> Pre-processing

Hi all,

A question about pre-processing, will the images be pre-processed similar to the Monai Tutorial on the challenge or will be using original DICOM images? Information about this is confusing, the github tutorial says it will be pre-processed (and in fixed folders "DATASET_ROOT" and "DATALIST_PREFIX"), where as in the challenge web it states "Anonymized DICOM images will be available for the challenge. Data will not be provided for download. No pre-processing will be performed.". Could you please clarify?

Thank you!


Posted by: robertmarti @ July 22, 2022, 3:15 p.m.

Hello, thanks for your feedback. To clarify, the challenge data is in the same format (i.e. folder structure, file type, basic preprocessing steps) as the sample data used in the breast density challenge MONAI tutorial. For more information on this data format, please see here: https://github.com/Project-MONAI/tutorials/tree/main/federated_learning/breast_density_challenge/data#:~:text=code/pt/utils/preprocess_dicomdir.py.

If your submission runs on the sample data and you were mindful of parts of MONAI tutorial implementation that should not be changed (indicated with comments), then your submission should work with the challenge data.

By the way, you should have received an email from breastdensityfl2022@gmail.com with submission instructions.

Posted by: challenge-organizer @ July 25, 2022, 3:45 p.m.
Post in this thread