Advice on scan acquisition and/or volume processing

I am having issues with generating a smooth segment with sufficient enough resolution to serve my purposes. I have a skull scanned in a medical setting. The Volumes module reports the dimensions are 512x512x44 and the spacing is 0.488281mm x 0.488281mm x 5mm. Based on similar support threads, this seems to be related to the fact that this is anisotropic (?). I have found support threads that seem to indicate that this may be an intrinsic property of the scan related to common radiology strategies to reduce radiation/scan time, but (unfortunately for me) results in an anisotropic scan which is not useful for segmentation purposes in 3Dslicer. Here is my question:

Can one somehow fix this problem by using the “Resample Scalar Volume” module and/or the “Crop Volume” module (or some other technique)? Based on feedback on a similar thread, I attempted using the Crop Volume module, selecting “Isotropic spacing” and “B-spline” (this resulted in spacing: 0.48mmx0.48mmx0.48mm & dimensions: 512x512x450), but unfortunately the resulting segment is still isn’t great (see green segment).

I really enjoy using 3DSlicer and I have found there to be many potential research opportunities using patient scans; however, I frequently find this to be a hurdle in obtaining segments which are morphologically accurate enough for research purposes. I suppose for scans that I acquire myself I can simply request an isometric scan(?); however, this may not help with other databases of patient scan data.

Any help or advice would be very greatly appreciated. Also, I would be very interested in learning more from resources which go into further detail on this subject so I can develop better strategies for scan acquisition and/or scan processing prior to segmentation etc. THANK YOU SO MUCH :slight_smile:

Unfortunately you have 10 times less information in the Z plane (0.48mm vs 5mm). Interpolation techniques are not going to give you what is not present in the data. that’s reality of working with clinical data for research projects.

If you can find datasets, 1-2mm slice spacing, then you will have much better looking interpolated data.

1 Like

Understood! Thank you, Murat!

I wanted to exhaust my options to see if I could get a “good enough” segment from this scan, but it seems very clear that a difference of this magnitude simply can’t be reconciled for my purposes. Thank you very much, this information will greatly inform my strategies moving forward.

With gratitude,
Jon