A Tongue Image Recognition Method Based on Type II Diabetes Traditional Chinese Medicine Syndrome Classification
Kan Hongxing1# Zhang Luyao1* Dong Changwu2
1School of Medical Information Technology, Anhui University Traditional Chinese Medicine, Hefei 230012, China 2Clinical College of Traditional Chinese Medicine, Anhui University Traditional Chinese Medicine, Hefei 230012, China
Abstract Tongue diagnosis is one of the methods oftype II diabetes traditional Chinese medicine (TCM) syndrome differentiation. In order to reduce the diagnostic deviation caused by external light and the doctor′s experience, a tongue image recognition method was proposed in this paper according to the TCM for syndrome classification of type II diabetes. Firstly, the corrected tongue image was obtained by color-calibration. Secondly, based on the difference between the color space, tongue image was divided into background and tongue area using fast K-mean clustering method,and then tongue area was separated into tongue body area and tongue fur area using Ohta color threshing method. Thirdly, according to the type IIdiabetes TCM tongue features, the color features, crack features and dental indentations features were extracted form tongue body area, tongue fur and tongue nature area. Finally, the random forest was used as the classifier to train tongue features data, and compared with the support vector machine method. The proposedmethod has been tested in 218 tongue image cases. Experimental results showed that the average recognition accuracy of random forest methods was 90.37%, which was increased by 10.74% compared with that of the support vector machine method, suggesting the random forest method was more accurate and efficient for type II diabetes TCM syndrome classifications.
Kan Hongxing Zhang Luyao Dong Changwu. A Tongue Image Recognition Method Based on Type II Diabetes Traditional Chinese Medicine Syndrome Classification[J]. Chinese Journal of Biomedical Engineering, 2016, 35(6): 658-664.