Automated subcortical and cortical segmentations of BigBrains 1 and 2 using FreeSurfer v8.0 beta
Poster No:
1630
Submission Type:
Abstract Submission
Authors:
Lindsay Lewis1, Claude Lepage1, Alan Evans1
Institutions:
1MCIN, McGill University, Montreal, QC
First Author:
Co-Author(s):
Introduction:
In 2013, we published BigBrain 1 (BB1), a high-resolution (20µm3) histological 3D-reconstructed model of the human brain (Amunts et al., 2013). Over the past several years, advances have been made on the reconstruction of BigBrain 2 (BB2) (Mohlberg et al., 2024), with preliminary segmented volumes [white matter (WM) and gray matter (GM)] now available.
Earlier this year, we (Lewis et al., 2024) produced automated subcortical and cortical segmentations as well as surface extractions of BB1 and BB2 using an adapted FreeSurfer (FS) v7.4 pipeline (Fischl, 2012).
Here, we have extended our pre-processing approach to directly submit the histological intensity volume (eliminating the need for an MRI simulator). Furthermore, we are able to harness the benefits of the recently released beta version of FS v8.0, which integrates a new subcortical segmentation tool, as well as additional segmentation tools for hippocampal subfields, nuclei of the amygdala, brainstem, and thalamic nuclei (Iglesias et al., 2015, 2018).
Earlier this year, we (Lewis et al., 2024) produced automated subcortical and cortical segmentations as well as surface extractions of BB1 and BB2 using an adapted FreeSurfer (FS) v7.4 pipeline (Fischl, 2012).
Here, we have extended our pre-processing approach to directly submit the histological intensity volume (eliminating the need for an MRI simulator). Furthermore, we are able to harness the benefits of the recently released beta version of FS v8.0, which integrates a new subcortical segmentation tool, as well as additional segmentation tools for hippocampal subfields, nuclei of the amygdala, brainstem, and thalamic nuclei (Iglesias et al., 2015, 2018).
Methods:
(1) Preprocessing:
Inhomogeneity correction: the histological intensity BB volume (200µm3) was submitted to AntsN4BiasFieldCorrection (Tustison et al., 2010) with the following parameters: shrink factor = 2; 3 cycles of 1000 iterations; spline distance = 80mm (80mm, 40um, 20um for each of 3 cycles); and a background mask. This volume is 'T2w-like' in that GM is brighter than WM, with a black background. Since FS only accepts in vivo T1w-like input, it was necessary to invert the histological intensities.
(2) FreeSurfer v8.0 main processing pipeline (recon-all) at 200µm3:
In previous versions of FS, we found that at 200µm3, mri_ca_label took several days to process (and only on a single core), with suboptimal results. Now, mri_ca_label has been replaced with the deep learning tool mri_synthseg, for which there are pros and cons. Cons: it unavoidably downsamples to 1mm3. Pros: it processes quickly, on multiple cores, and with improved accuracy relative to mri_ca_label. Following cortical surface extraction, the GM and WM volumes are refined at the resolution at which the surfaces were extracted, but the subcortical segmentation unfortunately remains at 1mm3.
Recon-all parameters were as follows:
recon-all -noskullstrip -no-ants-denoise -no-mprage
mri_synthseg v1
(3) Hippocampal subfields and nuclei of the amygdala [segmentHA_T2.sh, subsequent to recon-all (Iglesias et al., 2015)]:
Following recon-all at 200µm3, it is possible to extract the hippocampal subfields and nuclei of the amygdala within the hippocampal and amygdala regions defined at 1mm3 by mri_synthseg. Here, we supply the original histological intensity volume at 100µm3 as a supplemental T2w-like volume to improve the subfields segmentation.
Inhomogeneity correction: the histological intensity BB volume (200µm3) was submitted to AntsN4BiasFieldCorrection (Tustison et al., 2010) with the following parameters: shrink factor = 2; 3 cycles of 1000 iterations; spline distance = 80mm (80mm, 40um, 20um for each of 3 cycles); and a background mask. This volume is 'T2w-like' in that GM is brighter than WM, with a black background. Since FS only accepts in vivo T1w-like input, it was necessary to invert the histological intensities.
(2) FreeSurfer v8.0 main processing pipeline (recon-all) at 200µm3:
In previous versions of FS, we found that at 200µm3, mri_ca_label took several days to process (and only on a single core), with suboptimal results. Now, mri_ca_label has been replaced with the deep learning tool mri_synthseg, for which there are pros and cons. Cons: it unavoidably downsamples to 1mm3. Pros: it processes quickly, on multiple cores, and with improved accuracy relative to mri_ca_label. Following cortical surface extraction, the GM and WM volumes are refined at the resolution at which the surfaces were extracted, but the subcortical segmentation unfortunately remains at 1mm3.
Recon-all parameters were as follows:
recon-all -noskullstrip -no-ants-denoise -no-mprage
mri_synthseg v1
(3) Hippocampal subfields and nuclei of the amygdala [segmentHA_T2.sh, subsequent to recon-all (Iglesias et al., 2015)]:
Following recon-all at 200µm3, it is possible to extract the hippocampal subfields and nuclei of the amygdala within the hippocampal and amygdala regions defined at 1mm3 by mri_synthseg. Here, we supply the original histological intensity volume at 100µm3 as a supplemental T2w-like volume to improve the subfields segmentation.
Results:
Fig 1: BigBrain 2 FS v8.0 beta automated cortical surface extraction (top row); subcortical and cortical parcellations ('wmparc.mgz' / DKT atlas, bottom row).
Fig 2: BigBrain 2 FS v8.0 beta automated segmentations of hippocampal subfields and nuclei of the amygdala (100µm3; Iglesias et al., 2015), thalamic nuclei (Iglesias et al., 2018), and brainstem (Iglesias et al., 2015).
Fig 2: BigBrain 2 FS v8.0 beta automated segmentations of hippocampal subfields and nuclei of the amygdala (100µm3; Iglesias et al., 2015), thalamic nuclei (Iglesias et al., 2018), and brainstem (Iglesias et al., 2015).
Conclusions:
To our knowledge, this is the first whole-brain parcellation of BigBrains 1 and 2. It provides regions of interest for higher-resolution analyses, such as FS' integrated hippocampal subfield segmentation (100µm3), as well as other external analyses. However, caution should be taken, particularly due to FS' internal subcortical downsampling at 1mm3.
A potential benefit of FS-derived BB surfaces is to improve the multimodal surface matching (MSM) registration (Robinson et al., 2018) approach - which was developed and parameterized on FS surfaces - to canonical surface templates in other software ecosystems (Lewis et al., 2023), e.g., FS' fsaverage and Human Connectome Project (HCP)'s fs_LR (Van Essen et al., 2012).
A potential benefit of FS-derived BB surfaces is to improve the multimodal surface matching (MSM) registration (Robinson et al., 2018) approach - which was developed and parameterized on FS surfaces - to canonical surface templates in other software ecosystems (Lewis et al., 2023), e.g., FS' fsaverage and Human Connectome Project (HCP)'s fs_LR (Van Essen et al., 2012).
Modeling and Analysis Methods:
Image Registration and Computational Anatomy
Methods Development
Segmentation and Parcellation 1
Neuroanatomy, Physiology, Metabolism and Neurotransmission:
Cortical Anatomy and Brain Mapping 2
Neuroinformatics and Data Sharing:
Brain Atlases
Keywords:
Atlasing
Cortex
Data analysis
Data Registration
MRI
STRUCTURAL MRI
Workflows
Other - BigBrain
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
Which processing packages did you use for your study?
Provide references using APA citation style.
Fischl, B., (2012). FreeSurfer. NeuroImage, 62(2), 774-781.
Iglesias, J.E., et al. (2015). Bayesian segmentation of brainstem structures in MRI. NeuroImage, 113, 184-195.
Iglesias, J.E., et al. (2018). A probabilistic atlas of the human thalamic nuclei combining ex vivo MRI and histology. NeuroImage, 183, 314-326.
Lewis, L.B., et al. (2023). Advances in MSM surface registration to bridge data across the BigBrain and FS / HCP ecosystems. OHBM poster, Montreal.
Lewis, L.B., et al. (2024). Adaptation of FreeSurfer v7.4 pipeline for automated volumetric parcellation and cortical surface extraction of BigBrains 1 and 2. BigBrain Workshop talk, Padua, Italy.
Mohlberg, H., et al. (2024). 3D reconstruction of BigBrain2: Progress report on semi-automated repairs of histological sections. BigBrain Workshop poster, Padua, Italy.
Robinson, E.C., et al. (2018). Multimodal surface matching with higher-order smoothness constraints. Neuroimage, 167, 453-465
Tustison, N.J., et al. (2010). N4ITK: improved N3 bias correction. IEEE Trans Med Imaging, 29(6), 1310-1320.
Van Essen, D.C., et al. (2012). Parcellations and hemispheric asymmetries of human cerebral cortex analyzed on surface-based atlases. Cerebral Cortex, 2012(10), 2241–2262.