HomeWhy Challenges?All ChallengesCreate Your Own ChallengeContributorsForum
Sign in / Register

Submission of the Results

Remember that there are two sub-challenges, one regarding polyp localization (where you should use ETIS-Larib database) and another on polyp localization (ASU-Mayo Testing Video Database is the one to use here).

In order us to accurately evaluate your methods, we ask you to provide the results in the following format:

  • Polyp Localization: 2 files. 1 CSV file and 1 (optional) compressed file (.rar or .zip).
    • The CSV file should include, for each frame of ETIS-Larib database, the following information: Each row represent a frame and, for each row, the first column should include the number of frame, the second the x-position of the polyp, the third column the y-position of the polyp.
      • Notes:
        • we assume that for each polyp you only provide one position, such as the center or the point in the image with highest likelihood to be part of the polyp
        • If you give more than one potential polyp location per image you should present each location in a new row in the csv file (e.g. 1st row 1 250 400, 2nd row 1 500 650)
        • if your algorithm has some kind of confidence value to give the output, you can use a 4th column in the csv file to provide this confidence parameter value so we can draw ROC and FROC curves to evaluate the different configurations of your methods
    • The compressed file should contain the output of your method so we can evaluate your output. In this context, we expect the output of your method to be some kind of likelihood or energy map which, once thresholded, can lead to obtain polyp locations.
      • Notes:
        • the output of the method should be either on .mat , .bmp or .tif format. You can use TIF format to store the results if you are using C/C++.
        • even if your method has different confidence parameter, just send us output results of the best configuration (under your opinion)
    • Both CSV file and .rar output file should be included in one final compressed file named polyplocalization_TEAMNAME(.rar o .zip) (eg, polyplocalization_UAB) and sent to polypchallengemiccai@gmail.com BEFORE SEPTEMBER 27TH 23H59 CET
  • Polyp Detection: 1 compressed file containing one CSV file for each of the videos, following videos naming (e.g. testVD1.csv for the first video)
    • The CSV file should include, for each frame of ASU-Mayo database, the following information: Each row represent a frame in the video and, for each row, the first column should include the number of video, the second the number of frame within the video, the third  x-position of the polyp, the fourth column the y-position of the polyp.
      • Notes:
        • we assume that for each polyp you only provide one position, such as the center or the point in the image with highest likelihood to be part of the polyp
        • If you give more than one potential polyp location per image you should present each location in a new row in the csv file (e.g. first video 1st row 1 1 250 400, 2nd row 1 1 500 650)
        • in case your method does not provide a polyp location for a given frame, the csv file should present the following information for a given row (eg. frame 150 of first video has no polyp under your method's opinion, the row goes like this 1 150 0 0)
        • if your algorithm has some kind of confidence value to give the output, you can use a 5th column in the csv file to provide this confidence parameter value so we can draw ROC and FROC curves to evaluate the different configurations of your methods)
    • The compressed file should be named polypdetection_TEAMNAME(.rar o .zip) (eg, polypdetection_UAB) and sent to polypchallengemiccai@gmail.com BEFORE SEPTEMBER 27TH 23H59 CET
  • You can submit results for both localization or detection or to any of them.

Consortium for Open Medical Image Computing © 2012-