Structural MRI Preprocessing for Biological Brain-Age Estimation

Abstract

Brain age is an MRI-derived estimate of an individual’s biological brain age based on structural and functional characteristics of the brain. Unlike chronological age, it reflects inter-individual variation in brain maturation, ageing, and possible disease-related alterations. The difference between predicted and chronological age, known as the brain-age gap, may provide a non-invasive marker of accelerated or delayed brain ageing. Brain-age modelling can therefore support the study of normal ageing, early detection of pathology, risk stratification, and monitoring of treatment effects.
This research aims to develop a reproducible pipeline for brain-age prediction from structural T1-weighted MRI. The current focus is the preparation and quality assessment of MRI data before neural network training. The implemented preprocessing workflow includes input validation, defacing, gradient-distortion handling, linear and nonlinear registration to the MNI space, brain masking, bias-field correction, tissue segmentation using FAST, head-size normalization with SIENAX, and subcortical segmentation with FIRST. In addition, regional grey-matter measures are extracted from a 139-region atlas, while technical quality-control metrics and visual montages are generated for each scan. These standardized derivatives will be used to train and evaluate a neural network to predict biological brain age.
The research was financially supported by the Ministry of Science and Higher Education of the Russian Federation (Agreement on the provision from the federal budget of a grant in the form of subsidies for state support of development programs for world-class research centers carrying out research and development in priority areas of scientific and technological development dated June 3, 2026, No. 075-15-2026-383).

Speaker

Mikhail Sokolov
Bauman Moscow State Technical University
Russia

Discussion

Ask question