Meeting Banner
Abstract #2095

An Objective Autoselection of Resting-State Networks Based on Time Course Correlation

Hsin-Long Hsieh1, 2, Pin-Yu Chen3, Jhih-Wei He1, Yao-Chia Shih1, 2, Fu-Shan Jaw1, Wen-Yih Isaac Tseng

1Institute of Biomedical Engineering, National Taiwan University, Taipei, Taiwan, Taiwan; 2Center for Optoelectronic Biomedicine, National Taiwan Univerity College of Medicine , Taipei, Taiwan, Taiwan; 3 Center for Optoelectronic Biomedicine, National Taiwan Univerity College of Medicine, Taipei, Taiwan, Taiwan

Independent component analysis (ICA) has recently been employed in the detection of the resting-state networks (RSNs) which are consistent and highly reproducible across healthy subjects. However, the identification of RSNs often involves visual inspection and/or correlating spatial maps derived from ICA with templates or seed-based results. To avoid bias caused by investigators, we employed a more objective and template-free approach to select and classify components derived from ICA as RSNs based on the component's time course correlation. Our proposed method adds value to the data-driven approach in defining RSNs, and is potentially useful in the connectome research.

Abstract PDF Poster PDF

Keywords

according adds adults applied asked asleep avoid belong bias biomedical biomedicine blue channel classified classify closed clustered coded coefficient coefficients coil college color component components consistent correction correlated correlating correlation course courses datasets decompose defined defining derived descending detection determined discarded distinguished drift driven easily employed engineering eyes fall filtering free frequency front functional green handed head healthy highly identification independent inspection institute intensity involves kernel labels long males maps matrix maximally medicine minute motion national natl negatively network networks noise noisy normalization nothing nuisance objective often optoelectronic pairs partial participants pass physiological potentially power preprocessed preprocessing proposed recently recruited registration relationship removal removing repeated reproducible resolution resting scanner scanning seed select selection series several significantly slice slices smith smoothing sorted spatial spectra steps still strong structures subjective subjects template templates temporal think timing useful visual volumes young zero