深度学习在舌象分割中的应用综述OA

The objectification and quantification of traditional Chinese medicine(TCM)tongue diagnosis is an important topic in the modernization of TCM.Tongue image segmentation,as the foundation and prerequisite for automated tongue image analysis,directly impacts the accuracy of subsequent feature analysis.Deep learning technologies have made signifi-cant progress in the field of tongue image segmentation,greatly improving segmentation accuracy and automation.This article briefly introduces commonly used public datasets and data preprocessing methods for tongue image segmentation.It then categorizes existing methods based on the characteristics of their network architectures into four main types:seg-mentation methods based on classic convolutional neural network(CNN)structures,methods based on modular design,methods based on generative adversarial networks,and methods based on Transformers.The advantages and limitations of each type are discussed.Next,the applications of segmented tongue images in disease diagnosis are summarized,emphasizing the role of tongue image segmentation in supporting intelligent disease diagnosis.Finally,this article points out the challenges faced by current deep learning techniques in the field of tongue image segmentation,such as limited dataset size,and insufficient model generalization ability.It also provides prospects for future research directions in this field.