Machine learning based Computer Aided Tongue Diagnosis system for illness Prognosis

Abstract

Examining a patient's pulse, eyes, face, tongue, etc. has been a standard practice in medicine since ancient times. Because the tongue contains so much information, observing it is a hard task. The medical professional may learn something new by studying the Tongue's many regions, colours, and coatings. To diagnose internal organ problems by studying the tongue calls for a great deal of training and expertise. Disease prediction using tongue image analysis (DPTIA) was the central emphasis of the suggested model's artificial intelligence architecture. As a first step, the test picture undergoes pre-processing, various noise reduction processes, and colour improvements via the use of Fast Non-Local Mean (FNLM) filtering. To do this, the picture resolutions in the dataset are pre-processed. To further extract textural information, the Grey Level Cooccurrence Matrix (GLCM) is also used. The last step is to utilise the retrieved characteristics to identify the various illnesses using a Hybrid Extreme Learning Machine (HELM) classifier. A variety of diseases, including appendicitis, bronchitis, gastritis, heart disease, and pancreatitis, may be predicted using the DPTIA model. The proposed Computer Aided Tongue Diagnosis System (CATDS) model outperforms state-of-the-art methods like Random Forest and Support Vector Machine (SVM).