[vtkusers] Problem rotating vtkImageSlice
David Gobbi
david.gobbi at gmail.com
Sun Aug 28 19:32:24 EDT 2016
Also see the ImageResliceMapperOffAxis test (and the other tests in
Rendering/Image/Testing/Cxx):
https://gitlab.kitware.com/vtk/vtk/blob/master/Rendering/Image/Testing/Cxx
On Sun, Aug 28, 2016 at 5:27 PM, Richard Frank <rickfrank at me.com> wrote:
> Thanks I'll take a look!
>
> Rick Frank
>
> On Aug 28, 2016, at 7:21 PM, David Gobbi <david.gobbi at gmail.com> wrote:
>
> Rotating the vtkImageSlice (i.e. the actor) rotates the data with respect
> to world coords. Rotating the SlicePlane of the vtkImageResliceMapper
> rotates the cut plane wrt world coords. So between these, you have full
> control over both the data-to-world transformation, and over the cut plane
> that you want to use to extract an oblique slice from the rotated data. So
> what you are trying to do is definitely possible.
>
> Here is the code that I use to display images for registration:
> https://github.com/dgobbi/AIRS/blob/master/Programs/register.cxx#L1994
>
> - David
>
> On Sun, Aug 28, 2016 at 4:52 PM, Richard Frank <rickfrank at me.com> wrote:
>
>> Hi David
>>
>> I have a preop CT typically 600 - 900 slices. The intra-op ct ( DICOM
>> says modality is XA , essentially a lower res non - HU calibrated spin
>> volume) is from a c-arm right before surgical procedure. I need to align
>> the intra op with the preop. This is manual registration by moving 3
>> orthogonal slice pairs to create a 6 DOF registration matrix. My initial
>> implementation of the slices ( Axial, sagittal, etc) has them all "facing"
>> the user. Rolling the scroll wheel moves through the slices.
>>
>> This is all parallel projection.
>>
>> We have a dual volume renderer written in OpenCL that shows the alignment
>> in 3D of the two ( fixed, moving ) volumes.
>>
>> Now I would like to, for simplicity, align the
>> slices so the world coordinates line up with my ijkLPS transform.
>> In the case of direction cosines == identity, HFS position, my sagittal
>> slice can be aligned with world coordinates by turning it so the patient is
>> head pointing out of screen (+Z Inferior - Superior), face down ( + Y
>> Anterior to Posterior) ( +X Left to Right).
>>
>> The I would move the slice to proper location in space based on image
>> position patient. ( one coordinate is always 0 ) And then move the camera
>> so the slice normal is aligned with the camera view vector.
>>
>> This is all working correctly with our 3D views and associated surgical
>> equipment. I thought I would be able to manipulate the slices in space as
>> any other actor ( rotate, translate, etc) but perhaps that's not possible?
>> When I try to rotate the image slice ( which is a Prop3D) the image seems
>> to be resliced as you say, instead of the prop being rotated.
>>
>> Thanks
>>
>> Rick Frank
>>
>> On Aug 28, 2016, at 5:41 PM, David Gobbi <david.gobbi at gmail.com> wrote:
>>
>> Okay, let me try to understand this. Please let me know if I'm wrong on
>> any of the following points.
>>
>> 1) You have a CT scan, consisting of several slices (I assume around 100
>> slices) with uniform spacing.
>>
>> 2) You have a single XA image. An XA is a projection image, in contrast
>> to CT which is a tomographic image.
>>
>> 3) You need to find the direction-of-projection through the CT volume
>> such that it matches the XA.
>>
>> This is generally done by creating a DRR a MIP of the CT. I'm not sure
>> how you plan to achieve this by rotating a slice.
>>
>> - David
>>
>> On Sun, Aug 28, 2016 at 3:09 PM, Richard Frank <rickfrank at me.com> wrote:
>>
>>> I thought I coil rotate the image slice in 3 space as defined here
>>>
>>> http://www.vtk.org/doc/nightly/html/classvtkImageSlice.html#details
>>>
>>> Rick Frank
>>>
>>> On Aug 28, 2016, at 3:42 PM, David Gobbi <david.gobbi at gmail.com> wrote:
>>>
>>> Then remind me again why you're using vtkImageResliceMapper instead of
>>> vtkImageSliceMapper?
>>>
>>> On Sun, Aug 28, 2016 at 1:23 PM, Richard Frank <rickfrank at me.com> wrote:
>>>
>>>> I'm just trying to rotate the plane in 3 space, not reslicing...
>>>>
>>>> Rick Frank
>>>>
>>>> On Aug 28, 2016, at 3:16 PM, David Gobbi <david.gobbi at gmail.com> wrote:
>>>>
>>>> If you only had one input slice, then why would you be surprised that
>>>> reslicing at 90 degrees gives an output that is only one pixel wide?
>>>>
>>>> On Sun, Aug 28, 2016 at 12:53 PM, Richard Frank <rickfrank at me.com>
>>>> wrote:
>>>>
>>>>> Hi
>>>>>
>>>>> I'm using the VTK example which creates an in memory single cyan
>>>>> slice.....
>>>>>
>>>>> Thanks
>>>>>
>>>>> Rick Frank
>>>>>
>>>>> On Aug 28, 2016, at 2:42 PM, David Gobbi <david.gobbi at gmail.com>
>>>>> wrote:
>>>>>
>>>>> Check the extent of the input image to make sure all of the slices
>>>>> were loaded.
>>>>>
>>>>> On Sun, Aug 28, 2016 at 12:37 PM, Richard Frank <rickfrank at me.com>
>>>>> wrote:
>>>>>
>>>>>> Hi David,
>>>>>>
>>>>>> Sorry about the attachment.
>>>>>>
>>>>>> Well, I'm pretty far along with my work on this, so I'd like to try
>>>>>> the vtkPlane orientation as a solution if possible.
>>>>>>
>>>>>> using the example code, If I create a vtkPlane, set its origin to
>>>>>> 0,0,0 and set its normal to 1 0 0 (instead of 0,0,1) and call
>>>>>> imageResliceMapper->SetSlicePlane(plane);
>>>>>>
>>>>>> I see a thin line. Seems correct. If I change the interactor style to
>>>>>> Trackball camera, and rotate the camera around it seems the image is only 1
>>>>>> unit wide however.
>>>>>>
>>>>>> Is there something else I need to do to get the image to map to a
>>>>>> plane with different orientation?
>>>>>>
>>>>>> THanks
>>>>>> Rick Frank
>>>>>>
>>>>>> On Aug 28, 2016, at 12:51 PM, David Gobbi <david.gobbi at gmail.com>
>>>>>> wrote:
>>>>>>
>>>>>> Hi Frank,
>>>>>>
>>>>>> It looks like you are doing a data transformation when you should be
>>>>>> doing a view transformation. Don't rotate the actor if your goal is to
>>>>>> rotate the view. When you rotate the actor, you are changing the
>>>>>> relationship between data coordinates and world coordinates.
>>>>>>
>>>>>> If you want a sagittal view, then you can rotate the view by moving
>>>>>> the VTK camera. The mapper has a method SliceFacesCameraOn() that can be
>>>>>> used to make the mapper automatically extract a slice in the correct
>>>>>> orientation.
>>>>>>
>>>>>> Also, please limit the width of inlined images to around 1000 pixels
>>>>>> or less. Large inlined images make your email hard to read on my system (I
>>>>>> wish that gmail would auto-shrink oversized inlined images, but
>>>>>> unfortunately it displays them full-size, even if they're several times the
>>>>>> size of the display).
>>>>>>
>>>>>> - David
>>>>>>
>>>>>> On Sun, Aug 28, 2016 at 9:56 AM, Richard Frank <rickfrank at me.com>
>>>>>> wrote:
>>>>>>
>>>>>>> Hi David,
>>>>>>>
>>>>>>> Thanks for the reply. I'm still a bit confused about what I've tried
>>>>>>> and why it doesn't work....
>>>>>>>
>>>>>>> So, I have 3 slice views and in the screenshot the Axial View is
>>>>>>> lined up appropriately (image position patient is in the correct place, and
>>>>>>> each slice is in correct place etc) with the world coordinates by using the
>>>>>>> image reslice axis and camera etc to align the DICOM coordinates and the
>>>>>>> VTK World coordinates.
>>>>>>>
>>>>>>> I now want to orient the sagittal slice so that it aligns with the
>>>>>>> world coordinates.
>>>>>>>
>>>>>>> (see screen shots)
>>>>>>>
>>>>>>>
>>>>>>> I believe I need to rotate the slice (if I could rotate the actor)
>>>>>>> about Z 90 CCW and then about its new X 0- degrees and then move the camera
>>>>>>> about world Y 90 and the roll the camera 90 to get the sagittal voxels
>>>>>>> DICOM aligned with the World Coordinates.
>>>>>>>
>>>>>>> I tried various parameters for the resliceMapper->SetSlicePlane()
>>>>>>> but the slice just dissapears off screen somewhere so I must be getting
>>>>>>> sometime wrong.
>>>>>>>
>>>>>>> How would I use the SetSlicePlane to achieve this?
>>>>>>>
>>>>>>> Thanks
>>>>>>>
>>>>>>> Rick
>>>>>>>
>>>>>>> On Aug 27, 2016, at 02:51 PM, David Gobbi <david.gobbi at gmail.com>
>>>>>>> wrote:
>>>>>>>
>>>>>>> You should also try vtkImageSliceMapper instead of
>>>>>>> vtkImageResliceMapper,
>>>>>>> perhaps its behavior will better suit your requirements.
>>>>>>>
>>>>>>>
>>>>>>> On Sat, Aug 27, 2016 at 11:51 AM, David Gobbi <david.gobbi at gmail.com
>>>>>>> > wrote:
>>>>>>>
>>>>>>>> Hi Rick,
>>>>>>>>
>>>>>>>> You're seeing this because you are doing an out-of-plane rotation
>>>>>>>> on an image that has only one slice. The rotation is around the image
>>>>>>>> origin, which (in this case) is at the bottom-left corner of the image.
>>>>>>>>
>>>>>>>> The vtkImageResliceMapper has a SetSlicePlane() method:
>>>>>>>> http://www.vtk.org/Wiki/VTK/Image_Rendering_Classes#vtkImage
>>>>>>>> ResliceMapper
>>>>>>>> You must rotate this SlicePlane when you rotate the vtkImageSlice.
>>>>>>>>
>>>>>>>> - David
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> On Sat, Aug 27, 2016 at 10:54 AM, Richard Frank <rickfrank at me.com>
>>>>>>>> wrote:
>>>>>>>>
>>>>>>>>> Hi,
>>>>>>>>>
>>>>>>>>> If I build the example
>>>>>>>>>
>>>>>>>>> VTK/Examples/Cxx/Images/ImageSlice
>>>>>>>>> and add
>>>>>>>>>
>>>>>>>>> imageSlice->RotateX(1);
>>>>>>>>>
>>>>>>>>> Which should, I think, rotate the image actor about its X axis by
>>>>>>>>> 1 degree, I instead get
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> which seems incorrect.
>>>>>>>>>
>>>>>>>>> I can rotate about Z.
>>>>>>>>>
>>>>>>>>> Is there something I'm not doing correctly, or is this a bug?
>>>>>>>>>
>>>>>>>>> I also tried RotateWXYZ(1,1,0,0) and get the same result.
>>>>>>>>>
>>>>>>>>> This is VTK 7
>>>>>>>>>
>>>>>>>>> Thanks
>>>>>>>>>
>>>>>>>>> Rick
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtkusers/attachments/20160828/d7f57667/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: TestImageResliceMapperOffAxis.png
Type: image/png
Size: 46693 bytes
Desc: not available
URL: <http://public.kitware.com/pipermail/vtkusers/attachments/20160828/d7f57667/attachment-0001.png>
More information about the vtkusers
mailing list