V Congreso Latinoamericano de Ingeniería Biomédica (CLAIB2011)

Toolbox de Matlab para la Detección de Crisis Epilépticas
Márquez Bocalandro Yuniel, Martínez Montes Eduardo, Lage Castellanos Agustín

Última modificación: 2011-03-26 11:40


In this work we introduce a new Matlab Toolbox for the detection/classification of epileptic events. We aimed at designing a neuroinformatics tool with a user friendly interface in order to be used by researchers and users not familiarized with the statistical and computational methods behind the classification methodology. The toolbox consists in modules for data loading and management, for selecting the seizure patterns and for visualizing the main results. The current classification method is based on the simultaneous space-time-frequency characterization of the pattern electrical activity with Parallel Factor Analysis, although the modular implementation allow for including new methods for feature extraction and classification. In order to evaluate different classification methods, the toolbox also allows for an ROC analysis if a gold standard of classification is available.