editing CT lung metadata to create new sequence of cuts

[murray baron]

  As a rheumatologist I am pretty naive about this area. 

I want to have a large number of high res lung CTs read by computer. The programs require contiguous slices but many of the CTs available to me have only 30 slices. I thought that perhaps I could just expand that to look like there are 250 slices by duplicating each of the 30 cuts  8 times and hope that if those were thought to be contiguous then reading that set of dicom files might approximate pretty well a true 250 set of cuts from that patient. 

Now I understand  that to do that would require not only duplicating each of the 30 files about 8 times but then changing the metadata of each duplicated file so that it looks like a unique cut and the whole 250 new files look like they come form one true CT scan.. 

Now I think that to do that would require not only duplicating each of the 30 files about 8 times but then changing the metadata of each duplicated file so that it looks like a unique cut. 

Do you think this would be possible  ?   Is there anyone out there who could do this for me?  willing to pay for help.

[Stuart Swerdloff]

You will need to interpolate the values in the image position patient ( https://dicom.innolitics.com/ciods/ct-image/image-plane/00200032 )

If the CT is "perfectly lined up on the Z axis" then you need to get the 3rd value in image position patient and interpolate between the adjacent original CT.

It's certainly possible.  It would be very time consuming to do it manually.

Doing it programatically is entirely possible.  packages like pydicom can be used in python scripts and the amount of code involved is under 100 lines of code for the simple case.

You might not get what you are hoping for though.

If you have large gaps between slices (e.g. spaced by 3 cm, but each CT was actually only 1mm thick), it won't change what it looks like, it will be a sharp transition every Nth slice still.  If the slices are just very thick (e.g. 1cm thick, and space by 1 cm), then it will be just as chunky.

If what your situation is calling for is that your voxels need to be isotropic (the thickness is the same as the pixel size in a single slice), then you have to "re-slice" your volume, and the hack you mention isn't going to do the job.  

Maybe a tool like 3D slicer would be helpful: https://slicer.readthedocs.io/en/latest/user_guide/modules/resamplescalarvectordwivolume.html

Best of luck (sorry, I'm overwhelmed with my day job to moonlight...)