[Rtk-users] About Kaiser-Bessel Filter

Wed Nov 19 08:26:17 EST 2014

Hi,Simon and Chao,
i want to test the cuda performance in SART algorithm.
but when i looked in the parameters in forwardprojection and
backprojection. i got the  enum:

enum enum_fp { fp__NULL = -1, fp_arg_Joseph = 0,
fp_arg_RayCastInterpolator, fp_arg_CudaRayCast };
enum enum_bp { bp__NULL = -1, bp_arg_VoxelBasedBackProjection = 0,
bp_arg_Joseph, bp_arg_CudaVoxelBased, bp_arg_NormalizedJoseph };

So if i want to use the trilinear intrtpolation forwardprojection you
mentioned , how should i set the -b and -p in SART? or how to choose -b and
-p that  i  can get a fastest reconstruction in SART in this case??

thanks for your help in advance.
Best
Guangming

*Guangming Zang (Alex)*
*King Abdullah University of Science and Technology(KAUST)*
*University of Chinese Academy of Sciences(UCAS)*

2014-11-18 19:47 GMT+03:00 Simon Rit <simon.rit at creatis.insa-lyon.fr>:

> The trilinear version is available with CUDA:
>
> http://www.openrtk.org/Doxygen/classrtk_1_1CudaForwardProjectionImageFilter.html
> Simon
>
> On Tue, Nov 18, 2014 at 5:43 PM, Guangming Zang <
> guangming.zang at kaust.edu.sa> wrote:
>
>> Hi Chao,
>> Thanks very much for your prompt and so kind reply.
>> I will recheck the process of setting carefully with your suggestions and
>>  a short report may be given later.
>>  BTW, do you have any plan recently to add other interpolation filter(e.g.Kaiser-Bessel
>> or Trilinear)??
>> Thanks for your time.
>> Regards
>> Guangming
>>
>>
>>
>> 2014-11-18 19:09 GMT+03:00 Chao Wu <wuchao04 at gmail.com>:
>>
>>> Hi Guangming,
>>>
>>> To reconstruct a volume from a series of tiff files you have to let the
>>> algorithm know the geometrical relationship among the source, the volume
>>> and the projections stored in the tiffs. The idea is the same: RTK uses
>>> projection matrices and image coordinates to understand the positioning of
>>> those projections. The projection matrices are based on a lot of
>>> geometrical parameters like gantry angle, source position, detector
>>> position, in-plane and out-of-plane rotation angles of the detector etc. I
>>> won't go though all since most of them have no differences if you use tiff
>>> files or whatever format as input. I focus on some tiff tips.
>>>
>>> In the simplest situation, the tiff files contain correct resolution
>>> (DPI) information, then you need to do nothing on these files. You have to
>>> specify the correct detector offset though, since in the tiff image
>>> coordinates the origin is the first pixel (corner pixel) which is
>>> apparently not the intersection of the central ray and the detector. The
>>> detector offsets in x and y actually mean the position of the image origin
>>> (the first pixel) in the coordinate system centred at the intersection.
>>> Since the direction cosines is an identity matrix and pixel sizes are
>>> positive, the detector offsets in x and y will be negative.
>>>
>>> If the resolution information of the tiff files is not correct, you have
>>> to rewrite this information. You can of course perform a pre-processing to
>>> change the DPI of all tiff files with a third-party software, or you can
>>> alter the RTK code to allow such changes after reading the images, by using
>>> a itk::ChangeInformationImageFilter filter.
>>>
>>> Another non-geometrical issue with tiff files is that when their integer
>>> pixel values are converted into attenuation, the maximum possible integer
>>> is used as the reference value. This may not be correct in practice, e.g.
>>> in your 16-bit tiffs you use 62000 to represent air (no attenuation)
>>> leaving the integer range from 62001 to 65535 for statistical fluctuation.
>>> If for instance you only use the lower 14 bits for your data then this
>>> problem is more severe. For sure you can correct this by changing the code
>>> and specify the correct reference value.
>>>
>>> Hope these can help.
>>>
>>> Regards,
>>> Chao
>>>
>>> 2014-11-18 15:09 GMT+01:00 Guangming Zang <guangming.zang at kaust.edu.sa>:
>>>
>>>> Thanks
>>>>
>>>> *Guangming Zang (Alex)*
>>>> *King Abdullah University of Science and Technology(KAUST)*
>>>> *University of Chinese Academy of Sciences(UCAS)*
>>>>
>>>>
>>>> 2014-11-18 16:46 GMT+03:00 Guangming Zang <guangming.zang at kaust.edu.sa>
>>>> :
>>>>
>>>>> Hi,
>>>>> I am using the sart reconstruction in RTK, it works well.Thanks for
>>>>> your great work.
>>>>> and i was wondering do you have any plan to add other filters except
>>>>> Joseph, such as Kaiser-Bessel???  ( described here
>>>>> http://wscg.zcu.cz/wscg2003/papers_2003/herman.pdf)
>>>>> BTW, i have read the messages sent by other users(
>>>>> http://public.kitware.com/pipermail/rtk-users/2014-October/000573.html
>>>>> )
>>>>> but i still can not make it. can u help me and explain again about
>>>>> using a series of .tiff images ans sart algorithms to get the volume.
>>>>> Thanks in advance
>>>>> Best
>>>>> *King Abdullah University of Science and Technology(KAUST)*
>>>>> *University of Chinese Academy of Sciences(UCAS)*
>>>>>
>>>>>
>>>>
>>>> ------------------------------
>>>> This message and its contents, including attachments are intended
>>>> solely for the original recipient. If you are not the intended recipient or
>>>> have received this message in error, please notify me immediately and
>>>> delete this message from your computer system. Any unauthorized use or
>>>> distribution is prohibited. Please consider the environment before printing
>>>> this email.
>>>> _______________________________________________
>>>> Rtk-users mailing list
>>>> Rtk-users at public.kitware.com
>>>> http://public.kitware.com/mailman/listinfo/rtk-users
>>>>
>>>>
>>>
>>
>> ------------------------------
>> This message and its contents, including attachments are intended solely
>> for the original recipient. If you are not the intended recipient or have
>> received this message in error, please notify me immediately and delete
>> this message from your computer system. Any unauthorized use or
>> distribution is prohibited. Please consider the environment before printing
>> this email.
>>
>> _______________________________________________
>> Rtk-users mailing list
>> Rtk-users at public.kitware.com
>> http://public.kitware.com/mailman/listinfo/rtk-users
>>
>>
>

-- 

------------------------------
This message and its contents, including attachments are intended solely 
for the original recipient. If you are not the intended recipient or have 
received this message in error, please notify me immediately and delete 
this message from your computer system. Any unauthorized use or 
distribution is prohibited. Please consider the environment before printing 
this email.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/rtk-users/attachments/20141119/a2e05a2a/attachment-0009.html>