Fig 1.
Functional MRI activation for a hand tapping task (top), and corresponding T1 images (bottom) for two children with unilateral cerebral palsy.
The top-left of each image represents the anterior-left of each brain. Left Column: a higher-functioning child tapping their less-impaired (right) hand. Robust unilateral activation can be seen in the 'typical' hand knob location of the sensorimotor cortex, as may be expected from a typically developing child. Right Column: a child with higher degrees of brain damage tapping their more-impaired (left) hand. Activation appears more medially, bilaterally, and anteriorly; predominantly in the assumed supplementary motor areas rather than in the typical hand-knob locus. These differences may illustrate potentially abnormal brain organisation occuring in response to the early-life brain injury sustained. Images are not to scale.
Fig 2.
Examples of participants with unilateral cerebral palsy, for whom neurological abnormalities include distortion of the midbrain, and/or loss of contrast between the thalamus and posterior limb of the internal capsule.
Subtler losses in contrast are indicated by arrows. Such abnormalities can make reliable delineation of these structures difficult to perform. The top row displays two slices from a single participant. The remaining images are from six different participants. All displayed participants contributed data that was utilized for one or more analyses.
Fig 3.
Examples of pathology seen in the enrolled cohort.
The top row displays one coronal and one axial image from two participants with severe brain pathology. The remaining two rows display images of eight unique participants. No participants are duplicated from Fig 2. All displayed participants contributed data that was utilized for one or more analyses. These images, combined with those in Fig 2, are representative of the range of pathology severity seen across enrolled participants.
Table 1.
Participant details for each stage of analysis.
Fig 4.
Summary of the pipeline for the Surface-based fMRI-guided dMRI tractography.
Images and meshes associated with each major step depict real results from a single participant enrolled in this study. Major steps of the pipeline include tissue segmentation, mesh generation, projection of fMRI echo-planar image values onto the mesh, statistical fMRI calculation on the mesh, fMRI-seeded tractography, and tract clustering. See text for additional details.
Fig 5.
Example tractography from the surface-based (left column) and voxel-based (right column) methods, in a single participant with left-hemispheric pathology.
The top row shows all tracks from all slices, with thalamocortical tracts in green or blue and corticomotor tracts in red or yellow. In this participant, the superior sections of the tracts were considerably more coherent when delineated with the surface-based method. The bottom row shows an axial slice at the level of the dotted white line. For this participant, at this level, track clustering (left) provided visually similar results to the region-of-interest based approach to track classification. Arrows indicate the approximate location of the midbrain regions of interest that were used to filter tracks.
Fig 6.
Examples of pathology and impacts on analyses.
Row A: The child for whom a surface could not be generated. Tissue segmentation failed as the allowed magnitude of warping was insufficient to match any atlas, leading to cerebrospinal fluid being classed as grey and white matter. Voxelwise analyses were successful. Rows B and C: Two participants for whom surface analyses were successful, but voxelwise analyses failed to detect significant fMRI activation. Row D: A child with severe pathology for whom both voxelwise and surface analyses detected fMRI activation. Both methods found no corticomotor or thalamocortical tracks in the hemisphere with pathology; no genuine connections of this type were probably present.
Fig 7.
Typical significant fMRI activation detected through surface (top row) and voxelwise (middle and bottom rows) methods for tapping of the ‘impaired’ hand in a single participant.
The middle and bottom rows show coronal, sagittal, and right-facing axial sections in the left, middle, and right columns respectively. Both methods show activation (red) in the approximate pre- and post-central gyri of the left hemisphere, and the right anterior lobe of the cerebellum. The voxelwise analysis resulted in approximately oval shaped activations that include grey-matter, white-matter, and cerebrospinal fluid. The surface-based method resulted in less-uniformly shaped activation patterns and two activation sites on the cerebellum.
Table 2.
Counts of Participants Displaying Significant Activation.
Table 3.
Correlations (R2) between clinical scores and age-adjusted diffusion metrics recorded via tractography for two tracts superior to the midbrain ROIs.
Table 4.
Correlations (R2) between clinical scores and age-adjusted diffusion metrics recorded via tractography for two tracts superior to the midbrain ROIs.