[Rtk-users] Simple FDK reconstruction problem

Simon Rit simon.rit at creatis.insa-lyon.fr
Mon Aug 28 07:33:16 CEST 2023


Hi,
Please stick to the mailing list.
Well done with the ProjectionsReader. To minimize the Python package, we
indeed avoid wrapping GPU types when it's not needed. Conversion to CPU
image type is rather easy as you've experienced.
It's a bit hard to help with one projection only and without knowing what
the result is.
RTK indeed converts ushort files to float because it attempts to convert
input files to line integrals. In theory, everything is templated but we
only test float/double data types.
Simon

On Fri, Aug 25, 2023 at 6:12 PM Victor Hu <victorhu233 at gmail.com> wrote:

> After changing `rtk.ProjectionsReader`'s type as CPUImageType, it works.
> But the reconstruction result is very odd. I tuned a number of parameters
> but still failed. The attachment includes FirstCudaReconstruction.py file
> and one of my .tif image. I'd appreciate it if you could help.
>
> By the way, the data in my .tif file is 16bit (usigned short, USHORT), but
> it seems that RTK only accept float type. Therefore, the final .raw file is
> in float type. Is there a way to read, reconstruct and save in USHORT?
>
> Thanks in advance.
>
> Victor Hu <victorhu233 at gmail.com> 于2023年8月24日周四 15:25写道:
>
>> Hello Simon,
>>
>> `rtk.ProjectionsReader` seems to be not able to read type GPUImageType
>> which is the same as the example `FirstCudaReconstruction.py`. Any
>> suggestions?
>>
>>
>>
>> Simon Rit <simon.rit at creatis.insa-lyon.fr> 于2023年8月20日周日 13:36写道:
>>
>>> Hi,
>>> The rotation axis in ITK is the Y axis.
>>> Yes, it does not matter if the size in not the same in every direction.
>>> You need to use the Cuda version of the package, itk-rtk-cuda116, see
>>> INSTALLATION.md
>>> <https://github.com/SimonRit/RTK/blob/master/INSTALLATION.md>.
>>> Simon
>>>
>>> On Sun, Aug 20, 2023 at 7:31 AM Victor Hu <victorhu233 at gmail.com> wrote:
>>>
>>>> Hello RTK team and Simon,
>>>>
>>>> I've corrected sid and sdd, which are 200 and 1850. However, the
>>>> reconstructed result (3d tif image) is so strange. Is it a series of slice
>>>> along with Z axis?
>>>>
>>>> I tried change the output size to 500*500*250. Is it OK if it's not the
>>>> same in z dimension?
>>>>
>>>> The CPU version was time-costing, so I turned to GPU version. However,
>>>> an error just occurred:
>>>> AttributeError: module 'RTK' has no attribute 'CudaImage' when executing '
>>>> GPUImageType = rtk.CudaImage[itk.F,3] '. What's wrong and how to fix it?
>>>>
>>>> Thanks in advance.
>>>> _______________________________________________
>>>> Rtk-users mailing list
>>>> rtk-users at openrtk.org
>>>> https://www.creatis.insa-lyon.fr/mailman/listinfo/rtk-users
>>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.creatis.insa-lyon.fr/pipermail/rtk-users/attachments/20230828/2dda6436/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: FirstCudaReconstruction.py
Type: text/x-python
Size: 3511 bytes
Desc: not available
URL: <http://www.creatis.insa-lyon.fr/pipermail/rtk-users/attachments/20230828/2dda6436/attachment.py>


More information about the Rtk-users mailing list