Machine learning of neuroimaging for assisted diagnosis of cognitive impairment and dementia: A systematic review. Issue 1 (10th August 2018)

• Record Type: Journal Article
• Authors: Pellegrini, Enrico; Ballerini, Lucia; Hernandez, Maria del C. Valdes; Chappell, Francesca M.; González‐Castro, Victor; Anblagan, Devasuda; Danso, Samuel; Muñoz‐Maniega, Susana; Job, Dominic; Pernet, Cyril; Mair, Grant; MacGillivray, Tom J.; Trucco, Emanuele; Wardlaw, Joanna M.
• Is Part Of: Alzheimer's & dementia. Volume 10:Issue 1(2018)
• Page Start: 519
• Page End: 535
• Publication Date: 2018-08-10
• Subjects: Dementia -- Cerebrovascular disease -- Pathological aging -- Small vessel disease -- MRI -- Machine learning -- Classification -- Segmentation; Alzheimer's disease -- Periodicals; Alzheimer's disease -- Diagnosis -- Periodicals; Dementia -- Periodicals; Dementia -- Diagnosis -- Periodicals; 616.831
• DOI: 10.1016/j.dadm.2018.07.004
• Languages: English
• ISSNs: 2352-8729

Abstract

Abstract: Introduction: Advanced machine learning methods might help to identify dementia risk from neuroimaging, but their accuracy to date is unclear. Methods: We systematically reviewed the literature, 2006 to late 2016, for machine learning studies differentiating healthy aging from dementia of various types, assessing study quality, and comparing accuracy at different disease boundaries. Results: Of 111 relevant studies, most assessed Alzheimer's disease versus healthy controls, using AD Neuroimaging Initiative data, support vector machines, and only T1‐weighted sequences. Accuracy was highest for differentiating Alzheimer's disease from healthy controls and poor for differentiating healthy controls versus mild cognitive impairment versus Alzheimer's disease or mild cognitive impairment converters versus nonconverters. Accuracy increased using combined data types, but not by data source, sample size, or machine learning method. Discussion: Machine learning does not differentiate clinically relevant disease categories yet. More diverse data sets, combinations of different types of data, and close clinical integration of machine learning would help to advance the field. Highlights: Systematic review of machine learning methods of neuroimaging was performed. Machine learning to predict risk of dementia does not seem ready for clinical use. Methods have high accuracy to differentiate Alzheimer's disease versus healthy control. Performances were poorer when assessing more clinically relevant distinctions.