Investigators interested in brain functional connectivity.Functional magnetic resonance imaging (fMRI) has been widely recognized as being effective in detecting neural activity in brain cortex¹. Detection of neural activity in white matter, however, has been rarely reported in the fMRI literature to date². This is presumably attributable to much lower vascular density in white matter than in gray matter, and therefore much less blood oxygenation level dependent (BOLD) contrast that is necessary for fMRI. Recently we have observed correlational anisotropy in white matter BOLD signals, and proposed a concept of “functional tensors” for its full descriptions³. We found that directional preferences of functional tensors along many white matter tracts are grossly consistent with those revealed by diffusion tensors, and that evoked functions selectively enhance visualization of relevant fiber pathways. These tend to suggest that BOLD signals in white matter may encode neural activity as well, and may be detectable provided the availability of sensitive imaging and analysis techniques. This is in fact partly supported by the observation that MRI signals from T₂*-sensitive acquisitions in a resting state exhibit structure-specific temporal correlations in white matter^{4, 5}. In this work, we further analyze temporal correlations of T₂*-weighted signals along fiber pathways in task conditions, to examine whether there exist temporal correlations along white matter pathways that are relevant to evoked functions.

Full brain MRI data were acquired from 12 healthy, gender-matched, and left-handed adult volunteers (mean age = 28.7 yrs, stdev = 4.8 yrs). Prior to imaging, informed consent was given by each subject according to protocols approved by the Vanderbilt University Institutional Review Board. All human imaging was performed on a 3T Philips Achieva scanner (Philips Healthcare, Inc., Best, Netherlands) using a 32-channel head coil. For each subject, the following datasets were obtained: a) High resolution anatomical images using T₁-weighted multi-shot gradient echo (GE) sequence with TR/TE=8.9/4.6 ms, matrix size= 256x256x150, and voxel size=1x1x1 mm³. b) Resting state T₂*-sensitive images using a multi-echo GE, echo planar imaging sequence with TR/TE=3000/45 ms, matrix size= 80x80, FOV=240x240 mm², 43 axial slices of 3 mm thick with zero gap, and 145 volumes. c) T₂*-sensitive images using the same parameters as in b) with subjects performing right-hand finger tapping. d) T₂*-sensitive images using the same parameters as in c) with subjects performing left-hand finger tapping. The finger tapping task started with 5 volumes of resting state, followed by 10 volumes of index finger tapping and then 10 volumes of resting state and so on.

Once acquired, the datasets from each subject underwent the following procedures: a) all fMRI time series were corrected for slice timing and head motion and smoothed with FWHM=4mm using SPM12. b) The smoothed fMRI time series in task conditions were used to identify the left and right primary motor cortex (PMC) respectively. c) All smoothed fMRI time series were detrended and then band-pass filtered to retain frequencies only of 0.01-0.08 Hz (which contained the peak frequency of the finger tapping task of 0.017Hz). d) T₁-weighted images were segmented into gray matter, white matter and cerebrospinal fluid, which were coregistered to the mean fMRI volume using SPM12.

Finally, three seed voxels in the PMC of each hemisphere identified above were manually chosen near the center location, and the coefficient of correlation of all white matter voxels to these seed voxels was computed by linear regression using the six seed time series from both hemispheres as the regressors and the time series in resting state, left-hand and right-hand finger tapping as the regressand respectively.

For all the subjects, we have found high correlations in many regions of projection pathways, and clear differences among left- and right-hand finger tapping and resting state. Figure 1 shows typical correlation maps from one subject. It can be seen that with right-hand finger tapping (left column), the left projection pathways (Path1) have more significant correlations than the right counterparts and the right thalamus-operculum pathways (Path4) also have significant correlations; with left-hand finger tapping (middle column), the right projection pathways (Path2) have more significant correlations than the left counterparts and the left thalamus-operculum pathways (Path5) also have significant correlations; in the resting state (right column), correlations in the projection pathways are generally weaker than under task conditions, but there are still significant correlations in the right thalamus-operculum pathways (Path3) and the left thalamus-PMC pathways (Path6).This study demonstrates that white matter may contain physiologically meaningful BOLD signals that may be detectable using sensitive methods. Efforts are being made to construct connectivity atlases for the projection pathways.