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

Results Format

The results must be submitted as CSV files, one for each image of the testing set. The name of the CSV file must be the same as the name of the corresponding image file. Example: for image A06_00Aa.tiff, the result file name will be A06_00Aa.csv

CSV File Format for Nuclear Atypia

  • One result CSV file for each image of the testing set at x20 magnification.
  • Each file contains only a single information: the nuclear atypia score (1, 2 or 3) for this image, optionally followed by a confidence degree.
  • The nuclear atypia score and the confidence degree are separated by a comma.
  • Please provide an empty file if the nuclear atypia cannot be evaluated for the image, or if the confidence degree is too low.

Example:

3,0.857438

CSV File Format for Mitosis

  • One result CSV file for each image of the testing set at x40 magnification.
  • Each line of the CSV file gives the coordinates of the centroid of one detected object you consider as mitosis followed by an optional confidence degree.
  • It is up to you to select the best confidence degree threshold and to export only the objects having a confidence degree higher than this threshold.
  • Coordinates and confidence degree are separated by a comma. One separate line for each mitosis.
  • If there are no mitosis detected in an image, provide an empty CSV file for that image.

The format should look like this:

mitosis_1_x_coord,mitosis_1_y_coord,mitosis_1_confidence_degree
mitosis_2_x_coord,mitosis_2_y_coord,mitosis_2_confidence_degree

Example:

375,784,0.952342
878,13,0.899445

Submit your results by email: ludovic17.roux@gmail.com

 

Metrics for Nuclear Atypia

The goal is to give the correct nuclear atypia score (1, 2 or 3) on each frame at X20 magnification.

A correct score score will give one point.

An incorrect score will give zero point if the absolute value of difference between the proposed score and the ground truth score is equal to one. Example: proposed score = 2, ground truth score = 3. ABS(2-3) = 1

An incorrect score will give a negative point (-1) if the absolute value of difference between the proposed score and the ground truth score is equal to 2. Example: proposed score = 1, ground truth score = 3. ABS(1-3) = 2

The contestants will be ranked according to the sum of points.

Metrics for Detection of Mitosis

The goal is to give the list of all mitosis visible on each frame at X40 magnification. A candidate mitosis would be accepted as correctly detected if the coordinates of its centroid are within a range of 8 µm from the centroid of a ground truth mitosis.

The metrics for detection of mitosis are as follows:

  • D = number of mitosis detected (centroid within a range of 8 µm from the centroid of a ground truth mitosis)
  • TP = number of True Positives, that is the number of mitosis that are ground truth mitosis among the D mitosis detected
  • FP = number of False Positives, that is the number of mitosis that are not ground truth mitosis among the D mitosis detected
  • FN = number of False Negatives, that is the number of ground truth mitosis that have not been detected
  • recall (sensitivity) = TP / (TP+FN)
  • precision (positive predictive value) = TP / (TP+FP)
  • F-measure = 2 * (precision * sensitivity) / (precision + sensitivity)

The contestants will be ranked according to the F-measure.


Consortium for Open Medical Image Computing © 2012-