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

• Published in: Human brain mapping Vol. 42; no. 8; pp. 2332 - 2346
• Main Authors: Baecker, Lea, Dafflon, Jessica, Costa, Pedro F., Garcia‐Dias, Rafael, Vieira, Sandra, Scarpazza, Cristina, Calhoun, Vince D., Sato, João R., Mechelli, Andrea, Pinaya, Walter H. L.
• Format: Journal Article
• Language: English
• Published: Hoboken, USA John Wiley & Sons, Inc 01.06.2021
• Subjects: Abnormalities; Age; Age Factors; Aged; Aging; biological ageing; Brain; Brain - anatomy & histology; Brain - diagnostic imaging; Female; Gaussian process; healthy ageing; Humans; Learning algorithms; Machine Learning; Magnetic resonance imaging; Magnetic Resonance Imaging - methods; Magnetic Resonance Imaging - standards; Male; Middle Aged; Morphology; Neuroimaging; Neuroimaging - methods; Neuroimaging - standards; Predictions; Principal components analysis; Regression; Regression Analysis; Support Vector Machine; Support vector machines; biological ageing; support vector machine; machine learning; healthy ageing; regression analysis
• ISSN: 1065-9471; 1097-0193; 1097-0193
• DOI: 10.1002/hbm.25368

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. We compared the machine learning models support vector regression, relevance vector regression and Gaussian process regression for brain age prediction using different types of morphometric input and sample sizes of more than 10,000 subjects. The mean absolute error across the different models ranged from 3.7 to 4.7 years. The type of data input (region‐ or voxel‐level) had a greater impact on performance than the choice of model.