Classification of late- and early-onset Alzheimer’s disease using grey matter density measures

A wide range of studies show that there are significant differences between early- and late-onset Alzheimer’s disease (AD), notably in symptomology and physiological biomarkers. Since early-onset AD patients show, on average, more severe grey matter atrophy in their brain compared to late-onset AD, this paper examined whether separate classification models trained on early- and late-onset AD improve the respective group’s classification accuracy. In this study, grey matter density predictors obtained with MRI scans were used, corresponding to 48 different cortex regions. The dataset utilized in this study consisted of N = 233 participants, including AD patients (N = 60) and controls (N = 173). The AD patients were split into an early-onset group (<65) and a late-onset group (≥65), both groups consisting of N = 30 participants. In order to rule out effects of age-related decline, the control group was split into a young control group (N = 90) and an old control group (N = 83). LASSO...Show moreA wide range of studies show that there are significant differences between early- and late-onset Alzheimer’s disease (AD), notably in symptomology and physiological biomarkers. Since early-onset AD patients show, on average, more severe grey matter atrophy in their brain compared to late-onset AD, this paper examined whether separate classification models trained on early- and late-onset AD improve the respective group’s classification accuracy. In this study, grey matter density predictors obtained with MRI scans were used, corresponding to 48 different cortex regions. The dataset utilized in this study consisted of N = 233 participants, including AD patients (N = 60) and controls (N = 173). The AD patients were split into an early-onset group (<65) and a late-onset group (≥65), both groups consisting of N = 30 participants. In order to rule out effects of age-related decline, the control group was split into a young control group (N = 90) and an old control group (N = 83). LASSO logistic regression was used to create three classification models, one based on the whole group, one based on early-onset AD and one based on late-onset AD. All three models were tested separately on early- and late-onset AD. Results showed that the classification model trained on the whole group had a higher classification accuracy for the early-onset AD group compared to the late-onset AD group. Additionally, a dedicated model did improve the classification accuracy for the late-onset AD group, but this was not the case for the early-onset AD group. Memory-related cortex regions were found to be more important in the late-onset AD model, and language-related cortex regions were found to be more important in the early-onset AD model. These results provide initial evidence on the important impact of onset-age on grey matter density classification performance.Show less

Leiden University Student Repository

Documents

In Collections

Classification of late- and early-onset Alzheimer’s disease using grey matter density measures