Data mining in the process of localization and classification of subcorticals structures

Abstract

This study presents a set of statistical indexes which allow quantifying the quantity of information contained in physiological signals and its classification for better diagnosis. The physiological signals to considering are constituted by records of microelectrode (MER) obtained during deep brain stimulation (DBS) in parkinsonians patients. The MER corresponds to the subcorticals structures: Thalamus nucleus, Zone Incerta, Subthalamic nucleus and Substantia Nigra. The results show that by means of the statistical indexes obtained it is achieved to locate the different subcorticals structures and using as classifier the algorithm C4.5 of decision trees, is obtained a classification of 98.8 % between the structures. In conclusion, in view of the high precision obtained in the classification, the application of this type of statistical indexes could be used in the process of localization and classification of subcorticals structures, and mainly the subthalamic nucleus for neurostimulation.