Surface-based atlases and subsections

Anatomical atlas

An anatomical atlas has been created in surface space for SubCortexMesh (SCM)’s ASeg-derived “fsaverage” template to help users identify more precisely where their statistical results are located in subcortical regions-of-interest (ROIs). This anatomical atlas is based on existing volume-based atlases that were projected to surface templates with a number of manual fine-tuning to produce a rough estimate of which ROIs are surface vertices located in. We detail here how it was built and how to use it in SCM.

⚠️ Because it is a graphical projection, not delineated sections directly derived from brain scans, and since the template surfaces are asymmetrical, the ROIs must be interpreted as a broad guidance for reference rather than precisely certain locations.

Summary

• Development details

  • Volumetric atlas extraction

  • Volume to surface projection

• How to use them in SubCortexMesh

• Lookup tables

• References

---

Development details

Volumetric atlas extraction

Depending on the subcortical region, labels from various volumetric atlases were selected. Because of high granularity lost or scrambled upon conversion to surface, some ROIs were merged, and others excluded, especially when buried inside regions or too subtle to be correctly captured on the surface “shell”. We also took advantage of FreeSurfer freeview’s “3D isosurface” view to guide assessment of the converted surfaces.

The atlas includes the following subsegmentations:

• Globus pallidus interna and externa: we relied on the CIT168 Reinforcement Learning Atlas (Pauli, Nili & Tyszka, 2018), using the MNI 152 space. Specifically, “CIT168_Reinf_Learn_v1.1.0/MNI152-Nonlin-Asym-2009c/CIT168toMNI152-2009c_det.nii” in OSF. The atlas was split to distinguish left and right ROIs.

• Cerebellar layers: we relied on the Diedrichsen 2009 atlas (Diedrichsen et al., 2009; Diedrichsen et al., 2011). Specifically, the probability atlas in MNI space (“atl-Anatom_space-MNISym_probseg.nii”) was converted to volumetric segmentations with minimal threshold (whichever label had the strongest value above 0). Regions excluded as outside of the FreeSurfer ASeg segmentation were Vernis Crus I, Vermis X, dentate nuclei, interposed nuclei and fastigial nuclei. Other Vermis areas, although not strictly inside SCM’s shapes either, were kept to label the medial sides of the cerebellar surfaces connected to them.

• Thalamic nuclei: we relied on FreeSurfer (Fischl, 2012)’s Subcortical Segmentations, which provide Iglesias et al. (2018)’s probability atlas in MNI space (specifically, “ThalamusProbs.MNIsymSpace.nii.gz”). The atlas was converted to volumetric segmentations with minimal threshold (whichever label had the strongest value above 0). Due to their granularity, several ROIs were merged per broader categories (see Iglesias et al. (2018), Table 2): Lateral (LD, LP), Ventral anterior (VA, VAmc), Ventral lateral (VLa, VLp), Intralaminar (CM, Pf). Excluded ROIs included: Lateral and medial geniculate nuclei (LGN and MGN) as outside of the classic ASeg segmentation; Mediodorsal lateral parvocellular (MDl), Pulvinar anterior (PuA) as deep inside the volume, so not corresponding to any outer surface; Reticular nucleus (R) since it is a thin outer layer covering multiple nuclei and cutting through other volumes creating artefacts; Ventromedial nucleus (VM), paracentral (Pc) and paratenial (Pt) which did not end in the probability atlas when converted to volumetric segmentation at all; Central lateral intralaminar (CL), Reuniens (MV-re) as too subtle for meaningful surface projection.

• Hippocampal subfields: we also relied on the FreeSurfer Subcortical Segmentations, using the atlas from Iglesias et al. (2015a) (specifically, “HippoAmygProbs.MNIsymSpace.left.nii.gz” and “HippoAmygProbs.MNIsymSpace.right.nii.gz”), with the same probability thresholding approach. Excluded ROIs included: GC-DG, molecular layer, and hippocampal fissure, as deep inside the volume, so not really matching the outer surface; alveus and fimbria, as too subtle for surface projection; and HATA, as outside of the classic ASeg segmentation.

• Amygdalar nuclei: we also relied on the FreeSurfer Subcortical Segmentations, using the atlas from Saygin & Kliemann et al. (2017) (likewise, “HippoAmygProbs.MNIsymSpace.left.nii.gz” and “HippoAmygProbs.MNIsymSpace.right.nii.gz”), with the same probability thresholding approach. Excluded ROIs included: Anterior-amygdaloid-area, as outside the ASeg volume; Medial nucleus as too subtle for surface projection; and Paralaminar nucleus, as it’s a thin outer layer covering multiple nuclei, cutting through other volumes and creating artefacts.

• Brain-Stem main sections: we also relied on the FreeSurfer Subcortical Segmentations, using the atlas from Iglesias et al. (2015b) (specifically, using “BrainstemProbs.MNIsymSpace.nii.gz”) with the same probability thresholding approach.

• Ventral diencephalon (DC) subcomponents: we created our own volumetric atlas, likewise made of different volumetric atlases, stitched together to “populate” the parts of the ventral DC, called “Ventral DC Frankenstein”. The atlas included Brainstem Navigator’s atlas (García-Gomar et al., 2019; Singh et al., 2021) for the medial and lateral geniculate nuclei (specifically taken from “2b.DiencephalicNucleiAtlas_MNI”), with >0 probability threshold as opposed to 0.35 (which was justified by the Ventral DC’s own rough delineation as opposed to the high resolution segmentation the Brainstem Navigator is based upon). It also included the John Hopkins University (JHU)’s DTI atlas (Hua et al. 2008; Wakana et al. 2007) for the cerebral peduncles, (specifically, the neurovault ICBM 1mm version). Finally, it made use of the same CIT168 atlas as for the pallidum (Pauli et al., 2018) due to its overlapping labels (Hypothalamus, Red nucleus etc.). The subtantia nigras’s two pars were merged, as well as the parabrachial pigmented and ventral tegmental area, to facilitate volume-to-surface projection due to difficulty distinguishing them in the projected surface.

Note about MNI space: SCM’s surface template are based on ASeg which is strictly speaking in MNI305 space. CIT168 and Diedrichsen 2009 were resampled to MNI305 (ASeg’s space) using the Neuroatlas R package’s MNI152_to_MNI305 transformation matrices (Buchsbaum, 2006; see R/coordinate_spaces.R). FreeSurfer’s subcortical segmentations were simply resampled with the same software’s “mri_vol2vol –regheader” which gave satisfying results.

Volume to surface projection

Volume labels were projected to their corresponding template surface object, for each ROI one by one, using the Visualization Toolkit (VTK v9.5.2; Schroeder, Martin & Lorensen, 2006) in Python. The volume regions were aligned to their respective SCM surface template as consistently and closely as possible, with automated centroid alignment, manual fine-tuning of the coordinates, and manual tweaking of their scale. Once satisfying spatial correspondance was achieved, we projected their atlas voxel-wise label values to the nearest vertices in the surface, using SciPy (v1.15.2; Virtanen et al. 2020)’s Euclidean Distance Transform. Further smoothing was applied to minimise presence of sparse triangles.

The following screenshot shows an example of a surface-based right hippocampus next to its respective voxel-based volume:

---

How to use them in SubCortexMesh

As of version 1.1.0, the data fetched via template_data_fetch() also downloads the “atlas” subdirectory for the fsaverage template, which includes an array with values corresponding to each vertex in template space (“anatomical_atlas_fsaverage.npy”) - following the order of the allaseg_roi_id.txt reference - and a lookup table with the labels matching each value (“anatomical_atlas_fsaverage_names”).

Now, mesh_metrics() will not only print summary statistics for the broad ROIs in native space, but also for the atlas subsegmentations in template space. A .txt called “[measure]_stats_atlas.txt” will show rows such as:

Label	Mean	SD	Min	Max	Range	n_vert
Brain_Stem_Medulla	7.818	3.500	1.419	17.271	15.852	2431
Brain_Stem_Pons	13.417	2.161	4.965	18.040	13.075	4716
Brain_Stem_Superior_Cerebellar_Peduncle	14.934	0.496	13.759	15.870	2.111	117
Brain_Stem_Midbrain	9.272	3.291	0.425	16.580	16.155	2188
…

⚠️ When interpreting thickness, one must keep in mind that it is a radial distance from the medial curve, not from the true end of any subfield/nucleus’ shape. Thicker vertices in a given ROI could but not necessarily would mean that the area is thicker, as it would still appear thicker if an inner layer was “pushing out” the outer layer that the atlas is delineating.

The vis_merged(), vis_merged_flat() and slm_plot() all include the option to show atlas colours and/or outlines via the atlas_map boolean argument. When set to True, the 3D visualisers will include two buttons, one to show the labelling values, a second their outline, on top of the main mesh values. Hovering the cursor on top of each ROI will display the label name in the bottom left corner of the window. The 2D vis_merged_flat() output only includes the outlines, along with a legend in the bottom left box.

Here is an example with the bilateral thalami example statistical outputs from the website’s main page and overlaid atlas labels:

slm_plot(slm_model, 't_rft', cmap='Blues_r', smooth_mesh=10, threshold=.05, atlas_map=True)

Those also appear in the cluster_summary() function, indicating the subregion where a cluster’s peak t statistics vertex is located, if applicable:

cluster_summary(slm_model, template='fsaverage')

{'Positive contrast': 'No significant clusters',
 'Negative contrast':    clusid  nverts       P      X      Y      Z  tstat          region                          atlas
 0       1  2713.0  <0.001  152.0  120.4  153.6  -4.08   left-thalamus    Thalamus_Pulvinar-Lateral_L
 1       2  1846.0  <0.001  125.3  132.8  142.0  -3.53  right-thalamus  Thalamus_Mediodorsal-Medial_R}

Lookup tables

The table below lays out every region included in the anatomical atlas along with their corresponding IDs and labels:

Subcortical Atlas Labels

new_id	new_label	old_id	roi	R	G	B
1	Brain_Stem_Medulla	175	brain-stem	137	198	71
2	Brain_Stem_Pons	174	brain-stem	43	111	239
3	Brain_Stem_Superior_Cerebellar_Peduncle	178	brain-stem	229	66	126
4	Brain_Stem_Midbrain	173	brain-stem	57	66	191
5	Accumbens_Area_L		left-accumbens-area	120	70	206
6	Accumbens_Area_R		right-accumbens-area	120	70	206
7	Amygdala_Lateral_Nucleus_L	7001	left-amygdala	198	115	59
8	Amygdala_Lateral_Nucleus_R	7001	right-amygdala	198	115	59
9	Amygdala_Basal-Nucleus_L	7003	left-amygdala	221	202	79
10	Amygdala_Basal-Nucleus_R	7003	right-amygdala	221	202	79
11	Amygdala_Accessory-Basal-Nucleus_L	7008	left-amygdala	42	88	204
12	Amygdala_Accessory-Basal-Nucleus_R	7008	right-amygdala	42	88	204
13	Amygdala_Central-Nucleus_L	7005	left-amygdala	170	219	48
14	Amygdala_Central-Nucleus_R	7005	right-amygdala	170	219	48
15	Amygdala_Cortical-Nucleus_L	7007	left-amygdala	164	211	33
16	Amygdala_Cortical-Nucleus_R	7007	right-amygdala	164	211	33
17	Amygdala_Corticoamygdaloid-Transitio_L	7009	left-amygdala	239	62	216
18	Amygdala_Corticoamygdaloid-Transitio_R	7009	right-amygdala	239	62	216
19	Caudate_L		left-caudate	74	68	191
20	Caudate_R		right-caudate	74	68	191
21	Cerebellum_I_IV_L	1	left-cerebellum-cortex	71	224	155
22	Cerebellum_I_IV_R	2	right-cerebellum-cortex	71	224	155
23	Cerebellum_V_L	3	left-cerebellum-cortex	221	44	88
24	Cerebellum_V_R	4	right-cerebellum-cortex	221	44	88
25	Cerebellum_VI_L	5	left-cerebellum-cortex	48	232	103
26	Cerebellum_Vermis_VI	6	cerebellum-cortex	210	86	216
27	Cerebellum_VI_R	7	right-cerebellum-cortex	48	232	103
28	Cerebellum_CrusI_L	8	left-cerebellum-cortex	44	197	224
29	Cerebellum_CrusI_R	10	right-cerebellum-cortex	44	197	224
30	Cerebellum_CrusII_L	11	left-cerebellum-cortex	34	86	216
31	Cerebellum_Vermis_CrusII	12	cerebellum-cortex	206	47	50
32	Cerebellum_CrusII_R	13	right-cerebellum-cortex	34	86	216
33	Cerebellum_VIIb_L	14	left-cerebellum-cortex	200	96	224
34	Cerebellum_Vermis_VIIb	15	cerebellum-cortex	190	89	204
35	Cerebellum_VIIb_R	16	right-cerebellum-cortex	200	96	224
36	Cerebellum_VIIIa_L	17	left-cerebellum-cortex	57	97	229
37	Cerebellum_Vermis_VIIIa	18	cerebellum-cortex	60	60	201
38	Cerebellum_VIIIa_R	19	right-cerebellum-cortex	57	97	229
39	Cerebellum_VIIIb_L	20	left-cerebellum-cortex	224	78	219
40	Cerebellum_Vermis_VIIIb	21	cerebellum-cortex	186	65	216
41	Cerebellum_VIIIb_R	22	right-cerebellum-cortex	224	78	219
42	Cerebellum_IX_L	23	left-cerebellum-cortex	186	55	191
43	Cerebellum_Vermis_IX	24	cerebellum-cortex	71	239	164
44	Cerebellum_IX_R	25	right-cerebellum-cortex	186	55	191
45	Cerebellum_X_L	26	left-cerebellum-cortex	214	187	36
46	Cerebellum_X_R	28	right-cerebellum-cortex	214	187	36
47	Hippocampus_Tail_L	226	left-hippocampus	204	55	186
48	Hippocampus_Tail_R	226	right-hippocampus	204	55	186
49	Hippocampus_Subiculum_Body_L	236	left-hippocampus	68	214	148
50	Hippocampus_Subiculum_Body_R	236	right-hippocampus	68	214	148
51	Hippocampus_CA1_Body_L	238	left-hippocampus	36	201	198
52	Hippocampus_CA1_Body_R	238	right-hippocampus	36	201	198
53	Hippocampus_Subiculum_Head_L	235	left-hippocampus	196	62	160
54	Hippocampus_Subiculum_Head_R	235	right-hippocampus	196	62	160
55	Hippocampus_Presubiculum_Head_L	233	left-hippocampus	76	214	57
56	Hippocampus_Presubiculum_Head_R	233	right-hippocampus	76	214	57
57	Hippocampus_CA1_Head_L	237	left-hippocampus	204	71	224
58	Hippocampus_CA1_Head_R	237	right-hippocampus	204	71	224
59	Hippocampus_Presubiculum_Body_L	234	left-hippocampus	167	221	48
60	Hippocampus_Presubiculum_Body_R	234	right-hippocampus	167	221	48
61	Hippocampus_Parasubiculum_L	203	left-hippocampus	229	48	175
62	Hippocampus_Parasubiculum_R	203	right-hippocampus	229	48	175
63	Hippocampus_CA3_Body_L	240	left-hippocampus	43	53	196
64	Hippocampus_CA3_Body_R	240	right-hippocampus	43	53	196
65	Hippocampus_CA4_Head_L	241	left-hippocampus	158	224	100
66	Hippocampus_CA4_Head_R	241	right-hippocampus	158	224	100
67	Hippocampus_CA4_Body_L	242	left-hippocampus	167	80	229
68	Hippocampus_CA4_Body_R	242	right-hippocampus	167	80	229
69	Hippocampus_CA3_Head_L	239	left-hippocampus	62	189	224
70	Hippocampus_CA3_Head_R	239	right-hippocampus	62	189	224
71	Globus_Pallidus_Externa_L	1	left-pallidum	232	129	60
72	Globus_Pallidus_Externa_R	3	right-pallidum	232	129	60
73	Globus_Pallidus_Interna_L	2	left-pallidum	117	68	191
74	Globus_Pallidus_Interna_R	4	right-pallidum	117	68	191
75	Putamen_L		left-putamen	214	70	197
76	Putamen_R		right-putamen	214	70	197
77	Thalamus_Pulvinar-Inferior_L	8121	left-thalamus	239	172	79
78	Thalamus_Pulvinar-Inferior_R	8221	right-thalamus	239	172	79
79	Thalamus_Pulvinar-Medial_L	8123	left-thalamus	219	126	83
80	Thalamus_Pulvinar-Medial_R	8223	right-thalamus	219	126	83
81	Thalamus_Limitans-Suprageniculate_L	8111	left-thalamus	137	59	214
82	Thalamus_Limitans_Suprageniculate_R	8211	right-thalamus	137	59	214
83	Thalamus_Ventral-Posterolateral_L	8133	left-thalamus	52	126	211
84	Thalamus_Ventral-Posterolateral_R	8233	right-thalamus	52	126	211
85	Thalamus_Centromedian-Parafascicular_L	8106	left-thalamus	107	84	221
86	Thalamus_Centromedian-Parafascicular_R	8206	right-thalamus	107	84	221
87	Thalamus_Ventral-Lateral_L	8128	left-thalamus	47	206	58
88	Thalamus_Ventral-Lateral_R	8228	right-thalamus	47	206	58
89	Thalamus_Mediodorsal-Medial_L	8113	left-thalamus	156	219	74
90	Thalamus_Mediodorsal-Medial_R	8213	right-thalamus	156	219	74
91	Thalamus_Ventral-Anterior_L	8126	left-thalamus	114	63	204
92	Thalamus_Ventral-Anterior_R	8226	right-thalamus	114	63	204
93	Thalamus_Central-Medial_L	8104	left-thalamus	84	191	148
94	Thalamus_Central-Medial_R	8204	right-thalamus	84	191	148
95	Thalamus_Pulvinar-Lateral_L	8122	left-thalamus	198	198	51
96	Thalamus_Pulvinar-Lateral_R	8222	right-thalamus	198	198	51
97	Thalamus_Anteroventral_L	8103	left-thalamus	234	222	51
98	Thalamus_Anteroventral_R	8203	right-thalamus	234	222	51
99	Thalamus_Laterodorsal-Lateral-Posterior_L	8108	left-thalamus	63	234	137
100	Thalamus_Laterodorsal-Lateral-Posterior_R	8208	right-thalamus	63	234	137
101	Ventral_DC_Lateral_Geniculate_L	1	left-ventraldc	48	204	126
102	Ventral_DC_Lateral_Geniculate_R	2	right-ventraldc	48	204	126
103	Ventral_DC_Medial_Geniculate_L	3	left-ventraldc	219	104	94
104	Ventral_DC_Medial_Geniculate_R	4	right-ventraldc	219	104	94
105	Ventral_DC_Cerebral_Peduncle_L	6	left-ventraldc	122	60	193
106	Ventral_DC_Cerebral_Peduncle_R	5	right-ventraldc	122	60	193
107	Ventral_DC_Substantia_Nigra_L	47	left-ventraldc	59	212	229
108	Ventral_DC_Substantia_Nigra_R	67	right-ventraldc	59	212	229
109	Ventral_DC_Red_Nucleus_L	48	left-ventraldc	154	198	43
110	Ventral_DC_Red_Nucleus_R	68	right-ventraldc	154	198	43
111	Ventral_DC_Ventral_Tegmental_Area_L	51	left-ventraldc	107	76	201
112	Ventral_DC_Ventral_Tegmental_Area_R	71	right-ventraldc	107	76	201
113	Ventral_DC_Hypothalamus_L	54	left-ventraldc	141	237	64
114	Ventral_DC_Hypothalamus_R	74	right-ventraldc	141	237	64
115	Ventral_DC_Subthalamic_Nucleus_L	56	left-ventraldc	72	176	214
116	Ventral_DC_Subthalamic_Nucleus_R	76	right-ventraldc	72	176	214

References

• Buchsbaum B (2026). neuroatlas: Neuroimaging Atlases and Parcellations. R package version 0.1.0, https://github.com/bbuchsbaum/neuroatlas.

• Diedrichsen, J., Balsters, J. H., Flavell, J., Cussans, E., & Ramnani, N. (2009). A probabilistic atlas of the human cerebellum. Neuroimage.

• Diedrichsen, J., Maderwald, S., Kuper, M., Thurling, M., Rabe, K., Gizewski, E. R., et al. (2011). Imaging the deep cerebellar nuclei: A probabilistic atlas and normalization procedure. Neuroimage.

• Fischl, B. (2012). FreeSurfer. Neuroimage, 62(2), 774-781.

• García-Gomar, M. G., Strong, C., Toschi, N., Singh, K., Rosen, B. R., Wald, L. L., & Bianciardi, M. (2019). In vivo probabilistic structural atlas of the inferior and superior colliculi, medial and lateral geniculate nuclei and superior olivary complex in humans based on 7 Tesla MRI. Frontiers in neuroscience, 13, 764.

• Hua, K., Zhang, J., Wakana, S., Jiang, H., Li, X., Reich, D. S., … & Mori, S. (2008). Tract probability maps in stereotaxic spaces: analyses of white matter anatomy and tract-specific quantification. Neuroimage, 39(1), 336-347.

• Iglesias, J. E., Augustinack, J. C., Nguyen, K., Player, C. M., Player, A., Wright, M., … & Alzheimer’s Disease Neuroimaging Initiative. (2015a). A computational atlas of the hippocampal formation using ex vivo, ultra-high resolution MRI: Application to adaptive segmentation of in vivo MRI. Neuroimage, 115, 117-137.

• Iglesias, J. E., Van Leemput, K., Bhatt, P., Casillas, C., Dutt, S., Schuff, N., … & Alzheimer’s Disease Neuroimaging Initiative. (2015b). Bayesian segmentation of brainstem structures in MRI. Neuroimage, 113, 184-195.

• Iglesias, J. E., Insausti, R., Lerma-Usabiaga, G., Bocchetta, M., Van Leemput, K., Greve, D. N., … & Alzheimer’s Disease Neuroimaging Initiative. (2018). A probabilistic atlas of the human thalamic nuclei combining ex vivo MRI and histology. Neuroimage, 183, 314-326.

• Pauli, W. M., Nili, A. N., & Tyszka, J. M. (2018). A high-resolution probabilistic in vivo atlas of human subcortical brain nuclei. Scientific data, 5(1), 180063.

• Saygin, Z. M., Kliemann, D., Iglesias, J. E., van der Kouwe, A. J., Boyd, E., Reuter, M., … & Alzheimer’s Disease Neuroimaging Initiative. (2017). High-resolution magnetic resonance imaging reveals nuclei of the human amygdala: manual segmentation to automatic atlas. Neuroimage, 155, 370-382.

• Schroeder W, Martin K, Lorensen B. The Visualization Toolkit (4th Ed.). Kitware; 2006.

• Singh, K., García-Gomar, M. G., & Bianciardi, M. (2021). Probabilistic atlas of the mesencephalic reticular formation, isthmic reticular formation, microcellular tegmental nucleus, ventral tegmental area nucleus complex, and caudal–rostral linear raphe nucleus complex in living humans from 7 Tesla magnetic resonance imaging. Brain Connectivity, 11(8), 613-623.

• Virtanen P, Gommers R, Oliphant TE, et al. SciPy 1.0: fundamental algorithms for scientific computing in Python. Nat Methods. 2020;17(3):261-272. doi:10.1038/s41592-019-0686-2

• Wakana, S., Caprihan, A., Panzenboeck, M. M., Fallon, J. H., Perry, M., Gollub, R. L., … & Mori, S. (2007). Reproducibility of quantitative tractography methods applied to cerebral white matter. Neuroimage, 36(3), 630-644.