Radiomics: Is it possible to import DICOM image and a binary mask as a segmentation to extract features?

I have a DICOM image along with a binary mask of the segmented region generated by another software. I noticed that Radiomics reads an Input Image Volume and Input Regions. Is there any way to use a volumetric (DICOM) binary mask as an Input Region? Or should I create a contour by, for example, a region growing or thresholding of the binary mask?

PyRadiomics (optionally available through the SlicerRadiomics extension) uses binary masks to identify the regions of interest, so you don’t need to define a contour. The only requirement in PyRadiomics as standalone is that both image and mask are single files, readable by SimpleITK.

If you use the SlicerRadiomics extension, if you load both the image and mask, you can select them (the mask should either be loaded as a labelmap (legacy support) or a segmentation).

1 Like