[vtkusers] CT Gantry Tilt correction

David Gobbi david.gobbi at gmail.com
Thu Nov 30 12:08:59 EST 2017


The time will be proportional to the number of slices in the volume.

I've pushed a change to github so that you can call the following method to
speed things up:

  rectify->SetInterpolationModeToLinear();

Linear interpolation will be at least 10 times faster than sinc
interpolation.
Just grab the latest master branch and this method will be available.

When I have some time, I'll dig into the sinc interpolation code to speed
it up.
Right now it's just using vtkImageReslice and a 3D sinc kernel, but CT
rectification really only needs a one-dimensional kernel.

 - David



On Thu, Nov 30, 2017 at 8:24 AM, satwik k <satwik.k8 at gmail.com> wrote:

> Last time when I used this API.. By mistake added in loop.. so it was
> taking more time... But now I moved it some where else method.. Now it is
> taking around 10sec... My doubt is, for 144 images.. it is taking this much
> extra time.. wht if have 1000images..?
>
> will it take same time or more?
>
>
> Regards,
> Satwik k
>
> On Nov 30, 2017 19:31, "David Gobbi" <david.gobbi at gmail.com> wrote:
>
>> Hi Satwik,
>>
>> I tested a 512x512x144 volume on my laptop, and it took 15 seconds.
>> So, no, that's not very fast.  The class uses a windowed sinc
>> interpolator,
>> which is computationally expensive.  I could add an option to use linear
>> interpolation.  With linear interpolation it only takes 1 second.
>>
>>  - David
>>
>>
>> On Wed, Nov 29, 2017 at 10:29 PM, Satwik <satwik.k8 at gmail.com> wrote:
>>
>>> Hi,
>>>
>>> Dimensions of my volume is (512, 512, 144).
>>>
>>> ///My Code///
>>> vtkDICOMCTRectifier* m_rectify = vtkDICOMCTRectifier::New();
>>> m_rectify->SetEnableSMP(true);
>>> m_rectify->SetVolumeMatrix(d->m_initOrient); //passing orientation
>>> matrix
>>> m_rectify->SetInputData(d->m_imageData); //passing image data
>>> m_rectify->Update();
>>>
>>> Regards,
>>> Satwik K.
>>>
>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtkusers/attachments/20171130/cec4c2a2/attachment.html>


More information about the vtkusers mailing list