Relative Brain Signature (RBS): A population-based feature extraction technique to identify functional biomarkers in the brain of alcoholic subjects

Monday, September 22, 2014 — Poster Session I
12:00 p.m. – 2:00 p.m.	FAES Academic Center	NICHD	BIOENG-6

* FARE Award Winner

Authors

• N Karamzadeh
• M Kellman
• A Anderson
• F Chowdhry
• Y Ardeshirpour
• D Chorlian
• E Wegman
• A Gandjbakhche

Abstract

Constructing a brain signature that uniquely characterizes subjects’ relationship to populations with distinctive brain functionality is an unexplored and challenging problem. We present a novel feature extraction technique, Relative-Brain-Signature (RBS), which signifies subjects’ similarity to populations with distinctive neuronal activity. We demonstrate that shared features of the signature vectors among the subjects from a similar population introduce a set of biomarkers for the population. In this study, we have applied our technique on an EEG data of 77 abstinent alcoholics and 43 control subjects. For every subject, one RBS vector with respect to alcoholic and one for control population were computed which describe the relationship of the subject to a population. The extracted RBS vectors were employed to identify functional biomarkers over the cortical area of the alcoholics that are functionally distinctive from controls. The corresponding functionality of the identified regions in the abstinent alcoholics’ brain points to the areas that are known to be affected in the brain of social drinkers shortly after alcohol consumption. A machine-learning algorithm was also employed to evaluate the efficacy of the RBS vectors to be used as feature vectors to classify subjects into alcoholics and controls which resulted in 87% of accuracy.

back to top