Human tongue image analysis by sign change reflections for effective diagnosis of diseases

Abstract

Image analysis of the human tongue has been found to be useful in detecting various diseases in the body. The tongue image indicates the condition of different parts of the body and the changes in the tongue reflect the misbehavior of the internal parts of the body. So, the diagnosis of diseases is very much needed. The human tongue is detected and extracted effectively by implementing adaptive threshold segmentation. Gabor filter is used to identify the color and texture of the image. From the factors such as color, texture (coating), smoothness/cracks and size, the healthiness of the tongue is analyzed effectively. The threshold value taken from a healthy human tongue is used to classify a person s tongue whether it is normal or abnormal. If the tongue is abnormal, diseases, such as, thyroid, ulcer and diabetes can be diagnosed. In the first part of the work, Semi Supervised Learning (SSL) for the classification of tongue image is proposed. In the second part of the work, Adaptive Semi Supervised Learning (ASSL) for the classification of tongue image is proposed. In the third part of the work, diagnosis of diabetes in tongue image using Versatile Tooth-Marked Region (VTMR) classification is proposed. The three proposed methods are tested on the 96 BioHit tongue images collected from Sri Muthukumaran Medical College Hospital and Research Institute and 97 UV scanned tongue images captured from the patients by using IPhone with HD camera. newline