摘要
目的舌诊是中医诊断痤疮的有效途径,目前中医通过观察舌象来确定痤疮患者的证候类型。由于痤疮患者众多,完全基于人工诊断的效率较低。本文提出一种基于图像处理的痤疮证型识别方法来辅助中医诊断。方法首先,分别提取舌象的颜色、纹理和齿痕特征,然后使用贝叶斯网络建模,找出特征与特征,特征与证型之间的关系,其中将齿痕提取算法进行改进,将计算凸包面积改进为找到每个凸包的关键点,最后使用该算法对舌象进行齿痕数量提取,并与中医诊断结果相比较。结果对比医生诊断结果,基于图像处理的痤疮证型自动分类,分3类的正确率达83.87%,并直观地表示出特征与特征,特征与证型之间的关系。结论使用图像处理的方法进行痤疮证型的识别具有可行性,对计算机辅助痤疮诊断的发展有一定帮助。
Objective Tongue diagnosis is one of the effective ways to diagnose acne. Traditional Chinese medicine（ TCM） in treatment of acne is currently mainly determined by doctors to be observed in patients＇ tongues. Because the number of patients with acne is numerous,the efficiency completely based on the artificial diagnosis is low. This paper presents a novel method for the diagnosis of acne syndromes with a computer to assist TCM diagnosis. Methods First of all,features of color,texture and teeth marks were extracted from acne tongues. Then we used Bayesian network to simulate the relationship between the features and syndromes,and completed the classification. The original extraction algorithm for teeth marks was improved from calculation of convex hull area to finding each key point of convex hull. Finally we used the algorithm to extract the number of teeth marks and compared with TCM diagnosis. Results The extraction results were consistented with TCM diagnosis. Compared with the TCM diagnosis,the accuracy of acne syndrome classification based on image processing was 83. 87%,and intuitively showed the interaction between features and features,features and syndromes. Conclusions The results show the feasibility of image processing for the diagnosis of acne syndromes,which is beneficial to the development of auxiliary diagnosis of acne syndromes in TCM.
出处
《北京生物医学工程》
2016年第5期464-468,共5页
Beijing Biomedical Engineering
基金
北京市属高等学校高层次人才引进与培养计划项目(CIT&TCD201504018)资助
关键词
中医诊断
痤疮
证型分类
舌象
图像处理
贝叶斯网络
traditional Chinese medicine diagnosis
acne
syndrome classification
tongue
image processing
Bayesian network
