Question regarding converting a model back to a segmentation node

Support
1

Hello,

My question pertains to the conversion of a model back to a segmentation.

Currently i want to overlay 2 CT scans to calculate the dice score between 2 femoral segmentations (preoperative and postoperative).
These CT’s are not in the same orientation and do not overlap.

I first segment both, convert them to a model and use SlicerIGT’s model registration to overlap these models. Then convert them back to a segmentation node and us slicerRT segment comparison to compute the dice score and hausdorff distances.

If i return both models back to a segmentation node, will there be a loss in quality or will the original segmentation stay intact?

Any suggestion to my way of approaching this would also be quite welcome.

Thank you in advance.

2

You wouldn’t need to convert the models back to segmentations. You can just apply the transform to the segmentations and calculate your metrics in binary labelmap space.

1 Like
3

Truly overcomplicated it. Thank you, such a simple solution.