Visuospatial dysfunction in AD is common and a direct contributor to falls, car accidents and loss of independence. Better understanding of visuospatial network dysfunction in AD allows for earlier intervention and improved safety. Atrophy mapping studies have localized visuospatial dysfunction to multiple brain regions suggesting it maps to a network. Here, cortical thickness and atrophy network maps were created and compared to clock-copy and clock-draw scores to identify the network underlying visuospatial dysfunction in AD using the Alzheimer’s Disease Neuroimaging Initiative (ADNI). 

Initial analyses were run using ADNI1 then validated using independent subjects in ADNI2. Cortical thickness was directly compared to clock-draw and clock-copy scores identifying brain regions directly associated with performance. A general-linear-model was generated from each cohort’s healthy controls and used to estimate expected cortical thickness for AD subjects. Atrophy maps were generated using multiple Z-score cutoffs at a voxel-wise level. Atrophy network maps were generated using the normative connectome and compared to clock-copy and clock-draw scores.

Single-subject cortical thickness maps compared directly to clock-copy or clock-draw scores did not identify any voxels that correlated with scores meeting FWE p<0.05. Atrophy network maps compared to clock-draw scores identified a network but not voxels meeting FWE p<0.05. However, atrophy network mapping identified a map centered on the bilateral angular gyri significantly correlating with clock-copy scores. Clock-copy atrophy network maps showed higher reproducibility than clock-draw. 

Our results show that clock-draw and clock-copy scores map to networks rather than single brain regions. Clock-draw maps to a distributed brain network while clock-copy maps to the angular gyrus and is more reproducible. Clock-copy is a more specific measure of parietal lobe function in AD than clock-draw.