Meeting Banner
Abstract #3283

Experimental Validation of Dynamic Granger Causality for Inferring Sub-100ms Neuronal Timing Differences from fMRI Without the Confounding Effect of Hemodynamic Variability

Yunzhi Wang1, Santosh Katwal2, 3, Baxter Rogers2, 4, John C . Gore, 4, Gopikrishna Deshpande1, 5

1AU MRI Research Center, Department of Electrical and Computer Engineering, Auburn University, Auburn, AL, United States; 2Vanderbilt University Institute of Imaging Science (VUIIS), Nashville, TN, United States; 3Department of Electrical Engineering and Computer Science, Vanderbilt University, Nashville, TN, United States; 4Radiology and Radiological Sciences, Vanderbilt University, Nashville, TN, United States; 5Department of Psychology, Auburn University, Auburn, AL, United States

Dynamic Granger causality (DGC) was performed to infer sub-100ms neuronal timing differences in BOLD responses from the visual cortex. DGC connectivities were calculated from experimental fMRI data which introduced a known delay between right and left hemified stimuli and a GLM was fit between the experimental paradigm and dynamic Granger causality difference series. The results come out that dGCD significantly covaried with the paradigm for all delays except zero. Our experimental validation of DGC to detect sub-100 ms neuronal timing differences without the confounding effect of hemodynamic variability provide a reliable data-driven method for effective connectivity analysis of fMRI data.

Abstract PDF Presentation Video

Keywords

activated activation activations around aspect attempted auburn black blocks blue bold boxcar brain canonical causality computer confound confounding connectivity consequently consisted constant convolution coronal cortex cortical covariance curves delay delays denoted described details detect detected detecting directly discernible driven dynamic effective electrical engineering even event examine experimental external fissure function functional generally generated goal gore gradient granger greater healthy hemisphere hence human hundreds illustrates increasing indirectly infer inferring initial institute introducing investigation john known larger metric milliseconds mitigate model modulation near needs neuronal notable organizing paradigm please poor possibility precedence press previously problems propose proposed psychology radiological radiology recently refer regular related reliable relies remain resolution respectively response responses rogers sample scanner science sciences selected self series side sided significance significantly simulations slice slices sources start stimulation stimuli stimulus structural studies subject subjects suffer summary table temporal though timing traditional trends trial trials validation variability versus visual whether zero