Plaque Lesion Classification Fuzzy Model Based on Various Color Models
Yuslinda Wati Mohamad Yusof, Hadzli Hashim, Khairul Anam Sulaiman, S. Subahir, Noor Ezan Abdullah, Fairul Nazmie Osman · International Conference on Advances in System Simulation · 2011
This paper investigates discrimination of plaque lesion from other types of psoriasis lesions using fuzzy logic technology. The proposed intelligent model can aid dermatologist in doing pre-diagnosis of psoriasis lesion particularly in hospitals that are scarce of expert persons. Skin lesions can be represented in terms of enhanced image pixel indices from various color models such as RGB, HSV and YCbCr. These indices are used as inputs in designing an intelligent model based on fuzzy algorithm. However prior to that, numerical analysis is done statistically in order to select only significant color components that would infer plaque discrimination from the non-plaque group samples. The outcome of the designed fuzzy model has produced sensitivity and specificity of 72.72% and 90.09% respectively. Eventually, the overall accuracy of the fuzzy model is 81.82%, and is about 7% higher when compared to the optimized ANN model developed earlier from previous work.