NA-MIC Project Weeks

Back to Projects List

Integrating 3D Landmark Detection Models

Key Investigators

• Alexandra Ertl (German Cancer Research Center DKFZ, Germany)

Project Description

We developed a framework for 3D medical landmark detection, based on a U-Net and heatmap regression. The method was evaluated on several public datasets. We want to make these trained models available within a slicer extension.

Objective

Within the extension, users should be able to: 1) Choose and run a model for prediction. 2) Visualize predicted landmarks.

Approach and Plan

No response

Progress and Next Steps

We have adjusted the SLicerNNUnet extension for landmark detection.

Illustrations

No response

Background and References

The respective publication is currently under review for MIDL 2026 (OpenReview submission)