2023 Fiscal Year Final Research Report

Establishment of a New Lung Cancer Image Analysis Method for Ultra-High Resolution CT: Search for Image Factors Contributing to Diagnostic Performance and AI Analysis

Research Project

PDF

Project/Area Number	21K07672
Research Category	Grant-in-Aid for Scientific Research (C)
Allocation Type	Multi-year Fund
Section	一般
Review Section	Basic Section 52040:Radiological sciences-related
Research Institution	Osaka University
Principal Investigator	Yanagawa Masahiro 大阪大学, 大学院医学系研究科, 准教授 (00546872)
Co-Investigator(Kenkyū-buntansha)	森井英一大阪大学, 大学院医学系研究科, 教授 (10283772) 鈴木裕紀大阪大学, 大学院医学系研究科, 特任助教(常勤) (20845599) 富山憲幸大阪大学, 大学院医学系研究科, 教授 (50294070) 木戸尚治大阪大学, 大学院医学系研究科, 特任教授(常勤) (90314814)
Project Period (FY)	2021-04-01 – 2024-03-31
Keywords	超高精細CT / 肺腺癌 / PD-L1 / 人工知能 / Vision Transformer / LIME / 病理学的浸潤成分
Outline of Final Research Achievements	Ultra-high resolution CT is a device that has twice the spatial resolution of conventional CT in both the in-plane and body axis directions, and can obtain information on minute anatomical structures and histopathological image that could not be observed with conventional CT. In this study, we aimed to search for imaging factors and establish image analysis methods that are useful for accurate diagnosis of lung cancer and selection of a treatment plan by collecting image data of lung cancers with ultra-high resolution CT.　Specifically, CT images of lung cancer were compared with histopathological images and genetic information, and quantitative analysis software was developed utilizing artificial intelligence.　The visualization of the site that the developed artificial intelligence-based software focused on during diagnosis was examined, and the predictive diagnosis of the software for pathological invasiveness components was examined using clinical cases of lung adenocarcinoma.
Free Research Field	放射線医学
Academic Significance and Societal Importance of the Research Achievements	放射線診断領域では、画像データをバイオマーカーとして活用し、画像定量値と多彩な臨床情報や分子学的情報と関連付けた研究が進んでおり、従来CTよりも情報量の多い超高精細CT画像を定量解析することは、遺伝子情報を含めた病理学的因子や予後因子を予測するための新しい画像学的因子の発見に繋がる可能性が高い。本研究の結果は、肺癌の的確な診断や治療方針の選択に役立つ新たな情報を提供し、肺癌の画像診断の発展につながる可能性が高く、その社会的意義は大きい。