Tesis de doctorado: Statistical modeling and quantification of EEG signals: application to the characterization and onset detection in epileptic seizures
Fecha
2019-04
Autores
Título de la revista
ISSN de la revista
Título del volumen
Editor
Resumen
"Identificar la actividad cerebral epiléptica utilizando señales de electroencefalografía (EEG) en tiempo real es un problema difícil. Los métodos modernos de detección basados en técnicas avanzadas de aprendizaje automático, son efectivos pero requieren grandes conjuntos de datos de entrenamiento y son difíciles de implementar en sistemas de monitoreo en tiempo real, debido a su costo computacional relativamente alto. Esta tesis se centra en dos problemas centrales vinculados a la caracterización de las crisis epilépticas con señales de EEG. El primero se relaciona con la detección de inicio y el otro se refiere al reconocimiento de patrones epileptiformes. Usando el nuevo método de caracterización presentado en el capítulo 2, ambos problemas pueden implementarse en tiempo real y lograr un alto rendimiento de detección. En general, esta tesis permitió aportar cinco nuevas contribuciones para tratar los problemas desafiantes de la epilepsia."
"Identifying epileptic brain activity using electroencephalography signals (EEG) in real-time is a difficult problem. Modern detection methods based on advanced machine learning techniques are effective but require large training datasets, and are difficult to implement in real-time monitoring systems because of their relatively high computational cost. This thesis focuses on two central problems linked to the characterization of epileptic seizures with EEG signals. The first one is related to onset detection and the other one is about epileptiform pattern recognition. Using the new characterization method presented in chapter 2, both can be implemented in real-time and achieve a high detection performance. In general, this thesis brings five new contributions to deal with challenging epilepsy problems."
"Identifying epileptic brain activity using electroencephalography signals (EEG) in real-time is a difficult problem. Modern detection methods based on advanced machine learning techniques are effective but require large training datasets, and are difficult to implement in real-time monitoring systems because of their relatively high computational cost. This thesis focuses on two central problems linked to the characterization of epileptic seizures with EEG signals. The first one is related to onset detection and the other one is about epileptiform pattern recognition. Using the new characterization method presented in chapter 2, both can be implemented in real-time and achieve a high detection performance. In general, this thesis brings five new contributions to deal with challenging epilepsy problems."
Descripción
Palabras clave
EPILEPSIA, ELECTROENCEFALOGRAFIA, PROCESAMIENTO DE SEÑALES, ESTADISTICA, ALGORITMOS