" "identificador" => "aff0020" ] 4 => array:3 [ "entidad" => "Departamento Radiología, Hospital Clínic, Barcelona, Spain" "etiqueta" => "e" "identificador" => "aff0025" ] 5 => array:3 [ "entidad" => "Servicio de Informática Clínica, Hospital Clínic, Barcelona, Spain" "etiqueta" => "f" "identificador" => "aff0030" ] ] "correspondencia" => array:1 [ 0 => array:3 [ "identificador" => "cor0005" "etiqueta" => "⁎" "correspondencia" => "Corresponding author." ] ] ] ] "titulosAlternativos" => array:1 [ "es" => array:1 [ "titulo" => "Clasificadores de aprendizaje supervisado no lineales basados en radiómica de la TC cerebral sin contraste para predecir el pronóstico funcional en pacientes con hematoma intracerebral espontáneo" ] ] "resumenGrafico" => array:2 [ "original" => 0 "multimedia" => array:8 [ "identificador" => "fig0015" "etiqueta" => "Figure 3" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => true "mostrarDisplay" => false "figura" => array:1 [ 0 => array:4 [ "imagen" => "gr3.jpeg" "Alto" => 1318 "Ancho" => 3018 "Tamanyo" => 232340 ] ] "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0015" "detalle" => "Figure " "rol" => "short" ] ] "descripcion" => array:1 [ "en" => "<p id="spar0015" class="elsevierStyleSimplePara elsevierViewall">Patient selection flowchart.</p>" ] ] ] "textoCompleto" => "<span class="elsevierStyleSections"><span id="sec0005" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0065">Introduction</span><p id="par0005" class="elsevierStylePara elsevierViewall">Cerebrovascular accidents (CVAs) are the second leading cause of death worldwide and one of the main causes of disability.<a class="elsevierStyleCrossRef" href="#bib0005"><span class="elsevierStyleSup">1</span></a> Spontaneous intracerebral haematomas (SICH) are the second most common type of CVA after ischaemic strokes and account for 10%–15% of all CVAs.<a class="elsevierStyleCrossRef" href="#bib0010"><span class="elsevierStyleSup">2</span></a></p><p id="par0010" class="elsevierStylePara elsevierViewall">Although the prognosis varies, SICHs are still a major cause of mortality and morbidity worldwide,<a class="elsevierStyleCrossRef" href="#bib0015"><span class="elsevierStyleSup">3</span></a> with a mortality rate of approximately 40% at one month, 54% at one year and 75% at five years. Only 12%–39% of patients regain functional independence.<a class="elsevierStyleCrossRef" href="#bib0015"><span class="elsevierStyleSup">3</span></a> The identification of patients at risk of poor functional prognosis improves triaging so these patients can be offered intensive therapies tailored to their needs.<a class="elsevierStyleCrossRef" href="#bib0020"><span class="elsevierStyleSup">4</span></a></p><p id="par0015" class="elsevierStylePara elsevierViewall">Since non-contrast computed tomography (NCCT) of the brain is the preferred diagnostic modality for acute SICHs,<a class="elsevierStyleCrossRef" href="#bib0025"><span class="elsevierStyleSup">5</span></a> several qualitative parameters (radiological signs) associated with SICH growth and poor prognosis have been identified. NCCT radiological signs provide a way of visualising the morphology (irregularity) and density (heterogeneity) of SICHs.</p><p id="par0020" class="elsevierStylePara elsevierViewall">However, these radiological signs<a class="elsevierStyleCrossRefs" href="#bib0030"><span class="elsevierStyleSup">6–12</span></a> may be interpreted with a degree of subjectivity and some of the definitions overlap,<a class="elsevierStyleCrossRef" href="#bib0065"><span class="elsevierStyleSup">13</span></a> with variable interobserver agreement having been published.<a class="elsevierStyleCrossRef" href="#bib0070"><span class="elsevierStyleSup">14</span></a> Furthermore, low sensitivity rates have been reported for the prognostic prediction of these signs (between 14.3% and 39.2% for functional prognosis according to Law et al.),<a class="elsevierStyleCrossRef" href="#bib0075"><span class="elsevierStyleSup">15</span></a> which may lead to erroneous estimates and eventually affect decision making.</p><p id="par0025" class="elsevierStylePara elsevierViewall">These limitations highlight the need to develop and use reproducible quantitative parameters (biomarkers) that minimise the subjective component of qualitative radiological assessment and improve the prognostic predictive performance of radiological signs.</p><p id="par0030" class="elsevierStylePara elsevierViewall">Radiomics is a quantitative approach in which a large number of biomarkers are extracted from radiological images.<a class="elsevierStyleCrossRefs" href="#bib0080"><span class="elsevierStyleSup">16,17</span></a></p><p id="par0035" class="elsevierStylePara elsevierViewall">By applying supervised machine-learning algorithms<a class="elsevierStyleCrossRef" href="#bib0090"><span class="elsevierStyleSup">18</span></a> to biomarkers, models can be built that are capable of predicting an outcome variable on previously unassessed data. Our hypothesis is that radiomics-based nonlinear supervised learning classifiers can predict poor functional prognosis in SICH patients by targeting features not visible to the human eye, such as intensity, shape and texture.</p><p id="par0040" class="elsevierStylePara elsevierViewall">The aim of this study is to evaluate whether nonlinear supervised learning classifiers based on radiomics of NCCTs of the brain can predict functional prognosis at discharge for SICH patients.</p></span><span id="sec0010" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0070">Material and methods</span><p id="par0045" class="elsevierStylePara elsevierViewall">This paper has been structured in accordance with the CLAIM (Checklist for Artificial Intelligence in Medical Imaging) initiative.<a class="elsevierStyleCrossRef" href="#bib0095"><span class="elsevierStyleSup">19</span></a></p><p id="par0050" class="elsevierStylePara elsevierViewall">Each SICH was segmented using 3D Slicer software (version 4.10.2). The data processing software used was Orange data mining, version 3.31 (<a href="https://orangedatamining.com/">https://orangedatamining.com/</a>).</p><span id="sec0015" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0075">Study design</span><p id="par0055" class="elsevierStylePara elsevierViewall">This study is based on a single-centre retrospective observational analysis of consecutive patients with a diagnosis of SICH confirmed by NCCT of the brain in a tertiary stroke centre, between January 2016 and April 2018. The objective is to create a nonlinear supervised learning algorithm for screening to predict functional prognosis at discharge for SICH patients. Given the objective, the target is to optimise sensitivity and negative predictive value metrics.</p></span><span id="sec0020" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0080">Data</span><p id="par0060" class="elsevierStylePara elsevierViewall">The study protocol was approved by the local Clinical Research Ethics Committee (registration number HCB/2020/0180) in accordance with national laws and regulations (Law 14/2007 of 3 July 2007 on Biomedical Research) and their international counterparts (Declaration of Helsinki, last updated in Fortaleza, Brazil, 2013). Given the retrospective nature of the study, specific informed consent was not required to include data in the study. To ensure the anonymity of the study participants, pseudonymisation was carried out. The dataset used in the study has not been used previously. The data supporting its conclusions are available from the author upon justified request.</p><p id="par0065" class="elsevierStylePara elsevierViewall">A total of 128 patients was initially assessed. All were over 18 years of age, had been diagnosed with SICH and had undergone an NCCT of the brain within 24 h of symptom onset. We excluded patients with secondary SICH from the sample, along with those for whom some radiomic variables were not available. We collected data on demographics (age and sex), toxic habits (alcohol and tobacco), cerebrovascular and cardiovascular risk factors (hypertension, dyslipidaemia, diabetes mellitus, atrial fibrillation and ischaemic heart disease), medical history of SICH or previous stroke, and concomitant antiplatelet or anticoagulant drug treatments.</p><p id="par0070" class="elsevierStylePara elsevierViewall">On admission, levels were recorded for systolic and diastolic blood pressure (mmHG), blood glucose (mmol/l) and a baseline neurological assessment was conducted using the National Institute of Health Stroke Scale (NIHSS). Functional prognosis at discharge was determined using the modified Rankin scale (mRS) and was divided into two categories: good prognosis (mRS 0–2) and poor prognosis (mRS 3–6).</p><p id="par0075" class="elsevierStylePara elsevierViewall">A sequential NCCT study of the brain was performed on two multislice CT scanners (Somatom Definition Flash and Somatom Sensation 64, Siemens Healthcare, Erlangen, Germany). Sequential axial images were obtained parallel to the orbitomeatal line from skull base to the vertex, using standard parameters of 140 kV, 230 mAs and axial reconstructions with a thickness of 5 mm.</p><p id="par0080" class="elsevierStylePara elsevierViewall">The location of the SICH was analysed (basal ganglia, lobar, brainstem and cerebellum), along with the presence of intraventricular haemorrhage and haematoma volume (ml). The latter was calculated using the validated A × B × C/2 method.<a class="elsevierStyleCrossRef" href="#bib0100"><span class="elsevierStyleSup">20</span></a></p><p id="par0085" class="elsevierStylePara elsevierViewall">The NCCT brain images included in the study for each patient were imported from the Picture Archiving and Communication System (PACS) to the 3D Slicer software (version 4.10.2), where the ‘Segment Editor’ module was used for segmentation.</p><p id="par0090" class="elsevierStylePara elsevierViewall">The segmentation process was performed by a qualified radiologist who was blinded to all clinical information. The contours of all the SICHs were manually drawn, slice by slice and three-dimensional volumes of interest (VOI) were formed for each SICH.</p><p id="par0095" class="elsevierStylePara elsevierViewall">Using 3D Slicer’s ‘Radiomics’ module, a total of 105 variables were automatically obtained from each of the VOI, related to the intensity, shape and texture of the haematoma. The ‘Radiomics’ module of the 3D Slicer is based on the PyRadiomics<a class="elsevierStyleCrossRef" href="#bib0105"><span class="elsevierStyleSup">21</span></a> library, which complies with the standards set out in the Image Biomarker Standardisation Initiative (IBSI).<a class="elsevierStyleCrossRef" href="#bib0110"><span class="elsevierStyleSup">22</span></a><a class="elsevierStyleCrossRef" href="#fig0005">Fig. 1</a> summarises the segmentation process.</p><elsevierMultimedia ident="fig0005"></elsevierMultimedia></span><span id="sec0025" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0085">Principal objective</span><p id="par0100" class="elsevierStylePara elsevierViewall">The main objective of this work was to evaluate whether supervised learning classifiers based on radiomics of NCCTs of the brain are able to predict poor functional prognosis at discharge for SICH patients. Poor prognosis was defined as the dependence of patients on others to perform day-to-day activities at discharge or death following the SICH (mRS 3–6). The outcome variable (mRS at discharge) was divided into two categories: good prognosis (functional independence at discharge: mRS 0–2) and poor prognosis (functional dependence or death at discharge: mRS 3–6). The mRS scale assesses global disability after a stroke and is the most comprehensive and widely used functional outcome measure in stroke trials. However, a certain level of interobserver variability has been reported, which may lead to misclassification and limit the validity of the results.<a class="elsevierStyleCrossRef" href="#bib0115"><span class="elsevierStyleSup">23</span></a> Therefore, the mRS score at discharge was determined by six vascular neurologists with more than five years of experience and who are qualified to perform this task. In addition, the decision was made to divide the mRS scores into two categories instead of using the original scale, as it has been shown that classification error rates are lower with dichotomisation than when ordinal scoring is used.<a class="elsevierStyleCrossRef" href="#bib0120"><span class="elsevierStyleSup">24</span></a></p></span><span id="sec0030" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0090">Data distribution</span><p id="par0105" class="elsevierStylePara elsevierViewall">The sample was divided into two stratified cohorts of patients, a training and testing cohort (70%, <span class="elsevierStyleItalic">n</span> = 70) and a validation cohort (30%, <span class="elsevierStyleItalic">n</span> = 29). The stratification ensured that the proportion of patients with good and poor functional prognoses in each cohort reflected that of the overall sample.</p></span><span id="sec0035" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0095">Model</span><p id="par0110" class="elsevierStylePara elsevierViewall">Orange data mining software, version 3.31 (<a href="https://orangedatamining.com/">https://orangedatamining.com/</a>), was used for data processing. All the radiomic variables obtained (both in the training and testing cohort and in the validation cohort) underwent initial processing to ensure the classifiers functioned correctly. This processing consisted, firstly, of excluding patients with missing values for any of the 105 variables. Secondly, the Isolation Forest algorithm was used to eliminate 5% of patients with extreme radiomics values. Finally, we carried out a standardisation and normalisation process to convert the radiomics variables into normally distributed variables with values between 0 and 1.</p></span><span id="sec0040" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0100">Dimensionality reduction — selection of variables</span><p id="par0115" class="elsevierStylePara elsevierViewall">Once initial processing had been completed for the radiomics variables, we selected the variables to be included in the study. First, we used all the variables without selecting or reducing their dimensionality (evaluating the 105 variables for each of the patients). We then applied dimensionality reduction techniques in order to identify and remove irrelevant and redundant information.<a class="elsevierStyleCrossRef" href="#bib0125"><span class="elsevierStyleSup">25</span></a> Since we had a sample of 70 patients to train the model, the dimensionality was reduced by seven variables to minimise the overfitting effect. The following methods were used for variable selection and dimensionality reduction:<ul class="elsevierStyleList" id="lis0005"><li class="elsevierStyleListItem" id="lsti0005"><span class="elsevierStyleLabel">-</span><p id="par0120" class="elsevierStylePara elsevierViewall">ANOVA method: we selected the seven variables with the greatest difference in mean values in the different groups, as long as there was no correlation between them (Spearman's correlation coefficient <0.5).</p></li><li class="elsevierStyleListItem" id="lsti0010"><span class="elsevierStyleLabel">-</span><p id="par0125" class="elsevierStylePara elsevierViewall">Linear dimensionality reduction algorithms:<ul class="elsevierStyleList" id="lis0010"><li class="elsevierStyleListItem" id="lsti0015"><span class="elsevierStyleLabel">•</span><p id="par0130" class="elsevierStylePara elsevierViewall">Principal Component Analysis (PCA)-80:<a class="elsevierStyleCrossRef" href="#bib0130"><span class="elsevierStyleSup">26</span></a> is a statistical procedure that orthogonally transforms the original n numerical dimensions of a data set into a new set of n dimensions called principal components. In this case, the new set will have as many dimensions as necessary to preserve 80% of the variation in the data.</p></li><li class="elsevierStyleListItem" id="lsti0020"><span class="elsevierStyleLabel">•</span><p id="par0135" class="elsevierStylePara elsevierViewall">PCA-90: in this case, the new set will have as many dimensions as are required to retain 90% of the variation in the data.</p></li></ul></p></li><li class="elsevierStyleListItem" id="lsti0025"><span class="elsevierStyleLabel">-</span><p id="par0140" class="elsevierStylePara elsevierViewall">Non-linear dimensionality reduction algorithms with which we selected seven new dimensions (variables) to represent our original variables:<ul class="elsevierStyleList" id="lis0015"><li class="elsevierStyleListItem" id="lsti0030"><span class="elsevierStyleLabel">•</span><p id="par0145" class="elsevierStylePara elsevierViewall"><span class="elsevierStyleItalic">tSNE</span>-7 (t-distributed stochastic neighbour embedding):<a class="elsevierStyleCrossRef" href="#bib0135"><span class="elsevierStyleSup">27</span></a> this algorithm calculates the probability that datapoint pairs in the high-dimensional space are related and then maps it into lower dimensions to produce a similar distribution.</p></li><li class="elsevierStyleListItem" id="lsti0035"><span class="elsevierStyleLabel">•</span><p id="par0150" class="elsevierStylePara elsevierViewall">Isomap-7 (isometric feature mapping):<a class="elsevierStyleCrossRef" href="#bib0140"><span class="elsevierStyleSup">28</span></a> an algorithm that projects data to a lower dimension, preserving the shortest distance between two points of a curve.</p></li><li class="elsevierStyleListItem" id="lsti0040"><span class="elsevierStyleLabel">•</span><p id="par0155" class="elsevierStylePara elsevierViewall">LLE-7 (locally linear embedding):<a class="elsevierStyleCrossRef" href="#bib0145"><span class="elsevierStyleSup">29</span></a> an algorithm that maintains the locally linear characteristics of the samples, so that each point can be represented as a linear, weighted sum of its neighbours.</p></li></ul></p></li></ul></p></span><span id="sec0045" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0105">Model construction</span><p id="par0160" class="elsevierStylePara elsevierViewall">To build the model, we used algorithms that had been used previously in the literature with good results in the assessment of SICH growth and prognosis.<a class="elsevierStyleCrossRefs" href="#bib0150"><span class="elsevierStyleSup">30–32</span></a> The following algorithms were applied:<ul class="elsevierStyleList" id="lis0020"><li class="elsevierStyleListItem" id="lsti0045"><span class="elsevierStyleLabel">-</span><p id="par0165" class="elsevierStylePara elsevierViewall">K-nearest neighbours (KNN) is one of the most basic classification algorithms. It is a non-parametric learning algorithm, i.e. it makes no assumptions about the functional form of the data. On the contrary, it is an instance-based algorithm, i.e. the algorithm does not learn a model, but memorises training instances that are used as a ‘knowledge base’ to make predictions.<a class="elsevierStyleCrossRefs" href="#bib0165"><span class="elsevierStyleSup">33,34</span></a><ul class="elsevierStyleList" id="lis0025"><li class="elsevierStyleListItem" id="lsti0050"><span class="elsevierStyleLabel">•</span><p id="par0170" class="elsevierStylePara elsevierViewall">Euclidean distance (KNN-E): is the straight line distance, or the shortest possible path, between two points.</p></li><li class="elsevierStyleListItem" id="lsti0055"><span class="elsevierStyleLabel">•</span><p id="par0175" class="elsevierStylePara elsevierViewall">Manhattan distance (KNN-M): the Manhattan distance between two points is the sum of the absolute differences between their coordinates. In other words, it is the sum of the lengths of the two legs of a right triangle. It is the distance between two points on a city street grid, where it is not possible to travel between the two points in a straight line.</p></li></ul></p></li><li class="elsevierStyleListItem" id="lsti0060"><span class="elsevierStyleLabel">-</span><p id="par0180" class="elsevierStylePara elsevierViewall">Support Vector Machine (SVM): an algorithm that represents the instances of the sample in the space, separating the classes into two spaces by means of a separation hyperplane. When new instances are introduced into the model, the space they belong to is used to classify them into one class or the other. An SVM constructs a hyperplane or set of hyperplanes in a high-dimensional space that can be used to solve classification or regression problems. The simplest way to perform the separation is by using a straight line—a straight plane—but the problems to be studied do not usually involve two dimensions. Typically, an SVM algorithm must deal with non-linear separating curves and more than two predictor variables. Kernel function representation offers a solution to this problem as it takes a low dimensional input space and transforms it into a higher dimensional space, i.e. it converts the non-separable problem into a separable one.<a class="elsevierStyleCrossRefs" href="#bib0165"><span class="elsevierStyleSup">33,35,36</span></a> We used Polynomial Kernel (P-SVM); Radial Kernel (R-SVM) and Sigmoid Kernel (S-SVM).</p></li><li class="elsevierStyleListItem" id="lsti0065"><span class="elsevierStyleLabel">-</span><p id="par0185" class="elsevierStylePara elsevierViewall">Random Forest (RF): This algorithm follows the bagging method of running several decision tree algorithms, i.e. the different trees see different proportions of data so each tree is trained on different samples of data for the same problem. To classify a new instance, each decision tree gives a classification, and finally the decision with the highest number of ‘votes’ is the prediction of the algorithm.<a class="elsevierStyleCrossRefs" href="#bib0165"><span class="elsevierStyleSup">33,37</span></a> We used algorithms based on the decision of 10 trees (RF-10) and 50 trees (RF-50).</p></li><li class="elsevierStyleListItem" id="lsti0070"><span class="elsevierStyleLabel">-</span><p id="par0190" class="elsevierStylePara elsevierViewall">CatBoost gradient boosting (GB): an algorithm built with individual decision trees trained sequentially, so that each new tree tries to improve on the errors of the previous one (boosting). The prediction of a new instance is obtained by aggregating the predictions of all the individual trees that make up the model.<a class="elsevierStyleCrossRefs" href="#bib0165"><span class="elsevierStyleSup">33,38</span></a></p></li></ul></p><p id="par0195" class="elsevierStylePara elsevierViewall"><a class="elsevierStyleCrossRef" href="#fig0010">Fig. 2</a> summarises the method of data processing, variable selection and model building.</p><elsevierMultimedia ident="fig0010"></elsevierMultimedia></span><span id="sec0050" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0110">Training and testing</span><p id="par0200" class="elsevierStylePara elsevierViewall">In the training and testing cohort, stratified 10-fold cross validation was performed. That is, the model was trained with 90% of the cases from the training and testing cohort and the remaining 10% of the sample was used for prediction. This was performed ten times and the mean was calculated using the values from the ten predictions that appear in the area under the ROC curve (AUC). No data augmentation strategies were used in our sample.</p><p id="par0205" class="elsevierStylePara elsevierViewall">Once the algorithms in the training and testing cohort had been trained, we used the validation cohort to make predictions. We calculated the sensitivity and negative predictive value of the classifiers in the validation cohort. The CI was defined at 95%.</p></span></span><span id="sec0055" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0115">Results</span><span id="sec0060" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0120">Data</span><p id="par0210" class="elsevierStylePara elsevierViewall">A total of 105 patients met the final inclusion and exclusion criteria and were analysed. The flow chart of the patients is shown in <a class="elsevierStyleCrossRef" href="#fig0015">Fig. 3</a>. The main demographic, clinical and imaging characteristics are shown in <a class="elsevierStyleCrossRefs" href="#tbl0005">Tables 1 and 2</a>.</p><elsevierMultimedia ident="fig0015"></elsevierMultimedia><elsevierMultimedia ident="tbl0005"></elsevierMultimedia><elsevierMultimedia ident="tbl0010"></elsevierMultimedia><p id="par0215" class="elsevierStylePara elsevierViewall">Data processing and training and testing cohort (<a class="elsevierStyleCrossRef" href="#fig0020">Fig. 4</a>)</p><elsevierMultimedia ident="fig0020"></elsevierMultimedia><p id="par0220" class="elsevierStylePara elsevierViewall">After excluding patients with missing values and applying the Isolation Forest algorithm (which eliminated 5% of patients with extreme values), we analysed the radiomics variables of 99 SICH patients. The sample was divided into a training and testing cohort (70%, <span class="elsevierStyleItalic">n</span> = 70) and a validation cohort (30%, <span class="elsevierStyleItalic">n</span> = 29). In the training and testing cohort, stratified 10-fold cross validation was carried out (the model was trained on 90% of the cases in the cohort and the remaining 10% of the sample was used for prediction). This was carried out 10 times and the mean was calculated using the values from the ten predictions that appear in the AUC. Once the algorithms were trained on the training and testing cohort, the validation cohort was used to make predictions. We calculated the sensitivity and negative predictive value of the classifiers in the validation cohort (<a class="elsevierStyleCrossRef" href="#fig0020">Fig. 4</a>).</p><p id="par0225" class="elsevierStylePara elsevierViewall">The mean AUC of the different variable and classifier selection methods with the training and testing cohort is summarised in <a class="elsevierStyleCrossRef" href="#tbl0015">Table 3</a>. KNN-E, P-SVM and RF-10, in combination with the ANOVA correlation feature selection method, were the best performing classifiers in the training and testing cohort (AUC of 0.752, 0.798 and 0.742, respectively). The seven radiomics variables selected using the ANOVA method were: ‘Run Length Non-Uniformity’, ‘Gray Level Non-Uniformity’, ‘High Gray Level Emphasis’, ‘Run Entropy’, ‘Busyness’, ‘Long Run Emphasis’ and ‘Interquartile Range’.</p><elsevierMultimedia ident="tbl0015"></elsevierMultimedia></span><span id="sec0065" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0125">Model performance</span><p id="par0230" class="elsevierStylePara elsevierViewall">After training the different models, the classifiers were evaluated using the data from the validation cohort (<span class="elsevierStyleItalic">n</span> = 29) to predict poor patient functional prognosis at discharge, and the classifier prediction was compared to the functional prognosis of these patients. <a class="elsevierStyleCrossRef" href="#tbl0020">Table 4</a> shows the sensitivity of the classifiers in the validation cohort. KNN-E, KNN-M, P-SVM and RF-10 in combination with the ANOVA correlation feature selection method were the best performing classifiers in the validation cohort, with a sensitivity of 0.897 (CI 95%: 0.778−1), with no false negatives, a positive predictive value of 89% and a negative predictive value of 100%. In combination with the Isomap-7 variable selection method, the RF-10 classifier achieved the same sensitivity rate. <a class="elsevierStyleCrossRef" href="#tbl0025">Table 5</a> shows the confusion matrix of the five models with the best sensitivity results.</p><elsevierMultimedia ident="tbl0020"></elsevierMultimedia><elsevierMultimedia ident="tbl0025"></elsevierMultimedia></span></span><span id="sec0070" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0130">Discussion</span><p id="par0235" class="elsevierStylePara elsevierViewall">In this retrospective study, we developed different radiomics-based nonlinear supervised learning models to predict functional prognosis at discharge for SICH patients. In combination with the ANOVA variable selection method, the P-SVM, KNN-E and RF-10 algorithms were the best performing classifiers in the training and testing cohort (AUC of 0.798, 0.752 and 0.742, respectively). In the validation cohort, the predictions of these models had a sensitivity of 0.897 (95% CI: 0.778−1), with a false negative rate of 0% for predicting poor functional prognosis at discharge. Identifying patients at risk of poor functional prognosis improves triaging, ensuring that these patients can be offered intensive therapies tailored to their needs.</p><p id="par0240" class="elsevierStylePara elsevierViewall">The main strength of this analysis is that it is the first study based on non-linear supervised learning algorithms in which the principal objective is to predict functional prognosis at discharge for SICH patients.</p><p id="par0245" class="elsevierStylePara elsevierViewall">There are two previously published studies based on nonlinear supervised learning algorithms, whose objective was the prediction of SICH growth. The first was published in 2019 by Hui Li et al.<a class="elsevierStyleCrossRef" href="#bib0195"><span class="elsevierStyleSup">39</span></a> The study investigated whether the radiomics values obtained from the NCCT of the brain could predict the growth of SICH. After the data selection process, they analysed four radiomics variables to build the model and applied 23 supervised learning algorithms. The best performing predictor of SICH growth was the Linear Support Vector Classifier (<a class="elsevierStyleCrossRef" href="#tbl0030">Table 6</a>).</p><elsevierMultimedia ident="tbl0030"></elsevierMultimedia><p id="par0250" class="elsevierStylePara elsevierViewall">The second study was carried out by Song,<a class="elsevierStyleCrossRef" href="#bib0150"><span class="elsevierStyleSup">30</span></a> and aimed to determine whether the NCCT of the brain models based on radiomics values and supervised learning algorithms could improve the prediction of early haematoma expansion in SICH patients. The authors built several models to predict SICH growth: the radiological model, radiomic model, clinical-radiological model, radiomics-radiological model and a combined model. From their results they observed that the radiomic model (in particular the logistic regression algorithm) demonstrated better performance and higher sensitivity than the clinical-radiological and radiological models (<a class="elsevierStyleCrossRef" href="#tbl0030">Table 6</a>).</p><p id="par0255" class="elsevierStylePara elsevierViewall">Another strength of this study is that, despite its retrospective nature, the image acquisition and reconstruction protocol is standardised and there is no variability. Due to the retrospective nature of most radiomics studies, imaging protocols—including acquisition—and reconstruction settings are not often controlled or standardised.</p><p id="par0260" class="elsevierStylePara elsevierViewall">A number of investigations have evaluated the impact of these parameters (such as voltage, milliamperage, pitch, field of view, slice thickness, acquisition, manufacturer and movement) and their influence on radiomics variables. In 2016, Lu et al.<a class="elsevierStyleCrossRef" href="#bib0200"><span class="elsevierStyleSup">40</span></a> evaluated agreement in radiomics values when slice thickness parameters and the NCCT of the brain reconstruction algorithm varied. They concluded that the use of different reconstruction algorithms and slice thickness led to variation, highlighting the importance of standardising image acquisition. In our analysis we used two different devices, both manufactured by Siemens Healthcare (the Somatom Definition Flash and Somatom Sensation 64). Future research should investigate whether the radiomics characteristics vary if different brands are used.</p><p id="par0265" class="elsevierStylePara elsevierViewall">One of the limitations of this study is that, as mentioned in the introduction, SICH is one of the main causes of disability, so a high percentage of patients included in this study (84.8%) had a poor functional prognosis at discharge (mRS 3–6). This means that the groups in the training and testing cohort and the validation cohort are unbalanced.</p><p id="par0270" class="elsevierStylePara elsevierViewall">SICH is a pathology with high morbidity and mortality rates, which means that the pre-test probability of poor functional prognosis at discharge in these patients is high. The pre-test probability of poor functional prognosis after SICH is 0.86 in the overall sample, and after applying the radiomics-based supervised learning classifiers, we obtained a sensitivity of 0.89 in the validation cohort. This is the main limitation of the study, as the probability of predicting prognosis does not increase substantially once the model is applied. The performance of the model could be improved by increasing the sample size in order to achieve better balanced cohorts.</p><p id="par0275" class="elsevierStylePara elsevierViewall">According to results published by Pszczolkowski et al.<a class="elsevierStyleCrossRef" href="#bib0205"><span class="elsevierStyleSup">41</span></a> and Huang et al.,<a class="elsevierStyleCrossRef" href="#bib0210"><span class="elsevierStyleSup">42</span></a> another measure that could be applied to improve the performance of our model is to create combined models in which information on demographic and clinical factors are incorporated into the radiomics model.</p><p id="par0280" class="elsevierStylePara elsevierViewall">Pszczolkowski et al.<a class="elsevierStyleCrossRef" href="#bib0205"><span class="elsevierStyleSup">41</span></a> evaluated the predictive performance of radiomics-based variables from NCCTs of the brain to predict not only SICH expansion, but also poor functional prognosis using generalised linear models. They also investigated the predictive performance of radiological signs and clinical factors independently and in combination with radiomics-based variables. They concluded that models which use radiomics-based variables from NCCTs of the brain outperform individual models that use radiological signs or clinical factors in isolation. In addition, they found that combined models, which incorporate demographic and clinical factors into the radiomics model, improved the prediction of poor prognosis for SICH patients (<a class="elsevierStyleCrossRef" href="#tbl0030">Table 6</a>).</p><p id="par0285" class="elsevierStylePara elsevierViewall">Similarly, in 2022, Huang et al.<a class="elsevierStyleCrossRef" href="#bib0210"><span class="elsevierStyleSup">42</span></a> evaluated the predictive performance of radiomics-based variables taken from NCCTs of the brain for SICH and perihaematomal oedema, and developed several models based on radiomics and clinical features to predict functional prognosis at three months, using generalised linear models. They showed that the combined radiomic and clinical model performed better and was more sensitive when it came to predicting poor prognosis in the training and testing cohort as well as in the internal and external validation cohort (<a class="elsevierStyleCrossRef" href="#tbl0030">Table 6</a>).</p><p id="par0290" class="elsevierStylePara elsevierViewall">Finally, another limitation of the study is that, while the SICH segmentation in the NCCT of the brain was performed by a specialist radiologist, the robustness of radiomics values with various segmentations was not analysed. Manual and semi-automated segmentation introduce observer bias, and studies have shown that many radiomics variables are not robust to intra- and inter-observer variations in ROI/VOI delineation.<a class="elsevierStyleCrossRef" href="#bib0215"><span class="elsevierStyleSup">43</span></a> Consequently, it would be advisable to perform intra- and inter-observer reproducibility assessments of the derived radiomics variables and to exclude non-reproducible variables. Nevertheless, the segmentation method we used is the same as that used previously in the literature.<a class="elsevierStyleCrossRef" href="#bib0220"><span class="elsevierStyleSup">44</span></a> We also consider that the high contrast between the SICH (hyperdense) and the rest of the adjacent structures in the NCCT facilitates the identification and segmentation of the SICH, and we do not believe that this invalidates our results.</p><p id="par0295" class="elsevierStylePara elsevierViewall">In conclusion, radiomics-based nonlinear supervised learning classifiers with machine learning methods are a promising diagnostic tool for predicting functional outcome at discharge for SICH patients, with a low false negative rate. However, larger, balanced studies combining radiomics and clinical features are still needed to develop and improve their performance.</p></span><span id="sec0075" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0135">Funding</span><p id="par0300" class="elsevierStylePara elsevierViewall">This research has not received any external funding.</p></span><span id="sec0080" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0140">Author contributions</span><p id="par0305" class="elsevierStylePara elsevierViewall"><ul class="elsevierStyleList" id="lis0030"><li class="elsevierStyleListItem" id="lsti0075"><span class="elsevierStyleLabel">1</span><p id="par0310" class="elsevierStylePara elsevierViewall">Research coordinators: ES, JM, LL, AR, CZ, SA, LO, ALR.</p></li><li class="elsevierStyleListItem" id="lsti0080"><span class="elsevierStyleLabel">2</span><p id="par0315" class="elsevierStylePara elsevierViewall">Development of study concept: ES, ALR, SA, LO.</p></li><li class="elsevierStyleListItem" id="lsti0085"><span class="elsevierStyleLabel">3</span><p id="par0320" class="elsevierStylePara elsevierViewall">Study design: ES, ALR, SA, LO.</p></li><li class="elsevierStyleListItem" id="lsti0090"><span class="elsevierStyleLabel">4</span><p id="par0325" class="elsevierStylePara elsevierViewall">Data collection: ES, JM, LL, AR, CZ, SA, LO, ALR.</p></li><li class="elsevierStyleListItem" id="lsti0095"><span class="elsevierStyleLabel">5</span><p id="par0330" class="elsevierStylePara elsevierViewall">Data analysis and interpretation: ES, JM, LL, AR, CZ, SA, LO, ALR.</p></li><li class="elsevierStyleListItem" id="lsti0100"><span class="elsevierStyleLabel">6</span><p id="par0335" class="elsevierStylePara elsevierViewall">Data processing: ALR.</p></li><li class="elsevierStyleListItem" id="lsti0105"><span class="elsevierStyleLabel">7</span><p id="par0340" class="elsevierStylePara elsevierViewall">Literature search: ES, ALR.</p></li><li class="elsevierStyleListItem" id="lsti0110"><span class="elsevierStyleLabel">8</span><p id="par0345" class="elsevierStylePara elsevierViewall">Writing of article: ES, ALR.</p></li><li class="elsevierStyleListItem" id="lsti0115"><span class="elsevierStyleLabel">9</span><p id="par0350" class="elsevierStylePara elsevierViewall">Critical review of the manuscript with intellectually relevant contributions: ES, JM, LL, AR, CZ, SA, LO, ALR.</p></li><li class="elsevierStyleListItem" id="lsti0120"><span class="elsevierStyleLabel">10</span><p id="par0355" class="elsevierStylePara elsevierViewall">Approval of the final version: ES, JM, LL, AR, CZ, SA, LO, ALR.</p></li></ul></p></span><span id="sec0085" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0145">Conflicts of interest</span><p id="par0360" class="elsevierStylePara elsevierViewall">The authors declare that they have no conflicts of interest.</p></span></span>" "textoCompletoSecciones" => array:1 [ "secciones" => array:12 [ 0 => array:3 [ "identificador" => "xres2079421" "titulo" => "Abstract" "secciones" => array:4 [ 0 => array:2 [ "identificador" => "abst0005" "titulo" => "Purpose" ] 1 => array:2 [ "identificador" => "abst0010" "titulo" => "Methods" ] 2 => array:2 [ "identificador" => "abst0015" "titulo" => "Results" ] 3 => array:2 [ "identificador" => "abst0020" "titulo" => "Conclusion" ] ] ] 1 => array:2 [ "identificador" => "xpalclavsec1773962" "titulo" => "Keywords" ] 2 => array:3 [ "identificador" => "xres2079420" "titulo" => "Resumen" "secciones" => array:4 [ 0 => array:2 [ "identificador" => "abst0025" "titulo" => "Objetivo" ] 1 => array:2 [ "identificador" => "abst0030" "titulo" => "Material y método" ] 2 => array:2 [ "identificador" => "abst0035" "titulo" => "Resultados" ] 3 => array:2 [ "identificador" => "abst0040" "titulo" => "Conclusión" ] ] ] 3 => array:2 [ "identificador" => "xpalclavsec1773961" "titulo" => "Palabras clave" ] 4 => array:2 [ "identificador" => "sec0005" "titulo" => "Introduction" ] 5 => array:3 [ "identificador" => "sec0010" "titulo" => "Material and methods" "secciones" => array:8 [ 0 => array:2 [ "identificador" => "sec0015" "titulo" => "Study design" ] 1 => array:2 [ "identificador" => "sec0020" "titulo" => "Data" ] 2 => array:2 [ "identificador" => "sec0025" "titulo" => "Principal objective" ] 3 => array:2 [ "identificador" => "sec0030" "titulo" => "Data distribution" ] 4 => array:2 [ "identificador" => "sec0035" "titulo" => "Model" ] 5 => array:2 [ "identificador" => "sec0040" "titulo" => "Dimensionality reduction — selection of variables" ] 6 => array:2 [ "identificador" => "sec0045" "titulo" => "Model construction" ] 7 => array:2 [ "identificador" => "sec0050" "titulo" => "Training and testing" ] ] ] 6 => array:3 [ "identificador" => "sec0055" "titulo" => "Results" "secciones" => array:2 [ 0 => array:2 [ "identificador" => "sec0060" "titulo" => "Data" ] 1 => array:2 [ "identificador" => "sec0065" "titulo" => "Model performance" ] ] ] 7 => array:2 [ "identificador" => "sec0070" "titulo" => "Discussion" ] 8 => array:2 [ "identificador" => "sec0075" "titulo" => "Funding" ] 9 => array:2 [ "identificador" => "sec0080" "titulo" => "Author contributions" ] 10 => array:2 [ "identificador" => "sec0085" "titulo" => "Conflicts of interest" ] 11 => array:1 [ "titulo" => "References" ] ] ] "pdfFichero" => "main.pdf" "tienePdf" => true "fechaRecibido" => "2023-05-21" "fechaAceptado" => "2023-08-03" "PalabrasClave" => array:2 [ "en" => array:1 [ 0 => array:4 [ "clase" => "keyword" "titulo" => "Keywords" "identificador" => "xpalclavsec1773962" "palabras" => array:6 [ 0 => "Acute cerebrovascular accident" 1 => "Cerebral hemorrhage" 2 => "CT scanner" 3 => "X-ray" 4 => "AI (artificial intelligence)" 5 => "Biomarker" ] ] ] "es" => array:1 [ 0 => array:4 [ "clase" => "keyword" "titulo" => "Palabras clave" "identificador" => "xpalclavsec1773961" "palabras" => array:6 [ 0 => "Accidente cerebrovascular agudo" 1 => "Hemorragia intracerebral" 2 => "CT scanner" 3 => "X-ray" 4 => "Inteligencia artificial" 5 => "Biomarcadores" ] ] ] ] "tieneResumen" => true "resumen" => array:2 [ "en" => array:3 [ "titulo" => "Abstract" "resumen" => "<span id="abst0005" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0010">Purpose</span><p id="spar0070" class="elsevierStyleSimplePara elsevierViewall">To evaluate if nonlinear supervised learning classifiers based on non-contrast CT can predict functional prognosis at discharge in patients with spontaneous intracerebral hematoma.</p></span> <span id="abst0010" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0015">Methods</span><p id="spar0075" class="elsevierStyleSimplePara elsevierViewall">Retrospective, single-center, observational analysis of patients with a diagnosis of spontaneous intracerebral hematoma confirmed by non-contrast CT between January 2016 and April 2018. Patients with HIE > 18 years and with TCCSC performed within the first 24 h of symptom onset were included. Patients with secondary spontaneous intracerebral hematoma and in whom radiomic variables were not available were excluded. Clinical, demographic and admission variables were collected. Patients were classified according to the Modified Rankin Scale (mRS) at discharge into good (mRS 0−2) and poor prognosis (mRS 3–6). After manual segmentation of each spontaneous intracerebral hematoma, the radiomics variables were obtained. The sample was divided into a training and testing cohort and a validation cohort (70−30% respectively). Different methods of variable selection and dimensionality reduction were used, and different algorithms were used for model construction. Stratified 10-fold cross-validation were performed on the training and testing cohort and the mean area under the curve (AUC) were calculated. Once the models were trained, the sensitivity of each was calculated to predict functional prognosis at discharge in the validation cohort.</p></span> <span id="abst0015" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0020">Results</span><p id="spar0080" class="elsevierStyleSimplePara elsevierViewall">105 patients with spontaneous intracerebral hematoma were analyzed. 105 radiomic variables were evaluated for each patient. P-SVM, KNN-E and RF-10 algorithms, in combination with the ANOVA variable selection method, were the best performing classifiers in the training and testing cohort (AUC 0.798, 0.752 and 0.742 respectively). The predictions of these models, in the validation cohort, had a sensitivity of 0.897 (0.778−1;95%CI), with a false-negative rate of 0% for predicting poor functional prognosis at discharge.</p></span> <span id="abst0020" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0025">Conclusion</span><p id="spar0085" class="elsevierStyleSimplePara elsevierViewall">The use of radiomics-based nonlinear supervised learning classifiers are a promising diagnostic tool for predicting functional outcome at discharge in HIE patients, with a low false negative rate, although larger and balanced samples are still needed to develop and improve their performance.</p></span>" "secciones" => array:4 [ 0 => array:2 [ "identificador" => "abst0005" "titulo" => "Purpose" ] 1 => array:2 [ "identificador" => "abst0010" "titulo" => "Methods" ] 2 => array:2 [ "identificador" => "abst0015" "titulo" => "Results" ] 3 => array:2 [ "identificador" => "abst0020" "titulo" => "Conclusion" ] ] ] "es" => array:3 [ "titulo" => "Resumen" "resumen" => "<span id="abst0025" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0035">Objetivo</span><p id="spar0090" class="elsevierStyleSimplePara elsevierViewall">Evaluar si clasificadores de aprendizaje supervisado no lineales basados en radiómica de la TC cerebral sin contraste (TCCSC), pueden predecir el pronóstico funcional al alta en pacientes con Hematoma intracerebral espontáneo (HIE).</p></span> <span id="abst0030" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0040">Material y método</span><p id="spar0095" class="elsevierStyleSimplePara elsevierViewall">Análisis observacional retrospectivo y unicéntrico de pacientes con diagnóstico de HIE confirmado por TCCSC entre enero 2016 y abril 2018. Se incluyeron pacientes con HIE > 18 años y con TCCSC realizado dentro de las primeras 24 horas del inicio de los síntomas. Se excluyeron los HIE secundarios y en los que no se disponía de las variables de radiómica. Se recogieron datos clínicos, demográficos y variables al ingreso. Los pacientes se clasificaron según la Escala Modificada de Rankin (mRS) al alta en buen (mRS 0−2) y mal pronóstico (mRS 3–6). Tras la segmentación manual de la TCCSC de cada HIE se obtuvieron las variables de radiómica. La muestra se dividió en una cohorte de entrenamiento y prueba y otra cohorte de validación (70−30% respectivamente). Se usaron diferentes métodos de selección de variables y reducción de dimensionalidad, así como diferentes algoritmos para la construcción del modelo. Se realizaron 10 iteraciones de validación cruzada estratificada en la cohorte de entrenamiento y prueba y se calculó la media de los valores de área bajo la curva (AUC). Una vez entrenados los modelos, se calculó la sensibilidad de cada uno para predecir el pronóstico funcional al alta en la cohorte de validación.</p></span> <span id="abst0035" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0045">Resultados</span><p id="spar0100" class="elsevierStyleSimplePara elsevierViewall">105 pacientes con HIE fueron analizados. Se evaluaron 105 variables de radiómica de cada paciente. Los algoritmos P-SVM, KNN-E y RF-10, en combinación con el método de selección de variables ANOVA, fueron los clasificadores con mejor rendimiento en la cohorte de entrenamiento y prueba (AUC 0.798, 0.752 y 0.742 respectivamente). Las predicciones de estos modelos, en la cohorte de validación, tuvieron una sensibilidad de 0,897 (0,778−1;95%IC), con una tasa de falsos negativos del 0% para la predicción de mal pronóstico funcional al alta.</p></span> <span id="abst0040" class="elsevierStyleSection elsevierViewall"><span class="elsevierStyleSectionTitle" id="sect0050">Conclusión</span><p id="spar0105" class="elsevierStyleSimplePara elsevierViewall">El uso de clasificadores de aprendizaje supervisado no lineales basados en radiómica son una herramienta de diagnóstico prometedora para predecir el resultado funcional al alta en pacientes con HIE, con una baja tasa de falsos negativos, aunque todavía son necesarios estudios con mayor tamaño muestral y balanceados para desarrollar y mejorar su rendimiento.</p></span>" "secciones" => array:4 [ 0 => array:2 [ "identificador" => "abst0025" "titulo" => "Objetivo" ] 1 => array:2 [ "identificador" => "abst0030" "titulo" => "Material y método" ] 2 => array:2 [ "identificador" => "abst0035" "titulo" => "Resultados" ] 3 => array:2 [ "identificador" => "abst0040" "titulo" => "Conclusión" ] ] ] ] "multimedia" => array:10 [ 0 => array:8 [ "identificador" => "fig0005" "etiqueta" => "Figure 1" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => true "mostrarDisplay" => false "figura" => array:1 [ 0 => array:4 [ "imagen" => "gr1.jpeg" "Alto" => 1221 "Ancho" => 3510 "Tamanyo" => 415539 ] ] "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0005" "detalle" => "Figure " "rol" => "short" ] ] "descripcion" => array:1 [ "en" => "<p id="spar0005" class="elsevierStyleSimplePara elsevierViewall">Segmentation process. NCCT images of the brain imported from Picture Archiving and Communication System (PACS) to 3D Slicer software (version 4.10.2), using the Segment Editor module for segmentation. The contours of each SICH were drawn manually slice by slice and the three-dimensional volumes of interest (VOI) of each SICH were created. From the ‘Radiomics’ module of the 3D Slicer software, a total of 105 variables were automatically obtained from each of the VOI, related to the intensity, shape and texture of the haematoma.</p>" ] ] 1 => array:8 [ "identificador" => "fig0010" "etiqueta" => "Figure 2" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => true "mostrarDisplay" => false "figura" => array:1 [ 0 => array:4 [ "imagen" => "gr2.jpeg" "Alto" => 1189 "Ancho" => 3010 "Tamanyo" => 381052 ] ] "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0010" "detalle" => "Figure " "rol" => "short" ] ] "descripcion" => array:1 [ "en" => "<p id="spar0010" class="elsevierStyleSimplePara elsevierViewall">Summary of data processing method, variable selection and model building.</p>" ] ] 2 => array:8 [ "identificador" => "fig0015" "etiqueta" => "Figure 3" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => true "mostrarDisplay" => false "figura" => array:1 [ 0 => array:4 [ "imagen" => "gr3.jpeg" "Alto" => 1318 "Ancho" => 3018 "Tamanyo" => 232340 ] ] "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0015" "detalle" => "Figure " "rol" => "short" ] ] "descripcion" => array:1 [ "en" => "<p id="spar0015" class="elsevierStyleSimplePara elsevierViewall">Patient selection flowchart.</p>" ] ] 3 => array:8 [ "identificador" => "fig0020" "etiqueta" => "Figure 4" "tipo" => "MULTIMEDIAFIGURA" "mostrarFloat" => true "mostrarDisplay" => false "figura" => array:1 [ 0 => array:4 [ "imagen" => "gr4.jpeg" "Alto" => 1249 "Ancho" => 3010 "Tamanyo" => 229048 ] ] "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0020" "detalle" => "Figure " "rol" => "short" ] ] "descripcion" => array:1 [ "en" => "<p id="spar0020" class="elsevierStyleSimplePara elsevierViewall">Training and validation procedure. A total of 105 patients met the inclusion criteria. Patients with missing values were excluded and the Isolation Forest algorithm eliminated 5% of patients with extreme values. After the initial processing, 99 patients were finally analysed. The sample was divided into two stratified cohorts of patients: a training and testing cohort (70%, <span class="elsevierStyleItalic">n</span> = 70) and a validation cohort (30%, <span class="elsevierStyleItalic">n</span> = 29). In the training and testing cohort, we performed stratified 10-fold cross-validation. Once the algorithms were trained on the training and testing cohort, predictions were made with the validation cohort.</p>" ] ] 4 => array:8 [ "identificador" => "tbl0005" "etiqueta" => "Table 1" "tipo" => "MULTIMEDIATABLA" "mostrarFloat" => true "mostrarDisplay" => false "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0025" "detalle" => "Table " "rol" => "short" ] ] "tabla" => array:1 [ "tablatextoimagen" => array:1 [ 0 => array:2 [ "tabla" => array:1 [ 0 => """ <table border="0" frame="\n \t\t\t\t\tvoid\n \t\t\t\t" class=""><tbody title="tbody"><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Age (years), mean (IQR)</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">78 (66−84) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Sex (male)</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">59 (56.2) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">History, n (%)</span></td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Alcohol \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">7 (6.7) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Tobacco \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">7 (6.7) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>High BP \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">66 (62.9) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Dyslipidaemia \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">41 (39) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Diabetes Mellitus \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">23 (21.9) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Atrial fibrillation \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">22 (21) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Ischaemic heart disease \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">8 (7.6) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Previous SICH \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">5 (4.8) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Previous stroke \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">14 (13.3) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Antiplatelet medications \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">26 (24.8) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Anticoagulants \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">25 (23.8) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Variables on admission</span></td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Blood glucose, median(IQR), Mmol/l \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">138 (114−173) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>SBP, median (IQR), mmHG \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">159 (141−188) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>DBP, median (IQR), mmHG \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">81 (67−100) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>NIHSS baseline score, median (IQR) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">15 (5−21) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">mRS on discharge, n (%)</span></td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>mRS 0−2 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">16 (15.2) \t\t\t\t\t\t\n \t\t\t\t</td></tr></tbody></table> """ ] "imagenFichero" => array:1 [ 0 => "xTab3443112.png" ] ] ] ] "descripcion" => array:1 [ "en" => "<p id="spar0025" class="elsevierStyleSimplePara elsevierViewall">Demographic characteristics.</p>" ] ] 5 => array:8 [ "identificador" => "tbl0010" "etiqueta" => "Table 2" "tipo" => "MULTIMEDIATABLA" "mostrarFloat" => true "mostrarDisplay" => false "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0030" "detalle" => "Table " "rol" => "short" ] ] "tabla" => array:1 [ "tablatextoimagen" => array:1 [ 0 => array:2 [ "tabla" => array:1 [ 0 => """ <table border="0" frame="\n \t\t\t\t\tvoid\n \t\t\t\t" class=""><thead title="thead"><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black"> \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Total, <span class="elsevierStyleItalic">n</span> (%) \t\t\t\t\t\t\n \t\t\t\t\t\t</th></tr></thead><tbody title="tbody"><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Location</span></td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Lobar \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">47 (44.8) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Deep \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">47 (44.8) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Cerebellar \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">8 (7.6) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Brainstem \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">3 (2.9) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Volume (ml), median (IQR)</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">17 (8−47) \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Ventricular extension</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">47 (44.8) \t\t\t\t\t\t\n \t\t\t\t</td></tr></tbody></table> """ ] "imagenFichero" => array:1 [ 0 => "xTab3443110.png" ] ] ] ] "descripcion" => array:1 [ "en" => "<p id="spar0030" class="elsevierStyleSimplePara elsevierViewall">Image characteristics.</p>" ] ] 6 => array:8 [ "identificador" => "tbl0015" "etiqueta" => "Table 3" "tipo" => "MULTIMEDIATABLA" "mostrarFloat" => true "mostrarDisplay" => false "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0035" "detalle" => "Table " "rol" => "short" ] ] "tabla" => array:2 [ "leyenda" => "<p id="spar0040" class="elsevierStyleSimplePara elsevierViewall">GB: Gradient boosting CatBoost; Isomap-7: Isometric feature mapping; KNN: K-nearest neighbours; KNN-E: Euclidean distance; KNN-M: Manhattan distance; LLE-7: locally linear embedding; PCA: Principal component analysis; P-SVM: Polynomial kernel; RF: Random forest; R-SMV: Radial kernel; S-SVM: Sigmoid kernel; SVM: Support vector machine; tSNE: t-distributed stochastic neighbour embedding.</p>" "tablatextoimagen" => array:1 [ 0 => array:2 [ "tabla" => array:1 [ 0 => """ <table border="0" frame="\n \t\t\t\t\tvoid\n \t\t\t\t" class=""><thead title="thead"><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col">AUC mean \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " colspan="8" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Classifiers</th></tr><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black"> \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">KNN-E \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">KNN-M \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">P-SVM \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">R-SVM \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">S-SVM \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">RF-10 \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">RF-50 \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">GB \t\t\t\t\t\t\n \t\t\t\t\t\t</th></tr></thead><tbody title="tbody"><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " colspan="9" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Variable selection methods</span></td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>No selection \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.554 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.472 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.693 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.575 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.330 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.514 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.607 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.709 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>ANOVA + SPERAMAN (7) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.752</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.629 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.798</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.636 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.522 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.742</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.715 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.690 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>PCA-80 (23) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.558 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.466 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.487 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.462 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.425 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.572 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.536 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.601 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>PCA-90 (34) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.494 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.528 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.384 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.419 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.512 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.548 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.581 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.505 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>tSNE-7 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.600 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.590 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.462 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.288 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.328 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.574 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.638 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.596 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Isomap-7 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.585 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.475 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.530 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.399 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.411 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.355 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.311 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.268 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>LLE-7 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.462 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.491 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.433 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.468 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.482 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.486 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.482 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.513 \t\t\t\t\t\t\n \t\t\t\t</td></tr></tbody></table> """ ] "imagenFichero" => array:1 [ 0 => "xTab3443111.png" ] ] ] ] "descripcion" => array:1 [ "en" => "<p id="spar0035" class="elsevierStyleSimplePara elsevierViewall">Mean AUC of classifiers after stratified 10-fold cross validation in the training and testing cohort.</p>" ] ] 7 => array:8 [ "identificador" => "tbl0020" "etiqueta" => "Table 4" "tipo" => "MULTIMEDIATABLA" "mostrarFloat" => true "mostrarDisplay" => false "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0040" "detalle" => "Table " "rol" => "short" ] ] "tabla" => array:2 [ "leyenda" => "<p id="spar0050" class="elsevierStyleSimplePara elsevierViewall">GB: Gradient Boosting CatBoost; Isomap-7: isometric feature mapping; KNN: K-nearest neighbours; KNN-E: Euclidean distance; KNN-M: Manhattan distance; LLE-7: locally linear embedding; PCA: Principal component analysis; P-SVM: Polynomial kernel; RF: Random forest; R-SMV: Radial kernel; S-SVM: Sigmoid kernel; SVM: Support vector machine; tSNE: t-distributed stochastic neighbour embedding.</p>" "tablatextoimagen" => array:1 [ 0 => array:2 [ "tabla" => array:1 [ 0 => """ <table border="0" frame="\n \t\t\t\t\tvoid\n \t\t\t\t" class=""><thead title="thead"><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col">AUC mean \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " colspan="8" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Classifiers</th></tr><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black"> \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">KNN-E \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">KNN-M \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">P-SVM \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">R-SVM \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">S-SVM \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">RF-10 \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">RF-50 \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">GB \t\t\t\t\t\t\n \t\t\t\t\t\t</th></tr></thead><tbody title="tbody"><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " colspan="9" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Variable selection methods</span></td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>No selection \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>ANOVA + SPERAMAN (7) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.897</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.897</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.897</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.897</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.793 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>PCA-80 (23) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>PCA-90 (34) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>tSNE-7 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.793 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.793 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Isomap-7 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleBold">0.897</span> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>LLE-7 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.793 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.862 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0.828 \t\t\t\t\t\t\n \t\t\t\t</td></tr></tbody></table> """ ] "imagenFichero" => array:1 [ 0 => "xTab3443115.png" ] ] ] ] "descripcion" => array:1 [ "en" => "<p id="spar0045" class="elsevierStyleSimplePara elsevierViewall">Sensitivity of classifiers in validation cohort.</p>" ] ] 8 => array:8 [ "identificador" => "tbl0025" "etiqueta" => "Table 5" "tipo" => "MULTIMEDIATABLA" "mostrarFloat" => true "mostrarDisplay" => false "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0045" "detalle" => "Table " "rol" => "short" ] ] "tabla" => array:1 [ "tablatextoimagen" => array:1 [ 0 => array:2 [ "tabla" => array:1 [ 0 => """ <table border="0" frame="\n \t\t\t\t\tvoid\n \t\t\t\t" class=""><thead title="thead"><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " colspan="4" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Prediction</th></tr><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black"> \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Good prognosis \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Poor prognosis \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Total \t\t\t\t\t\t\n \t\t\t\t\t\t</th></tr></thead><tbody title="tbody"><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " colspan="4" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleItalic">Reality</span></td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Good prognosis \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">1 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">3 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">4 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Poor prognosis \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">0 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">25 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">25 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"><span class="elsevierStyleHsp" style=""></span>Total \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">1 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">28 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">29 \t\t\t\t\t\t\n \t\t\t\t</td></tr></tbody></table> """ ] "imagenFichero" => array:1 [ 0 => "xTab3443113.png" ] ] ] ] "descripcion" => array:1 [ "en" => "<p id="spar0055" class="elsevierStyleSimplePara elsevierViewall">Confusion matrix of the five models with best sensitivity results in the validation cohort.</p>" ] ] 9 => array:8 [ "identificador" => "tbl0030" "etiqueta" => "Table 6" "tipo" => "MULTIMEDIATABLA" "mostrarFloat" => true "mostrarDisplay" => false "detalles" => array:1 [ 0 => array:3 [ "identificador" => "at0050" "detalle" => "Table " "rol" => "short" ] ] "tabla" => array:2 [ "leyenda" => "<p id="spar0065" class="elsevierStyleSimplePara elsevierViewall">AUC: area under curve; Sp: specificity; KNN-E: Euclidean distance; P-SVM: Polynomial kernel; RF: Random forest; S: sensitivity.</p>" "tablatextoimagen" => array:1 [ 0 => array:2 [ "tabla" => array:1 [ 0 => """ <table border="0" frame="\n \t\t\t\t\tvoid\n \t\t\t\t" class=""><thead title="thead"><tr title="table-row"><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Outcome variable \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Study \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Method \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Model \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black">Results \t\t\t\t\t\t\n \t\t\t\t\t\t</th><th class="td" title="\n \t\t\t\t\ttable-head\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t" scope="col" style="border-bottom: 2px solid black"> \t\t\t\t\t\t\n \t\t\t\t\t\t</th></tr></thead><tbody title="tbody"><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " rowspan="6" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">SICH growth</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Hui Li et al.<a class="elsevierStyleCrossRef" href="#bib0195"><span class="elsevierStyleSup">39</span></a> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Non-linear supervised learning algorithms \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics: Linear Support Vector Classifier \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.726; Sp 0.717; AUC 0.729</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " rowspan="5" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Song et al.<a class="elsevierStyleCrossRef" href="#bib0150"><span class="elsevierStyleSup">30</span></a></td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " rowspan="5" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Non-linear supervised learning algorithms</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiological: Black Hole \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.367; Sp 0.853; AUC 0.610</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Clinical-radiological \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.645; Sp 0.775; AUC 0.766</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics: Logistic regression \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.761; Sp 0.818; AUC 0.850</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics + radiological \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.795; Sp 0.879; AUC 0.867</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Combined (radiomics + radiological + clinical) \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.804; Sp 0.881; AUC 0.867</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " rowspan="6" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Prognosis + SICH growth</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " rowspan="6" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Pszczolkowski et al.<a class="elsevierStyleCrossRef" href="#bib0205"><span class="elsevierStyleSup">41</span></a></td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " rowspan="6" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Generalised linear models</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t"> \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Prognosis \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Growth \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.698; Sp 0.741; AUC 0.783 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.635; Sp 0.690; AUC 0.693 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiological \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.318; Sp 0.880; AUC 0.621 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.467; Sp 0.711; AUC 0.609 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics + radiological \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.698; Sp 0.741; AUC 0.783 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.635; Sp 0.69; AUC0.693 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Clinical \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.620; Sp 0.815; AUC 0.789 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.350; Sp 0.839; AUC 0.668 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics + clinical \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.694; Sp 0.826; AUC 0.818 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.650; Sp 0.711; AUC 0.704 \t\t\t\t\t\t\n \t\t\t\t</td></tr><tr title="table-row"><td class="td-with-role" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t ; entry_with_role_rowhead " rowspan="4" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">SICH prognosis</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " rowspan="3" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Huang et al.<a class="elsevierStyleCrossRef" href="#bib0210"><span class="elsevierStyleSup">42</span></a></td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " rowspan="3" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Generalised linear models</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.705; Sp 0.725; AUC 0.773</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Clinical \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.767; Sp 0.725; AUC 0.828</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics + clinical \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.775; Sp 0.739; AUC 0.844</td></tr><tr title="table-row"><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Our results \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Non-linear supervised learning algorithms \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">Radiomics:P-SVM, KNN-E and RF-10 \t\t\t\t\t\t\n \t\t\t\t</td><td class="td" title="\n \t\t\t\t\ttable-entry\n \t\t\t\t " colspan="2" align="left" valign="\n \t\t\t\t\ttop\n \t\t\t\t">S 0.897 (95%CI: 0.778−1)</td></tr></tbody></table> """ ] "imagenFichero" => array:1 [ 0 => "xTab3443114.png" ] ] ] ] "descripcion" => array:1 [ "en" => "<p id="spar0060" class="elsevierStyleSimplePara elsevierViewall">Summary of the results of the internal validation cohorts of radiomics work aimed at predicting growth and prognosis of spontaneous intracerebral haematoma (SICH), as mentioned in this manuscript.</p>" ] ] ] "bibliografia" => array:2 [ "titulo" => "References" "seccion" => array:1 [ 0 => array:2 [ "identificador" => "bibs0005" "bibliografiaReferencia" => array:44 [ 0 => array:3 [ "identificador" => "bib0005" "etiqueta" => "1" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Global, regional, and national burden of stroke and its risk factors, 1990-2019: A systematic analysis for the Global Burden of Disease Study 2019" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:1 [ 0 => "G.S. Collaborators" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1016/S1474-4422(21)00252-0" "Revista" => array:5 [ "tituloSerie" => "Lancet Neurol" "fecha" => "2021" "volumen" => "20" "paginaInicial" => "795" "paginaFinal" => "820" ] ] ] ] ] ] 1 => array:3 [ "identificador" => "bib0010" "etiqueta" => "2" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "International epidemiology of intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:3 [ 0 => "M.A. Ikram" 1 => "R.G. Wieberdink" 2 => "P.J. Koudstaal" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1007/s11883-012-0252-1" "Revista" => array:6 [ "tituloSerie" => "Curr Atheroscler Rep" "fecha" => "2012" "volumen" => "14" "paginaInicial" => "300" "paginaFinal" => "306" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/22538431" "web" => "Medline" ] ] ] ] ] ] ] ] 2 => array:3 [ "identificador" => "bib0015" "etiqueta" => "3" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Epidemiology, risk factors, and clinical features of intracerebral hemorrhage: an update" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:3 [ 0 => "S.J. An" 1 => "T.J. Kim" 2 => "B.W. Yoon" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.5853/jos.2016.00864" "Revista" => array:6 [ "tituloSerie" => "J Stroke" "fecha" => "2017" "volumen" => "19" "paginaInicial" => "3" "paginaFinal" => "10" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/28178408" "web" => "Medline" ] ] ] ] ] ] ] ] 3 => array:3 [ "identificador" => "bib0020" "etiqueta" => "4" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Predictors of hematoma expansion predictors after intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:6 [ 0 => "S. Chen" 1 => "B. Zhao" 2 => "W. Wang" 3 => "L. Shi" 4 => "C. Reis" 5 => "J. Zhang" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.18632/oncotarget.19366" "Revista" => array:6 [ "tituloSerie" => "Oncotarget" "fecha" => "2017" "volumen" => "8" "paginaInicial" => "89348" "paginaFinal" => "89363" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/29179524" "web" => "Medline" ] ] ] ] ] ] ] ] 4 => array:3 [ "identificador" => "bib0025" "etiqueta" => "5" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Guidelines for the management of spontaneous intracerebral hemorrhage: A guideline for healthcare professionals from the American Heart Association/American Stroke Association" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "J.C. Hemphill" 1 => "S.M. Greenberg" 2 => "C.S. Anderson" 3 => "K. Becker" 4 => "B.R. Bendok" 5 => "M. Cushman" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/STR.0000000000000069" "Revista" => array:6 [ "tituloSerie" => "Stroke" "fecha" => "2015" "volumen" => "46" "paginaInicial" => "2032" "paginaFinal" => "2060" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/26022637" "web" => "Medline" ] ] ] ] ] ] ] ] 5 => array:3 [ "identificador" => "bib0030" "etiqueta" => "6" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Island sign: an imaging predictor for early hematoma expansion and poor outcome in patients with intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "Q. Li" 1 => "Q.J. Liu" 2 => "W.S. Yang" 3 => "X.C. Wang" 4 => "L.B. Zhao" 5 => "X. Xiong" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/STROKEAHA.117.017985" "Revista" => array:6 [ "tituloSerie" => "Stroke" "fecha" => "2017" "volumen" => "48" "paginaInicial" => "3019" "paginaFinal" => "3025" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/29018128" "web" => "Medline" ] ] ] ] ] ] ] ] 6 => array:3 [ "identificador" => "bib0035" "etiqueta" => "7" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Swirl sign in intracerebral haemorrhage: Definition, prevalence, reliability and prognostic value" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:4 [ 0 => "E. Selariu" 1 => "E. Zia" 2 => "M. Brizzi" 3 => "K. Abul-Kasim" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1186/1471-2377-12-109" "Revista" => array:5 [ "tituloSerie" => "BMC Neurol" "fecha" => "2012" "volumen" => "12" "paginaInicial" => "109" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/23013418" "web" => "Medline" ] ] ] ] ] ] ] ] 7 => array:3 [ "identificador" => "bib0040" "etiqueta" => "8" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Black Hole Sign: Novel imaging marker that predicts hematoma growth in patients with intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "Q. Li" 1 => "G. Zhang" 2 => "X. Xiong" 3 => "X.C. Wang" 4 => "W.S. Yang" 5 => "K.W. Li" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/STROKEAHA.116.013186" "Revista" => array:6 [ "tituloSerie" => "Stroke" "fecha" => "2016" "volumen" => "47" "paginaInicial" => "1777" "paginaFinal" => "1781" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/27174523" "web" => "Medline" ] ] ] ] ] ] ] ] 8 => array:3 [ "identificador" => "bib0045" "etiqueta" => "9" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Blend sign predicts poor outcome in patients with intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "Q. Li" 1 => "W.S. Yang" 2 => "X.C. Wang" 3 => "D. Cao" 4 => "D. Zhu" 5 => "F.J. LV" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1371/journal.pone.0183082" "Revista" => array:3 [ "tituloSerie" => "PLoS One" "fecha" => "2017" "volumen" => "12" ] ] ] ] ] ] 9 => array:3 [ "identificador" => "bib0050" "etiqueta" => "10" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Noncontrast computed tomography hypodensities predict poor outcome in intracerebral hemorrhage patients" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "G. Boulouis" 1 => "A. Morotti" 2 => "H.B. Brouwers" 3 => "A. Charidimou" 4 => "M.J. Jessel" 5 => "E. Auriel" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/STROKEAHA.116.014425" "Revista" => array:6 [ "tituloSerie" => "Stroke" "fecha" => "2016" "volumen" => "47" "paginaInicial" => "2511" "paginaFinal" => "2516" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/27601380" "web" => "Medline" ] ] ] ] ] ] ] ] 10 => array:3 [ "identificador" => "bib0055" "etiqueta" => "11" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Noncontrast computed tomography markers of intracerebral hemorrhage expansion" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:5 [ 0 => "G. Boulouis" 1 => "A. Morotti" 2 => "A. Charidimou" 3 => "D. Dowlatshahi" 4 => "J.N. Goldstein" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/STROKEAHA.116.015062" "Revista" => array:6 [ "tituloSerie" => "Stroke" "fecha" => "2017" "volumen" => "48" "paginaInicial" => "1120" "paginaFinal" => "1125" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/28289239" "web" => "Medline" ] ] ] ] ] ] ] ] 11 => array:3 [ "identificador" => "bib0060" "etiqueta" => "12" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "The new Hematoma Maturity Score is highly associated with poor clinical outcome in spontaneous intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "E. Serrano" 1 => "A. López-Rueda" 2 => "J. Moreno" 3 => "A. Rodriguez" 4 => "L. Llull" 5 => "C. Zwanzger" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1007/s00330-021-08085-4" "Revista" => array:6 [ "tituloSerie" => "Eur Radiol" "fecha" => "2022" "volumen" => "32" "paginaInicial" => "290" "paginaFinal" => "299" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/34148109" "web" => "Medline" ] ] ] ] ] ] ] ] 12 => array:3 [ "identificador" => "bib0065" "etiqueta" => "13" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Standards for detecting, interpreting, and reporting noncontrast computed tomographic markers of intracerebral hemorrhage expansion" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "A. Morotti" 1 => "G. Boulouis" 2 => "D. Dowlatshahi" 3 => "Q. Li" 4 => "C.D. Barras" 5 => "C. Delcourt" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1002/ana.25563" "Revista" => array:6 [ "tituloSerie" => "Ann Neurol" "fecha" => "2019" "volumen" => "86" "paginaInicial" => "480" "paginaFinal" => "492" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/31364773" "web" => "Medline" ] ] ] ] ] ] ] ] 13 => array:3 [ "identificador" => "bib0070" "etiqueta" => "14" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Noncontrast computed tomography markers of outcome in intracerebral hemorrhage patients" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:6 [ 0 => "M. Quintas-Neves" 1 => "L. Marques" 2 => "L. Silva" 3 => "J.M. Amorim" 4 => "C. Ferreira" 5 => "J. Pinho" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1080/01616412.2019.1673279" "Revista" => array:6 [ "tituloSerie" => "Neurol Res" "fecha" => "2019" "volumen" => "41" "paginaInicial" => "1083" "paginaFinal" => "1089" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/31578942" "web" => "Medline" ] ] ] ] ] ] ] ] 14 => array:3 [ "identificador" => "bib0075" "etiqueta" => "15" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Noncontrast computed tomography signs as predictors of hematoma expansion, clinical outcome, and response to tranexamic acid in acute intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "Z.K. Law" 1 => "A. Ali" 2 => "K. Krishnan" 3 => "A. Bischoff" 4 => "J.P. Appleton" 5 => "P. Scutt" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/STROKEAHA.119.026128" "Revista" => array:5 [ "tituloSerie" => "Stroke" "fecha" => "2020" "volumen" => "51" "paginaInicial" => "121" "paginaFinal" => "128" ] ] ] ] ] ] 15 => array:3 [ "identificador" => "bib0080" "etiqueta" => "16" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Radiomics: Extracting more information from medical images using advanced feature analysis" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "P. Lambin" 1 => "E. Rios-Velazquez" 2 => "R. Leijenaar" 3 => "S. Carvalho" 4 => "R.G.P.M. van Stiphout" 5 => "P. Granton" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1016/j.ejca.2011.11.036" "Revista" => array:6 [ "tituloSerie" => "Eur J Cancer" "fecha" => "2012" "volumen" => "48" "paginaInicial" => "441" "paginaFinal" => "446" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/22257792" "web" => "Medline" ] ] ] ] ] ] ] ] 16 => array:3 [ "identificador" => "bib0085" "etiqueta" => "17" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Radiomics: Images are more than pictures, they are data" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:3 [ 0 => "R.J. Gillies" 1 => "P.E. Kinahan" 2 => "H. Hricak" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1148/radiol.2015151169" "Revista" => array:6 [ "tituloSerie" => "Radiology" "fecha" => "2016" "volumen" => "278" "paginaInicial" => "563" "paginaFinal" => "577" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/26579733" "web" => "Medline" ] ] ] ] ] ] ] ] 17 => array:3 [ "identificador" => "bib0090" "etiqueta" => "18" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Bits and bytes: The future of radiology lies in informatics and information technology" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:5 [ 0 => "J.A. Brink" 1 => "R.L. Arenson" 2 => "T.M. Grist" 3 => "J.S. Lewin" 4 => "D. Enzmann" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1007/s00330-016-4688-5" "Revista" => array:6 [ "tituloSerie" => "Eur Radiol" "fecha" => "2017" "volumen" => "27" "paginaInicial" => "3647" "paginaFinal" => "3651" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/28280932" "web" => "Medline" ] ] ] ] ] ] ] ] 18 => array:3 [ "identificador" => "bib0095" "etiqueta" => "19" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Checklist for Artificial Intelligence in Medical Imaging (CLAIM): A guide for authors and reviewers" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:3 [ 0 => "J. Mongan" 1 => "L. Moy" 2 => "C.E. Kahn" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1148/ryai.2020200029" "Revista" => array:3 [ "tituloSerie" => "Radiol Artif Intell" "fecha" => "2020" "volumen" => "2" ] ] ] ] ] ] 19 => array:3 [ "identificador" => "bib0100" "etiqueta" => "20" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "The ABCs of measuring intracerebral hemorrhage volumes" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "R.U. Kothari" 1 => "T. Brott" 2 => "J.P. Broderick" 3 => "W.G. Barsan" 4 => "L.R. Sauerbeck" 5 => "M. Zuccarello" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/01.str.27.8.1304" "Revista" => array:6 [ "tituloSerie" => "Stroke" "fecha" => "1996" "volumen" => "27" "paginaInicial" => "1304" "paginaFinal" => "1305" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/8711791" "web" => "Medline" ] ] ] ] ] ] ] ] 20 => array:3 [ "identificador" => "bib0105" "etiqueta" => "21" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Computational radiomics system to decode the radiographic phenotype" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "J.J.M. van Griethuysen" 1 => "A. Fedorov" 2 => "C. Parmar" 3 => "A. Hosny" 4 => "N. Aucoin" 5 => "V. Narayan" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1158/0008-5472.CAN-17-0339" "Revista" => array:6 [ "tituloSerie" => "Cancer Res" "fecha" => "2017" "volumen" => "77" "paginaInicial" => "e104" "paginaFinal" => "e107" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/29092951" "web" => "Medline" ] ] ] ] ] ] ] ] 21 => array:3 [ "identificador" => "bib0110" "etiqueta" => "22" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "The Image Biomarker Standardization Initiative: Standardized quantitative radiomics for high-throughput image-based phenotyping" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "A. Zwanenburg" 1 => "M. Vallières" 2 => "M.A. Abdalah" 3 => "H.J.W.L. Aerts" 4 => "V. Andreanczyk" 5 => "A. Apte" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1148/radiol.2020191145" "Revista" => array:6 [ "tituloSerie" => "Radiology" "fecha" => "2020" "volumen" => "295" "paginaInicial" => "328" "paginaFinal" => "338" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/32154773" "web" => "Medline" ] ] ] ] ] ] ] ] 22 => array:3 [ "identificador" => "bib0115" "etiqueta" => "23" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Reliability of the modified Rankin Scale: A systematic review" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:4 [ 0 => "T.J. Quinn" 1 => "J. Dawson" 2 => "M.R. Walters" 3 => "K.R. Lees" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1161/STROKEAHA.109.557256" "Revista" => array:6 [ "tituloSerie" => "Stroke" "fecha" => "2009" "volumen" => "40" "paginaInicial" => "3393" "paginaFinal" => "3395" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/19679846" "web" => "Medline" ] ] ] ] ] ] ] ] 23 => array:3 [ "identificador" => "bib0120" "etiqueta" => "24" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Quantification of errors in ordinal outcome scales using shannon entropy: Effect on sample size calculations" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:5 [ 0 => "P. Mandava" 1 => "C.S. Krumpelman" 2 => "J.N. Shah" 3 => "D.L. White" 4 => "T.A. Kent" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1371/journal.pone.0067754" "Revista" => array:3 [ "tituloSerie" => "PLoS One" "fecha" => "2013" "volumen" => "8" ] ] ] ] ] ] 24 => array:3 [ "identificador" => "bib0125" "etiqueta" => "25" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Feature selection for machine learning: Comparing a correlation-based filter approach to the wrapper" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:2 [ 0 => "M.A. Hall" 1 => "L.A. Smith" ] ] ] ] ] "host" => array:1 [ 0 => array:1 [ "Libro" => array:1 [ "fecha" => "1999" ] ] ] ] ] ] 25 => array:3 [ "identificador" => "bib0130" "etiqueta" => "26" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Machine learning: Algorithms, real-world applications and research directions" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:1 [ 0 => "I.H. Sarker" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1007/s42979-021-00592-x" "Revista" => array:5 [ "tituloSerie" => "SN Comput Sci" "fecha" => "2021" "volumen" => "2" "paginaInicial" => "160" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/33778771" "web" => "Medline" ] ] ] ] ] ] ] ] 26 => array:3 [ "identificador" => "bib0135" "etiqueta" => "27" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Visualizing data using t-SNE" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:2 [ 0 => "L. Van der Maaten" 1 => "G. Hinton" ] ] ] ] ] "host" => array:1 [ 0 => array:1 [ "Revista" => array:3 [ "tituloSerie" => "J Mach Learn Res" "fecha" => "2008" "volumen" => "9" ] ] ] ] ] ] 27 => array:3 [ "identificador" => "bib0140" "etiqueta" => "28" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "A global geometric framework for nonlinear dimensionality reduction" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:3 [ 0 => "J.B. Tenenbaum" 1 => "V. de Silva" 2 => "J.C. Langford" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1126/science.290.5500.2319" "Revista" => array:6 [ "tituloSerie" => "Science" "fecha" => "2000" "volumen" => "290" "paginaInicial" => "2319" "paginaFinal" => "2323" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/11125149" "web" => "Medline" ] ] ] ] ] ] ] ] 28 => array:3 [ "identificador" => "bib0145" "etiqueta" => "29" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Nonlinear dimensionality reduction by locally linear embedding" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:2 [ 0 => "S.T. Roweis" 1 => "L.K. Saul" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1126/science.290.5500.2323" "Revista" => array:6 [ "tituloSerie" => "Science" "fecha" => "2000" "volumen" => "290" "paginaInicial" => "2323" "paginaFinal" => "2326" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/11125150" "web" => "Medline" ] ] ] ] ] ] ] ] 29 => array:3 [ "identificador" => "bib0150" "etiqueta" => "30" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Noncontrast computed tomography-based radiomics analysis in discriminating early hematoma expansion after spontaneous intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "Z. Song" 1 => "D. Guo" 2 => "Z. Tang" 3 => "H. Liu" 4 => "X. Li" 5 => "S. Luo" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.3348/kjr.2020.0254" "Revista" => array:5 [ "tituloSerie" => "Korean J Radiol" "fecha" => "2021" "volumen" => "22" "paginaInicial" => "415" "paginaFinal" => "424" ] ] ] ] ] ] 30 => array:3 [ "identificador" => "bib0155" "etiqueta" => "31" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Introduction to radiomics" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "M.E. Mayerhoefer" 1 => "A. Materka" 2 => "G. Langs" 3 => "I. Läggström" 4 => "P. Szczypinski" 5 => "P. Gibbs" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.2967/jnumed.118.222893" "Revista" => array:5 [ "tituloSerie" => "J Nucl Med" "fecha" => "2020" "volumen" => "61" "paginaInicial" => "488" "paginaFinal" => "495" ] ] ] ] ] ] 31 => array:3 [ "identificador" => "bib0160" "etiqueta" => "32" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Clinical and radiographic predictors of intracerebral hemorrhage outcome" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:6 [ 0 => "F. Al-Mufti" 1 => "A.M. Thabet" 2 => "T. Singh" 3 => "M. El-Ghanem" 4 => "K. Amuluru" 5 => "C.D. Gandhi" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1159/000484571" "Revista" => array:6 [ "tituloSerie" => "Interv Neurol" "fecha" => "2018" "volumen" => "7" "paginaInicial" => "118" "paginaFinal" => "136" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/29628951" "web" => "Medline" ] ] ] ] ] ] ] ] 32 => array:3 [ "identificador" => "bib0165" "etiqueta" => "33" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Machine learning for medical imaging" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:4 [ 0 => "B.J. Erickson" 1 => "P. Korfiatis" 2 => "Z. Akkus" 3 => "T.L. Kline" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1148/rg.2017160130" "Revista" => array:5 [ "tituloSerie" => "Radiographics" "fecha" => "2017" "volumen" => "37" "paginaInicial" => "505" "paginaFinal" => "515" ] ] ] ] ] ] 33 => array:3 [ "identificador" => "bib0170" "etiqueta" => "34" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Improving nearest neighbor classification with cam weighted distance" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:2 [ 0 => "C.Y. Zhou" 1 => "Y.Q. Chen" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1016/j.patcog.2005.09.004" "Revista" => array:5 [ "tituloSerie" => "Pattern Recognition" "fecha" => "2006" "volumen" => "39" "paginaInicial" => "635" "paginaFinal" => "645" ] ] ] ] ] ] 34 => array:3 [ "identificador" => "bib0175" "etiqueta" => "35" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "A tutorial on support vector machines for pattern recognition" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:1 [ 0 => "C.J.C. Burges" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1023/A:1009715923555" "Revista" => array:5 [ "tituloSerie" => "Data Min Knowl Discov" "fecha" => "1998" "volumen" => "2" "paginaInicial" => "121" "paginaFinal" => "167" ] ] ] ] ] ] 35 => array:3 [ "identificador" => "bib0180" "etiqueta" => "36" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "An introduction to support vector machines and other Kernel-based learning methods" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:2 [ 0 => "N. Cristianini" 1 => "J. Shawe-Taylor" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1017/CBO9780511801389" "Libro" => array:2 [ "fecha" => "2000" "editorial" => "Cambridge University Press" ] ] ] ] ] ] 36 => array:3 [ "identificador" => "bib0185" "etiqueta" => "37" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Random forests" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:1 [ 0 => "L. Breiman" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1023/A:1010933404324" "Revista" => array:5 [ "tituloSerie" => "Machine Learning" "fecha" => "2001" "volumen" => "45" "paginaInicial" => "5" "paginaFinal" => "32" ] ] ] ] ] ] 37 => array:3 [ "identificador" => "bib0190" "etiqueta" => "38" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "CatBoost: Gradient boosting with categorical features support" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:3 [ 0 => "A.V. Dorogush" 1 => "V. Ershov" 2 => "A. Gulin" ] ] ] ] ] "host" => array:1 [ 0 => array:1 [ "Revista" => array:2 [ "tituloSerie" => "ArXiv" "fecha" => "2018" ] ] ] ] ] ] 38 => array:3 [ "identificador" => "bib0195" "etiqueta" => "39" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Radiomics features on non-contrast computed tomography predict early enlargement of spontaneous intracerebral hemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:6 [ 0 => "H. Li" 1 => "Y. Xie" 2 => "X. Wang" 3 => "F. Chen" 4 => "J. Sun" 5 => "X. Jiang" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1016/j.clineuro.2019.105491" "Revista" => array:3 [ "tituloSerie" => "Clin Neurol Neurosurg" "fecha" => "2019" "volumen" => "185" ] ] ] ] ] ] 39 => array:3 [ "identificador" => "bib0200" "etiqueta" => "40" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Assessing agreement between radiomic features computed for multiple CT imaging settings" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:4 [ 0 => "L. Lu" 1 => "R.C. Ehmke" 2 => "L.H. Schwartz" 3 => "B. Zhao" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1371/journal.pone.0166550" "Revista" => array:3 [ "tituloSerie" => "PLoS One" "fecha" => "2016" "volumen" => "11" ] ] ] ] ] ] 40 => array:3 [ "identificador" => "bib0205" "etiqueta" => "41" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Quantitative CT radiomics-based models for prediction of haematoma expansion and poor functional outcome in primary intracerebral haemorrhage" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "S. Pszczolkowski" 1 => "J.P. Manzano-Patrón" 2 => "Z.K. Law" 3 => "K. Krishnan" 4 => "A. Ali" 5 => "P.M. Bath" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1007/s00330-021-07826-9" "Revista" => array:6 [ "tituloSerie" => "Eur Radiol" "fecha" => "2021" "volumen" => "31" "paginaInicial" => "7945" "paginaFinal" => "7959" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/33860831" "web" => "Medline" ] ] ] ] ] ] ] ] 41 => array:3 [ "identificador" => "bib0210" "etiqueta" => "42" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Radiomics for prediction of intracerebral hemorrhage outcomes: A retrospective multicenter study" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "X. Huang" 1 => "D. Wang" 2 => "Q. Zhang" 3 => "Y. Ma" 4 => "H. Zhao" 5 => "S. Li" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1016/j.nicl.2022.103242" "Revista" => array:3 [ "tituloSerie" => "Neuroimage Clin" "fecha" => "2022" "volumen" => "36" ] ] ] ] ] ] 42 => array:3 [ "identificador" => "bib0215" "etiqueta" => "43" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Radiomics in medical imaging-"how-to" guide and critical reflection" "autores" => array:1 [ 0 => array:2 [ "etal" => false "autores" => array:5 [ 0 => "J.E. van Timmeren" 1 => "D. Cester" 2 => "S. Tanadini-Lang" 3 => "H. Alkadhi" 4 => "B. Baessler" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1186/s13244-020-00887-2" "Revista" => array:5 [ "tituloSerie" => "Insights Imaging" "fecha" => "2020" "volumen" => "11" "paginaInicial" => "91" "link" => array:1 [ 0 => array:2 [ "url" => "https://www.ncbi.nlm.nih.gov/pubmed/32785796" "web" => "Medline" ] ] ] ] ] ] ] ] 43 => array:3 [ "identificador" => "bib0220" "etiqueta" => "44" "referencia" => array:1 [ 0 => array:2 [ "contribucion" => array:1 [ 0 => array:2 [ "titulo" => "Admission computed tomography radiomic signatures outperform hematoma volume in predicting baseline clinical severity and functional outcome in the ATACH-2 trial intracerebral hemorrhage population" "autores" => array:1 [ 0 => array:2 [ "etal" => true "autores" => array:6 [ 0 => "S.P. Haider" 1 => "A.I. Qureshi" 2 => "A. Jain" 3 => "H. Tharmaseelan" 4 => "E.R. Berson" 5 => "T. Zeevi" ] ] ] ] ] "host" => array:1 [ 0 => array:2 [ "doi" => "10.1111/ene.15000" "Revista" => array:5 [ "tituloSerie" => "Eur J Neurol" "fecha" => "2021" "volumen" => "28" "paginaInicial" => "2989" "paginaFinal" => "3000" ] ] ] ] ] ] ] ] ] ] ] "idiomaDefecto" => "en" "url" => "/21735107/0000006500000006/v2_202401250617/S2173510723001404/v2_202401250617/en/main.assets" "Apartado" => array:4 [ "identificador" => "66551" "tipo" => "SECCION" "en" => array:2 [ "titulo" => "Original articles" "idiomaDefecto" => true ] "idiomaDefecto" => "en" ] "PDF" => "https://static.elsevier.es/multimedia/21735107/0000006500000006/v2_202401250617/S2173510723001404/v2_202401250617/en/main.pdf?idApp=UINPBA00004N&text.app=https://www.elsevier.es/" "EPUB" => "https://multimedia.elsevier.es/PublicationsMultimediaV1/item/epub/S2173510723001404?idApp=UINPBA00004N" ]
