MER Feature Analysis and Selection for Computer-Aided GPe/GPi Navigation

Y. Thesen, P. Gemmar (Germany), and F. Hertel (Luxembourg)


Biomedical Signal Processing, Dystonia, Deep Brain Stimulation, Self-Organizing Maps, Feature Extraction, GPi.


Using microelectrode recordings (MER) to aid the navi gation of stimulation electrodes to the optimal target po sition during deep brain stimulation (DBS) surgery in pa tients suffering from Parkinson’s disease proved effective in the past and has become common practice. In recent years, pallidal DBS for severe cases of dystonia with the internal part of the globus pallidus (GPi) as the preferred stimulation target has become increasingly popular. How ever, recordings from dystonic patients have been found to be less conclusive in delineating the target structure from its surroundings. We studied microelectrode recordings obtained from 19 patients undergoing surgery for pallidal DBS and ob served several types of recordings exhibiting distinct fir ing characteristics. Various features were extracted and analyzed in order to find measures that characterize these signal types. Using statistic analysis and Self-Organizing Maps (SOMs), those features allowing the best separation between non-neuronal and neuronal recordings were iden tified. The information gained was used to generate a fuzzy classifier which is capable of differentiating between MERs of the two classes in most of the cases.

Important Links:

Go Back