Research about Tongue Image of Traditional Chinese Medicine(TCM) Based on Artificial Intelligence Technology

Abstract

Artificial intelligence has successfully integrated the frontier research in many fields, which is gradually mature. Among them, the intelligent diagnosis of traditional Chinese medicine(TCM) has entered the initial stage. Tongue diagnosis is one of the traditional methods of clinical history collection in TCM, as one part of the “Four Diagnosis of Traditional Chinese Medicine”, playing an irreplaceable role in diagnosing. Due to the influence of external environment such as light and temperature, tongue diagnosis lacks objective, quantitative and standard evaluation and application. Based on the Yolo deep learning technology, this paper used a classification method to construct a multi task learning model of tongue image, which realized the simultaneous identification of tongue color, fur color, crack and tooth mark in traditional Chinese tongue diagnosis. The tongue images of 200 subjects were labeled by labelImg, including 160 training data sets and 40 testing data sets. The model effect was evaluated by precision rate(P), recall rate(R) and accuracy rate(A). It could reliably complete the task of tongue feature identification and had a good migration ability, providing a theoretical basis for the application of intelligent diagnosis technology in the medical field.