Abstract Details

Title Functional Connectivity Measured with Magnetoencephalography Distinguishes Mild Cognitive Impairment from Normal Aging

Topic Aging and Dementia

Presentation(s) S44 - (-)

Poster/Presentation
Number 004

Objective

Background Alzheimer's disease (AD) and the prodromal syndrome of MCI represent disconnection syndromes. We evaluated different FC measures to determine the extent to which abnormalities in functional networks could reliably identify MCI.

Design/Methods Eyes-closed, resting state MEG data were obtained from sites in Spain, Finland, USA, England, and Japan participating in the Magnetoencephalography International Consortium-AD; 96 participants with MCI and 75 healthy controls (age - 72; 43% female) were included. A data mining process evaluated different analysis alternatives combining preprocessing tasks and machine learning techniques (support vector machines, random forests, bayesian networks, multilayer perceptron, logistic regression, k-nearest neighbor, induction trees). The procedure included: (1) synchronization metrics (synchronization likelihood (SL), phase locking value (PLV), and mutual information (MI)) calculated from delta, theta, alpha, beta and gamma frequency bands, (2) preprocessing tasks (performing epoch outlier detection, the derivation of series of statistical estimators as a result of the time-scale flattering, data standardization, feature subset selection to keep those attributes more informative and less redundant according to the classification task), (3) classification model extraction, and (4) internal evaluation of models, using leave-one-out cross validation method (using accuracy as a validation measure).

Results Data from the learning set (83 MCI, 54 controls) yielded sensitivity to MCI in excess of 0.83 and specificity in excess of 0.80. The critical altered connections were along the anterior-posterior axis. Data from the test set (using the models from the learning set) yielded a positive predictive rate of 1.00 and a negative predictive rate of 0.84; the random forest method was the most robust.

Conclusions FC data from MEG appears to be able to reliably distinguish MCI from normal cognition across multiple independent sites.

Authors/Disclosures
PRESENTER	No disclosure on file
	No disclosure on file
Santosh Kesari, MD, PhD, FAAN (Saint John's Cancer Institute)	Dr. Kesari has received personal compensation in the range of $50,000-$99,999 for serving as a Consultant for xcures. Dr. Kesari has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Duke.
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
	No disclosure on file
Anto Bagic, MD, PhD (UPMC/Univesrity of Pittsburgh)	Dr. Bagic has nothing to disclose.
	No disclosure on file
	No disclosure on file
Eero Pekkonen, MD, PhD (Helsinki University Hospital)	No disclosure on file
	No disclosure on file
Edward Y. Zamrini, MD (Irvine Clinical Research)	Dr. Zamrini has received personal compensation for serving as an employee of George Washington University. Dr. Zamrini has received personal compensation for serving as an employee of Irvine Clinical Research. Dr. Zamrini has received personal compensation in the range of $500-$4,999 for serving as a Consultant for NIH. Dr. Zamrini has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Eli Lilly. The institution of Dr. Zamrini has received research support from NIH. The institution of Dr. Zamrini has received research support from NIH.
	No disclosure on file
	No disclosure on file
James T. Becker, PhD (University of Pittsburgh)	No disclosure on file
	No disclosure on file

��ɫ��

��ɫ��