應用影像辨識及中醫概念之舌象檢測

Tongue Diagnosis Using Image Recognition and Traditional Chinese Medicine

數千年來，舌診在中醫醫學（TCM，Traditional Chinese Medicine）中一直都擁有舉足輕重的作用。舌質特徵，與身體器官的健康有關，對辨證和治療選擇有很大的幫助。一般民眾通常並不具備辨別舌頭異常狀況的專業能力，大多數人認為給醫生看診就是唯一且精確的方法。不過，醫師若僅是以傳統主觀的判斷看診流程，有可能不易進行精確的診斷；近年來，隨著深度卷積神經網絡（CNN）應用、人工智慧框架技術的急速發展，¬¬更完善的方式─應該是設計一套可以「藉由圖像來檢測舌苔及舌質各式樣貌並判斷症狀的系統」。
有鑑於前述所探討，本計畫所開發的系統採用開放式平台來實現，期望透過使用者的網路攝像頭自行拍攝出的影像利用網際網路，將影像回傳伺服端進行分析，先篩選出「舌象部分」再加以分析，並結合事先所蒐集的中醫相關資訊，再將分析結果回傳至使用者端，當作病例的參考，同時將其資料儲存進伺服端，以供未來進行紀錄的調閱以及追蹤使用。
使用者端可利用任何具有瀏覽器功能的裝置，連接本計畫所開發的網站做使用。此計畫所開發的網頁將採用響應式網頁設計（Responsive Web Design，RWD），使得使用者無論在行動裝置或電腦，都能得到最佳的視覺體驗。

Since thousands of years ago, tongue diagnosis has played a crucial role in Traditional Chinese Medicine (TCM). The characteristics of the tongue are related to the health of the body's organs. In addition, they're also helpful in diagnosing and treatment choices. Unfortunately, the public generally does not have the professional ability to distinguish abnormal tongue conditions. Most people think that seeing a doctor is the unique and accurate method. However, if the doctor only uses traditional subjective judgment for the examination process, it may not be easy to make an accurate diagnosis. In recent years, with the rapid development of deep convolutional neural networks (CNN) applications and artificial intelligence framework technology, it's considered that designing a system that can ''detect tongue coating and tongue texture through images and diagnose symptoms'' should be a more appropriate method.
Given the above discussion, the system developed in this project is constructed on an open platform. Using the image taken by the user's network camera through the internet, then return the image to the server for analysis. First, it will filter out the ''tongue image'' and then analyze it. Then after combining it with the previously collected Chinese medicine-related information, the system will return the analysis results to the user as a reference case and store the data into the server for future record review and tracking simultaneously.
Users can use any devices with a browser function to connect to the website developed by this project. The website developed by this project adopts responsive web design (RWD), making it possible for users to use it on mobile devices.