Fig 1.
The challenge: Motion and eddy-current correction of diffusion MRI data in the elderly brain.
Top row: Non diffusion-weighted MR-image (left), and diffusion-weighted images, encoded using a b-value of 1000 s/mm2 (middle) and 2750 s/mm2 (right) and averaged across multiple directions. Note that the rim of CSF surrounding the anterior part of the brain visualized in the non-diffusion weighted image to the left is absent in the diffusion weighted images. Bottom: Normalized plot of the logarithm of the MR signal as a function of position along the lines indicated in the images at the top. The anterior position where the MR signal turns into background is shifted posteriorly when comparing the zero b-value profile (blue) to the high b-value profile (orange). If the correction employs a non diffusion-encoded reference, the difference in contrast may cause the high b-value data to be erroneously scaled in the antero-posterior direction to fit the low b-value signal outline.
Fig 2.
Flowchart of the motion and eddy-current correction (MEC) procedures.
Two conventional registration procedures (C-MEC) utilizing different registration softwares were explored, and two extrapolation-based correction procedures (EB-MEC) using different extrapolation techniques.
Fig 3.
Comparison of image contrasts.
A: Acquired without diffusion encoding. B: Acquired using a high b-value. C: Extrapolated using the CHARMED model. D: Extrapolated using the proposed method with CSF correction. The two bottom rows show magnifications of anterior segments of the brain for two diffusion encoding directions. The yellow lines show the outline of the brain at a fixed position across all of the images. Note how the anterior rim of CSF is completely attenuated in the high b-value image (compare A and B). Both extrapolation methods yielded images with gross contrast similar to the acquired images. However, the CHARMED model introduced a shift in the outline of the brain for some encoding directions (compare outlines between bottom two rows in column C). For the proposed method, the outline of the brain did not vary substantially with encoding direction (column D).
Fig 4.
Illustration of the registration error between low and high b-value volumes.
The left and middle columns show FA-volumes obtained from a low and a high b-value data set, respectively, processed using the conventional motion and eddy-current correction method (C-MEC). The mismatch between the two data sets becomes apparent when the position of the corpus callosum and surrounding tracts (second, third, and fourth row) are compared to a fixed position indicated by the red line. The third column shows FA projections from high b-value data processed using the CSF-corrected extrapolation-based motion and eddy-current corrections (EB-MEC), where no apparent mismatch between the low and high b-value volumes is visible.
Table 1.
Accuracy of motion and eddy-current correction methods.
The table shows mean (standard deviation) of rotation, translation, scale and skew parameters obtained by registering FA-volumes calculated using high b-values to those calculated using low b-values. Top two sub tables show results from conventional motion and eddy-current correction (C-MEC) using FSL and ElastiX whereas the bottom two subpanels show results from CSF-corrected and CHARMED-based extrapolation-based motion and eddy-current correction (EB-MEC).
Fig 5.
Transform parameters from the motion and eddy-current correction versus volume number.
Data were averaged across all subjects, for the conventional correction (A, red) and the CSF-corrected extrapolation-based procedure (B, blue). Rows and columns show values of the rotation, translation, scaling and skewing parameters around and along the x, y, and z axes, respectively. Deviations from zero can be expected, for example, due to eddy-currents or misregistration. The value of R2, i.e., the amount of variability explained by regressing gradient amplitudes onto the data, is shown in the top right corner of each panel. High values of R2 indicate that eddy-currents explain much of the variation in the transform parameters. Acquisition of high b-values started at volume number 36, which explains the discontinuity of, for example, the z-scaling in the C-MEC data.
Fig 6.
Parameter maps from the model fitting.
Panel A shows the fraction of CSF (fCSF from Eq 7). Panel B shows the mean diffusivity (MD) weighted by w = 1–fCSF. Panel C shows the fractional anisotropy (FA). Panel D and E show the mean kurtosis (MW) weighted by w, estimated from data corrected using the CSF-corrected extrapolation-based, and the conventional method, respectively.
Fig 7.
Histograms of DKI and DTI parameters in white matter.
The lines represent the conventional correction method (red), extrapolation-based method using the CSF-corrected approach for extrapolation (blue), and the extrapolation-based method using CHARMED for extrapolation (green). With the conventional approach, there was a clear negative bias in the mean kurtosis, MW, but also a large fraction of approximately 10% of all voxels where FA was equal to one (spike not shown in histogram).
Fig 8.
Effect of the registration error on tractography based on high b-value data.
Panel A shows the mean diffusivity map of a patient with atrophy. Panels B–D shows a zoomed in section, delineated in panel A. Point clouds in white represent a coronal cross section of the retrosplenial cingulum obtained from tractography using high b-value HARDI data. The tract points are overlaid on top of a colour FA-volume, which was calculated from data acquired with low b-values. When using data corrected with the conventional method (panel B), the point cloud appears in a region approximately two voxels above the expected region (green voxels). Data corrected with the CHARMED-based extrapolation method resulted in a point cloud slightly below the expected region (panel C). For data processed with the CSF-corrected extrapolation-based motion and eddy-current correction the point cloud corresponds well to the anatomical structure (panel D).
Fig 9.
Tractography of the analysed WM bundles.
The three segments of the cingulum, i.e., the parahippocampal, retrosplenial and subgenual segment, are shown in blue, orange and red, respectively. Tracks projecting anteriorly from the thalamus are shown in green.
Fig 10.
Mean kurtosis (MW) in the retrosplenial segment of the cingulum.
The swarm plots show parameters obtained using the conventional and extrapolation-based methods in the healthy controls and PDD patient group. A significant difference was observed for the CSF-corrected extrapolation-based method (p = 0.018), which vanished for the data corrected using the conventional method (p = 0.84). The extrapolation-based method also resulted in higher values of MW.
Table 2.
Effect of motion and eddy-current correction method on group comparison.
Results are presented as mean (standard deviation), and were for the left pair of columns obtained using conventional motion and eddy-current correction (C-MEC) whereas the right pair of columns show values obtained using extrapolation-based motion and eddy-current correction. Significant differences were observed more frequently between the healthy controls and PDD patients when using the extrapolation-based correction. Results were similar but not identical when each hemisphere was compared separately.