AortaSeg_2024

Contribution to the AortaSeg Challenge (MICCAI 2024) by lWM.

How to reproduce?

Option 1:

• Download the models (the link contains all the models used for the final submission - after the final retraining): Link
• Save the models to a chosen folder and copy the absolute path
• Navigate to the /reproduce folder
• There is a file reproduce.ipynb. Set the three variables:
  • 'input_path' - direct path to the input CT volume
  • 'output_path' - direct path where the segmentation mask should be saved
  • 'models_path' - path to the directory where the models are saved
• Run the inference by sequentially running all the cells in the notebook. To run the inference you need GPU with at least 10 GB of VRAM.
• If you want to run the inference for numerous files - just introduce for loop to the notebook and iteratve over numerous files.

Option 2:

Directly download and use the Docker used for the final submission: Link

The Docker follows all the conventions from the Grand Challenge platform - use it as a Grand Challenge Algorithm.

Option 3:

Request access to the algorithm on the Grand Challenge platform and test the method directly using the GC infrastructure.

Technical description

The technical description of the proposed method is available at: Link