Development of automated classification of glomeruli with clinical information

2019 Fiscal Year Final Research Report

• Project/Area Number: 19K21115
• Project/Area Number (Other): 18H05959 (2018)
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Multi-year Fund (2019); Single-year Grants (2018)
• Review Section: 0403:Biomedical engineering and related fields
• Research Institution: Kyoto University
• Principal Investigator: Uchino Eiichiro 京都大学, 医学研究科, 特定助教 (20820905)
• Project Period (FY): 2018-08-24 – 2020-03-31
• Keywords: 腎病理 / 人工知能

Outline of Final Research Achievements

We developed a system that automatically determines pathological findings using an artificial intelligence model for glomerular images obtained from pathological images of renal biopsy. It was confirmed that the model showed similar classification performance to that of clinicians. Moreover, it was shown that the final classification performance might be improved if these models were used in a majority vote of the artificial intelligence model and clinicians.

Free Research Field

腎臓内科学

Academic Significance and Societal Importance of the Research Achievements

近年応用への取り組みが進められている人工知能技術について、医療分野、特に腎臓病診療における病理画像診断にも有用である可能性と、現在の標準的な手法におけるベンチマークが得られた。また同種のモデルを実際の臨床現場に使えうるかという観点においても評価が行われ、今後の臨床応用に向けた可能性が示された。