Automatic Sublingual Vein Feature Extraction System

• Main Authors: Yi-Jing Chen, 陳怡靜
• Other Authors: John Y. Chiang
• Format: Others
• Language: zh-TW
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/37693028352093646870

碩士 === 國立中山大學 === 資訊工程學系研究所 === 102 === The essence of TCM diagnosis is “Syndrome Differentiation and Treatment”, in which differentiation is based on four methods of observation, smell, inquiry and palpation. The examination of the observing is the most important procedure in the method of “tongue diagnosis”. In recent years, sublingual vein is proved to be related closely with human organs in medical researches. By observing pathological changes of sublingual vein with features to determine if lesion happens to human organs and then give treatments as soon as possible. Due to the clinical tongue diagnosis usually depends on the factors of doctor’s subjective opinions, accumulated experiences and environment at time, the result is likely to be limited to and influenced by subjectively judgments of knowledge, experiences, thinking patterns, diagnosis skills, senses and explanations of color. Different doctors may not have the same diagnosis to one tongue pattern which leads to inconsistency. By using technical method as assistance to diagnose, along with standard judgment progress to acquire reliable diagnosis is an important issue to improve the value of TCM clinical application. This research is about using one of the developments of image process technology-Automatic sublingual vein feature extraction system-to reach targets of objective and quantitative by computer interpretation. An important prerequisite is that the result of sublingual vein features also can be applied in clinical assistance and used in sublingual vein date base. 　　The features of sublingual arteries and veins are captured mainly from separating the parts of the back of tongue, and then grab the area of sublingual arteries and veins to identify its features. First, take the photos of patients’ back tongues and grab the exact images of their sublingual arteries and veins via adjusting the unreal color parts of the tongue photos. This research grabs the images of back tongues via analyzing the RGB performances of tongue back, lips, teeth, and mouth skin, and transferring these into HIS color, which is easier for human eyes to recognize the features of these parts, and also combined with the removals of mouth skin area, teeth area, and black area, rectangle detection, and detection of control points to further improve the images of the back of tongue. After the separation of the back of the tongue image, process histogram equalization and hue offset to enhance the color contrast level, and use the changes in RGB color components, hue, saturation, and brightness characteristics to separate sublingual vein, and in accordance with sublingual vein colors and locations to distinguish the color light, the length and the vein branches to analyze whether there is a varicose vein. At the same time, the research is intended to capture the tongue features such as sublingual sac column, vesicles, petechiae, and bleeding silk, and integrate and analyze these features to assist and provide practitioners as clinical diagnostic references.