Room: AAPM ePoster Library
Purpose: Weisskoff analysis is a common quality assurance method for measuring functional MRI instability. By plotting temporal coefficient of variation over a range of regions of interest (ROI) sizes from dynamic echo-planar imaging (EPI) data of a uniform phantom, signal instability can be detected based on increased intervoxel correlations. Resting-state (rs-) fMRI measures functional connectivity by the correlations of the blood oxygenation level-dependent signals between brain regions. Therefore we hypothesis that the signal instability characterized by the Weisskoff analysis will be associated with false correlations in the functional connectivity analysis.
Methods: We scanned a uniform spherical phantom on two clinical 3T scanners (A & B). Two different multi-channel receive coils (1 & 2) were used on Scanner B. Three tests of single-shot gradient-echo EPI rs-fMRI scan were performed for each Scanner/Coil combination. Weisskoff analysis was performed using square ROI with width from 1 to 23 voxels and the radius of decorrelation (RDC) determined. For the functional connectivity analysis, the EPI data was preprocessed with slice timing correction and a bandpass filter. The correlation maps were calculated with a 6-mm diameter spherical seed located at the center of the imaging volume and converted to z-value maps. Mean absolute z-value was then calculated from a 30 mm diameter spherical ROI to indicate the amount of false correlations.
Results: In general, Scanner B performed better than Scanner A. Scanner B with Coil 1 performed most consistently between the three repeated tests. Significant correlation (p = 0.01) was found between the mean z-value and the RDC.
Conclusion: This phantom experiment demonstrates that fMRI instability varies between scanners, radiofrequency coils used and repeat measurements. The system instability can lead to false correlations in the rs-fMRI analysis, and the RDC obtained from the Weisskoff analysis is an effective indicator of the false correlations.