How do we extract the features for each independent object in an image?

I have a tissue image and there are around 100 cells in it. I also have the corresponding mask image which includes those 100 cells. How do I calculate the radiomics features for each cell?

Hi @mozdag

Yes, you can do this using SlicerRadiomics extension module.

  1. Load your tissue image and the corresponding mask as a lablel image in Slicer.

  2. In the SlicerRadiomics module, you can then specify your Input Image Volume (tissue image), and Input Regions (your converted labelimage) and then run the module.


1 Like

Note that if you want to have features separately for the individual cells, you will need to assign a unique label value for each of the cell segmentations.

You can do this using the “Islands” effect of Segment Editor:

Thank you for your response. I won’t be using the Slicer app. I’ll have to use some library such as connected component:

from cc3d import connected_components
labels_out = connected_components(data_mask)

I believe, labels_out is a set of those ROI regions for each cell? Another point is that my images will be 2D. Have any of you got an idea on this?

I still run this code and got error: Attribute Error; image has no attribute “shape”.

1 Like

There are some additional attributes you need to make sure are set, did you check this documentation page: ?

As to splitting connected components and extracting features, check out this issue, which deals with the same problem.

1 Like