SeqSeg, From New Dataset to Training to Inference

A tutorial on setting up SeqSeg for medical image segmentation

1. Data Preparation and Structure

The first step in using SeqSeg is to preprocess your data.

We require the following data:

• A directory containing the images named images
• A directory containing the masks named truths
• A directory containing the centerlines named centerlines
  • as .vtp files

A few things to note:

• The images and masks and centerlines should have the same name
• If you need to extract centerlines from masks, you can use the SeqSeg/centerlines.py script, or use VMTK or other tools
  • Centerlines must contain radius information in the .vtp file
• Make sure that the masks, images and centerlines align correctly, for example open and view together in an image viewer e.g. Paraview
• Make sure the images contain origin, spacing and direction information in the metadata
  • This is important for correct alignment before centerline extraction

2. Data Preprocessing (for Training)

The next step is to preprocess the data for training. SeqSeg requires a model trained on local patches, so we need to extract patches from the images and masks based on centerlines.

The repository for this is vascular-segment-sampler. Use main_with_nnunet.py to extract patches and convert them to nnU-Net format in one step. First, edit the configuration YAML in config/ (e.g. config/global.yaml) so it matches your dataset, then run:

python3 main_with_nnunet.py \
    --config_name global \
    --data_dir /path/to/data \
    --outdir ./extracted_data/ \
    --num_cores 4 \
    --modality CT \
    --nnunet_name AORTAS \
    --nnunet_dataset_number 1

Key arguments:

• --config_name — name of the configuration file in config/ (without .yaml)
• --data_dir — directory containing images/, truths/, and centerlines/
• --outdir — where extracted patches and the nnU-Net dataset are written (default: ./extracted_data/)
• --modality — imaging modality (CT, MR, or comma-separated, e.g. CT,MR)
• --nnunet_name — dataset name for nnU-Net (default: AORTAS)
• --nnunet_dataset_number — nnU-Net dataset number (default: 1)

For a quick test on a subset of cases:

python3 main_with_nnunet.py \
    --config_name global \
    --data_dir /path/to/data \
    --outdir ./extracted_data/ \
    --modality CT \
    --nnunet_name AORTAS \
    --nnunet_dataset_number 1 \
    --max_samples 100 \
    --testing

The new data can be output anywhere, but we recommend writing directly into the nnU-Net Raw directory (or copying the resulting DatasetXXX_* folder there).

3. Training

The next step is to train the model with nnU-Net (see the documentation for more details). Make sure nnUNet_raw, nnUNet_preprocessed, and nnUNet_results are set, and that the dataset from step 2 is in nnUNet_raw.

Preprocessing

Run fingerprint extraction, experiment planning, and preprocessing:

nnUNetv2_plan_and_preprocess -d 1 --verify_dataset_integrity

Replace 1 with the --nnunet_dataset_number you used above. Use --verify_dataset_integrity the first time you run this. To preprocess only a specific configuration:

nnUNetv2_plan_and_preprocess -d 1 -c 3d_fullres

Training

Train a fold with:

nnUNetv2_train DATASET_NAME_OR_ID CONFIGURATION FOLD

For example, with dataset 1 (Dataset001_AORTAS), train fold 0 of a 3D full-resolution model:

nnUNetv2_train 1 3d_fullres 0

Repeat for folds 0–4 (or train fold all for a single model on all training cases). Other common configurations:

nnUNetv2_train 1 2d 0
nnUNetv2_train 1 3d_lowres 0
nnUNetv2_train 1 3d_fullres 0

Add --npz if you plan to use nnUNetv2_find_best_configuration later. Resume an interrupted run with --c.

4. Inference

The final step is to run SeqSeg inference on new data. This is done with the SeqSeg/seqseg.py script. You need direct access to the directory containing the images and seed points, and another containing the trained model weights.