Dear All,

Thanks to the notification of a participant, we realized that one part of the script that we were using to compute the FROC curve was not completely correct.

The bug involved the situation that predictions were not filtered based on the thresholds when there were no ground truth points in an ROI, which artificially increased the number of false positives and caused a shift at the start of the FROC curve. We have fixed the bug now and thoroughly tested the FROC code again.

We have updated the evaluation script on GitHub link, and we are going to update all scores within the next few hours.

We want to thank the participants for their active discussion and questions in the forum, which led to fixing this problem, and want to encourage them to continue with this community-driven approach

To allow participants to potentially re-calibrate their scores based on this update, we have decided to increase the number of submissions for L1 to 3 per week.

We are looking forward to the remaining of this challenge!

Best wishes, TIGER organization team