IoT-Enabled Biomedical Tongue Color Image Analysis using Optimized Dynamic Global-Structure Enhanced Multi-Channel Graph Neural Network

In the paper, an Optimized Dynamic global structure (DGS) improved multi-channel graph neural network (MGNN) dependent Biomedical Tongue Color Image Analysis in Internet of Things (DGSMGNN-BTCIA-IOT) is proposed. Initially, the input image are gathered from Tongue Color Image dataset. The gathered data is fed into preprocessing stage to reduce the noise and enhance the quality of the image. Then the preprocessing data is given into the feature extraction stage using High-Order Synchroextracting transform (HOST) to extract geometric features, like area, slope, centroids and perimeter. The extracted images are supplied to the DGSMGNN to efficiently classify tongue color classifications as Mixed Hemorrhoids (MH), Coronary Heart Diseases (CHD), Chronic Bronchitis (CB), Nephritis (NT), Verrucous Gastritis (VG), Pneumonia (PN), Nephritis Syndrome (NS), Chronic Kidney Disease (CKD), Nephritis Syndrome (NT), Chronic Cerebral Circulation Insufficiency (CCCI), Upper Respiratory Tract Infection (URT), Erosive Gastritis (EG), Coronary Heart Disease. Finally, Stock Exchange Trading Optimization Algorithm (SETOA) is used to optimize DGSMGNN which accurately classifies the tongue color. Then the proposed DGSMGNNBTCIA-IOT is implemented and some performance metrics are evaluated: accuracy, precision, recall, specificity, f1-score, Kappa, ROC and calculation time. The performance of the DGSMGNNBTCIA-IOT approach achieves 17.20%, 25.81% and 32.18% better accuracy, 17.22%, 22.01% and 30.08% higher precision and 18.62%, 28.21% and 31.38% higher recall when compared with existing techniques: IoT and harmonious DL based biomedical tongue color picture analysis for infection analysis and organization (IOT-DTBCI-DDC), Tongue picture division and tongue color organization depend on DL (TIC-TCC-DL) and Computational tongue color simulation in tongue diagnosis (CTC-TD) methods respectively.