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
1.16k stars 499 forks source link

Feature extraction from 2D image #619

Closed hellorp1990 closed 2 years ago

hellorp1990 commented 4 years ago

Hi, I am working with 2D images currently. I can use pyradiomics to generate features. But I have one issue or doubt. My images are in .png format (8bit grayscale). So when I use these 2d images directly, I can have a set of radiomics features.

In some older issues in pyradiomics, I found the moderator was asking to use sitk.joinseries () for 2d images. So I used that and the code is running and generates features as well. But the problem is the features obtained from direct 2d images and after joinseries() are not same. I understood that it may not be same. But my question is which one is the correct way or calculating features, as I will have different results using two different method.

JoostJM commented 4 years ago

Which features are different?

There should be no difference in features. The only special case is shape, for which there is a 3D en 2D equivalent.

In those older issues, joinseries was required, as PyRadiomics did not yet support 2D input, which is not an issue anymore.

rcuocolo commented 4 years ago

I can confirm that since 2D extraction has been officially supported I have found no issues extracting features from truly 2D png image files (exported from an ultrasound scanner).