Objectives: The weight and volume of battery-powered wireless electroencephalographyud(EEG) systems are dominated by the batteries. Battery dimensions are inudturn determined by the required energy capacity, which is derived from the systemudpower consumption and required monitoring time. Data reduction may be carriedudout to reduce the amount of data transmitted and thus proportionally reduceudthe power consumption of the wireless transmitter, which dominates system powerudconsumption. This thesis presents two new data selection algorithms that, in additionudto achieving data reduction, also select EEG containing epileptic seizures andudspikes that are important in diagnosis.udMethods: The algorithms analyse short EEG sections, during monitoring, touddetermine the presence of candidate seizures or spikes. Phase information fromuddifferent frequency components of the signal are used to detect spikes. For seizureuddetection, frequencies below 10 Hz are investigated for a relative increase in frequencyudand/or amplitude.udSignificant attention has also been given to metrics in order to accurately evaluateudthe performance of these algorithms for practical use in the proposed system.udAdditionally, signal processing techniques to emphasize seizures within the EEGudand techniques to correct for broad-level amplitude variation in the EEG have beenudinvestigated.udResults: The spike detection algorithm detected 80% of spikes whilst achievingud50% data reduction, when tested on 992 spikes from 105 hours of 10-channel scalpudEEG data obtained from 25 adults. The seizure detection algorithm identified 94%udof seizures selecting 80% of their duration for transmission and achieving 79% dataudreduction. It was tested on 34 seizures with a total duration of 4158 s in a databaseudof over 168 hours of 16-channel scalp EEG obtained from 21 adults. These algorithmsudshow great potential for longer monitoring times from miniaturised wirelessudEEG systems that would improve electroclinical diagnosis of patients.
展开▼