Parcellations and hemispheric asymmetries of human cerebral cortex analyzed on surface-based atlases - PubMed (original) (raw)
Parcellations and hemispheric asymmetries of human cerebral cortex analyzed on surface-based atlases
David C Van Essen et al. Cereb Cortex. 2012 Oct.
Abstract
We report on surface-based analyses that enhance our understanding of human cortical organization, including its convolutions and its parcellation into many distinct areas. The surface area of human neocortex averages 973 cm(2) per hemisphere, based on cortical midthickness surfaces of 2 cohorts of subjects. We implemented a method to register individual subjects to a hybrid version of the FreeSurfer "fsaverage" atlas whose left and right hemispheres are in precise geographic correspondence. Cortical folding patterns in the resultant population-average "fs_LR" midthickness surfaces are remarkably similar in the left and right hemispheres, even in regions showing significant asymmetry in 3D position. Both hemispheres are equal in average surface area, but hotspots of surface area asymmetry are present in the Sylvian Fissure and elsewhere, together with a broad pattern of asymmetries that are significant though small in magnitude. Multiple cortical parcellation schemes registered to the human atlas provide valuable reference data sets for comparisons with other studies. Identified cortical areas vary in size by more than 2 orders of magnitude. The total number of human neocortical areas is estimated to be ∼150 to 200 areas per hemisphere, which is modestly larger than a recent estimate for the macaque.
Figures
Figure 1.
Analysis of distortions arising during registration of an individual human hemisphere to the fsaverage atlas. (A) Individual pial, white, and midthickness surface (native mesh) created by averaging the rh.white and rh.pial surfaces (FreeSurfer Case RI from Tsao et al. 2008). (B) A map of average convexity (rh.sulc) on the spherical surface (rh.sphere). (C) Areal distortions (rh.sphere vs. rh.midthickness) are generally less than 2-fold. FreeSurfer's method reduces distortions on the sphere relative to the white matter surface (Fischl et al. 1999a); it achieves lower distortions than does Caret's multiresolution morphing in regions such as the occipital and frontal poles (cf. Van Essen et al. 2011; Fig. 4 in Van Essen 2005). (D) Population-average average convexity (rh.avg_sulc). (E) The average convexity map displayed on the individual sphere registered (deformed) to the fsaverage sphere (rh.sphere.avg). Arrows in B and C point to locations that have undergone highly nonlinear local deformations. (F) Areal distortions (rh.sphere.reg vs. rh.sphere). (G) Individual midthickness surface on the 164k_fs_R mesh, generated with a deformation map created using rh.sphere.avg as the source sphere and rh.sphere from the fsaverage midthickness surface atlas directory as the target sphere shaded by the map of folding (rh.curv). (H) The fsaverage midthickness surface is based on averaging rh.white_avg and rh.pial_avg, each representing a population average from 40 subjects. Highlighted nodes in panels G and H represent geographic correspondences specified by the registration process. (I) A distortion map relating surface area on the individual-subject midthickness surface to the fsaverage left hemisphere population.
Figure 2.
Interhemispheric registration of fsaverage surfaces to a left–right hybrid target using Landmark-SBR. (A) Fsaverage midthickness surfaces can be jointly visualized in Caret to demonstrate the lack of geographic correspondence. The left and right hemisphere meshes are specified as 164k_fs_L and 164k_fs_R, respectively. (B) Fifty-five landmark contours identified along corresponding geographic locations in the left and right hemisphere fsaverage surfaces. (C) Landmarks projected to the left and right spherical standard surfaces. (D) Hybrid landmarks generated by averaging the right and mirror-flipped left spherical landmarks. (E) Geographic correspondences of the resampled fsaverage midthickness and inflated surfaces on the hybrid 164k_fs_LR mesh.
Figure 3.
Interatlas registration between fsaverage and PALS-B12 surfaces using Landmark-SBR. (A) Most nodes are in very similar locations relative to local features of PALS-B12 left and right average midthickness surfaces. However, nodes in lateral temporal cortex (1 and 5), a region of known asymmetry (cf. Figure 4), are in discernibly different locations relative to local features. The imperfect geographic correspondence in this region arises because the dorsal superior temporal gyrus is difficult to delineate reliably in individual subjects, owing to its inconsistent trajectory, and was not one of the “Core 6” landmarks used for the PALS-B12 registration process (Van Essen 2005). (B) Landmarks on the PALS-B12 surfaces used for registration to fs_LR were the same 55 as in Figure 2. (C) Target landmarks on the fs_LR atlas sphere. (D) Node correspondences between PALS-B12 and fs_LR after interatlas registration.
Figure 4.
Hemispheric symmetries and asymmetries in population-average midthickness surfaces. (A) Lateral views of the left and right fsaverage midthickness surfaces. (B) Corresponding view of the Conte-69 average midthickness surfaces. Axes (1 cm grid) in A and B are centered on the origin (anterior commissure). White arrows in A and B point to asymmetries near the ventral tip of the postcentral sulcus that are further analyzed in Figure 3. (C,D) Coordinate difference maps for the fsaverage (C) and Conte-69 (D) average midthickness surfaces, based on the distance from each right hemisphere node to the corresponding node in the mirror-flipped left hemisphere surface.
Figure 5.
(A) Surface asymmetry map based on the ratio of surface area associated with corresponding left and right hemisphere nodes (average area of all tiles containing the node) in each individual, averaged across all Conte-69 subjects and smoothed (70 iterations on the surface) to reduce noise. (B) Statistically significant asymmetries based on the TFCE method (see Materials and Methods). Hemispheric asymmetries in the Sylvian Fissure. (C) Average distortion map (ratio of individual midthickness to Conte-69 average midthickness surface area, averaged across all 69 individuals separately for the left and right hemispheres and used to compensate for surface area estimates). (D) Hemispheric areal asymmetry map displayed on an expanded view of the Conte-69 left very inflated surface along with landmarks used for registration of the left and right hemispheres to the fs_LR atlas (full hemisphere on bottom, expanded on top) To estimate the average areal asymmetry of the PT, its extent was delineated by landmark contour 1 (light blue), the posterior HG landmark (dark blue), and the superior temporal and supramarginal gyri (pink contour). (E) Landmark contours overlaid on mean curvature maps—computed on average midthickness but displayed on very inflated for the left (top) and right (bottom) perisylvian surfaces. (F) Same as E but shown on midthickness surfaces.
Figure 6.
VBR versus SBR of area 2. (A) Parasagittal slice through the postcentral sulcus (PoCeS) and neighboring regions of the Colin single-subject atlas volume. Arrows indicate the expected location of area 2 on the anterior bank of the PoCeS. (B) Probabilistic area 2 (VBR-based) overlaid on the same volume slice. (C) Probabilistic area 2 registered by additional VBR steps to the fsaverage volume, with the fsaverage surface contour overlaid. (D) VBR-based area 2 mapped to the fsaverage midthickness surface by volume-to-surface mapping. (E) The same map of VBR-based area 2 displayed on the inflated atlas surface. (F) Area 2 mapped to the fsaverage surface by Energy-SBR (Fischl et al. 2008) and to the fs_LR surface (see Materials and Methods). In panels E and F, white highlighted nodes were used for determining geodesic distances between the limits of area 2 mapped by VBR versus SBR.
Figure 7.
VBR versus SBR of hOc5. White highlighted node is the same in all panels. (A) Probabilistic area hOc5 mapped by nonlinear VBR to the left (panel A) and right (panel B) fsaverage midthickness surface (left) and inflated surface (right). (C,D) Energy-SBR maps of hOc5 for the left and right hemispheres. In all panels, contour outlines show the composite border for hOc5 based on the energy-SBR probabilistic map for each hemisphere (panels C and D).
Figure 8.
Probabilistic architectonic maps from Fischl et al. (2008) registered to FreeSurfer's fsaverage right hemisphere surface, lateral (top), inflated lateral (middle), and medial (bottom) surfaces. In a few instances, the areal map for an individual subject has little or no overlap with the main population distribution (green arrows for left and right hOc5 and right area 45).
Figure 9.
Differences in registration for Energy-SBR (via fs_R) versus Landmark-SBR (via PALS-B12) applied to the same hemisphere. (A) Case NJ right hemisphere (Tsao et al. 2008) registered to the fs_LR mesh via PALS-B12, using the Landmark Pin and Relax algorithm for consistency with other data sets (followed by pals-to-fs_LR transformation), with 2 nodes (1 and 2) highlighted (blue). (B) Same hemisphere registered to the fs_LR mesh via fs_R (followed by fs_L-to-fs_LR transformation), with the “same” nodes highlighted based on node number. (C) The fsaverage midthickness surface on the fs_LR mesh, with nodes 1 and 2 highlighted. Node 1 is at a similar geographic location in all 3 surfaces. Node 2 represent locations whose 3D coordinates differ by 24 mm in the PALS-B12 versus fs_R versions of the individual midthickness surface (A,B). (D) Coordinate difference maps (absolute value if separation between “corresponding” nodes in the PALS-B12_registered versus fs_registered midthickness surfaces) for the left and right hemispheres of case NJ. (E) Coordinate difference maps for the left and right hemispheres of case RI. (F) A map of average difference in _x_-axis coordinate values for surfaces registered via PALS-B12 versus fsaverage for all 4 hemispheres of cases NJ and RI (after inverting the sign of left hemisphere differences before averaging with the right hemisphere). (G,H) Maps of average differences in _y_-axis values (G) and _z_-axis values. In panels F_–_H, yellow/orange/red regions have coordinates that are on average more positive in PALS-registered hemispheres (more lateral for the _x_-coordinate) than for the corresponding vertices in the fs_L/fs_R hemispheres.
Figure 10.
Retinotopic areas V1 and V2 relative to probabilistic areas 17 and 18. (A) Map of retinotopic V1 and V2 from Case 1 of KHP10 (Kolster et al. 2010), displayed on the posterior (top) and medial (bottom) views of the right hemisphere midthickness surface, with architectonic area 17 and 18 boundary contours overlaid. (B) The same KHP10 case 1 retinotopic areas displayed on the fsaverage midthickness surface. Areal boundaries are in similar locations relative to the occipital pole and other geographic landmarks in the vicinity. (C,D) Retinotopically defined area V1 and V2 boundaries from case 1 of SHM07 (Swisher et al. 2007) displayed on the individual midthickness surface (C) and on the fsaverage midthickness surface (D). Areal boundaries are close to the occipital pole on both surfaces. (E,F). Probabilistic map of architectonic area 18 from Fischl et al. (2008) on posterior (E) and medial (F) views of the fsaverage midthickness surface. (G_–_I) Maps of retinotopic V1 and V2 of KHP10 case 1 (G), SHM07 case 1 (H), and KHP10 composite map (I). The eccentricity range spanned in these studies (7.75° for KHP10; 6–7.5° for SHM07) should have covered a little less than half of V1, based on previous studies in macaques (Van Essen et al. 1984) and humans (Hinds et al. 2009).
Figure 11.
(A,B) Extent of retinotopically mapped MT and its relation to architectonic hOC5, in case 1 (A) and case 5 (B) of Kolster et al. (2010). Maps are split roughly evenly between gyral and sulcal cortex in the individual hemispheres. (C) Retinotopic areas of KPO10 include area V4t (purple), MSTv (blue), FST (brown), PITd (green), and LO2 (yellow). (D) Probabilistic area hOc5 from Fischl et al. (2008) mapped to the right (top) and left (bottom) very inflated atlas surfaces. Black contours in each panel indicates the most likely boundary of hOc5 and are also shown in panels C and E. (E) Composite maps of retinotopic areas from KPO10 cases 1 and 5 shown in relation to area hOc5 boundaries.
Figure 12.
(A) Retinotopic maps in human extrastriate cortex. A composite of the left and right hemisphere maps for one subject shown in medial (top) and dorsal-posterior views. (B) BLW05 (Brewer et al. 2005) retinotopic areas shown in the same views, the surfaces were smoother than a typical human anatomical surface. To improve registration to the PALS-B12 atlas, additional landmarks were added that were readily discernible in the individual and atlas surfaces. (C) KPO10 areas (Kolster et al. 2010). (D) Area V6 from PGH06 (Pitzalis et al. 2006).
Figure 13.
Human OMPFC maps from Ongur et al. (2003). (A) Ventral view of midthickness surface from case WBR. (B) The same areas from case WBR after registration to the PALS-B12 atlas. (C) Probabilistic map (n = 4) for areas 11l, Iai, and 13a. (D_–_F) Composite map of OMBFC areas after registration to the fs_LR atlas.
Figure 14.
(A) A composite map of nonoverlapping or minimally overlapping SBR-based human cortical areas on lateral and medial views of fsaverage midthickness. Areas from the following 5 schemes: FRB08 (Fischl et al. 2008), OFP03 (Ongur et al. 2003), SHM07 (Swisher et al. 2007), PGH08 (Pitzalis et al. 2006), and BWE08 (Burton et al. 2008). Several areal boundaries were adjusted slightly to deal with modest overlap (between V3 and 18, V3A, V3B). (B) Composite areal parcellation on fsaverage very inflated surface plus centers of gravity of 31 architectonic areas mapped to the atlas using nl-VBR (Eickhoff et al. 2005). (C) Brodmann (1909) parcellation scheme mapped to the fs_LR atlas surface. Note that insular areas 52, “J post” (granular insular) and “J ant” (agranular insula) were described by Brodmann (cf. Kurth et al. 2009) but were not on the figure from Polyak (1957) used to transpose the Brodmann areas to the atlas. (D) Histogram of surface areas for the composite parcellation shown in A.
Similar articles
- Cortical parcellations of the macaque monkey analyzed on surface-based atlases.
Van Essen DC, Glasser MF, Dierker DL, Harwell J. Van Essen DC, et al. Cereb Cortex. 2012 Oct;22(10):2227-40. doi: 10.1093/cercor/bhr290. Epub 2011 Nov 2. Cereb Cortex. 2012. PMID: 22052704 Free PMC article. - A Population-Average, Landmark- and Surface-based (PALS) atlas of human cerebral cortex.
Van Essen DC. Van Essen DC. Neuroimage. 2005 Nov 15;28(3):635-62. doi: 10.1016/j.neuroimage.2005.06.058. Epub 2005 Sep 19. Neuroimage. 2005. PMID: 16172003 - Hemispherically-unified surface maps of human cerebral cortex: reliability and hemispheric asymmetries.
Kang X, Herron TJ, Cate AD, Yund EW, Woods DL. Kang X, et al. PLoS One. 2012;7(9):e45582. doi: 10.1371/journal.pone.0045582. Epub 2012 Sep 18. PLoS One. 2012. PMID: 23029115 Free PMC article. - Functional and structural mapping of human cerebral cortex: solutions are in the surfaces.
Van Essen DC, Drury HA, Joshi S, Miller MI. Van Essen DC, et al. Proc Natl Acad Sci U S A. 1998 Feb 3;95(3):788-95. doi: 10.1073/pnas.95.3.788. Proc Natl Acad Sci U S A. 1998. PMID: 9448242 Free PMC article. Review. - Why data coherence and quality is critical for understanding interareal cortical networks.
Kennedy H, Knoblauch K, Toroczkai Z. Kennedy H, et al. Neuroimage. 2013 Oct 15;80:37-45. doi: 10.1016/j.neuroimage.2013.04.031. Epub 2013 Apr 18. Neuroimage. 2013. PMID: 23603347 Review.
Cited by
- A hierarchical atlas of the human cerebellum for functional precision mapping.
Nettekoven C, Zhi D, Shahshahani L, Pinho AL, Saadon-Grosman N, Buckner RL, Diedrichsen J. Nettekoven C, et al. Nat Commun. 2024 Sep 27;15(1):8376. doi: 10.1038/s41467-024-52371-w. Nat Commun. 2024. PMID: 39333089 Free PMC article. - Children born very preterm experience altered cortical expansion over the first decade of life.
Gorham LS, Latham AR, Alexopoulos D, Kenley JK, Iannopollo E, Lean RE, Loseille D, Smyser TA, Neil JJ, Rogers CE, Smyser CD, Garcia K. Gorham LS, et al. Brain Commun. 2024 Sep 17;6(5):fcae318. doi: 10.1093/braincomms/fcae318. eCollection 2024. Brain Commun. 2024. PMID: 39329081 Free PMC article. - Predicting Task Activation Maps from Resting-State Functional Connectivity using Deep Learning.
Madsen SJ, Uddin LQ, Mumford JA, Barch DM, Fair DA, Gotlib IH, Poldrack RA, Kuceyeski A, Saggar M. Madsen SJ, et al. bioRxiv [Preprint]. 2024 Sep 14:2024.09.10.612309. doi: 10.1101/2024.09.10.612309. bioRxiv. 2024. PMID: 39314460 Free PMC article. Preprint. - Early Life Neuroimaging: The Generalizability of Cortical Area Parcellations Across Development.
Tu JC, Myers M, Li W, Li J, Wang X, Dierker D, Day TKM, Snyder AZ, Latham A, Kenley JK, Sobolewski CM, Wang Y, Labonte AK, Feczko E, Kardan O, Moore LA, Sylvester CM, Fair DA, Elison JT, Warner BB, Barch DM, Rogers CE, Luby JL, Smyser CD, Gordon EM, Laumann TO, Eggebrecht AT, Wheelock MD. Tu JC, et al. bioRxiv [Preprint]. 2024 Sep 11:2024.09.09.612056. doi: 10.1101/2024.09.09.612056. bioRxiv. 2024. PMID: 39314355 Free PMC article. Preprint. - Contracted functional connectivity profiles in autism.
Weber CF, Kebets V, Benkarim O, Lariviere S, Wang Y, Ngo A, Jiang H, Chai X, Park BY, Milham MP, Di Martino A, Valk S, Hong SJ, Bernhardt BC. Weber CF, et al. Mol Autism. 2024 Sep 11;15(1):38. doi: 10.1186/s13229-024-00616-2. Mol Autism. 2024. PMID: 39261969 Free PMC article.
References
- Amunts K, Schleicher A, Burgel U, Mohlberg H, Uylings HB, Zilles K. Broca's region revisited: cytoarchitecture and intersubject variability. J Comp Neurol. 1999;412:319–341. - PubMed