Project: Integrate MRSegmentator and DeepSpA models into mhub.ai

[FJDorfner]

Draft Status

Draft - team will hold off on page creation

Category

Segmentation / Classification / Landmarking

Presenter Location

In-person

Key Investigators

• Felix Dorfner (Charité Universitätsmedizin Berlin, Germany)
• Hartmut Häntze (Charité Universitätsmedizin Berlin, Germany)
• Keno Bressem (Technical University of Munich, Germany)

Project Description

This project will aim to integrate two models into mhub.ai.

1. MRSegmentator:

• Is a segmentation model that can accurately segment 40 organs and structures in human MRI scans of the abdominal, pelvic and thorax regions. The model works on different sequence types, including T1- and T2-weighted, Dixon sequences and even CT images.
• Paper: https://arxiv.org/abs/2405.06463

2. DeepSpA:

• Is classification model that incorporates anatomical awareness to detect radiographic sacroiliitis. Detecting radiographic sacroiliitis plays an essential role in diagnosing and classifying axial Spondyloarthritis (axSpA).
• Paper: https://arxiv.org/abs/2405.07369

Objective

1. Objective: Working implementation of MRSegmentator into the mhub.ai platform.
2. Objective: Working implementation of DeepSpA into the mhub.ai platform

Approach and Plan

TBD

Progress and Next Steps

TBD

Illustrations

Illustrations of both models can be seen on their respective GitHub pages, linked below:

Background and References

MRSegmentator:

• Code: https://github.com/hhaentze/MRSegmentator
• Paper: https://arxiv.org/abs/2405.06463

DeepSpA:

• Code: https://github.com/FJDorfner/Anatomy-Aware-Classification-axSpA
• Paper: https://arxiv.org/abs/2405.07369

[fedorov]

Very cool! I would love to try this out on some of the images from Imaging Data Commons! I would love to join this project as a tester!

We also have been working on a SynthSeg-based method for MRI segmentation (see preprint here: https://arxiv.org/abs/2403.15609). I think @deepakri201 and @ccosmin97 should be interested to learn more about your approach.

[FJDorfner]

Thank you for your interest, it would be great to have you as a tester!

This sounds very interesting. I would like to put you in touch with @hhaentze for that who actually developed the MRSegmentator!

[surajpaib]

@FJDorfner Happy to help with this effort as well. I think having some tutorials at the end, say, running the MRSeg on CT & MR images from IDC could be very nice!

[FJDorfner]

Thank you @surajpaib, I think that would be great!

[sjh26]

Hi @FJDorfner , this is currently marked as a draft. If it is ready to go please let us know in a comment and we will get the page created

[FJDorfner]

Hi @sjh26, thank you very much! The proposal should be ready to go.

[github-actions]

Project Page Pull Request Creation

✅ COMPLETED: See #1182