Can SlicerRadiomics perform Feature selection for 2D imaging?

Operating system:window 10
Slicer version: 4.10.2
Expected behavior:
Actual behavior: Can the SlicerRadiomics perform Feature selection for 2D imaging? And how to set?

Yes. just load a 2D image and Slicer will treat it as a one-slice volume.

Hi Steve!
Thank you for your response. I have other questions. If I load a 2D image, how do I set parameters, such as Resampled voxel size, LoG filter kernel sizes and Wavelet-based features? Or I don’t need to set these parameters.

Best wishes!
Lin Junpeng

I’m not sure, but I don’t think it’s any different than in the 3D case - maybe @JoostJM can comment. I think you

Thank you very much!

Indeed no difference. The only thing that changes is that for resampledPixelSpacing, you need to fill in only 2 instead of 3 values. Be aware that this assumes truly 2D images. If it fails, carefully read the error message, this often contains clues as to where and how it failed.

OK! Thank you! I benefited a lot from your reply!

Hello everyone!
I have a similar issue like @LinJP123 has. I am currently working on 2D ROI from a DICOM volume of CT scan in 3D Slicer. When I try to extract 2D features with voxel resampling as 1,1 and try to add a LoG filter, the system throws error as Wrong dimensionality (2-D) of resampledPixelSpacing, 3-D required.

How can I do pixel resampling and LoG filtering at the same time with 2D images.

Thank you for all your consideration :):slight_smile:

Good question and perfect answers to this question. I appreciate that, thanks! :handshake: