探討台灣牙科醫師對以人工智慧為基礎之電腦輔助放射影像解讀工具在臨床診療應用的觀點

Perspectives of dentists in Taiwan regarding the use of artificial intelligence in dental radiographic applications

王佳惠(Chiahui Victoria Wang)

指導教授 ： 鍾國彪

國立臺灣大學/公共衛生學院/健康政策與管理研究所/碩士(2021年)

https://doi.org/10.6342/NTU202104358

摘要

背景：人工智慧(AI)在醫學影像分析技術領域的應用已成為醫療創新領域中最具有發展前景的領域之一。結合高運算能力的電腦設備和廣大多元的健康數據，再經由人工智慧科技處理更多複雜醫學影像的分析與詮釋後，將可以提供醫師最新與更精確的診斷結果。其中在牙科診斷中，口腔放射學檢查是病患疾病管理的重要步驟也是牙醫師用於對與牙齒和周圍結構相關的病理進行臨床檢查的診斷方式之一。高速電腦計算功能和健康數據分析的日益進步將使以人工智慧技術進行牙科X光片影像分析的工作更為有效率並同時可以達到精確的影像判讀結果。牙科X光片的結果將包含在患者的臨床記錄中，用於口腔疾病篩檢、病理診斷、開立治療計劃以及預後追蹤。由於人工智慧一直是放射科醫生討論的中心，探索人工智慧如何影響放射科醫生的觀點在國際上已經有許多的相關研究。然而在牙科領域中，人工智慧以漸進式的方式引導牙科領域進入資訊導向的創新應用新時代，這樣的演進不但會改變傳統的臨床診斷方式並會藉此提升牙科患者的照護管理。
研究目的：深入了解牙科醫師對這一個趨勢的看法十分重要。這個趨勢啟發了研究者進行一系列的牙科醫師訪談，經由質性研究的方法，設定兩個目標進行探索牙科醫師對人工智慧的觀點以及找出導入牙科臨床流程的益處與挑戰。該研究將先找出牙科醫師對於人工智慧應用在牙科Ｘ光影像判讀，找出導入臨床實踐中的成功要素與挑戰。
研究方法：本研究在台灣北部的北北桃地區進行，以刻意抽樣方式在三個都市中選出11位牙科醫師，性別不拘，研究者提供了六個半結構性的開放問題，出於數據收集的目的，利用Google Meet在線上進行了深度訪談。受訪者被問及他們對人工智慧技術的看法，人工智慧在牙科領域的使用經驗，以及他們對當前與未來使用人工智慧為基礎的臨床應用了解他們認為益處，衝突與障礙。
研究結果：訪談結果提供牙科醫師對人工智慧技術與利用人工智慧輔助的放射學詮釋工具之觀點，包含了19個相互關聯的主題以及26個影響主題的因素。這19個主題整合了運用人工智慧協助牙科影像詮釋工具的希望與質疑。其中，數據安全與系統互通性問題被視為隱私威脅以及機器與軟體互通性的障礙。這可以解釋人工智慧解決方案為何還無法成為牙科臨床流程的常規。在計畫與執行任何以人工智慧技術成為牙科臨床工作流程的一環之前，參與研究的牙科醫師都認為數據安全與系統互通性之兩大問題必須首先得到解決。他們更強調，牙科影像應用工具要有明顯的商用附加價值，如可以有效提高工作效率以及強化與病患溝通的能力。在沒有呈現明顯商用價值之前，牙科醫師們普遍認為人工智慧的工具被廣泛採用的機率很難提高。
結論與建議：質性研究的結果可以清楚看到牙科醫師對人工智慧的認識與期待。牙科教育對於拓寬牙醫師對人工智慧深度認識頗為重要。牙科製造商應提供使用者友善的應用程式介面(API)來促使不同製造商之設備系統能夠彼此互通與整合，才能有效地連結現有的牙科放射影像設備。若是無法提供實用的臨床轉化結果，人工智慧的益處仍然只是屬於理論狀態。牙科界需要討論與正視這些實際的問題並找出解決的方式。唯有如此，牙科醫師才能利用人工智慧應用工具來提高工作效率，讓他們有更充裕的時間提供以患者為中心的服務，讓牙科醫師更有效地為患者的口腔健康帶來益處。

關鍵字

人工智慧 ； 牙科 ； 牙科影像 ； 臨床工作 ； 診斷 ； 觀點 ； 工作效率 ； 病患溝通

並列摘要

Background: As the application of artificial intelligence (AI), primarily in medical imaging has become one of the most promising and progressive areas of health innovation, the increasing use of high-power computing and health data will lead to more successful utilization of AI to carry out more radiographic interpretation for up-to-date and more precise results. In dentistry, radiological examination is an essential part of patient management and among all dental procedures, it is frequently used by dentists to make clinical diagnosis of pathology related to teeth and surrounding structures. The results of dental radiographs are included into patients’ clinical records for screening, diagnosis, treatment planning, follow up, and is a standard requirement for patients’ dental insurance filings. As AI has been at the center of discussion among medical radiologists, many studies were performed to explore how AI has impacted on the perspectives of radiologists and medical imaging analysis. Deep learning has become a promising technique for processing medical imaging data.
Research Objectives: Dentistry is a data-driven field. Millions of dental radiographs are taken every year. The dental industry routinely generates a large and heterogeneous amount of data. However, fewer studies were carried out to seek dentists’ perspectives. Yet, the emergence of AI is slowly leading dentistry to a new era of innovative technology with the potential to shape current clinical practice and improve patient care. It’s important to understand the opinions of dentists toward this trend. The researcher was therefore inspired to conduct a study to explore the perspectives of dentists with two objectives. First, find out what dentists perceive towards artificial intelligence technology and second, explore their perspectives to adapt AI-based dental radiographic applications into their clinical practice.
Research Methods: This study was carried out in Taiwan with a qualitative case study conducting six in-depth and semi-structured interviews. There were 11 participating dentists in the study. Purposive and snowball sampling was applied until data reached saturation. Each interview was scheduled and conducted by online interviews via Google Meet for the data collection purpose. Participants of the interviews were asked about their perspectives about AI technology, practical experiences with AI in dentistry as well as barriers and potential benefits brought by AI technology on the use of dental radiography in their current and future clinical practice. The research setting was targeted in a metropolitan area covering Taipei city, New Taipei City and Taoyuan city with a total combined population of and between 7-8 million residents sharing similar living and working environment connected by public transportation systems and highways. All interviews were conducted from August to September 2021.
Results: The results of the interviews provided various insightful information of dentists’ perspectives toward AI technology and AI-based radiographic interpretation tools.
Researcher identified a total of 19 themes, 8 themes as perceived benefits, 5 themes as perceived conflicts, and the other 6 themes as perceived challenges. These 19 themes were the collections of 27 inter-related codes which cover a wide range of topics from various AI technologies, functionalities, clinical needs, concerns over the use of AI-based radiographic applications and many others. These 19 themes were mixed with hopes and doubts on AI-based radiographic interpretation applications. These themes may explain why AI radiographic solutions have not by large become part of routine dental practice.
Participants regarded validation of data accuracy and proven commercial value to be addressed up front and center before any AI solutions become part of clinical workflows in their routine dental practice. Participants emphasized that AI-based dental radiographic applications should demonstrate tangible benefits of increasing work efficiency and empowering patient communication.
Conclusion and Recommendation: In conclusion, the participants to this study showed positive attitude towards the adoption of AI-based radiographic applications. Perceived benefits and expectation were mostly shared by dentists. There were also hidden benefits or unmet needs found through this study which are not being properly addressed in the dental community. Perceived conflicts such as the patient liability and data protection remain unsolved and more legal framework needed. Concerns about reliability and applicability of AI applications shall be studied and validated before transferring into clinical practice. It is recommended that more dental education is needed to broaden AI knowledge of dentists. Inter-professional coordination and collaboration among dentists, researchers, AI engineers shall be working together for the development of AI applications in dentistry. Without more practical clinical translations, the potential benefits of AI remain theoretical. The dental community needs to discuss these practical matters to provide effective AI applications in clinical practice which can be adopted by dentists without too many challenges. The use of AI applications shall be viewed as a complementary asset to assist dentists and shall be integrated in a safe and controlled manner.
Qualitative dental studies hold a crucial role to explore unique insight into dentists’ personal perspectives when a new technology or solution is being proposed. In this study, having a comprehensive understanding of the dentists’ perspectives will ensure the dental community to address key issues when developing suitable products or implementing new technology to suit the real needs of dentists. Consequently, dentists can work more efficiently and provide patient-centric services to improve treatment outcomes and increase the trust level of their patients.

並列關鍵字

artificial intelligence (AI) ； dentistry ； dental radiography ； clinical work ； diagnosis ； perspective ； work efficiency ； patient communication

參考文獻

A.J.E. Qualtrough, J. D. S., L.A. Morrow and P.A. Brunton. (2005). Principles of Operative Dentistry.

Abdolali, F., Zoroofi, R. A., Otake, Y., Sato, Y. (2017). Automated classification of maxillofacial cysts in cone beam CT images using contourlet transformation and Spherical Harmonics. Computer Methods and Programs in Biomedicine, 139, 197-207. doi:https://doi.org/10.1016/j.cmpb.2016.10.024

Ahmed, N., Abbasi, M. S., Zuberi, F., Qamar, W., Halim, M. S. B., Maqsood, A., Alam, M. K. (2021). Artificial Intelligence Techniques: Analysis, Application, and Outcome in Dentistry-A Systematic Review. Biomed Res Int, 2021, 9751564. doi:10.1155/2021/9751564

Ahuja, A. S. (2019). The impact of artificial intelligence in medicine on the future role of the physician. PeerJ, 7, e7702-e7702. doi:10.7717/peerj.7702

Alabi, R. O., Elmusrati, M., Sawazaki-Calone, I., Kowalski, L. P., Haglund, C., Coletta, R. D., . . . Leivo, I. (2020). Comparison of supervised machine learning classification techniques in prediction of locoregional recurrences in early oral tongue cancer. Int J Med Inform, 136, 104068. doi:10.1016/j.ijmedinf.2019.104068