[Insight-users] Possible LevelSet Bug Due To Filters' Default UseDirectionOn?

[Chad Stanford]

I do not see any text here or was the point to push this message to the top again?  This bug seems pretty significant to me as it could affect many 3D level set segmentation results. Has anyone considered it?

From: hans-johnson at uiowa.edu
To: chadz111 at hotmail.com; insight-users at itk.org
Date: Wed, 4 Jul 2012 12:29:25 +0000
Subject: Re: [Insight-users] Possible LevelSet Bug Due To Filters' Default UseDirectionOn?














From: Chad Stanford <chadz111 at hotmail.com>

Date: Tuesday, July 3, 2012 9:13 PM

To: "insight-users at itk.org" <insight-users at itk.org>

Subject: [Insight-users] Possible LevelSet Bug Due To Filters' Default UseDirectionOn?








Hello, 



I think have found a bug in many of the level set functions where the level set function derivatives could potentially not correspond with the advection derivatives and possibly other terms.  Essentially, many of the level set methods use derivative/gradient
 filters which take into account image direction by default to compute some of their terms e.g. advection.  The ComputeUpdate function in itkLevelSetFunction.hxx does not seem to take this into account when it computes the derivatives/gradients of the level
 set function.  In my case where the cosine direction matrix was diagonal and had -1 for the first 2 elements, the x and y advection gradient were off by a sign with respect to the sign of the level set function gradients.



Suggestion:  It'd be really great if the UseDirection option being set by default could be mentioned in the user guide because I was not aware of it until I'd spent a lot of time tracing a bug to one of the gradient filter's having that option set by default.
  Also, the option is easy to miss because the user guide does not mention the direction matrix once (as far as I can tell), which would also be a nice addition, amongst other things.






While we're on the topic of the direction matrix, this problem has raised some questions I have about it.  Any answers or general insight would be greatly appreciated:



1.) I just want to verify that the direction matrix does not need to be diagonal or have unit elements and is fine as long as it's orthogonal?




Correct.  It does not need to be orthogonal.







2.) What coordinate system is the direction matrix rotating the image-coordinate system to? Basically I'm wondering where it comes from, not how it's made, as I understand given another coordinate system how to generate it.




The direction cosign follows the conventions described in the DICOM standard.










Thanks!
- Chad








Notice: This UI Health Care e-mail (including attachments) is covered by the Electronic Communications Privacy Act, 18 U.S.C. 2510-2521, is confidential and may be legally privileged.  If you are not the intended recipient, you are hereby notified that any
 retention, dissemination, distribution, or copying of this communication is strictly prohibited.  Please reply to the sender that you have received the message in error, then delete it.  Thank you.




_____________________________________
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.php

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 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20120716/7b0eb343/attachment.htm>