SimCortex

Deep Learning for Cortical Surface Reconstruction from Brain MRI

Website · GitHub · 3D Slicer Extension

About

SimCortex is an open-source research project focused on fast, accurate, and topologically consistent cortical surface reconstruction from structural T1-weighted magnetic resonance imaging (MRI).

The framework reconstructs four cortical surfaces:

• Left hemisphere white matter surface
• Left hemisphere pial surface
• Right hemisphere white matter surface
• Right hemisphere pial surface

SimCortex combines deep learning, geometric modeling, and topology-aware surface reconstruction to produce anatomically meaningful cortical meshes while minimizing self-intersections and inter-surface collisions.

Research Focus

Our work explores deep learning for neuroimaging, 3D geometric modeling, cortical mesh reconstruction, diffeomorphic surface deformation, and anatomically constrained learning.

Open-Source Ecosystem

SimCortex Framework: Core implementation, inference tools, and research code.

SlicerSimCortex: A 3D Slicer extension that integrates SimCortex through Docker, enabling cortical surface reconstruction and visualization directly within 3D Slicer.

Pretrained Models: Versioned model checkpoints and configurations will be made available through Hugging Face.

Research Data: Supporting datasets, benchmark results, and reproducibility resources will be linked as they become available.

Resources

• Website: https://simcortex.app
• Source Code: https://github.com/Neuro-iX/SimCortex
• Slicer Extension: https://github.com/Kaveh-Moradkhani/SlicerSimCortex

Research and Collaboration

SimCortex is developed as part of academic research on deep learning and medical image analysis at École de technologie supérieure (ÉTS), Montréal, Canada.

We welcome research collaboration, reproducibility efforts, and community contributions.