Amyloid-beta biomarkers in Braak stages and their predictive relationships with cognitive impairment: Support vector machine and deep learning approaches

Amyloid-beta (A{beta}) and tau tangles are hallmarks of Alzheimers disease. A{beta} distributions in the tau-defined Braak staging regions and their multivariate predictive relationships with mild cognitive impairment (MCI) are not known. In this study, we used PiB PET data from 60 participants (33 with MCI and 27 healthy controls (HC)), quantified A{beta} as distribution volume ratio (DVR) in Braak regions, and compared between MCI and controls to test the hypothesis that DVR alters with declining cognition. We found elevated DVR in participants with MCI, especially in the spatial distribution of Braak stages III-IV and V-VII, while an alteration in Braak stage I-II was near the statistical significance. DVR markers correlated with cognitive status, especially in Braak stages III-IV and VI-V. To evaluate whether these markers are predictive of cognitive dysfunction, we designed support vector machine and artificial neural network models. These methods showed predictive multivariate relationships between A{beta} makers of Braak regions and cognitive impairment. Overall, these results highlight the importance of computer-aided research efforts for understanding AD pathophysiology.