Brain Structural Saliency Over The Ages

TL;DR

This study uses deep learning and saliency mapping to identify brain regions associated with aging, revealing patterns of relevance change across different brain structures over the lifespan.

Contribution

It introduces a method combining ResNet with Layer-wise Relevance Propagation and DeepLIFT to analyze brain age models, highlighting age-related relevance shifts in brain regions.

Findings

Relevance to certain brain regions increases with age.

Relevance to other regions decreases with age.

Some regions remain consistently relevant across ages.

Abstract

Brain Age (BA) estimation via Deep Learning has become a strong and reliable bio-marker for brain health, but the black-box nature of Neural Networks does not easily allow insight into the features of brain ageing.We trained a ResNet model as a BA regressor on T1 structural MRI volumes from a small cross-sectional cohort of 524 individuals. Using Layer-wise Relevance Propagation (LRP) and DeepLIFT saliency mapping techniques, we analysed the trained model to determine the most relevant structures for brain ageing for the network, and compare these between the saliency mapping techniques. We show the change in attribution of relevance to different brain regions through the course of ageing. A tripartite pattern of relevance attribution to brain regions emerges. Some regions increase in relevance with age (e.g. the right Transverse Temporal Gyrus); some decrease in relevance with age (e.g. the right Fourth Ventricle); and others are consistently relevant across ages. We also examine the effect of the Brain Age Gap (BAG) on the distribution of relevance within the brain volume. It is hoped that these findings will provide clinically relevant region-wise trajectories for normal brain ageing, and a baseline against which to compare brain ageing trajectories.