DETECTION OF BOLD FMRI SIGNALS USING COMPLEX DATA: FUNCTIONAL BRAIN MAPPING

Stanford University, Stanford CA

Investigators

Linked publications & trials

Abstract

Introduction: Most fMRI studies use only magnitude image. Although magnitude detection is immune to incidental phase variations, important functional information is often encoded into the NMR phase images, and is not detectable with magnitude image. Here we propose the use of complex data to create functional activation maps. Methods and Materials: Algorithm A complete MR image is essentially a complex one, i.e., I = x + i*y, where x = Re(I) and y = Im(I). With a typical activation time series of N frames, a complex activation map is obtained by calculating cross-correlation coefficient, cc, on a pixel-by-pixel basis as: , , where r is reference signal that has the same period as the activation task on/off cycle, s's are the standard deviations over the time series. We used sinusoidal functions as reference functions. MR data acquisition Motor cortex activation time series images were obtained by running a 3D spiral sequence (8 slices, TR/TE/FA/FOV/TH = 80ms/40ms/25o/240mm/4mm, 4 interleaves, 100 time frames in 256s with the resting and rght hand finger tapping being alternated every 20s.) Results and Discussion: Activation maps using complex data contained extra activation signals that were not seen in the activation maps using magnitude images only. These extra signals were also seen in the activation maps using the NMR phase time series. These pixels were seen as dark spots in the magnitude images, suggesting that they contain large venous blood volume fraction. Therefore, the above method represents an efficient approach to localize large veins in BOLD fMRI studies without worry about possible inter-scan patient movement. Conclusion: The activation maps from complex data essentially integrate functional information from both magnitude and NMR phase detection. This suggests that detection of BOLD fMRI signals using complex data is superior to the conventional magnitude detection.