search

AIM-Harvard / pyradiomics

Open-source python package for the extraction of Radiomics features from 2D and 3D images and binary masks. Support: https://discourse.slicer.org/c/community/radiomics
http://pyradiomics.readthedocs.io/
BSD 3-Clause "New" or "Revised" License
1.14k stars 492 forks source link

[FEAT EXTRACTION] Pyradiomics vs TCIA-GBM Datasets #543

Open GitHub-username-hyphen opened 4 years ago

GitHub-username-hyphen commented 4 years ago

Hi all,

New to programming but looking into using Pyradiomics for feature extraction. I am comparing the features generated from Pyradiomics to those released by the TCIA for glioblastomas.

Describe the bug Images/Labels were downloaded from TCIA. We changed the labels so that they were binary values for the different tumor parts.

Comparison of feature values from TCIA radiomics data do not match those generated by Pyradiomics, despite running on the same image/label sets.

PyRadiomics configuration None

PyRadiomics log file N/A

To Reproduce See above bug description

Version (please complete the following information):

Additional context Add any other context about your question here.

fedorov commented 4 years ago

Radiomics features are sensitive to a number of configuration parameters, and, in general, and unfortunately, it is hard in practice to achieve consistent feature values across different implementations. You may want to contact the authors of the feature set you are using in comparison to get access to their implementation. First step would be to check if you can obtain the same feature values as the authors reported using their own software. If they used pyradiomics, you should ask them about the version of pyradiomics they used and the exact parameters file they used.