Premium
P2‐004: Bayesian Network analysis reveals altered regional connectivity of the default mode network in Alzheimer's disease
Author(s) -
Fleisher Adam,
Chen Kewei,
Li Rui,
Guan Xiaoting,
Zhang Yumei,
Reiman Eric,
Yao Li,
Wu Xia
Publication year - 2009
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.2009.04.313
Subject(s) - default mode network , resting state fmri , neuroscience , posterior cingulate , voxel , cortex (anatomy) , psychology , functional magnetic resonance imaging , medicine , radiology
Background: A short, simple MEG scan that measures synchronous functional activity has been shown to distinguish among patients diagnosed with multiple neurological and psychiatric disorders (Georgopoulos et al., 2007; J. Neural Eng. 4:349). We conducted a retrospective, case-controlled study at two MEG centers to investigate specific patterns of synchronous brain activity that distinguish persons with probable AD from healthy control subjects (ageand gender-matched). Methods: Subjects were recruited from the community based on a previous diagnosis of Alzheimer’s disease (n 1⁄4 47). Healthy, ageand gender-matched volunteers also were included in the study (n 1⁄4 90). After a neurological exam and neuropsychological testing to confirm the diagnosis (probable AD by NINCDSADRDA criteria or healthy controls), a single MEG scan (one minute scan, eyes-open, at rest) was conducted using a 4D Neuroimaging WH 3600 gradiometer instrument. Time series of MEG data were prewhitened using an auto-regressive integrated moving average model, and cross-correlations were calculated for each pair of a 248-sensor array. Patterns of cross-correlations were characterized in each subject and analyzed using computer algorithms. Accuracy in distinguishing the subject groups within the cohort was estimated by jack-knife cross-validation. Results: The synchronous neural interaction test distinguished between MEG scans from healthy volunteers and scans from patients with probable AD. These differences were similar to disease-specific patterns of brain electrical activity previously observed using EEG and MEG. Disease-relevant cross-correlations were selected, and subsequent jack-knife cross-validation with the selected cross-correlations indicated that persons with probable AD could be distinguished from healthy volunteers with a sensitivity of 96% and a specificity of 90%. Conclusions: Brief MEG scans conducted under resting conditions combined with prewhitening and cross-correlation analysis (the synchronous neural interaction test) identified a signature of AD pathology based on functional measurements. Clinical studies are underway that will increase the number of subjects in our MEG scan database, support formally blinded analysis of diagnostic accuracy, and demonstrate the usefulness of our test for measuring the effect of treatment.