Artificial Intelligence-Based Multidimensional Image Analysis for the Auxiliary Identification of COVID-19 Infection in Tongue Images

This study aims to improve the accuracy of Traditional Chinese Medicine (TCM) diagnosis for adenovirus and COVID-19 patients using multidimensional image analysis technology to analyze tongue image characteristics. These features of tongue image were extracted by a deep neural network based on the SSD framework. In extracted tongue images, key features include tongue color, tooth marks, moisture levels, and texture properties, such as roughness, direction, and contrast. Fuzzy K-Means cluster analysis was used to categorize the features to differentiate between viral infections. Statistical analysis revealed significant differences in tongue color R/G/B values, texture roughness, and contrast among the groups, highlighting their diagnostic value. These findings suggest advanced image analysis techniques can enhance TCM diagnostics' precision, offering new insights into disease mechanisms and aiding clinical decision-making.