Brain age prediction: A comparison between machine learning models using region- and voxel-based morphometric data

Author(s)

Baecker, Lea ^[1] ; Dafflon, Jessica ^[2] ; da Costa, Pedro F. ^[2] ; Garcia-Dias, Rafael ^[1] ; Vieira, Sandra ^[1] ; Scarpazza, Cristina ^{[1, 3]} ; Calhoun, Vince D. ^{[4, 5]} ; Sato, Joao R. ^[6] ; Mechelli, Andrea ^[1] ; Pinaya, Walter H. L. ^{[6, 1, 7]}

Affiliation

^[1] Kings Coll London, Inst Psychiat Psychol & Neurosci, Dept Psychosis Studies, London - England

^[2] Kings Coll London, Inst Psychiat Psychol & Neurosci, Dept Neuroimaging, London - England

^[3] Univ Padua, Dept Gen Psychol, Padua - Italy

^[4] Georgia State Univ, Triinst Ctr Translat Res Neuroimaging & Data Sci, Atlanta, GA 30303 - USA

^[5] Emory Univ, Georgia Inst Technol, Atlanta, GA 30322 - USA

^[6] Univ Fed ABC, Ctr Math Comp & Cognit, Sao Paulo - Brazil

^[7] Kings Coll London, Sch Biomed Engn & Imaging Sci, Dept Biomed Engn, London - England

• Document type: Journal article
• Source: Human Brain Mapping; v. 42, n. 8 MAR 2021.
• Web of Science Citations: 1

Abstract

Brain morphology varies across the ageing trajectory and the prediction of a person's age using brain features can aid the detection of abnormalities in the ageing process. Existing studies on such ``brain age prediction{''} vary widely in terms of their methods and type of data, so at present the most accurate and generalisable methodological approach is unclear. Therefore, we used the UK Biobank data set (N = 10,824, age range 47-73) to compare the performance of the machine learning models support vector regression, relevance vector regression and Gaussian process regression on whole-brain region-based or voxel-based structural magnetic resonance imaging data with or without dimensionality reduction through principal component analysis. Performance was assessed in the validation set through cross-validation as well as an independent test set. The models achieved mean absolute errors between 3.7 and 4.7 years, with those trained on voxel-level data with principal component analysis performing best. Overall, we observed little difference in performance between models trained on the same data type, indicating that the type of input data had greater impact on performance than model choice. All code is provided online in the hope that this will aid future research. (AU)

• FAPESP's process: 18/21934-5 - Network statistics: theory, methods, and applications
• Grantee: André Fujita
• Support Opportunities: Research Projects - Thematic Grants
• FAPESP's process: 18/04654-9 - Time series, wavelets and high dimensional data
• Grantee: Pedro Alberto Morettin
• Support Opportunities: Research Projects - Thematic Grants