Analysis for the paper

[sandrinebedard]

Description

This issue describes the analysis we want to include in the paper.

Inference on other datasets

• [x] Run sct_deepseg_sc 2d, 3d and propseg on sci-colorado and basel-mp2rage
• [x] Run inference on EPI data (to be determined with dataset)

CSA analysis

• [x] Run sct_deepseg_sc 2d, 3d and propseg on spine-generic and compute csa (add to csa violin plots)
• [x] CSA T1w vs T2w 6X6 for final model, nnUnet, sct_deepseg, GT (other??)

[valosekj]

• [ ] Run sct_deepseg_sc 2d, 3d and propseg on sci-colorado and basel-mp2rage

You can reuse comparison_with_other_methods_sc.sh and generate_figure.py scripts from the SCIseg project. Briefly, the comparison_with_other_methods_sc.sh script runs sct_propseg, sct_deepseg_sc 2d, and sct_deepseg_sc 3d together with a proposed model (in SCIseg, we use nnUNet, thus this part has to be updated for the contrast-agnostic). Then, the script computes ANIMA metrics between the predictions and GT. The script also tracks the execution time for each method. Then, the generate_figures.py script is used to fetch the ANIMA outputs and create raincloud plots.

[naga-karthik]

@sandrinebedard as far as I know, all the above tasks have been completed. Can I close this issue now?

[sandrinebedard]

Yes we can close!