Challenge to automatic classification between Parkinson's disease and atypical Parkinson's syndrome by artificial intelligence

2021 Fiscal Year Final Research Report

• Project/Area Number: 18K15565
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 52040:Radiological sciences-related
• Research Institution: Fujita Health University (2021); Teikyo University (2018-2020)
• Principal Investigator: Shiiba Takuro 藤田医科大学, 医療科学部, 准教授 (30759501)
• Project Period (FY): 2018-04-01 – 2022-03-31
• Keywords: パーキンソン病 / 人工知能 / 自動分類

Outline of Final Research Achievements

This study aimed to focus on the image features of dopamine transporter (DAT) SPECT, a biofunctional imaging technique, and to develop the next-generation imaging diagnosis method, an automatic classification system for Parkinson's disease (PD) and atypical Parkinson syndrome (APS), by introducing machine learning. However, the organization of APS cases planned initially did not proceed due to the spread of the new coronavirus infection.
Therefore, the objectives were set as constructing a radiomics signature and developing an automated classification system for PD and healthy subjects. As a result of the research, a highly accurate automatic classification system and a radiomics signature for PD was constructed. The results are expected to be applied to APS.

Free Research Field

放射線科学

Academic Significance and Societal Importance of the Research Achievements

本研究では、パーキンソン病の高精度な自動分類システムの構築と新たな画像バイオマーカを提案することができた。本研究で構築した自動分類システムの使用によって従来のパーキンソン病の画像診断の正確さを向上させる可能性があり、今後パーキンソン病と非定型パーキンソン症候群への応用が期待できる。