search

MStarmans91 / WORC

Workflow for Optimal Radiomics Classification
Other
69 stars 19 forks source link

Use features_from_this_directory #75

Closed bgielen closed 1 year ago

bgielen commented 1 year ago

Hi!

I try to use WORC using the features that I already extracted using PyRadiomics. I made hdf5 files for every patient containing the features using the example that you provided. However, when I run WORC it says that the first column should be patient, do I add the patient number as a feature value and feature label?

image

MStarmans91 commented 1 year ago

Can you please send the complete error message? In the future, please use the bug report template to give more information on your problem, that makes it easier to help.

I don't think the issue is related to your feature files, as those do not need a patient / object name in there per se, as also illustrated in the example code you showed above from the FAQ. Rather, the error has to do with your label file, i.e., the file in which you supply the ground truth for your objects / patients to WORC. In that file, to be able to match the labels with your images / segmentations / features, the first column in there should be headed "Patient", so WORC will use the keys in that column for matching. For an example, see https://github.com/MStarmans91/WORCTutorial/blob/master/Data/Examplefiles/pinfo_HN.csv.

bgielen commented 1 year ago

Yes you are right, it was my label file. Thanks for you response! :)

MStarmans91 commented 1 year ago

No problem, glad we found the problem.