Development of computer-aided diagnosis system for ultrasonic images of infant hip joint

2018 Fiscal Year Final Research Report

• Project/Area Number: 16K09012
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Medical Physics and Radiological Technology
• Research Institution: Niigata University
• Principal Investigator: Lee Yongbum 新潟大学, 医歯学系, 准教授 (10334658)
• Research Collaborator: MINAGAWA yasuko
• Project Period (FY): 2016-04-01 – 2019-03-31
• Keywords: 乳児 / 股関節 / 超音波検査 / Graf法 / 深層学習 / コンピュータ支援診断

Outline of Final Research Achievements

A deep learning based method was adopted for the automated classification of hip types corresponding to the Graf method that was defacto standard method for ultrasonographic assessment of infant hip dysplasia. The accuracy for 99 ultrasound images was 75.8%. A deep learning based method for recognition of adequate diagnostic images from ultrasound video clips of infant hip was also proposed. The proposed method was implemented to 12 ultrasound vide clips of infant hip, and indicated high classification accuracies (> 85%).

Free Research Field

医用画像情報学

Academic Significance and Societal Importance of the Research Achievements

近年の深層学習の進歩は学術的にも社会的にも注目を集めており，様々な分野への応用が期待されている．本研究成果は，医用画像診断への深層学習の適用事例として学術的意義を有するものである．また，本研究成果は乳児股関節検診の質的診断支援に寄与するものであり，乳児健診時の股関節異常の発見率の向上に貢献できる可能性を示唆している．研究期間内に実施した研究が，先天性股関節脱臼などの乳児股関節異常の早期発見の一助になれば幸いである．