Skip to content

resources.md

Latest commit

 

History

History
57 lines (40 loc) · 2.4 KB

resources.md

File metadata and controls

57 lines (40 loc) · 2.4 KB

Resources

TCIA Dataset: Pediatric CT Segmentation

  • Source: TCIA Pediatric CT Segmentation Collection
  • Description: A collection of pediatric CT scans with expert-annotated organ and tumor segmentations, suitable for training and evaluating medical image segmentation models.
  • Metadata:
    • TCIA/meta.csv: Contains patient IDs, scan information, and basic demographic data for each case in the dataset.

TotalSegmentator Dataset

  • Source: TotalSegmentator on Zenodo
  • Description: A large-scale dataset of adult CT scans with comprehensive multi-organ segmentations, designed for general-purpose medical image segmentation tasks.
  • Metadata:
    • TotalSegmentator/meta.csv: Contains scan identifiers, acquisition parameters, and summary statistics for each scan in the dataset.

Using Pretrained nnU-Net Models

We provide two nnU-Net v2 model checkpoints:

  • mixed_model_continual_learning.zip
  • pure_pediatric_model.zip

You can use these as pretrained weights for inference or further fine-tuning with nnU-Net v2.

1. Download the Model Weights

Go to the GitHub Releases page and download the desired zip files. Place them in a directory of your choice (e.g., resources/checkpoints/).

2. Install the Pretrained Model

Use the nnU-Net v2 utility to install the model from the zip file:

# For the mixed model:
nnUNetv2_install_pretrained_model_from_zip resources/checkpoints/mixed_model_continual_learning.zip

# For the pure pediatric model:
nnUNetv2_install_pretrained_model_from_zip resources/checkpoints/pure_pediatric_model.zip

3. Run Inference with the Installed Model

After installation, you can run inference using nnU-Net v2. For example:

nnUNetv2_predict -d <DATASET_ID> -i <INPUT_FOLDER> -o <OUTPUT_FOLDER> -c <CONFIGURATION> -f <FOLD>
  • <DATASET_ID>: The dataset number or name (e.g., 297 for TotalSegmentator, 797 for the mixed dataset).
  • <INPUT_FOLDER>: Folder with your images (NIfTI format).
  • <OUTPUT_FOLDER>: Where predictions will be saved.
  • <CONFIGURATION>: Model configuration (e.g., 3d_fullres, 2d).
  • <FOLD>: Fold number (usually 0, or all for all folds).

Example:

nnUNetv2_predict -d 797 -i imagesTs/ -o predictions/ -c 3d_fullres -f all