Developing the Artifacts Removing System for OCTA using generative adversarial network

2023 Fiscal Year Final Research Report

• Project/Area Number: 20K18332
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 56060:Ophthalmology-related
• Research Institution: Asahikawa Medical College
• Principal Investigator: Takahashi Kengo 旭川医科大学, 大学病院, 助教 (40646064)
• Project Period (FY): 2020-04-01 – 2024-03-31
• Keywords: 光干渉断層計 / SRF / IRF / セグメンテーション / AI

Outline of Final Research Achievements

Retinal specialists pointed out that the artifact removal system using a generative adversarial network predicted images that differed from the facts and that applying the system to clinical practice would be difficult. After that, we performed automatic segmentation of optical coherence tomography (OCT) using U-net, a type of Convolutional neural network, to evaluate the activity of retinal disease by automatically segmenting sub-retinal fluid (SRF) and intraretinal fluid (IRF) and evaluating them as subretinal fluid scores. We developed a system for predicting and preventing SRF and IRF and for tailored medicine. We summarized the results and submitted them as a paper. We are currently considering applying the system to clinical practice.

Free Research Field

網膜

Academic Significance and Societal Importance of the Research Achievements

糖尿病網膜症（DR)および進出型加齢黄斑変性（wAMD)は、それぞれ、本邦における失明原因の２位と４位を占めており、その個別管理・治療は重要である。DRおよびwAMDは黄斑部に網膜下液を生じ、光干渉断層計を用いた病状管理が重要である。本研究では、網膜下液を下液スコアとして評価し、それを用いて、予防・予測観点で個別管理するシステムの開発に寄与した。今後はこのシステムの実臨床への応用について検討の上、本邦における失明者を減らすことに貢献する。