[Insight-users] Cerebral aneurysm Angiogram REGISTRATION DOUBT!

[Luis Ibanez]

Hi Gaurav,


You can do the following:


A) Create a 3D image whose size, spacing, origin and orientation
     are such that one of its faces will be parallel to one of the
     2D images that you have, and another face will be parallel to the
     other 2D image.

     In this configuration, for every pixel of one of the 2D images,
     you will have a corresponding row of pixels in the 3D image.


B)  Visit in synchrony the 2D image and the 3D image, and for
     every pixel of the 2D image, fill up the entire corresponding
     row of the 3D image with the intensity values of the 2D pixel.


C)  Repeat with the other 2D image, but this time add the intensity
      values to the 3D image row.


Once you are done, you will have higher intensity values in the
locations where pixels of the same structure intersect.

This will still be full of ambiguities, but will give you a starting
point for bringing the two 2D images into 3D space.

The vessels at the base of the skull should be rather easy to
match, so you may want to start tracking from there.


Note that if your angiograms have the usual presentation in
which the contrast agent appears dark in a bright background,
then you should be looking for the darkest pixels of the 3D image.


      Regards,


             Luis


-----------------------------------------------------------------
On Thu, Mar 25, 2010 at 6:29 PM, Gaurav Sharda
<gaurav.sharda.uiowa at gmail.com> wrote:
> Hi Luis,
> Hey I asked you earlier about the biplane image registration of 2D cerebral
> angiograms.
> In that one option you suggested was to project the 2D images into a 3D
> volume and then track the minimum
> values of that 3D volume.
> Could you elaborate on this. How do I project the 2D angiograms into 3D
> volume in ITK?
> Here the image set I have are two orthogonal views ( lateral and AP view) of
> cerebral angiograms. I need to get a 3D volume of the aneurysm specifically
> from both these 2D views. Means kinda register the two 2D views and align
> them perpendicularly. I hope this makes sense.
>
> Eagerly awaiting your reply.
> Anyone having ideas, would appreciate your help.
> Thanks,
> Gaurav.
> On Sat, Feb 13, 2010 at 3:50 PM, Luis Ibanez <luis.ibanez at kitware.com>
> wrote:
>>
>> Hi Gaurav,
>>
>>
>> This is a very interesting subject, that unfortunately
>> doesn't have a straight-forward solution.
>>
>> Typically this is done by:
>>
>>
>> A) Extracting centerlines from each
>>     one of the 2D projections.
>>
>> B) Finding bifurcation points (or in general critical
>>     points) on each view.
>>
>> C) Matching bifurcation (and/or critical points)
>>     between the views.
>>
>>
>> The nature of the problem is such that you can't
>> solve for all possible cases.
>>
>> Are the images acquired with a bi-plane Fluoroscopy
>> system ?
>>
>> If so, at least you can build geometrical assumptions
>> based on the fact that the images will be aligned in
>> the circle of rotation of the Fluoroscopy machine.
>>
>> --
>>
>> You may want to take a look at the application:
>>
>>               Curves2DExtractor
>>
>> in InsightApplications
>>
>> http://www.itk.org/ITK/applications/Curve2DExtraction.html
>>
>> This application was developed to detect centerlines
>> in 3D Angiograms, in a context related to brain aneurysms.
>>
>> You could use the resulting centerline points as the input
>> for a process that finds geometrical correspondences.
>>
>>
>> ---
>>
>> Another independent approach, is to project both of the
>> 2D images into a 3D volume, and then track the minimum
>> values of that 3D volume. This of course is far from being
>> free of ambiguities, but at least it should give you an easy
>> initialization for correspondences between the thick vessels
>> at the base of the skull.
>>
>>
>>
>>       Regards,
>>
>>
>>            Luis
>>
>>
>> ---------------------------------------------------------------------
>> On Fri, Feb 12, 2010 at 10:21 PM, Gaurav Sharda
>> <gaurav.sharda.uiowa at gmail.com> wrote:
>> > Hi there!
>> >
>> > I am Gaurav Sharda, a graduate student in University of Iowa,US.
>> >
>> > I am going to start my work on cerebral aneurysm Angiogram Registration.
>> > I am going to have 2D angiograms of side view and top view of brain.
>> > Could
>> > you suggest some registration techniques to find the co inciding
>> > portions?
>> > The images are roughly orthogonal.
>> > Please see if you could suggest some techniques and give me some useful
>> > links.
>> > Thanks a lot..
>> >
>> > Gaurav
>> >
>> > _____________________________________
>> > Powered by www.kitware.com
>> >
>> > Visit other Kitware open-source projects at
>> > http://www.kitware.com/opensource/opensource.html
>> >
>> > Kitware offers ITK Training Courses, for more information visit:
>> > http://www.kitware.com/products/protraining.html
>> >
>> > Please keep messages on-topic and check the ITK FAQ at:
>> > http://www.itk.org/Wiki/ITK_FAQ
>> >
>> > Follow this link to subscribe/unsubscribe:
>> > http://www.itk.org/mailman/listinfo/insight-users
>> >
>> >
>
>