Skip to content

ManskeLab/AutomatedErosionPipeline

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

17 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Atlas-Guided Erosion Segmentation

Prerequisites

  • Install our in-house customized nnUNet
    • https://github.com/ManskeLab/multichannel_attention_nnUNet.git
  • Soft-tissue–stripped MCP joint images from HR-pQCT or SR-CBCT
    • With MC and PP bone masks
  • Atlas, located in:
    • ATLAS_MC=/work/manske_lab/images/hrpqct/mcp_atlasses/$MCP/atlas_mc_aligned.nii.gz
    • ATLAS_PP=/work/manske_lab/images/hrpqct/mcp_atlasses/$MCP/atlas_pp.nii.gz

Model Checkpoints

MCP Stripping

  • /work/manske_lab/jobs/mcp_nnunet/nnUNet_results/Dataset001_hand/nnUNetTrainer__nnUNetPlans__3d_fullres/fold_all/checkpoint_best.pth

Bone edge mask

  • /work/manske_lab/jobs/mcp_edge/nnUNet_results/Dataset001_mcp/nnUNetTrainer__nnUNetPlans__3d_fullres/fold_all/checkpoint_final.pth

Bone closed edge mask

  • /work/manske_lab/jobs/mcp_closed_edge/nnUNet_results/Dataset001_mcp/nnUNetTrainerWithAttention__nnUNetPlans__3d_fullres/fold_all/checkpoint_final.pth

Erosion segmentation

  • /work/manske_lab/jobs/mcp_erosion/nnUNet_results/Dataset001_mcp/nnUNetTrainerWithAttention__nnUNetPlans__3d_fullres/fold_all/checkpoint_final.pth

Pipeline Steps

1. Strip MCP Image

sbatch strip_mcp/predict_mcp.sh \
  $MCP_IMAGE \
  $OUTPUT_DIR

Use the stripped image (output of above script $OUTPUT_DIR/stripped_<input_image>.nii.gz) as MCP_IMAGE.

2. Run both edge segmentation scripts to get closed edge and bone edge masks

sbatch edge_masking/predict_edge.sh \
  $MCP_IMAGE \
  $OUTPUT_DIR

3. Get closed edge segmentation

sbatch edge_masking/predict_closed_edge.sh \
  $MCP_IMAGE \
  $EDGE_MASK \
  $OUTPUT_DIR

4. Run registrations

sbatch registration/reg_atlas_to_img.sh \
  $MCP_IMAGE \
  $MC_OR_PP_BONE_MASK \
  $ATLAS \
  $OUTPUT_DIR \
  $BONE_NAME

5. Get candidate erosions

sbatch candidate_erosions/get_candidate_erosions.sh \
  $MCP_IMAGE \
  $REGISTERED_ATLAS \
  $EDGE_MASK \
  $CLOSED_EDGE_MASK \
  $OUTPUT_DIR

6. Go through erosions candidates and create a folder of erosion ROIs.

7. Run the folder through erosion nnUNet.

sbatch erosion_model/batch_predict.sh \
  $PATH_TO_EROSIONS_ROI_DIR

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors