Automatic differential dementia diagnosis using brain FDG-PET, MRI, and machine learning

2017 Fiscal Year Final Research Report

• Project/Area Number: 15K09982
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Multi-year Fund
• Section: 一般
• Research Field: Radiation science
• Research Institution: Tokyo Metropolitan Geriatric Hospital and Institute of Gerontology
• Principal Investigator: Sakata Muneyuki 地方独立行政法人東京都健康長寿医療センター(東京都健康長寿医療センター研究所), 東京都健康長寿医療センター研究所, 研究員 (00403329)
• Co-Investigator(Renkei-kenkyūsha): ISHII Kenji 地方独立行政法人東京都健康長寿医療センター, 東京都健康長寿医療センター研究所, 研究部長 (10231135) ; WAGATSUMA Kei 地方独立行政法人東京都健康長寿医療センター, 東京都健康長寿医療センター研究所, 技術員 (40738283) ; KIMURA Yuichi 近畿大学, 生物理工学部, 教授 (60205002)
• Project Period (FY): 2015-04-01 – 2018-03-31
• Keywords: 認知症診断 / PET / MRI / 機械学習 / サポートベクターマシン / アルツハイマー病

Outline of Final Research Achievements

Differential diagnosis of dementia type is necessary for treatment and care of dementia. In this research, we provided an automatic discrimination using machine learning of brain FDG-PET images and MR images. Support vector machines were employed for training of extracted features from FDG-PET images as the index of hypometabolism, and MR images as the index of cerebral atrophy. To support the clinical diagnosis, the system automatically classifies the groups of Alzheimer’s disease, front-temporal dementia, dementia with Lewy bodies, and healthy subjects, and indicates a novel image index for severity or progression of the dementia from the classification.

Free Research Field

医用画像解析