Normalized Acquisition System of the Facial Diagnosis in Traditional Chinese Medicine

Abstract

Background: Facial diagnosis, an important part of clinical diagnosis in Traditional Chinese Medicine (TCM), is a method used to diagnose the functions of Zang-Fu organs by observing the color, luster, shape, and texture of faces. However, the traditional facial diagnosis mainly relies on doctors’ eyes, languages, and personal clinical experiences. Results are not only determined based on the doctors’ diagnostic skills but also by external conditions such as light and temperature. Without objective evaluation criteria, conducting studies on facial diagnosis to widen its application are difficult. To solve this problem, we should find new methods and technologies to realize the objectification and normalization of diagnosis in TCM. In this article, we discuss the results of our study on the normalized acquisition system of facial diagnosis in TCM. Some of the hardware used includes lights, image acquisition equipment, and dark boxes. The software used includes image acquisition and preprocessing. To the best of our knowledge, this study is the first to propose this system and no similar study has been reported yet. Methods: We initially introduced the hardware and the software that we used in this study. The key technologies in this system, including lighting equipment, facial diagnosis device, facial information acquisition scales, image acquisition, and preprocessing were then introduced one by one. The hardware of this system consists of a light emitting diode (LED), a digital camera, a dark box, and a computer. Each of this hardware has its special function: the LED imitates natural light; the camera records facial images; the dark box imitates the consultation room; and the computer stores the images. The software is used to acquire and adjust the images. The image acquisition system uses the computer to control the opening and the closing of the camera, the photography, and the setup of relevant index to obtain the fully automatic photography of faces and information transmission. Results: The normalized acquisition system of facial diagnosis in TCM was tested according to the following procedures. (1) The lighting uniformity of acquisition windows was tested. Results of the uniformity test showed an even distribution of the illumination on the opening of the dark box for facial complexion collection. (2) TCM experts valuated the acquisition environment. The four features basically tallied among the different diagnoses of doctors, in which the lowest consistency was 75% for the lip color and the highest was 95% for the moist/dry lips. These data showed that the LED the natural light can be used efficiently in the image collecting process. (3) The correction of collected images was tested. The sum of the Euclidean distance between the uncorrected color and the standard color was 1296.345, whereas the sum between the corrected color and the standard color was 403.527. The maximum distances before and after correction were 163.68 and 44.69, respectively. The minimum distances before and after correction were 13.3 and 5.9, respectively. (4) We collected 4050 photos of patients using this system, which was proven to be stable. Conclusions: This article introduces an automated acquisition system of facial diagnosis in TCM. The safety of the system can be ensured. By comparing TCM under natural light and in the dark box, this system meets the requirements of clinical application in all of the collected samples (more than 4050). The acquisition system of facial diagnosis in TCM has also been applied efficiently in a few hospitals.