doi: 10.1038/s41598-020-61326-2.
Parcellation of the neonatal cortex using Surface-based Melbourne Children's Regional Infant Brain atlases (M-CRIB-S)
Affiliations
- PMID: 32152381
- PMCID: PMC7062836
- DOI: 10.1038/s41598-020-61326-2
Item in Clipboard
Parcellation of the neonatal cortex using Surface-based Melbourne Children's Regional Infant Brain atlases (M-CRIB-S)
Sci Rep.
.
Abstract
Longitudinal studies measuring changes in cortical morphology over time are best facilitated by parcellation schemes compatible across all life stages. The Melbourne Children's Regional Infant Brain (M-CRIB) and M-CRIB 2.0 atlases provide voxel-based parcellations of the cerebral cortex compatible with the Desikan-Killiany (DK) and the Desikan-Killiany-Tourville (DKT) cortical labelling schemes. This study introduces surface-based versions of the M-CRIB and M-CRIB 2.0 atlases, termed M-CRIB-S(DK) and M-CRIB-S(DKT), with a pipeline for automated parcellation utilizing FreeSurfer and developing Human Connectome Project (dHCP) tools. Using T2-weighted magnetic resonance images of healthy neonates (n = 58), we created average spherical templates of cortical curvature and sulcal depth. Manually labelled regions in a subset (n = 10) were encoded into the spherical template space to construct M-CRIB-S(DK) and M-CRIB-S(DKT) atlases. Labelling accuracy was assessed using Dice overlap and boundary discrepancy measures with leave-one-out cross-validation. Cross-validated labelling accuracy was high for both atlases (average regional Dice = 0.79-0.83). Worst-case boundary discrepancy instances ranged from 9.96-10.22 mm, which appeared to be driven by variability in anatomy for some cases. The M-CRIB-S atlas data and automatic pipeline allow extraction of neonatal cortical surfaces labelled according to the DK or DKT parcellation schemes.
Conflict of interest statement
The authors declare no competing interests.
Figures
Vertex-wise parcellation mismatch rates for the M-CRIB-S(DKT) and M-CRIB-S(DK) atlases using the Leave-One-Out cross-validation method shown on the template inflated surfaces. Aspects shown are as follows: midline (i), lateral (ii), superior (iii), inferior (iv), frontal (v), and occipital (vi) for the left (LH) and right hemispheres (RH). Warmer colours indicate greater vertex-wise mismatch between automatic and manual labels.
Per-region Dice coefficients for the Leave-One-Out cross-validation test for labelled datasets comparing (i) automated with manual M-CRIB-S(DKT) parcellations, and (ii) automated with manual M-CRIB-S(DK) parcellations. The banks of the superior temporal sulcus (BSTS), frontal pole (FP), and temporal pole (TP) regions are greyed out in (i) because they are not present in the DKT parcellation scheme. CAC: Caudal anterior cingulate, CMF: Caudal middle frontal, CUN: Cuneus, ENT: Entorhinal, FUS: Fusiform, INFP: Inferior parietal, INS: Insula, ISTC: Isthmus cingulate, IT: Inferior temporal, LIN: Lingual, LOCC: Lateral occipital, LORB: Lateral orbitofrontal, MORB: Medial orbitofrontal, MT: Middle temporal gyrus, PARH: Parahippocampal, PARC: Paracentral lobule, POPE: Pars opercularis, PORB: Pars orbitalis, PCING: Posterior cingulate, PCAL: Pericalcarine, POSTC: Posterior cingulate, PCUN: Precuneus, PREC: Precentral, PTRI: Pars triangularis, RAC: Rostral anterior cingulate, RMF: Rostral middle frontal, SF: Superior frontal, SMAR: Supramarginal gyrus, SP: Superior parietal, ST: Superior temporal gyrus, TT: Transverse temporal gyrus.
Per-region Hausdorff distances, in mm units, for the Leave-One-Out cross-validation test for labelled datasets comparing (i) automated with manual M-CRIB-S(DKT) parcellations and (ii) automated with manual M-CRIB-S(DK) parcellations. The banks of the superior temporal sulcus (BSTS), frontal pole (FP), and temporal pole (TP) regions are greyed out in (i) because they are not present in the DKT parcellation scheme. CAC: Caudal anterior cingulate, CMF: Caudal middle frontal, CUN: Cuneus, ENT: Entorhinal, FUS: Fusiform, INFP: Inferior parietal, INS: Insula, ISTC: Isthmus cingulate, IT: Inferior temporal, LIN: Lingual, LOCC: Lateral occipital, LORB: Lateral orbitofrontal, MORB: Medial orbitofrontal, MT: Middle temporal gyrus, PARH: Parahippocampal, PARC: Paracentral lobule, POPE: Pars opercularis, PORB: Pars orbitalis, PCING: Posterior cingulate, PCAL: Pericalcarine, POSTC: Posterior cingulate, PCUN: Precuneus, PREC: Precentral, PTRI: Pars triangularis, RAC: Rostral anterior cingulate, RMF: Rostral middle frontal, SF: Superior frontal, SMAR: Supramarginal gyrus, SP: Superior parietal, ST: Superior temporal gyrus, TT: Transverse temporal gyrus.
Worst-case (i) and best-case (ii) instances of boundary discrepancies, measured by Hausdorff distances, between estimated (green) and manual (red) label boundaries. The star markers and white paths depict the traversal between nearest neighbours. Other surface vertices are shaded according to curvature, with light and dark grey denoting gyri and sulci, respectively.
(i) Indicative parcellations, (ii) curvature, and (iii) sulcal depth maps for the M-CRIB-S, University of North Carolina (UNC), and developing Human Connectome Project (dHCP) atlases displayed on right hemisphere smoothed white matter surfaces provided for each respective atlas. The M-CRIB-S and UNC parcellations in (i) are based on the DK parcellation scheme. See Bozek et al. for a description of the dHCP parcellation scheme. In (ii) and (iii) vertices are coloured according to whether they reside in a gyrus (green) or a sulcus (red).
Exemplary surface extraction pipeline output for one labelled subject. Panels show: (i) the original T2-weighted image; (ii) segmentations according to the DrawEM techniques; (iii) Deformable-extracted cortical surfaces, where the top panel shows inner (yellow) and outer (red) cortical surfaces overlaid onto the original image, and the middle and bottom panels show lateral aspects of the left hemisphere inner and outer surfaces in 3D; (iv) “inflated” and spherical versions of the white surface; (v) spherical surface registered to the template surface (a); (vi) automatic parcellation using the M-CRIB-S(DKT) scheme shown on the subject inflated surface for lateral (left) and medial (right) aspects. The label training data are depicted in volume format (b), and in 3D on the average inflated surface for lateral (c), and medial (d) aspects. Surface vertices in (iii), (iv) and (v) are coloured according to local mean curvature.
Illustrative surface projections of the manual parcellation for one subject from the labelled set using the M-CRIB-S(DKT) and M-CRIB-S(DK) labels for left (LH) and right (RH) hemispheres. The ellipses highlight some differences between M-CRIB-S(DKT) and M-CRIB-S(DK). The white ellipses highlight location disagreements of the lateral orbitofrontal (LORB) and pars orbitalis (PORB) regions between atlases. The black ellipses encompass the banks of the superior temporal sulcus region, which is not present in the DKT. BSTS: Banks of the superior temporal sulcus, CAC: Caudal anterior cingulate, CMF: Caudal middle frontal, CUN: Cuneus, ENT: Entorhinal, FP: Frontal pole, FUS: Fusiform, INFP: Inferior parietal, INS: Insula, ISTC: Isthmus cingulate, IT: Inferior temporal, LIN: Lingual, LOCC: Lateral occipital, LORB: Lateral orbitofrontal, MORB: Medial orbitofrontal, MT: Middle temporal gyrus, PARH: Parahippocampal, PARC: Paracentral lobule, POPE: Pars opercularis, PORB: Pars orbitalis, PCING: Posterior cingulate, PCAL: Pericalcarine, POSTC: Posterior cingulate, PCUN: Precuneus, PREC: Precentral, PTRI: Pars triangularis, RAC: Rostral anterior cingulate, RMF: Rostral middle frontal, SF: Superior frontal, SMAR: Supramarginal gyrus, SP: Superior parietal, ST: Superior temporal gyrus, TP: Temporal pole, TT: Transverse temporal gyrus.
Average white (i), pial (ii), and inflated (iii) left (LH) and right hemisphere (RH) surfaces for all subjects with the vertices coloured according to M-CRIB-S(DKT) labels. The average white matter curvature map is shown on the inflated surfaces in (iv). The lateral view of the M-CRIB-S(DK) atlas is shown in (v). The annotations in panel (iii) and (v) highlight selected differences between the atlases. The white ellipses focus on the lateral orbitofrontal and pars orbitalis regions. The black ellipses centre on the bank of the superior temporal sulcus (BSTS), which is absent in the M-CRIB-S(DKT) atlas. BSTS: Banks of the superior temporal sulcus, CAC: Caudal anterior cingulate, CMF: Caudal middle frontal, CUN: Cuneus, ENT: Entorhinal, FP: Frontal pole, FUS: Fusiform, INFP: Inferior parietal, INS: Insula, ISTC: Isthmus cingulate, IT: Inferior temporal, LIN: Lingual, LOCC: Lateral occipital, LORB: Lateral orbitofrontal, MORB: Medial orbitofrontal, MT: Middle temporal gyrus, PARH: Parahippocampal, PARC: Paracentral lobule, POPE: Pars opercularis, PORB: Pars orbitalis, PCING: Posterior cingulate, PCAL: Pericalcarine, POSTC: Posterior cingulate, PCUN: Precuneus, PREC: Precentral, PTRI: Pars triangularis, RAC: Rostral anterior cingulate, RMF: Rostral middle frontal, SF: Superior frontal, SMAR: Supramarginal gyrus, SP: Superior parietal, ST: Superior temporal gyrus, TP: Temporal pole, TT: Transverse temporal gyrus.
