Diabetes Mellitus Detection Based on Facial Block Texture Features Using the Gabor Filter

Abstract

Millions of people die from Diabetes Mellitus every year. Recently, researchers have discovered that Diabetes Mellitus can be detected in a non-invasive manner through the analysis of human facial blocks. Although algorithms have been developed to detect Diabetes Mellitus using facial block color features, use of its texture features to detect this disease has not been fully investigated. In this paper, we propose a novel method to detect Diabetes Mellitus based on facial block texture features using the Gabor filter. For Diabetes Mellitus detection we first select four blocks to represent a facial image. Next, we extract texture features using the Gabor filter from each facial block to represent the samples, where each facial block is defined by a single texture value. Afterwards, k-Nearest Neighbors and Support Vector Machine are applied for classification. Experimental results on a dataset show that the proposed method can distinguish Diabetes Mellitus and Healthy samples with an accuracy of 99.82%, a sensitivity of 99.64%, and a specificity of 100%, using a combination of facial blocks.