Feasibility study to unveil the potential: considerations of constrained spherical deconvolution tractography with unsedated neonatal diffusion brain MRI data

Abstract

Purpose: The study aimed to (1) assess the feasibility constrained spherical deconvolution (CSD) tractography to reconstruct crossing fiber bundles with unsedated neonatal diffusion MRI (dMRI), and (2) demonstrate the impact of spatial and angular resolution and processing settings on tractography and derived quantitative measures.

Methods: For the purpose of this study, the term-equivalent dMRIs (single-shell b800, and b2000, both 5 b0, and 45 gradient directions) of two moderate-late preterm infants (with and without motion artifacts) from a local cohort [Brain Imaging in Moderate-late Preterm infants (BIMP) study; Calgary, Canada] and one infant from the developing human connectome project with high-quality dMRI (using the b2600 shell, comprising 20 b0 and 128 gradient directions, from the multi-shell dataset) were selected. Diffusion tensor imaging (DTI) and CSD tractography were compared on b800 and b2000 dMRI. Varying image resolution modifications, (pre-)processing and tractography settings were tested to assess their impact on tractography. Each experiment involved visualizing local modeling and tractography for the corpus callosum and corticospinal tracts, and assessment of morphological and diffusion measures.

Results: Contrary to DTI, CSD enabled reconstruction of crossing fibers. Tractography was susceptible to image resolution, (pre-) processing and tractography settings. In addition to visual variations, settings were found to affect streamline count, length, and diffusion measures (fractional anisotropy and mean diffusivity). Diffusion measures exhibited variations of up to 23%.

Conclusion: Reconstruction of crossing fiber bundles using CSD tractography with unsedated neonatal dMRI data is feasible. Tractography settings affected streamline reconstruction, warranting careful documentation of methods for reproducibility and comparison of cohorts.

Keywords: constrained spherical deconvolution; diffusion MRI; diffusion tensor imaging; neonatal; tractography.

Figure 1

Schematic overview of experiment settings. (A) Shows the signal intensity relative to the applied b-values (with screenshots of the non-motion BIMP scan) and a local modeling reconstruction using diffusion tensor imaging (DTI) and constrained spherical deconvolution (CSD). (B) Displays the spatial and angular resolution and processing settings. Top left panel: The voxel size section shows a coronal snapshot of streamline reconstructions and the corresponding local modeling reconstructions for different voxel sizes with a grid overlay. Top right panel: Gradient directions were plotted on a sphere for BIMP-data (45 and 32 gradient directions) and dHCP-data (128 gradient directions). Bottom left panel: Head motion can result in signal dropout. As shown in this figure, tilting the head forward mid scanning (scanning caudal to cranial; red lines) results in double scanning of the posterior area (overlapping lines) and missing data of the frontal area (gap). Bottom right panel: Response function estimation with three algorithms results in minor differences, mainly in the amplitude of the response function. (C) Displays the tractography settings. The impact of fODF threshold 0 or 0.2 for displayed (2-dimensional) fODF, angle threshold for angle 30 of 60 and step size are show. A step size larger than the voxel size (blue dots) may result in streamlines being missed (black dashed line) by the region of interest (green area).

Figure 2

ROI placement for fiber bundle tractography, (A) corpus callosum, (B) left corticospinal tract, (C) single voxel tractography from a random ROI in the left posterior limb of internal capsule (white circle) and (D) left association fibers. Green: inclusion ROIs; red: exclusion ROIs.

Figure 3

DTI vs. CSD tractography using single-shell b800 and b2000 neonatal dMRI data with low angular resolution (45 directions). DTI resulted in clean and separated bundle reconstructions, while CSD resulted in more branched and intersected bundle reconstructions. Reconstructions were acquired from the BIMP scan without motion.

Figure 4

Spatial and angular resolution of unsedated neonatal diffusion MRI, including voxel (an)isotropy, voxel size and number of gradient directions affected streamline reconstructions. Crossing of corpus callosum streamlines with corticospinal tract streamlines were less apparent in the 2 mm isotropic scan (yellow circle) and different image resolution affected the reconstruction of a branch of the corpus callosum (white arrows). Tractography settings, such as step size (set to 0.5 mm), were not adapted to resolution. Reconstructions were acquired from the BIMP scan without motion.

Figure 5

High-quality diffusion MRI data from the developing human connectome project was down sampled from 128 (single-shell; upper row) to 45 (lower row) gradient directions. Branching/fanning of the corpus callosum was visibly affected by down sampling. Reconstructions were acquired from the dHCP scan.

Figure 6

Motion artifacts (reflected in signal dropout) affected tractography. The single voxel posterior limb of internal capsule reconstruction was severely affected by axial slice dropout, resulting in less and shorter tracts. However, these findings were not reflected in the CST reconstructions. Reconstructions were acquired from the BIMP scan with motion. The lower row visualizes differences between repol and no repol with overlapping streamlines visualized in white.

Figure 7

Response function method affected local modeling and tractography reconstructions. A higher fODF amplitude was found for FA, compared to Tournier, and for Tax, compared to FA. Reconstructions were acquired from the BIMP scan without motion.

Figure 8

fODF thresholding (indicated on the left) mostly affected more peripheral sections of fiber bundles. Reconstructions were acquired from the BIMP scan without motion.

Figure 9

An increasing angle threshold (indicated on the left) resulted in more branching of streamlines and a visually increased bundle diameter. Reconstructions were acquired from the BIMP scan without motion.

Figure 10

Step size (listed on left) mostly affected tract reconstruction with a step size larger than the voxel size (with increased region of interest thickness). Differences were highlighted with yellow circles. Reconstructions were acquired from the BIMP scan without motion.