Quantitative Assessment on the Severity Degree of Alzheimer Dementia by Algebraic Bigdata Analysis on Cortical Thickness Profiles of Human Brains

Abstract Background: Alzheimer disease(AD) affects profoundly the quality of human life. Quantifying the severity degree of AD for an individual person is critical for the early diagnose and prescription for delaying the further dementia progression. However, the quantitative diagnose for human subjects of the mild cognitively impairment (MCI) or AD with the different degree of dementia severity is still a difficult task due to both the very broad distribution of dementia severities and the lack of good quantitative determinant to assess it.Methods: We performed the bigdata analysis of cortical thickness of 1516 human brain images. Instead of dealing with cortical thickness at all 327,684 vertices on the whole cortex of a human brain from MRI, we extracted the essential cortical thickness data at a few hundred vertices to distinguish efficiently AD, MCI, and cognitively normal (CN) cohorts each other. We learned a statistical score matrix and a covariance correlation matrix of cortical thickness profile between human subjects as a set of classifier and predictor for diagnosing the dementia states of AD.Results: The subjects with AD were recognized with more than 91% accuracy in self-recognition test, and the independent validation subjects with AD were predicted correctly with more than 82% accuracy in stratified 3-fold cross validation test. Given a new person for diagnosing and provided with the covariance correlation matrix developed in this study as a predictor, more importantly one could estimate the personalized severity degree of dementia in terms of a quantitative value, ranging from 0 for the basin of CN state to 1 for the basin of AD state. Also, one could sort out the broad spectrum of the severity of dementia for MCI subjects in that whether they are prone to CN or how much they are progressed toward AD.Conclusions: Here, a new approach of algebraic bigdata analysis is presented for the quantitative assessment on both the severity degree of AD and the systematic classification of three cohorts based on cortical thickness data of 1516 human brain images from MRI. This new approach can facilitate the better diagnose of AD with the different degree of dementia.