Difference between revisions of "2010 Winter Project Week Congealing Testing Debugging"

From NAMIC Wiki
Jump to: navigation, search
 
(6 intermediate revisions by the same user not shown)
Line 2: Line 2:
 
<gallery>
 
<gallery>
 
Image:z15-aff.png|Affine registration result on subject 15.
 
Image:z15-aff.png|Affine registration result on subject 15.
Image:HammerSimilarity.png|Similarity map (middle column) computed on moving image (right) for the selected point on the fixed image (left, cross). Point indicated by a cross on the middle and right column is the point on the moving image that is the most similar to the picked point on the fixed image.
+
Image:z15-b8.png|B-spline 8 registration result on subject 15.
Image: namic.png|The output deformation filed by HAMMER
+
Image: z15-b16.png|B-spline 16 registration result on subject 15.
Image: hammer.png|HAMMER in Slicer3
+
Image: z15-b32.png|B-spline 32 registration result on subject 15.
 
</gallery>
 
</gallery>
  
Line 21: Line 21:
  
 
<h3>Approach, Plan</h3>
 
<h3>Approach, Plan</h3>
Initial issue: congealing does not work with recent ITK.  The goal is to figure out where the problem is. Run existing tests from sandbox and test on 26-subject FA dataset.  Debug.
+
Initial issue: congealing does not work with recent ITK.  The goal is to figure out where the problem is. The plan is to run existing tests from sandbox and also test on 26-subject FA dataset, and debug.
  
 
</div>
 
</div>
Line 28: Line 28:
  
 
<h3>Progress</h3>
 
<h3>Progress</h3>
We found the following flag is needed to run congealing: ITK_USE_CENTERED_PIXEL_COORDINATES_CONSISTENTLY:BOOL=OFFWe tested the code with existing tests from Serdar and also on a 26-subject FA dataset.  We debugged initial crashing problems and patched a bug assuming all input images had the same size.  To use this code in its current version however it may be advisable to pad all imput images to the same dimensions.
+
We got congealing to run without errors or crashesHowever more work would be needed for it to be compatible with recent changes/improvements in ITK.
  
 +
We found the following backwards-compatible flag is needed to run congealing: ITK_USE_CENTERED_PIXEL_COORDINATES_CONSISTENTLY:BOOL=OFF. 
 +
 +
We tested the code with existing tests from Serdar and also on a 26-subject FA dataset.  We debugged initial crashing problems and patched a bug assuming all input images had the same size.  To use this code in its current version however it may be advisable to pad all imput images to the same dimensions.
 +
 +
We found that the multiresolution B splines may have a bug as results were "smeared". Luis suggests the optimizer may not be using the correct size scale. However the initial size 8 grid seems to work well. See above image results.
 +
 +
Some congealing tests initially crashed with recent ITK. We are running against the backwards compatible ITK now to see if Serdar's original tests pass.  (So far 36/96 have passed including ones that failed with recent ITK.)
 
</div>
 
</div>
 
</div>
 
</div>
Line 35: Line 42:
 
<div style="margin: 20px;">
 
<div style="margin: 20px;">
 
<div style="width: 27%; float: left; padding-right: 3%; color: red;">
 
<div style="width: 27%; float: left; padding-right: 3%; color: red;">
 
  
 
==References==
 
==References==

Latest revision as of 18:56, 8 January 2010

Home < 2010 Winter Project Week Congealing Testing Debugging

Key Investigators

  • BWH: Lauren O'Donnell
  • Kitware: Luis Ibanez

Objective

Test and debug congealing registration implementation in ITK (in NAMIC Sandbox).

Approach, Plan

Initial issue: congealing does not work with recent ITK. The goal is to figure out where the problem is. The plan is to run existing tests from sandbox and also test on 26-subject FA dataset, and debug.

Progress

We got congealing to run without errors or crashes. However more work would be needed for it to be compatible with recent changes/improvements in ITK.

We found the following backwards-compatible flag is needed to run congealing: ITK_USE_CENTERED_PIXEL_COORDINATES_CONSISTENTLY:BOOL=OFF.

We tested the code with existing tests from Serdar and also on a 26-subject FA dataset. We debugged initial crashing problems and patched a bug assuming all input images had the same size. To use this code in its current version however it may be advisable to pad all imput images to the same dimensions.

We found that the multiresolution B splines may have a bug as results were "smeared". Luis suggests the optimizer may not be using the correct size scale. However the initial size 8 grid seems to work well. See above image results.

Some congealing tests initially crashed with recent ITK. We are running against the backwards compatible ITK now to see if Serdar's original tests pass. (So far 36/96 have passed including ones that failed with recent ITK.)

References