---
license: apache-2.0
pipeline_tag: image-segmentation
---

# WORD Dataset Queries and Analysis Results

This repository contains queries and analysis results for the **WORD** dataset, as presented in the papers:

- **Finally Outshining the Random Baseline: A Simple and Effective Solution for Active Learning in 3D Biomedical Imaging** ([Arxiv](https://arxiv.org/abs/2601.13677))
- **nnActive: A Framework for Evaluation of Active Learning in 3D Biomedical Segmentation** ([Arxiv](https://arxiv.org/abs/2511.19183))

Official Code: [GitHub - MIC-DKFZ/nnActive](https://github.com/MIC-DKFZ/nnActive)

## Summary
Active learning (AL) has the potential to drastically reduce annotation costs in 3D biomedical image segmentation. This work introduces **Class-stratified Scheduled Power Predictive Entropy (ClaSP PE)**, a query strategy that addresses class imbalance and redundancy. Within the **nnActive** framework, the authors demonstrate that ClaSP PE consistently outperforms improved random baselines in terms of both performance and annotation efficiency in realistic scenarios.

## Sample Usage

To setup and run experiments using the `nnactive` framework, you can use the following CLI commands:

```bash
# Setup an experiment
nnactive setup_experiment --experiment word__tr-nnActiveTrainer_500epochs__patch-29_74_87__sb-random-label2-all-classes__sbs-800__qs-800__unc-random__seed-12348

# Run the experiment
nnactive run_experiment --experiment word__tr-nnActiveTrainer_500epochs__patch-29_74_87__sb-random-label2-all-classes__sbs-800__qs-800__unc-random__seed-12348
```

To analyze existing experiments:

```bash
nnactive analyze_experiments --base_path $nnActive_results --raw_path $nnActive_data --output_path {OUTPUT_PATH}
```

## Citation
If you use these results or the nnActive framework, please cite the following work:

```bibtex
@article{luth2026finally,
    title={Finally Outshining the Random Baseline: A Simple and Effective Solution for Active Learning in 3D Biomedical Imaging},
    author={Carsten T. L{\"u}th and Jeremias Traub and Kim-Celine Kahl and Till J. Bungert and Lukas Klein and Lars Kr{\"a}mer and Paul F Jaeger and Klaus Maier-Hein and Fabian Isensee},
    journal={Transactions on Machine Learning Research},
    year={2026},
    url={https://openreview.net/forum?id=UamXueEaYW},
}

@article{luth2025nnactive,
    title={nnActive: A Framework for Evaluation of Active Learning in 3D Biomedical Segmentation},
    author={Carsten T. L{\"u}th and Jeremias Traub and Kim-Celine Kahl and Till J. Bungert and Lukas Klein and Lars Kr{\"a}mer and Paul F Jaeger and Fabian Isensee and Klaus Maier-Hein},
    journal={Transactions on Machine Learning Research},
    year={2025},
    url={https://openreview.net/forum?id=AJAnmRLJjJ},
}
```