Seminar DCSY - Kešner F.: Biological neuro-signla processing
In order to push forward understanding of the brain, many different techniques are being applied. This work will focus on electrical signals measured directly from the brain, specifically on detection of certain signal events such as inter-ictal spikes, which are one of the essential biomarkers used for an epilepsy diagnosis and research, since it is believed, that spikes participates in epileptiform process. The inter-ictal spikes can be recorded also by the scalp EEG technique but for better localization of their source, usually for surgical treatment of epilepsy, it is necessary to acquire intracranial recordings by depth electrodes and/or subdural electrode grids. Recordings are usually acquired in more than hundred channels simultaneously, and recording process runs for several hours per patient. With reasonable 5 kHz sampling rate, the generated data are of enormous size. These data would have to be analyzed by medical doctors - neurologists manually. It makes the need for an efficient automated detector, with good precision and sensitivity, obvious. Several algorithms for spike detection, from scalp EEG, already exist. But algorithms for spike detection in intracranial EEG (iEEG) are much more scarce. When the need for computational efficiency, or in other words, fast data processing is required in such algorithm, because of enormous data size and/or need for real-time detection capability, we are practically reaching a blank spot, which will be addressed by this work.