Plaque psoriasis diagnosis model with dominant pixel gradation from primary color space

In dermatology, characteristics of the "ABCD" information are useful features used by the expert domain in their morphological learning method for skin lesion identification. With the advancement of the computer vision technology, not only these features can be quantified in the digital image restoration and enhancement but also can be as input parameters for an intelligent diagnosis system. In this paper, clinical psoriasis lesion images are processed to produce the dominant pixel gradation indices in the primary color model. These reflectance indices gained under controlled environment are then used to design a ANN diagnosis model for plaque. The optimized model is evaluated and validated through analysis of the performance indicators regularly applied in medical research. Findings in this work have shown that the model has produced 75% in diagnostic accuracy with more than 80% achievement for both sensitivity and specificity.