Background A structural magnetic resonance imaging measure of combined neurovascular and neurodegenerative burden may be useful as these features often coexist in older people, stroke and dementia. Aim We aimed to develop a new automated approach for quantifying visible brain injury from small vessel disease and brain atrophy in a single measure, the brain health index. Materials and methods We computed brain health index in N = 288 participants using voxel-based Gaussian mixture model cluster analysis of T1, T2, T2*, and FLAIR magnetic resonance imaging. We tested brain health index against a validated total small vessel disease visual score and white matter hyperintensity volumes in two patient groups (minor stroke, N = 157; lupus, N = 51) and against measures of brain atrophy in healthy participants (N = 80) using multiple regression. We evaluated associations with Addenbrooke's Cognitive Exam Revised in patients and with reaction time in healthy participants. Results The brain health index (standard beta = 0.20-0.59, P < 0.05) was significantly and more strongly associated with Addenbrooke's Cognitive Exam Revised, including at one year follow-up, than white matter hyperintensity volume (standard beta = 0.04-0.08, P > 0.05) and small vessel disease score (standard beta = 0.02-0.27, P > 0.05) alone in both patient groups. Further, the brain health index (standard beta = 0.57-0.59, P < 0.05) was more strongly associated with reaction time than measures of brain atrophy alone (standard beta = 0.04-0.13, P > 0.05) in healthy participants. Conclusions The brain health index is a new image analysis approach that may usefully capture combined visible brain damage in large-scale studies of ageing, neurovascular and neurodegenerative disease.
Keywords: Atrophy; cerebral small vessel diseases; cognition; computer-assisted; image processing; magnetic resonance imaging; stroke.