Enhancing brain age estimation with structural MRI and synthesized cerebral blood volume maps

Abstract

Abstract Brain age gap estimation (BrainAGE) is a promising imaging-derived biomarker of neurobiological ageing and disease risk, yet current approaches rely predominantly on T1-weighted structural MRI, overlooking functional vascular changes that may precede tissue damage and cognitive decline. Deep learning-derived cerebral blood volume (DeepCBV) maps, synthesized from non-contrast MRI, offer a scalable alternative to contrast-enhanced perfusion imaging by capturing vascular information relevant to early neurodegeneration. We developed a multimodal BrainAGE framework that combines predictions from two separate three-dimensional convolutional neural networks: one trained only on structural MRI scans and another trained only on DeepCBV maps generated by a pre-trained three-dimensional patch-based deep learning model. Each model was trained and validated on 2851 scans (1507 females) from 13 open-source datasets and was evaluated for concordance with mild cognitive impairment (MCI) and Alzheimer’s disease (AD) using 1233 subjects. The combined model achieved the most accurate brain age gap for cognitively normal (CN) controls, with a mean absolute error of 3.95 years (R2 = 0.943), outperforming models trained on MRI (mean absolute error = 4.10) or DeepCBV alone (mean absolute error = 4.49). Saliency maps revealed complementary modality contributions: MRI emphasized white matter and cortical atrophy, while DeepCBV highlighted vascular-rich and periventricular regions implicated in hypoperfusion and early cerebrovascular dysfunction, consistent with known patterns of normal ageing. Next, we observed that BrainAGE increased stepwise across diagnostic strata (CN < MCI < AD) and correlated with cognitive impairment (Clinical Dementia Rating Sum of Boxes ⍴ = 0.403; Mini-Mental State Examination ⍴ = −0.310). DeepCBV-based BrainAGE showed a particularly strong separation between stable versus progressive MCI (Mann–Whitney U = 2.177 × 104, P = 4.43 × 10−8), suggesting sensitivity to prodromal vascular changes that precede overt atrophy. Integrating structural MRI with deep learning-derived vascular measures substantially enhances BrainAGE estimation and improves sensitivity to MCI and Alzheimer’s disease progression, supporting its potential role in risk stratification, early detection and monitoring of therapeutic response. By enabling a functional-like assessment from routine MRI, this approach lowers barriers to multimodal evaluation and provides a clinically actionable biomarker for large-scale ageing and dementia studies.