Queries regarding Evaluation and metrics

Dear Organizers,

I have few questions regarding evaluation and metric procedure.

• How do you aggregate the score for all images. I mean the final score which is average of DSC and JSC that we calculate for each image, how are they combined to get final score and ranking.

• If a image is fully negative, and prediction is also fully negative what will be the score in that case for that image.

Thanks

Dear Participants,

The score for each test phase is calculated using the following equation:

Score_phase = 0.5 * DSC_phase + 0.5 * JSC_phase

The final score is then computed as:

Final Score = 0.2 * Score_preliminary_test + 0.8 * Score_final_test

For more details on the model and evaluation process, please visit: https://cosas.grand-challenge.org/evaluation-ranking/

Additionally, if an image is fully negative and the prediction is also fully negative, the score for that image will be 1.

Best regards,

Xi Long

Dear Xi,

May I ask what happens if the image is fully negative but the model predicts, let's say, a single pixel as positive? Does that count as DSC=0?

Cheers,

Adrian