| 2021 | 2022 |
|---|---|
| link | link |
This series of challenges is split into two phases:
- Validation (data) phase
- Testing (data) phase
The Validation phase will utilize the CWL workflow,
validation-phase-workflow.cwl, accepting a group of prediction files and
scoring them against the ground truth files with CaPTk.
The Testing phase will utilize the CWL workflow,
testing-phase-workflow.cwl, accepting a Docker model to first infer the
predictions, then scoring those predictions with CaPTk.
Metrics are returned in a JSON file, results.json and are the same for
both workflows:
- Dice similarity coefficient
- Hausdorff distance, 95%
- Sensitivity
- Specificity
- Precision **new for 2022**