Anatomical Longitudinal Cortical Surface Registration

Aakash Saboo, Ashleigh Davies, Nashira Baena, Kaili Liang, Jiaxin Xiao, Yourong Guo, Renato Besenczi, Jonathan O'Muircheartaigh, Emma C. Robinson

Keywords: Geometric Deep Learning, Registration, Deformation

TL;DR: Register surfaces using native anatomical surfaces instead of spherical projection which improves registration metrics

Abstract: Longitudinal cortical surface registration is essential for accurately characterizing developmental and neurodegenerative trajectories, thereby facilitating a mechanistic understanding of cortical growth and the identification of biomarkers. This is hindered by current registration networks, which work on spherical projections of the cortical surface. Therefore, in this work, we present a novel longitudinal registration framework that operates directly on complex anatomical geometries by integrating a learning-based network with pairwise instance optimization. This hybrid strategy leverages the network to establish a robust initial alignment, which is subsequently refined through optimization to ensure high-fidelity registration. We demonstrate that this method yields growth maps with superior smoothness compared to baselines, enhancing their clinical utility, while rigorously preserving topological integrity as evidenced by analyses of self-intersecting faces, areal distortion, and anisotropic strain

Primary Subject Area: Geometric Deep Learning

Secondary Subject Area: Image Registration

Registration Requirement: Yes

Reproducibility: https://github.com/aakash-saboo/Anatomical-Longitudinal-Registration

Visa & Travel: Yes

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Midl Latex Submission Checklist: Ensure no LaTeX errors during compilation., Replace NNN with your OpenReview submission ID., Includes \documentclass{midl}, \jmlryear{2026}, \jmlrworkshop, \jmlrvolume, \editors, and correct \bibliography command., Did not override options of the hyperref package., Did not use the times package., Use the correct spelling and format, avoid Unicode characters, and use LaTeX equivalents instead., Any math in the title and abstract must be enclosed within $...$., Did not override the bibliography style defined in midl.cls and did not use \begin{thebibliography} directly to insert references., Avoid using \scalebox; use \resizebox when needed., Included all necessary figures and removed *unused* files in the zip archive., Removed special formatting, visual annotations, and highlights used during rebuttal., All special characters in the paper and .bib file use LaTeX commands (e.g., \'e for é)., No separate supplementary PDF uploads., Acknowledgements, references, and appendix must start after the main content.

Submission Number: 289