Intelligent Diagnostic System For Static Radionuclide Examination

The article presents the method of an informational synthesis of decision rules for radionuclide diagnostic system of static examination of the human on gamma camera in case of unbalanced training dataset. The making of input mathematical description for intelligent radionuclide diagnostic system is considered. The algorithms for segmentation and classification scintigraphy images based on informationextreme machine learning are developed. The algorithms are based on adaptive binary coding of feature vectors and optimization of geometrical parameters of feature space partition into classes with containers which build into a radial basis of binary Hamming space to maximize the information ability of system intended to radionuclide diagnostics. The modified S. Kulback's information criterion for estimate efficiency of diagnostic system is expressed in terms of positive and negative predictive value. The physical modeling of proposed algorithms is implemented by the example of bootstrap aggregating informationextreme classifiers intended to estimation of level of kidney lesion.