Establishment of diagnosis of diffuse lung diseases based on a data-integrated deep learning method

2021 Fiscal Year Final Research Report

• Project/Area Number: 18K11190
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Review Section: Basic Section 60030:Statistical science-related
• Research Institution: Seikei University
• Principal Investigator: Komori Osamu 成蹊大学, 理工学部, 准教授 (60586379)
• Co-Investigator(Kenkyū-buntansha): 江口 真透 統計数理研究所, 数理・推論研究系, 教授 (10168776)
• Project Period (FY): 2018-04-01 – 2022-03-31
• Keywords: 医療統計

Outline of Final Research Achievements

I was able to study a diffuse lung disease analysis study.(1) Applying the structure of quasi-linear options from behind k-means, fuzzy c-means, and normal mixed models to make statistical changes.(2)With Corona access, you can analyze clinical data and Easter data from above. The accuracy of the evaluation of bull learning is 10%.(3)By doing one Grad-Cam of visualization of deep learning method, it is said that the arrangement of the image is obtained.

Free Research Field

データサイエンス

Academic Significance and Societal Importance of the Research Achievements

本研究では深層学習，転移学習，アンサンブル学習を組み合わせることで，先行研究による分類精度を10％ほど改善することに成功した．分類がうまく行かない原因の考察や，分類結果の可視化による医学的な知見の獲得までには至らなかったものの，肺疾患の病変分類の精度改善には貢献しており，社会的な意義も大きいと思われる．