Premium
P3‐429: IS THE POSTERIOR SUBICULUM A KEY NODE IN THE BRAIN NETWORK THAT BECOMES PATHOLOGICAL IN ALZHEIMER'S DISEASE?
Author(s) -
Lindberg Olof,
Stomrud Erik,
Mårtensson Gustav,
Kern Silke,
Westman Eric,
Wahlund Lars-Olof,
Skog Ingmar,
Hansson Oskar
Publication year - 2018
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1016/j.jalz.2018.06.1792
Subject(s) - subiculum , posterior cingulate , precuneus , neuroscience , retrosplenial cortex , atrophy , temporal lobe , posterior cortical atrophy , psychology , cortex (anatomy) , hippocampus , superior parietal lobule , dementia , anatomy , medicine , pathology , functional magnetic resonance imaging , disease , dentate gyrus , epilepsy
and a FLAIR sequence. All hemispheres underwent neuropathologic examination by a board-certified neuropathologist (Fig.1). Logistic regression with L1 penaltywas used for training and testing two classifiers. The first classifier (referred to as baseline) used whole brain WMH burden as the sole feature. The second classifier was based on a feature vector including regional WMH burden, magnetic susceptibility of subcortical gray matter, number of microbleeds, and demographics (age, sex, education). 10-fold cross validation was used to assess generalization performance. Results: According to the receiver operating characteristic (ROC) (Fig.2) and precisionrecall (PR) curves (Fig.3), the proposed classifier showed superior performance over the baseline classifier, for all thresholds (p1⁄40.068). Conclusions: The proposed classifier based on regional WMH burden, magnetic susceptibility, microbleeds and demographics, was shown to have superior performance for identifying older adults with arteriolar sclerosis compared to a classifier based only on whole brain WMH burden. We will soon be able to dramatically increase the sample size, include additional features, and test the classifier in-vivo. Successful completion of this work will bring forward a much-needed biomarker for arteriolar sclerosis for use in clinical trials.