AUTHOR=Kautzky Alexander , Seiger Rene , Hahn Andreas , Fischer Peter , Krampla Wolfgang , Kasper Siegfried , Kovacs Gabor G. , Lanzenberger Rupert 

TITLE=Prediction of Autopsy Verified Neuropathological Change of Alzheimer’s Disease Using Machine Learning and MRI

JOURNAL=Frontiers in Aging Neuroscience

VOLUME=Volume 10 - 2018

YEAR=2018

URL=https://www.frontiersin.org/journals/aging-neuroscience/articles/10.3389/fnagi.2018.00406

DOI=10.3389/fnagi.2018.00406

ISSN=1663-4365

ABSTRACT=Background: Alzheimer´s disease (AD) is the most common form of dementia. While neuropathological changes pathognomonic for AD have been defined, early detection of AD prior to cognitive impairment in the clinical setting is still lacking. While pioneer studies applying machine learning to magnetic-resonance imaging (MRI) data to predict mild cognitive impairment or AD have yielded high accuracies, a classification algorithm for neuropathological change is still lacking.
Methods: Classification of autopsy verified neuropathological changes attributed to AD, as described by the National Institute on Aging - Alzheimer’s Association (NIAA) AD score, were predicted with RandomForest. Forty-nine patients of the prospective VITA study cohort were enrolled, using nested cross-validation on MRI data with a region of interest (ROI) design. Neuropathological assessment was performed for all subjects and MRI scans were performed at least two years prior to death. 
Results: The most informative ROIs included caudal and rostral anterior cingulate gyrus, entorhinal, fusiform and insular cortex and the subcortical ROIs anterior corpus callosum and the left vessel. The resulting classification models achieved an average accuracy for a three levelled NIAA AD score of 0.62 for the prediction in the decoding sets and of 0.61 for validation sets. Higher accuracies of 0.77 for both sets respectively were achieved when predicting presence or absence of neuropathological change.
Conclusion: This application of computer-aided prediction of autopsy verified neuropathological change according to the categorical NIAA score in AD may facilitate a more distinct and definite categorization of patients at risk for AD dementia. Reliable detection of neuropathological hallmarks of AD would enable risk stratification at an earlier level than prediction of MCI or clinical AD symptoms and advance precision medicine in neuropsychiatry.