Zikuan Chen1, Arvind Caprihan2, Eswar Damaraju2, Srinivas Rachakonda2, Vince Calhoun3. 1. The Mind Research Network and LBERI, Albuquerque, NM 87106, United States. Electronic address: zchen@mrn.org. 2. The Mind Research Network and LBERI, Albuquerque, NM 87106, United States. 3. The Mind Research Network and LBERI, Albuquerque, NM 87106, United States; University of New Mexico, Department of Electrical and Computer Engineering, Albuquerque, NM 87131, United States.
Abstract
BACKGROUND: The output of BOLD fMRI consists of a pair of magnitude and phase components. While the magnitude data has been widely accepted for brain function analysis, we can also make use of the phase data (unwrapped) since this is a good representation of the internal magnetic field. In this work, we discuss the use of fMRI phase data for brain function analysis. NEW METHODS: The fMRI phase data taken from 100 subjects are preprocessed using standard SPM approaches. Group independent component analysis (ICA) is applied to the magnitude and phase data separately. We then compare the spatial patterns for both magnitude and phase data using an empirical spatial smoothing procedure. We also evaluate the magnitude and phase functional network connectivity (FC) matrices. RESULTS: We observed the positive/negative correlation-balanced functional connectivity in phase data, which is distinct from the positive correlation prevalence in magnitude data. The phase FC (pFC) structure is quite different from the magnitude FC (mFC) in functional clusters (on-diagonal blocks or cliques) and inter-cluster couplings (off-diagonal blocks). COMPARISON WITH EXISTING: Methods since both the magnitude and phase data of the fMRI signals are generated from the same magnetic source, either can be useful for brain function analysis from different perspective (per different measurements). Herein, we report on making use of resting-state fMRI phase data for brain functional analysis in comparison with magnitude data. This exploration in phase fMRI may provide a new arena for more comprehensive brain function analysis.
BACKGROUND: The output of BOLD fMRI consists of a pair of magnitude and phase components. While the magnitude data has been widely accepted for brain function analysis, we can also make use of the phase data (unwrapped) since this is a good representation of the internal magnetic field. In this work, we discuss the use of fMRI phase data for brain function analysis. NEW METHODS: The fMRI phase data taken from 100 subjects are preprocessed using standard SPM approaches. Group independent component analysis (ICA) is applied to the magnitude and phase data separately. We then compare the spatial patterns for both magnitude and phase data using an empirical spatial smoothing procedure. We also evaluate the magnitude and phase functional network connectivity (FC) matrices. RESULTS: We observed the positive/negative correlation-balanced functional connectivity in phase data, which is distinct from the positive correlation prevalence in magnitude data. The phase FC (pFC) structure is quite different from the magnitude FC (mFC) in functional clusters (on-diagonal blocks or cliques) and inter-cluster couplings (off-diagonal blocks). COMPARISON WITH EXISTING: Methods since both the magnitude and phase data of the fMRI signals are generated from the same magnetic source, either can be useful for brain function analysis from different perspective (per different measurements). Herein, we report on making use of resting-state fMRI phase data for brain functional analysis in comparison with magnitude data. This exploration in phase fMRI may provide a new arena for more comprehensive brain function analysis.