Development of diagnostic system for neck lymph node metastasis in oral cancer using artificial intelligence

2023 Fiscal Year Final Research Report

• Project/Area Number: 20K18668
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 57060:Surgical dentistry-related
• Research Institution: Okayama University
• Principal Investigator: Obata Kyoichi 岡山大学, 大学病院, 助教 (80845389)
• Project Period (FY): 2020-04-01 – 2024-03-31
• Keywords: 口腔癌 / 頸部リンパ節 / 判別分析 / 予測

Outline of Final Research Achievements

Metastasis of oral squamous cell carcinoma to cervical lymph nodes (CLN) is a strong poor prognostic factor. Various imaging results obtained preoperatively and immunohistochemical staining results obtained postoperatively were used for statistical evaluation. Using the short diameter and CT values of CLN on preoperative CT and the SUVmax values of the primary tumor and CLN on FDG-PET/CT, we calculated a discriminant formula that could predict CLN metastasis preoperatively with a 92.2% discrimination rate. The 5-year recurrence rate was significantly higher (p<0.01) in patients who could not discriminate CLN metastasis from those who could discriminate CLN metastasis using the discrimination formula combined with immunohistochemical staining results. The same result was obtained in 96.0% of cases when new cases were entered by deep learning.

Free Research Field

口腔外科

Academic Significance and Societal Importance of the Research Achievements

口腔癌の頸部リンパ節(CLN)転移を評価する手法として、触診とともにCTやMRI、FDG-PET/CT、超音波検査等の画像検査が用いられる。しかし実臨床では、それぞれの検査間でCLNの評価が大きく異なることが散見される。CLN転移があるかないかでは手術の侵襲度が大きく異なるため、術前に正確なCLN評価を行うことが非常に重要である。本研究では90%以上の症例で術前にCLN転移を予測可能で、さらに免疫組織化学染色の結果を組み合わせることで、簡便に予後評価が可能となった。これは口腔癌の診断および治療における大きな進歩である。