ABSTRACT: Epilepsy is a neurological condition in which the electrical activity of groups of nerve cells or neurons in the brain becomes disturbed. In the extreme of cases, this abnormal activity causes convulsions, muscle spasms, and loss of consciousness in the patient. There are several factors that contribute to the triggering of an epileptic seizure. Among these factors are external stimuli such as flashing lights or sudden changes from dark to light, loud noises or monotonous sounds, or even certain musical notes.
This paper will focus on the development of a real-time DSP-based system that will assist a patient suffering from epileptic attacks that are specifically triggered by stimuli that consist of loud and sudden sounds from telephone ringing and pagers. It has been observed, however, that if the patient is forewarned of the impending occurrence of these sounds he is able to prepare himself for the event and avoid a seizure. Therefore, the system developed in this project aims to provide the kind of advanced warning needed by the patient.
The system will incorporate a parallel adaptive filter arrangement that will divide the available input audio spectrum into frequency bands that will be analyzed separately. An Adaptive Line Enhancer and Detector (ALE-D) structure will evaluate each frequency band for the purposes of extracting any periodic component present in the incoming input audio signal. This paper will discuss the offline performance of this structure or system model as well as the development of a real-time implementation.
The performance criterion for the offline implementation will be based on how accurately the system model will detect any audio signal including a ring component and how fast it can provide a timely warning signal to the patient that this seizure inducing sound is about to occur. To date, only the off-line system simulation model has been tested with pre-recorded data using Matlab and Simulink simulation packages. The results available for analysis in this paper have been produced using several recorded audio segments containing speech and different telephone rings. The challenges found in the ongoing real-time implementation of the system will be discussed.