Analysis of SIRM Fuzzy Systems in Classification of Epilepsy Risk Levels for Diabetic Neuropathy Patients using Cerebral Blood Flow and EEG Signals

摘要

This paper aims to compare two SIMM fuzzy systems,ie,(Mamdani and Sugeno methods) for classification of epilepsy risk levels in diabetic patients using Cerebral Blood Flow (CBF) and EEG signals.These fuzzy systems are having minimum number of fuzzy rules and tested with a group of 200 known diabetic patients.The-bell shaped membership function's slope function is derived from the parameters of EEG signals using aggregation operators.The performance and quality value are the parametric index of these fuzzy systems.Both methods are closely following each other in the four linguistic labels of the risk levels ie,normal,medium risk,high risk and very high risk.In the case of low epilepsy risk level,the first order Sugeno method out performs the Mamdani method.However,the problem of missed classification is removed in the both classifiers.But the false alarm is retained at 1.2% in the classifiers.The SIRM Sugeno fuzzy method has 99.5% of performance and quality value of 40.2 in compare with SIRM Mamdani fuzzy system's 98.58% and 36.56 respectively.A VLSI design and simulation of the SIRM Sugeno fuzzy model will be studied in future.