An Efficient Multiangle Weight Updated Haralick and Relevance Vector Machine Algorithm for Classifying Diabetic Foot from Medical Thermal Image

Abstract

Abstract Skin temperature assessment has gained attention in recent years for its ability to detect diabetes-related foot complications. Early detection of the complications can prevent devastating consequences. Hence, in this article, an efficient multiangle weight updated Haralick (MAWH) algorithm–based foot thermal image processing system is proposed for classification of features into diabetic and nondiabetic categories. Initially, the Gaussian noises in the medical infrared footprint images are preprocessed by the median filter. Then, the features from the preprocessed images are processed by the MAWH, primitive tint feature extraction, and convoluted Tamura pattern algorithms. From the extracted features, the optimal features are selected by the genetic algorithm–differential evolution–based feature subset algorithm. By exploiting the selected features, the relevance vector machine classifier classifies the features as diabetic or nondiabetic. To validate the performance of the proposed algorithm, it is compared with existing algorithms. The validation results prove that the proposed algorithm is more optimal than the existing algorithms for all metrics.