Wavelet transform and cross-correlation as tools for seizure prediction.

This paper describes the detection of preictal bursting using wavelet transform application and cross-correlation analysis. The wavelet transform is applied to data reduction and signal pre-processing. The extracted features provide simplified signals to process by means of the cross-correlation technique. The algorithm has been tested with a set of preictal data, interictal data and spontaneous crises, to determinate its sensitivity and its specificity (False Prediction Rate). The seizure occurrence period and the seizure prediction horizon are also calculated. The algorithm's merits are: 1) high sensitivity and 2) easy implementation.