Generally, seizures are known to be a sudden surge of electrical activity in the brain which enables the effected person to behave abnormally for a short time period. These seizures are not diseases in themselves but in fact, are the symptoms of various different disorders that might be affecting some areas of the brain.
Electroencephalography (EEG) is a device used to measure the fluctuations in the brain activity through the electrodes, pasted to different areas of the brain. This machine helps record the spontaneous brain activity for a short time span which makes it a handy tool for detecting brain abnormality that could help diagnose the disease causing it.
The ultimate aim of this project is to come up with a remote EEG machine which is able to automatically detect a seizure, without the help of an expert.
CHB-MIT Scalp EEG Database is being used as the main data set. The data found has been pre-processed already. So, the basic idea was to change it to the MATLAB readable format and then apply various different features to it. After that, the GMM and the SVM classifiers were applied to understand the classification accuracy of the features. Once the feature set with the best classifier results have been identified, an improved version of the classifiers will be worked upon and then tested on the available data.
Real-time processing of the data will be done on a device known as the ‘Raspberry pi’ and the preferred language of programming will be python.