[vtkusers] How does vtk create volume from dicom CT scans ?

[Mathieu Malaterre]

Nicolas,

[Please CC the vtkusers ML, thanks]

On Tue, Oct 20, 2009 at 9:25 AM, Gerber, Nicolas
<nicolas.gerber at istb.unibe.ch> wrote:
> Thank you very much :)
>
> I'm not sure we are talking about the same thing... or more or less !
>
> I was thinking about the problem of the machine that creates the dicom data and makes different slice thickness among the stack. I think that usually, the people compare the slice thickness given by the dicom tag and the calculated slice thickness between two slices (calculated with image position patient tag that we can rely on). Then they fix a tolerance (I don't know how to fix it, but generally it's between 1e-3 and 1e-5) and if the previous comparison is not inside this tolerance, I don't know what they do !

Please use proper vocabulary to avoid misunderstanding(*). The DICOM
standard refers to Spacing Between Slices and Slice Thickness. Those
are two *very* different measurements. Check post on this list for
clarification if needed.
If you use gdcm:IPPSorter, then -if I understand you- you should be
able to select what is your tolerance. The Spacing Between Slices is
computed from the Image Orientation (Patient) (IOP) and Image Position
(Patient) (IPP), but as you said we need a constant spacing in toolkit
such as VTK (or ITK). And is is up to the user to decide what
tolerance he/she means by 'constant' for a floating point value.

> - Do they popup an error message ?
> - Do they resample the hole volume to have perfect slices ?

If you are missing a slice you'll have to make a decision
(vtkImageReslice being one). But nothing is being done internally by
vtkGDCMImageReader.

> Same case with the rotation of slices that can vary...

Do you mean Gantry tilt ?

> And sorry, but I don't know what is the old storage for CT... it is the old dicom format ? If so, I'm using the new dicom format !

There is no such thing as 'new dicom format'.

> I think I'm entering in the wonderful world of dicom crap :) !

As someone else would say, there is no 'dicom crap', only poor
implementations of the DICOM standard.

Cheers
-- 
Mathieu

(*) Refer to section from PS 3.3 - 2008