Development of an artificial intelligence model to predict the extent of invasion of lung adenocarcinoma using autofluorescence microscopy images

2024 Fiscal Year Final Research Report

• Project/Area Number: 21K08905
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 55040:Respiratory surgery-related
• Research Institution: The University of Tokushima
• Principal Investigator: TAKIZAWA Hiromitsu 徳島大学, 大学院医歯薬学研究部(医学域), 教授 (90332816)
• Co-Investigator(Kenkyū-buntansha): 近藤 和也 徳島大学, 大学院医歯薬学研究部(医学域), 教授 (10263815) ; 宮本 直輝 徳島大学, 病院, 助教 (00865305)
• Project Period (FY): 2021-04-01 – 2025-03-31
• Keywords: 肺腺癌 / 浸潤診断 / 縮小手術 / 自家蛍光 / 人工知能

Outline of Final Research Achievements

This study aimed to develop a system for diagnosing the extent of lung adenocarcinoma invasion by capturing autofluorescence images of fresh-frozen surgical specimens using a digital slide scanner and analyzing them with artificial intelligence (AI). Sections were prepared from resected specimens, and autofluorescence images were captured using a digital slide scanner. Subsequently, the sections were stained with Elastica Van Gieson (EVG) and imaged. The EVG-stained images were analyzed using pathology image analysis software, where annotations were added to areas such as: (i) Invaded areas, (ii) Non-invaded areas, (iii) Normal lung. The autofluorescence and EVG-stained images were then subdivided, and a convolutional neural network model was trained using these annotated regions as supervised data to identify the extent of lung adenocarcinoma invasion. A program is currently under development to display the invaded areas as a heatmap.

Free Research Field

呼吸器外科学

Academic Significance and Societal Importance of the Research Achievements

肺腺癌浸潤範囲同定モデルにより微小な肺癌の浸潤範囲を術中に診断できるようになれば、より個別化した治療方針として肺部分切除や肺区域切除といった、より肺を温存できる術式を適応できる肺癌患者が増える可能性がある。また、このモデルは病理医にとっても判断が難しいとされる肺腺癌の浸潤範囲の決定において、病理医の診断を補完するシステムとなる可能性もある。