Pattern Recognition Approaches for Identifying Subcortical Targets during Deep Brain Stimulation Surgery

Feature selection for KNN classifier to improve accurate detection of subthalamic nucleus during deep brain stimulation surgery in Parkinson’s patients

“…The best clinical effect depends on precise placement of the electrode in the target nucleus by stereotactic neurosurgery [1] . The most common DBS target is the subthalamic nucleus (STN), a little structure with an approximated size of 5.9×3.7×5 mm [2].…”

“…The typical trajectory of MER in the DBS surgery is following the midbrain structures: the thalamus, zona incerta, STN, and substantia nigra. These structures have different neurophysiological spiking and spike background patterns [2].…”

“…To train this models it is necessary to calculate temporal [2,3] and spectral [3,4] features from MER. An important aspect to consider in the process of real-time classification is the possibility of reducing the feature dimension.…”

“…Chaovalitwongse W. et al [2] proposed three classifiers: Naive Bayes, K-nearest neighbours (KNN) and decision tree using 14 features obtained from MER. From the three proposed classifiers, those obtaining better ratios were with decision tree, achieving a sensitivity and specificity around 85% for the STN brain area.…”

“…They did not study a reduction of feature dimension. In [2] and [6], the classification results were provided without statistical comparative analyses.…”