Detection of Diabetes Mellitus using Tongue images

Abstract

One of the major health problem faced by people around the world is diabetes mellitus. Tongue diagnosis (a non-invasive approach) is made to detect DM & NPDR in its early stages. It uses color, texture & geometry features for diagnosis. A color gamut of the tongue is confirmed with 12 colors describing the color characteristics of the tongue. To delineate the tongue texture attributes, texture feature values of various blocks are employed. There are 13 features extorted from dialect images includes measurements, areas, distance and ratios of the tongue and they are called as geometry characteristic features. Using these features, it is possible to differentiate non-proliferative diabetic retinopathy, diabetes mellitus & healthy human being using their tongue images.