Structural complexity of brain regions in mild cognitive impairment and Alzheimers disease

Abstract

Early detection of Alzheimer’s disease (AD) has been a major focus of current research efforts to guide interventions at the earliest stages of the disease. Subtle changes to the brain might be observed with neuroimaging techniques, even before symptoms surface. We interrogated brain images obtained with Magnetic Resonance Imaging (MRI) from two large-scale dementia datasets (namely, ADNI and BioFIND) to establish the utility of fractal dimensionality (FD)—a relatively understudied measure that estimates the complexity of 3D structures (in this case, brain regions)—for the detection of AD. We show that FD can be used to detect group differences between patients and healthy controls, with the former showing significantly reduced complexity across multiple brain regions. Furthermore, these measures were successful when used as features for individual-based classification and were highly consistent across the two datasets. Finally, the contribution of specific brain regions to individual-based classification adhered to previous literature on the properties of the brain’s memory network. Taken together, the study offers novel and interpretable evidence for the utility of FD for the detection of AD.