Initial metrics release

Initial release SynthRad2023 Evaluation Metrics

This is the initial release of the SynthRad2023 metrics repository.
In this repository, all metrics that are used to score the submissions by participants is presented.

In particular, during validation the submissions by participants are evaluated based on image similarity.
During testing, the submissions of participants are evaluated based on image similarity and dose delivery.

The image similarity metrics include:

• Mean Squared Error (MSE)
• Peak Signal-to-Noise Ratio (PSNR)
• Structural Similarity (SSIM)

The dose metrics include:

• Gamma pass rate
• Dose mean absolute error (MAE)
• Dose-volume histogram difference (DVH)

All metrics for a specific patient can be computed by running, for example:

    metrics = ImageMetrics()
    ground_truth_path = "path/to/ground_truth.mha"
    predicted_path = "path/to/prediction.mha"
    print(metrics.score_patient(ground_truth_path, predicted_path))