[Rtk-users] Cone Beam Reconstruction using Open RTK

[Simon Rit]

Hi,
At first glance, no! The first step is indeed to check that you correctly
read the projections, i.e. that your images are converted to line integrals
(if needed). I would write to disk what you pass to FDK (with itk.imwrite)
and check, e.g., that air is at 0.
Simon

On Tue, Jan 24, 2023 at 12:58 AM Keven Qiu <keven.qiu at ucalgary.ca> wrote:

> Hello Simon,
>
> We are trying to use OpenRTK to reconstruct some knee images from cone
> beam CT. We have managed to run the example script in Python but haven't
> been able to reconstruct our knee images. We have all the geometry
> parameters but are still struggling to get it to work. Here is our
> pseudocode below:
> ```
> Load projection images from DICOM series using
> rtk.ProjectionsReader[ImageType] where ImageType = itk.Image[itk.F, 3]
>
> We construct geometry object with SID, SDD, and gantry angle, and source
> offset X/Y, as well as the number of projections
>
> We create a blank image
>
> We pass the projections, blank image, and geometry into the FDK
> reconstruction method
>
> We then pass the projections, FDK return value from previous step and
> geometry into the field of view mask
>
> Finally, we write the reconstructed image to file in NIfTI using
> rtk.ImageFileWriter[ImageType]
> ```
>
> We seem to be having a problem when we load any projection images instead
> of generating them like the FirstReconstruction example with the sphere. We
> get a blank image of 0's, or a grey image with a circle outline.
>
> At first glance, is there anything we are missing?
>
> Look forward to your response,
> Keven Q
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.creatis.insa-lyon.fr/pipermail/rtk-users/attachments/20230124/c511d922/attachment.htm>