Study On Resting ECoG Analysis Based Functional Mapping Of Brain

Intraoperative functional mapping is a major medical problem of neurosurgery operation,which urgently needs to be solved.Task-related ECoG intraoperative mapping has the advantage of accurate,quick and noninvasive so far.But since intraoperative patients can't cooperate well at data sampling,task-related ECoG has strong interference of the signal and the high rate of false detection.The research hypothesis that resting-state ECoG(RS-ECoG)based intraoperative functional mapping cerebral motor area could be used to reduce the false detection rate.Firstly,RS-ECoG data were collected and divided into training group and test group,and then independent component analysis is used in a multiple leads of signal to identify interference noise source of eye movement signal.Results demonstrated that the interference noise source of eye movement signal in the RS-ECoG was eliminated.Secondly,an algorithm for feature extraction and classification of RS-ECoG in the functional mapping utilizing experimental data was designed and verified.Above all,the discrete db3 wavelet transform was utilized for six-level wavelet decomposition and for reconstruction of signals filtered from the single sub-band of the original electrocorticogram data.Then,a 7-D characteristic vector is the energy ratio in the constructed signal filtered from each single sub-band.By applying 10-fold cross validation method,wavelet-support vector machines(SVM)classification algorithm is trained and tested to screen out the algorithm with low false detection rate and low omission rate.Compared with the gold standard ECS mapping,the spatial coincidence rate for functional mapping in the cortex based on RS-ECoG can be up to 94.4% with a false detection rate of 6.25%.Accordingly,it is believed that there exists specificity of RS-ECoG in the cerebral cortex,and intraoperative functional mapping cerebral area based on RS-ECoG are more accurate,safe and rapid,thus being feasible in clinical practice.Finally,this paper proposes a kind of combined algorithm of precise and real-time locating for subdural matrix electrode to map brain functional area boundary,including the electrode coordinate extraction and adding of corresponding mark.First of all,choose 2 or 4 starting-point electrodes and confirm the best electrode coordinate of those starting point via the coordinates and region pick-up algorithm for single electrode by histogram analysis,centroid method,region growing method and then add the mark.Then,take the coordinate of starting-point electrode as the starting point and meanwhile combine with the point-by-point sorting algorithm for extended electrode to confirm the coordinate and mark of all other electrodes by up,down,left and right point-by-point extension.The experiments indicate that this combined algorithm has excellent locating performance for matrix electrode,especially for the matrix electrode frequently placed in arc in the clinical application of neurosurgery operation.