re-slice dicom for ITK-snap project

[Karl Lindberg]

Hey!

I'm pretty new to ITK-snap. Have just finished watching the training videos.

I'm going to do a small project where I compare different ways of measuring pituitary gland adenoma volume (and size through only longest diameter which is often used clinically) depending on the slice thickness from the hospitals in the regions MRIs.

I have 18 sets of axial slices with contrast thats has 1mm thickness and 1mm spacing that was taken pre-operative for the neurosurgeons to navigate pre-op (it says voxel size 1*1*1 loaded into snap, I assume this is 1*1*1mm).

I'd like help to re-slice these sets to sets of 3mm thickness+spacing. Then I will do my comparisons between the different sets. It must be doable right? I assume I somehow can do it in convert3D but its above my technical expertise hah :) Is there anyway easier? Help is much appreciated!

--- --- ---

Bonus questions:

Do I understand it correctly that the DICOM sets with, say 3mm slice thickness which is pretty std for the regular MR-scans, and corresponding voxel intensity that intensity is actually a mean value from that space? If so it seems like a lot of information is lost, especially for smaller structures?

Would re-slicing my 1mm sets produce the exact same result as getting 3mm sets at the time of the MRI-scan?

Is there any ruler to measure distances (I'm interested in the longest diameter of the tumor and also the diameters that correspond to the three planes)? I can't find it. Maybe I have to use another application?

---

Regards

Karl Lindberg, med-student, Sahlgrenska University Hospital

[Karl Lindberg]

I think I solved it!

From ITK-snap i saved my dicom-series as a nifti file, i.e. converting it (this seems to have anonymized it too, there's not patient info in the meta-data, is that correct? Because in training video #2 I think it's said that converting from ITK-snap doesn't anonymize) . Then I used the resample command in convert3D putting it to 1*1*3mm.

This would be correct right? It looks exactly like what i wanted loading it up in ITK-snap. My question now would be what interpolation I should use for having the data as close to getting it with 3mm slices directly at the time of the scan. I guess linear since it's the default?

[Paul Yushkevich]

Hi Karl

Converting to nifti will strip all patient info from the header but will not remove facial features, hence it's not full anonymization for brain images. Otherwise it's fine!

Your approach is good, I would apply a small amount of smoothing along the z axis (0.8 mm) and then use linear interpolation. When sub sampling like you are, taking every third slice, linear interpolation will be identical to nearest neighbor.  What the scanner does is more complex but what I suggest is a simple approximation. 

Paul

--
You received this message because you are subscribed to the Google Groups "itksnap-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to itksnap-user...@googlegroups.com.
To post to this group, send email to itksna...@googlegroups.com.
Visit this group at http://groups.google.com/group/itksnap-users.
For more options, visit https://groups.google.com/d/optout.

--

Paul A. Yushkevich, Ph.D.
Associate Professor
Penn Image Computing and Science Laboratory
Department of Radiology
University of Pennsylvania

[Karl Lindberg]

Hi Paul!

Thank you, a simple approximation is what I want so I'll go with that :)