[Rtk-users] Question regarding RTK

[Simon Rit]

Hi Wen,
The offsets values depend on where the origin of the coordinate system
of your projections. With all angles at 0, the offsets give the
coordinates in the projection of the ray passing through the source
and the isocenter. There are some explanations on the wiki for the
geometry, it might help.
You might want to try using rtk::Reg23ProjectionGeometry. It is a
child of the other geometry with a different parameterization which
uses the row and column vectors of your projection images in the
coordinate system of the tomography to be reconstructed. It is
probably quite straightforward to extract this from your matrices. Let
us know if you can't figure it out, I will have a look then at your
data.
Good luck!
Simon

On Wed, Feb 6, 2013 at 9:41 PM, Wen P. Liu <wen.p.liu at gmail.com> wrote:
> Hi Simon,
>
> Thanks for the quick reply and the positivity! I was able to reconstruct
> (using SART) with raw and some simple rawToAttenuation projections (see .png
> in attachment). Through trial and error I found that setting the offset to
> image center was key. The dislocations of the sphere in reconstruction is
> probably due to my assumption of circular trajectories. I have the 3Dto2D
> projection matrices from Siemens for each angle that would project isocenter
> coordinates to detector but found that RTK is expecting the transformation
> at the source and detector (3DCircularProjectionGeometry->AddProjection()).
> If I wanted to inject these projection matrices directly into reconstruction
> what steps would you recommend? Thanks very much for your time. Just FYI I'm
> going to email the mailing list (minus the data) with the same question.
>
> Regards,
> Wen
>
>
> On Sat, Feb 2, 2013 at 10:47 AM, Simon Rit <simon.rit at creatis.insa-lyon.fr>
> wrote:
>>
>> Dear Wen,
>> I don't mind, that's why the project is open source! I think raw
>> images are already available via itk, you just have to create a mhd
>> header that points to that raw file. Look here for more information:
>> http://www.itk.org/Wiki/ITK/File_Formats. However, if there is some
>> information in a header or a joint file, you may want to do specific
>> I/O objects for Siemens. This is explained on the ITK documentation.
>> In my experience, it's hard to obtain information from constructor and
>> you have to guess or measure it yourself. In some cases, you even want
>> to do better than the constructor. However, the more information you
>> can get from them, the better. For example, I know I do something
>> close to the constructor for Elekta but I don't know how good is my
>> Varian stuff. I get the geometry from their files, it's usually simple
>> to understand. We may be able to help you if you can provide a sample
>> acquisition.
>> Discussions for RTK are done via the mailing list (cc), you can
>> suscribe on the website. There is also a wiki (click on help). There
>> are examples you can use there and some data are available on the RTK
>> midas community.
>> I hope this helps. Let me know if I can do more,
>> Simon
>>
>> On Sat, Feb 2, 2013 at 5:45 AM, Wen Liu <wen.p.liu at jhu.edu> wrote:
>> > Hi Simon,
>> >
>> > I hope you don't mind a random email regarding RTK. I'm a CS PhD student
>> > at
>> > Johns Hopkins working on integrating CBCT with medical robotics. We have
>> > a
>> > Siemen's C-arm system that provides raw (16 bit unsigned short) images
>> > that
>> > I wish to process and reconstruction using RTK. I've noticed your
>> > authorship
>> > on several classes (rtkImageXxxx, rtkVarianxxx, etc) which process
>> > platform
>> > specific outputs into projections/attuentation images that RTK can
>> > reconstruct. What would you recommend as my next steps? Port in raw
>> > images
>> > to RTK? Consult with Siemen's regarding their calibration data to map
>> > raw to
>> > attenuation? Is this typically done with lookup tables, i.e. based on
>> > geometric and calibration?
>> >
>> > Also if there are literature, discussion groups for RTK and these
>> > topics,
>> > please point me in the right direction!
>> >
>> > Thanks very much for your time and help.
>> >
>> > Regards,
>> > Wen
>> >
>> > https://www.lcsr.jhu.edu/User:Wliu25
>
>