Assessing Glymphatic System Impairment in Alzheimer's Disease Using Enlarged Perivascular Spaces with Automatic Quantification and DTI-ALPS Method

Abstract Objectives The glymphatic system has gained widespread attention in neurodegenerative diseases. The present study attempted to jointly assess the glymphatic system in Alzheimer’s disease (AD), amnestic mild cognitive impairment (aMCI), and normal controls (NC) using diffusion tensor imaging along the perivascular space (DTI-ALPS) and perivascular spaces (PVS) indexes as evaluation metrics. Materials & Methods A total of 89 AD, 24 aMCI, and 32 NC participants were recruited in this study. The bilateral ALPS index was calculated on the DTI image. A deep learning V-shape bottleneck network (VB-Net) model was employed to automatically segment the lesion of enlarged perivascular spaces (EPVS). The total volume and the number of EPVS lesions was estimated as EPVS burden, while the volume fractions of basal ganglia (BG) and centrum semiovale (CSO) EPVS were counted. All indexes above were compared among the three groups, followed by correlation analyses using cognitive scales including MMSE and MoCA scales. Results VB-Net model segmented the EPVS lesions automatically and precisely. In the stage of glymphatic assessment, AD and aMCI groups exhibited varying degrees of lower ALPS index values, higher EPVS burden and BG EPVS volume fraction in both hemispheres of the brain compared to NC. ALPS index values, EPVS burden, and BG EPVS volume fraction were significantly correlated with cognitive clinical scales. EPVS burden and BG EPVS volume fraction were also correlated with ALPS index. Conclusion Our study confirmed impairments in the glymphatic system parameters of patients with AD and aMCI with DTI-ALPS and EPVS, correlated with worse cognitive performance, which provided an in-depth understanding and facilitated early detection of the disease.