[Insight-users] Performance issues for single slices

[Simon Warfield]

Hi Luis,


Luis Ibanez wrote:

>
> Hi Nils,
>
> It is actually surprising that the pipeline
> ran at all  :-)
>
> When you configure a 3D pipeline and feed it
> with a degenerate volume ( a single sliced
> volume), you risk to have trouble with any/all
> of the filters requiring neighborhood information.
> Since these filters look for data in the neighbor
> slices (which happen not to exist in your case).
>
> ITK filters based on the neighborhood iterator
> will be able to manage the degenerate volume but
> at the price of spending most of the computing
> time checking for boundary conditions. 


I don't agree that it is reasonable to have an expectation that a 3D 
filter will run poorly on a single slice of a 3D volume. 

  Often it is valuable to be able to present to a user a single coronal, 
sagittal or axial slice, or perhaps all three, and illustrate rapidly a 
processing operation on these slices, and then ask the user if it should 
be carried out in 3D.  For example, this UI strategy is used in our 
3Dslicer environment.

I don't see what magic a 2D instantiation of template coded filter can 
work that a 3D filter running on a 2D slice could not also do.  Why 
could the 3D filter not determine it is running on an Xx1xZ data set, 
and know the boundary conditions as well as a 2D filter running on an 
XxZ data set ?

>
>
> It is unlikely that LevelSet filters will be able
> to process correctly a degenerate volume. These
> filters solve a PDE using finite differences and
> should be attempting to look for data in the
> (again, unexisting) neighbor slices.
>
> In order to get a reasonable speed in 2D,
> and a consistent processing of your data,
> you better instantiate a native 2D pipeline.
>
> If you really want to factorize your programming
> efforts and have a 2D and a 3D  pipeline, the
> simple way to go is to create a class holding
> your entire pipeline, and template this class
> over the image dimension. Then, in your application,
> you instantiate this class for both 2D and 3D, and
> at load time you switch from one pipeline to another.
>
>
>
> Regards,
>
>
>
>    Luis
>
>
> ----------------------------
> Nils Hanssen wrote:
>
>> Hi,
>>  
>> I put together a small application with a level-set pipeline on 
>> volumetric medical data. The scalar type of the images is float.
>> When I run the filters on a volume that consists of one slice only, 
>> the computation time is increasing significantly. The measured times 
>> (compiled in release) are:
>>  
>> Volume #1 (256x256x256 voxels):
>> Gradient magnitude: 6.6 sec.
>>  
>> Volume #2 (1984x1024x1 voxels):
>> Gradient magnitude: 39.9 sec.
>>  
>>  
>> Should I expect that the computation time is much lower if I use a 
>> 2-dimensional image for the slice images? The points is that I want 
>> to keep the application general; that is it should be able to load 
>> slices and volumetric images transparently.
>>  
>>  
>> Thanks in advance,
>> Nils
>>  
>> -------------------------
>> Nils Hanssen
>> Surgical Systems Laboratory
>> research center c ae sa r
>> Ludwig-Erhard-Allee 2
>> 53175 Bonn
>> fon: +49-228-9656-197
>> fax: +49-228-9656-117
>> ___http://www . caesar . de/ssl_
>>
>>  
>
>
>
>
> _______________________________________________
> Insight-users mailing list
> Insight-users at www . itk . org
> http://www . itk . org/mailman/listinfo/insight-users



-- 
Simon K. Warfield, Ph.D. warfield at bwh . harvard . edu Phone:617-732-7090
http://www . spl . harvard . edu/~warfield           FAX:  617-582-6033
Assistant Professor of Radiology,          Harvard Medical School
Director, Computational Radiology Laboratory
Thorn 329, Dept Radiology,  Brigham and Women's Hospital 
75 Francis St, Boston, MA, 02115