Premium
P1‐443: OPTIMIZING THE COMBINATION OF NEUROPSYCHOLOGICAL TESTS FOR EFFECTIVE COGNITIVE IMPAIRMENT CLASSIFICATION
Author(s) -
Gupta Abhay
Publication year - 2019
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.2019.06.1048
Subject(s) - support vector machine , feature selection , dementia , artificial intelligence , machine learning , random forest , cognitive test , episodic memory , test set , linear discriminant analysis , cognition , computer science , psychology , pattern recognition (psychology) , medicine , psychiatry , disease , pathology
were segmented using a previously validated pipeline based on location and intensity features from T1w and FLAIR images [Dadar et al., NeuroImage 2017]. WMH volumes were calculated per each brain lobe and hemisphere. Mixed-effects models were used to assess the WMH differences between controls and FTD patients (Model 1:WMHw1+Cohort+Age+1jID+1jSite) and their effect on cognition, as measured by the Mini-Mential State Examination (MMSE) scores (Model 2: MMSEw1+WMH+Cohort+WMH:Cohort+Age+1jID+1jSite). In Model 1, the variable of interest was Cohort (i.e. Control vs bv-FTD, sv-FTD, and pnfa-FTD). In Model 2, the variable of interest was the interaction between Cohort and WMH load, denoted byWMH:Cohort. ID and Site were considered as categorical random effects. Results: WMHs increased significantly with age for all lobes (p<0.01). Compared to controls, bvFTD patients had significantly higher WMH loads in the frontal (bRight1⁄40.805,pRight<0.0001, bLeft1⁄40.796,pLeft<0.0001), parietal (bRight1⁄40.647,pRight<0.0001, bLeft1⁄40.411,pLeft1⁄40.002), and occipital lobes (bRight1⁄40.345,pRight1⁄40.01, bLeft1⁄40.302,pLeft1⁄40.02), sv-FTD patients had higher WMH loads in the parietal (bRight1⁄40.465,pRight1⁄40.002, bLeft1⁄40.373,pLeft1⁄40.02) and left frontal (bLeft1⁄40.325,pLeft1⁄40.03) and temporal (bRight1⁄40.288,pRight1⁄40.05, bLeft1⁄40.469,pLeft1⁄40.003) lobes, and pnfa-FTD patients had higher WMH loads in the right parietal lobe (bRight1⁄40.423,pRight1⁄40.009) (Fig.1). Compared to controls, decrease in MMSE was associated with increased WMH load in frontal lobe in bv-FTD (bRight1⁄40.309,pRight1⁄40.001, bLeft1⁄4-0.373,pLeft1⁄40.0001) and pnfa-FTD patients (bRight1⁄4-0.686,pRight<0.0001, bLeft1⁄4-0.685,pLeft<0.0001), as well as parietal lobe in pnfa-FTD patients (bRight1⁄40.443,pRight1⁄40.02, bLeft1⁄4-0.449,pLeft1⁄40.001), and right parietal lobe in bv-FTD patients (bRight1⁄4-0.215,pRight1⁄40.02) (Fig.2). Conclusions: FTD subtypes present with different amounts of regional WMHs, matching their patterns of predominant cortical atrophy (frontal in bv-FTD and temporal in sv-FTD) but not restricted to fronto-temporal areas. Those WMHs are associated with increased cognitive deficits. P1-443 OPTIMIZING THE COMBINATION OF NEUROPSYCHOLOGICALTESTS FOR EFFECTIVE COGNITIVE IMPAIRMENT CLASSIFICATION