以深度學習為基礎之冠心病舌診影像辨識

Deep learning-Based Recognition of Coronary Artery Disease Tongue Diagnosis Images

鄭旻豪(Min-Hao Jheng)

指導教授 ： 王三元

共同指導教授 ： 歐陽振森(Chen-Sen Ouyang)

義守大學/智慧科技學院/資訊工程學系/碩士(2024年)

摘要

冠心病於現在的社會上已經層出不窮，隨著不健康飲食、外送的便利性與運動量的降低，我國國民現今普遍肥胖，並且人口也逐漸高齡化，使的罹患冠心病已成為死亡原因中前列。為檢測出疾病，中醫醫師經常會採用傳統中醫診查的「望聞問切」四診，即望診、聞診、問診、切診，以視覺診察為望診，以嗅覺、聽覺診察為聞診，以言語交談診察為問診，以觸覺診察為切診，依此四診，診察疾病於人體各處顯現的症狀和體徵，並由不同角度診察，相互印證、補充，才能達到全面的了解病情，又稱作四診合參。於傳統中醫診察中，往往需要醫師依靠經驗累積來進行主觀的判斷，導致不同醫師的判斷皆有異，於現今人工智慧成熟的年代，我們可以依靠訓練人工智慧，使用電腦來進行客觀的判斷，進而達到輔助醫師、減輕醫師負擔的作用。為達到上述目的，本研究依辨識冠心病為目標，結合深度學習與傳統中醫四診之望診，使人工
智慧將取得的舌下絡脈影像進行分類，並利用手機作為APP載體，製作APP作為快速篩檢工具，完成民眾可初步自檢之工具，降低醫療浪費成本，為台灣醫療與科技結合出一份力。

關鍵字

深度學習 ； 冠心病 ； 傳統中醫 ； 舌下絡脈

並列摘要

Coronary artery disease is increasingly prevalent today due to unhealthy diets, the convenience of food delivery, and decreased physical activity. The general population in our country is becoming increasingly obese, and with the aging population, coronary artery disease has become one of the leading causes of death. Traditional Chinese medicine (TCM) practitioners often employ the "Four Examinations" inspection, auscultation olfaction, inquiry, and palpation) to diagnose diseases. This method involves visual inspection, olfaction and listening, inquiry through conversation, and palpation to observe symptoms and signs throughout the body from different perspectives, confirming and complementing each other to achieve a comprehensive understanding of the disease, also known as the integration of the Four Examinations.
In TCM diagnosis, physicians often rely on accumulated experience to make subjective
judgments, leading to variations in diagnoses among different practitioners. In the era of mature artificial intelligence (AI), we can rely on AI training to achieve objective judgments using computers, thereby assisting physicians and reducing their workload. To achieve this goal, this study focuses on identifying coronary artery disease by combining deep learning with the visual examination component of traditional TCM. AI will classify images of the sublingual veins obtained, and a mobile application will be developed as a rapid screening tool, providing the public with a preliminary self check tool to reduce medical waste costs and contribute to the integration of medicine and technology in Taiwan.

並列關鍵字

deep learning ； coronary artery disease ； traditional Chinese medicine ； sublingual veins

參考文獻

[1] “Heart Disease and Stroke Prevention,” NEW YORK STATE Department of Health. Accessed: Aug.31, 2023.[Online]. https://www.health.ny.gov/diseases/cardiovascular/heart_disease/ Available:

[2] “111年國人死因統計結果,” 衛生福利部. Accessed: Jun.30, 2023.[Online]. Available: https://www.mohw.gov.tw/cp-16-74869-1.html

[3] “ 認識冠心病,” 衛生福利部. Accessed: Aug.29, 2022.[Online]. Available: https://www.hpa.gov.tw/Pages/Detail.aspx?nodeid=632&pid=1188

[4] A.V.Khera andS.Kathiresan, “Genetics of coronary artery disease: Discovery, biology and clinical translation,” 2017. doi:10.1038/nrg.2016.160.

[5] 陳琬青, “與冠心症有關的自費健康檢查,” 血管醫學防治季刊, no. 1, pp. 16–17, 2010, doi:10.6527/PVM.2010.1.10.