Massive Voxel-Based Morphometry of UK Biobank Data
Title: Massive Voxel-Based Morphometry of UK Biobank Data
SNIC Project: LiU-compute-2021-23
Project Type: LiU Compute
Principal Investigator: Paul Hamilton <>
Affiliation: Linköpings universitet
Duration: 2021-06-01 – 2022-06-01
Classification: 20603


Massive multivariate, machine learning techniques for estimating "brain age" from structural neuroimaging data have become popular in recent years. These techniques, however, have not been validated against other biological measures more proximate to the aging process. One such measure that has been applied in recent years is telomere length. Telomeres are regions of repeated nucleotide sequences that code for proteins at the ends of chromosomes. Telomeres are thought to protect the terminal regions of chromosomal DNA from degradation. Correspondingly, telomere shortening is robustly associated with aging, mortality and diseases that accompany aging. We propose using LiU compute to conduct voxel-based morphometry (VBM) on ~48,000, whole-brain, T1-weighted images downloaded from the UK Biobank. For this project, we intend: 1) to enter whole-brain VBM data into a convolutional neural network for estimating "brain age"; and 2) to then determine how well the network-derived estimates of brain age correlate with telomere length, a measure that is also available via the UK Biobank. By completing this project, we will be able to determine the degree of convergence between brain-age estimates and aging at a cellular level.