ExpACVO-Hybrid Deep learning: Exponential Anti Corona Virus Optimization enabled Hybrid Deep learning for tongue image segmentation towards diabetes mellitus detection

Abstract

A metabolic disease known as diabetes mellitus (DM) is primarily brought on by an increase in blood sugar levels. On the other hand, DM and the complications it causes, such as diabetic Retinopathy (DR), will quickly emerge as one of the major health challenges of the twenty-first century. This indicates a huge economic burden on health-related authorities and governments. The detection of DM in the earlier stage can lead to early diagnosis and a considerable drop in mortality. Therefore, in order to detect DM at an early stage, an efficient detection system having the ability to detect DM is required. An effective classification method, named Exponential Anti Corona Virus Optimization (ExpACVO) is devised in this research work for Diabetes Mellitus (DM) detection using tongue images. Here, the UNet-Conditional Random Field-Recurrent Neural Network (UNet-CRF-RNN) is used to segment the images, and the proposed ExpACVO algorithm is used to train the UNet-CRF-RNN. Deep Q Network (DQN) classifier is used for DM detection, and the proposed ExpACVO is used for DQN training. The proposed ExpACVO algorithm is a newly created formula that combines Anti Corona Virus Optimization(ACVO) with Exponential Weighted Moving Average (EWMA). With maximum testing accuracy, sensitivity, and specificity values of 0.932, 0.950, and 0.914, respectively, the developed technique thus achieved improved performance.