metricas
covid
Buscar en
Radiología (English Edition)
Toda la web
Inicio Radiología (English Edition) Radiomic-based nonlinear supervised learning classifiers on non-contrast CT to p...
Journal Information
Vol. 65. Issue 6.
Pages 519-530 (November - December 2023)
Share
Share
Download PDF
More article options
Visits
10
Vol. 65. Issue 6.
Pages 519-530 (November - December 2023)
Original articles
Radiomic-based nonlinear supervised learning classifiers on non-contrast CT to predict functional prognosis in patients with spontaneous intracerebral hematoma
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
Visits
10
E. Serranoa, J. Morenob, L. Llullc, A. Rodríguezc, C. Zwanzgerd, S. Amaroc, L. Oleagae, A. López-Ruedae,f,
Corresponding author
alrueda81@hotmail.com

Corresponding author.
a Departamento Radiología, Hospital Universitario Bellvitge, Hospitalet de Llobregat, Barcelona, Spain
b Clínica Iribas-IRM, Asunción, Paraguay
c Departamento de Neurología, Hospital Clínic, Barcelona, Spain
d Departamento Radiología, Hospital del Mar, Barcelona, Spain
e Departamento Radiología, Hospital Clínic, Barcelona, Spain
f Servicio de Informática Clínica, Hospital Clínic, Barcelona, Spain
This item has received
Article information
Abstract
Full Text
Bibliography
Download PDF
Statistics
Figures (4)
Show moreShow less
Tables (6)
Table 1. Demographic characteristics.
Table 2. Image characteristics.
Table 3. Mean AUC of classifiers after stratified 10-fold cross validation in the training and testing cohort.
Table 4. Sensitivity of classifiers in validation cohort.
Table 5. Confusion matrix of the five models with best sensitivity results in the validation cohort.
Table 6. 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.
Show moreShow less
Abstract
Purpose

To evaluate if nonlinear supervised learning classifiers based on non-contrast CT can predict functional prognosis at discharge in patients with spontaneous intracerebral hematoma.

Methods

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.

Results

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.

Conclusion

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.

Keywords:
Acute cerebrovascular accident
Cerebral hemorrhage
CT scanner
X-ray
AI (artificial intelligence)
Biomarker
Resumen
Objetivo

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).

Material y método

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.

Resultados

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.

Conclusión

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.

Palabras clave:
Accidente cerebrovascular agudo
Hemorragia intracerebral
CT scanner
X-ray
Inteligencia artificial
Biomarcadores

Article

These are the options to access the full texts of the publication Radiología (English Edition)
Subscriber
Subscriber

If you already have your login data, please click here .

If you have forgotten your password you can you can recover it by clicking here and selecting the option “I have forgotten my password”
Purchase
Purchase article

Purchasing article the PDF version will be downloaded

Price 19.34 €

Purchase now
Contact
Phone for subscriptions and reporting of errors
From Monday to Friday from 9 a.m. to 6 p.m. (GMT + 1) except for the months of July and August which will be from 9 a.m. to 3 p.m.
Calls from Spain
932 415 960
Calls from outside Spain
+34 932 415 960
E-mail
Article options
es en pt

¿Es usted profesional sanitario apto para prescribir o dispensar medicamentos?

Are you a health professional able to prescribe or dispense drugs?

Você é um profissional de saúde habilitado a prescrever ou dispensar medicamentos