Development of automatic radiographic image diagnosis support software for oral dysfunction using artificial intelligence (AI)

2023 Fiscal Year Final Research Report

• Project/Area Number: 20K10169
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 57060:Surgical dentistry-related
• Research Institution: Showa University
• Principal Investigator: Matsuda Yukiko 昭和大学, 歯学部, 教授 (50266178)
• Co-Investigator(Kenkyū-buntansha): 田口 亮 東京都市大学, 情報工学部, 教授 (40216825)
• Project Period (FY): 2020-04-01 – 2024-03-31
• Keywords: 舌骨 / パノラマエックス線写真 / 口腔機能低下症 / オーラルフレイル

Outline of Final Research Achievements

Low hyoid position was significantly associated with dysphagia. When the hyoid bone was below the imaginary line of the lower border of the mandible on panoramic radiography, the swallowing disorders were significantly observed, and this was decided as the cutoff level.
An AI program was created based on the cutoff level determined above. The position of the hyoid bone could be detected using a learning method that focused on the area below the lower border of the mandible, but the diagnostic accuracy was low in cases where the entire hyoid bone was not visible, which was an issue. When training the AI, the region of interest was expanded, the number of groups was reduced, and cases where the hyoid bone was partially or completely invisible were not trained. As a result, the diagnostic accuracy was improved from 0.83 to 0.97 compared to the conventional method.

Free Research Field

歯科放射線

Academic Significance and Societal Importance of the Research Achievements

以前の我々の研究で、舌骨が低位であることが嚥下障害と関連する可能性が高いことを示した。歯科治療においてパノラマエックス線検査は比較的頻繁に行われている検査ではあるが、歯や顎骨についての診断が主体であり、舌骨の位置に注目する歯科医師はほとんどいない。本研究により、AIを用いて舌骨の位置をより正確に検出できるようになった。舌骨が有意に低位であることを自動検出し、治療医師に提示できるようになれば、歯科治療時に誤飲や誤嚥の防止のために頭部の位置を調整するなどの配慮が可能となる。さらにオーラルフレイルへの警笛として、早期対策を促すことができるようになると考える。