This dataset has been retracted at the request of the authors on December 3, 2021. Do not cite this dataset.

Benchmarking Geometric Deep Learning for Cortical Segmentation and Neurodevelopmental Phenotype Prediction

, , , , , , , , , , , , ,
  1. 1Department of Biomedical Engineering, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom
  2. 2Centre for the Developing Brain, School of Biomedical Engineering and Imaging Sciences, King's College London, London, United Kingdom
  3. 3ETS Montreal
  4. 4Murdoch Children's Research Institute, Melbourne, Victoria, Australia
  5. 5Department of Neurosurgery, Royal Children's Hospital, Melbourne, Australia
  6. 6Florey Department of Neuroscience and Mental Health, The University of Melbourne, Melbourne, Australia
  7. 7Department of Paediatrics, The University of Melbourne, Melbourne, Australia
  8. 8Biomedical Image Analysis Group, Department of Computing, Imperial College London, London, United Kingdom

Published 04 Nov. 2021 | Retracted 03 Dec. 2021


The emerging field of geometric deep learning extends the application of convolutional neural networks into irregular domains such as graphs, meshes and surfaces. Several recent studies have explored the potential for using these techniques to analyse and segment the cortical surface. However, there has been no comprehensive comparison of these approaches to one another, nor to existing Euclidean methods,to date. This paper benchmarks a collection of geometric and traditional deep learning models on phenotype prediction and segmentation of sphericalised neonatal cortical surface data, from the publicly available Developing Human Connectome Project (dHCP). Tasks include prediction of postmenstrual age at scan, gestational age at birth and segmentation of the cortical surface into anatomical regions defined by the M-CRIB-S atlas. Performance was assessed not only in terms of model precision, but also in terms of network dependence on image registration, and model interpretation via occlusion. Networks were trained both on sphericalised and anatomical cortical meshes. Findings suggest that the utility of geometric deep learning over traditional deep learning is highly task-specific, which has implications for the design of future deep learning models on the cortical surface.


| Geometric deep learning | Benchmarking | Cortical surface | Cortical segmentation | Graph neural networks | Developing Human Connectome Project | Spherical convolutional networks |


  • Adamson, C. L., Alexander, B., Ball, G., Beare, R., Cheong, J. L., Spittle, A. J., ... & Thompson, D. K. (2020). Parcellation of the neonatal cortex using Surface-based Melbourne Children's Regional Infant Brain atlases (M-CRIB-S). Scientific reports, 10(1), 1-11.
  • Bozek, J., Makropoulos, A., Schuh, A., Fitzgibbon, S., Wright, R., Glasser, M. F., ... & Robinson, E. C. (2018). Construction of a neonatal cortical surface atlas using Multimodal Surface Matching in the Developing Human Connectome Project. NeuroImage, 179, 11-29.
  • Fawaz, A., Williams, L. Z. J., Alansary, A., Bass, C., Gopinath, K., da Silva, M., Dahan, S., Adamson, C., Alexander, B., Thompson, D., Ball, G., Lombaert, H., Rueckert, D., Edwards, A. D., & Robinson, E. C. (2021). Benchmarking Geometric Deep Learning for Cortical Segmentation and Neurodevelopmental Phenotype Prediction.
  • Fitzgibbon, S. P., Harrison, S. J., Jenkinson, M., Baxter, L., Robinson, E. C., Bastiani, M., ... & Andersson, J. (2020). The developing Human Connectome Project (dHCP) automated resting-state functional processing framework for newborn infants. NeuroImage, 223, 117303.
  • Makropoulos, A., Robinson, E. C., Schuh, A., Wright, R., Fitzgibbon, S., Bozek, J., ... & Rueckert, D. (2018). The developing human connectome project: A minimal processing pipeline for neonatal cortical surface reconstruction. NeuroImage, 173, 88-112.
  • Robinson, E. C., Jbabdi, S., Glasser, M. F., Andersson, J., Burgess, G. C., Harms, M. P., ... & Jenkinson, M. (2014). MSM: a new flexible framework for Multimodal Surface Matching. NeuroImage, 100, 414-426.
  • Robinson, E. C., Garcia, K., Glasser, M. F., Chen, Z., Coalson, T. S., Makropoulos, A., ... & Rueckert, D. (2018). Multimodal surface matching with higher-order smoothness constraints. NeuroImage, 167, 453-465.
  • Williams, L. Z. J., Fitzgibbon, S. P., Bozek, J., Winkler, A. M., Dimitrova, R., Poppe, T., ... & Robinson, E. C. (2021). Structural and functional asymmetry of the neonatal cerebral cortex. bioRxiv.


  • ERC 319,456