Skip to Main Content
This paper emphasizes on a VLSI design for SIRM fuzzy processor in biomedical application. A novel approach aims to identify and design a simple robust fuzzy system with minimum number of fuzzy rules to classify the epilepsy risk level of diabetic patients from cerebral blood flow and EEG signals is discussed in this paper. Four different types of fuzzy models are designed and tested with 200 patients. The better fuzzy system is identified based on the performance and quality value. The tuned SIRM fuzzy system with five rules is selected which has performance of 98.58% and quality value of 36.56. VLSI design and simulation of the tuned SIRM fuzzy system is undertaken. VLSI system is tested for the same 200 cases of known diabetic patients. The average performance obtained for VLSI system is 98.28. This result indicates that the VLSI simulation closely follows the original MATLAB derived fuzzy system.
Date of Conference: 11-14 Dec. 2006