Study of the relationships between quantitative measurements of cerebral vascularization obtained by artificial intelligence, cognitive functions, and biomarkers of Alzheimer’s disease
Vivian Ni1, Félix Janelle1, Félix Dumais1, Christian Bocti1, Kevin Whittingstall1
1Université de Sherbrooke
Objective:
Explore the relationship between the diameters and the volumes of the intracranial arteries and i) cognition; ii) Alzheimer’s Disease (AD) biomarkers, specifically hippocampal and cerebral atrophy and PET amyloid burden; iii) cognitive decline.
Background:
With the aging of the world population, approximately 78 million people worldwide will be affected by a major neurocognitive disorder (NCD) in 2030, with up to 80% of cases being of AD type. However, its etiology remains unclear. Given the growing evidence supporting a relationship between vascular factors and AD, we explored this hypothesis using a novel artificial intelligence (AI) algorithm.
Design/Methods:
Of the 409 participants (42-95 years old) selected from the OASIS-3 database, 338 had a normal cognition, 53, a questionable NCD, 16, a mild NCD, and 2, a major NCD. Intracranial arteries were evaluated by time-of-flight magnetic resonance angiography (TOF-MRA) and their measurements were calculated using a newly developed AI system. Cognitive status was assessed by MMSE, CDR, and several neuropsychological tests, including Logical Memory and Trail Making tests. Amyloid burden was visualized by amyloid PET with contrast (Pittsburgh Compound B or Florbetapir) and quantified using the Centiloid scale. Brain and hippocampal volumes were assessed by T1-weighted MRI and the resulting images were processed by the Freesurfer image analysis suite.
Results:
The anterior choroidal arteries (AchA) had the most associations (nonlinear and negative) with cognitive functions and AD markers. Conversely, the anterior cerebral (ACA), basilar, and 1st segment of the posterior cerebral arteries (PCA-P1) had very little relationship with cognition. Among the statistically significant results, the correlations were of low magnitude.
Conclusions:
Using TOF-MRA imaging and AI, we were able to calculate the quantitative measurements of the intracranial arteries and thus assess their association with certain AD elements. Future research to further explore the vascular hypothesis of AD should be conducted in order to confirm these findings.
10.1212/WNL.0000000000202670