Methodische Untersuchungen zur simultanen Detektion und Klassifikation Rolandischer Spikeaktivität

Abstract

In a previous study the authors showed that rolandic spikes are characterised by typical field distributions of the spectral parameters instantaneous power and instantaneous frequency [14]. According to localisation of the focus and lateralisation of instantaneous power seven topographic spike classes were determined visually and verified with a Neural Network classifier (multi layer perceptron - MLP) [9]. Based on these results an algorithm for simultaneous detection and classification of rolandic spike activity was developed [7]. Aim of this study was to check the results of visual spike classification by means of a global optimising cluster algorithm and to test additional classifiers - Linear Discriminant Analysis (LDA) and a Cascade Correlation net (CC) - for topographic spike classification and their application in the developed spike detection algorithm. Essentially, the results of cluster analysis confirmed the visual spike classification. The number of „correct” classifications of visually selected instantaneous power distributions of rolandic spikes (7 classes) and non-spike activities (alpha- and EMC-activities) of 10 Routine EEG records was nearly the same for the three classifiers LDA, MLP and CC. Routine EEG records of three further children containing more than 900 spikes were used to compare the performance of the spike detection algorithm using LDA, MLP or CC with the results of visual spike detection by two experienced electroencephalographers. The best results were obtained with the MLP as classifier in the developed detection algorithm. The number of „false/positive” detections was significant lower than when using LDA or CC.