Development of an ultra-high-precision diagnostic model using endoscopy and oral ultrasound with artificial intelligence in head and neck cancer.

2022 Fiscal Year Final Research Report

• Project/Area Number: 20K09713
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 56050:Otorhinolaryngology-related
• Research Institution: Hiroshima University
• Principal Investigator: Ueda Tsutomu 広島大学, 医系科学研究科(医), 准教授 (70522928)
• Co-Investigator(Kenkyū-buntansha): 樽谷 貴之 広島大学, 病院(医), 助教 (10569007) ; 卜部 祐司 広島大学, 病院(医), 寄附講座准教授 (10648033) ; 河原 大輔 広島大学, 病院(医), 助教 (20630461) ; 竹野 幸夫 広島大学, 医系科学研究科(医), 教授 (50243556)
• Project Period (FY): 2020-04-01 – 2023-03-31
• Keywords: 深達度診断 / 経口的咽喉頭手術 / Radionics / 人工知能（AI） / Deep learning / 咽喉頭癌

Outline of Final Research Achievements

Radiomics analysis of images of pharyngeal laryngeal carcinoma was performed to determine diagnostic performance for the presence of subepithelial invasion. The mean Accuracy, Sensitivity, Specificity, and AUC for cross-validation were 83.3%, 87.3%, 76.1%, and 0.868, respectively, suggesting that AI-based depth diagnosis complements the endoscopist's diagnosis. Transoral ultrasound depth diagnosis also complemented endoscopic findings. The positive predictive value was 65.6% for gross findings, 78.9% for magnified endoscopic findings, and 82.1% for oral ultrasound, but 100% when the three were combined.

Free Research Field

頭頸部癌

Academic Significance and Societal Importance of the Research Achievements

AIと医師の診断により、経口的咽喉頭手術を施行する頭頸部癌症例に対して，術前の超高精度の深達度診断をすることが可能となれば、触診による深達度診断が不可能な部位での適切な切除が可能になり，術後の嚥下障害を含めた合併症の回避が可能となる．その結果，特に高齢者の多い頭頸部患者の術後のQOLの向上に寄与すると考える。現在徐々に普及している経口的ロボット支援下手術にも応用が可能であり，更なる低侵襲手術の発展に寄与すると考える．