Difference between revisions of "Projects:RegistrationLibrary:RegLib C35"
From NAMIC Wiki
| Line 34: | Line 34: | ||
===Registration Challenges=== | ===Registration Challenges=== | ||
*The image content to be registered is less than 10% of the total image and of low energy (close to background); | *The image content to be registered is less than 10% of the total image and of low energy (close to background); | ||
| − | [[Image: | + | [[Image:RegLib_C35_Histogram.png|150px|lleft|histogram of image content: target structure (colon, red) is very small amount of total image energy, making intensity-based automated approach unlikely to succeed]] |
*there's large portions of the colon for which no matching/corresponding structures exist, i.e. a straightforward one-one mapping is not avail. | *there's large portions of the colon for which no matching/corresponding structures exist, i.e. a straightforward one-one mapping is not avail. | ||
*other landmarks (kidneys etc.) are very different in position and size | *other landmarks (kidneys etc.) are very different in position and size | ||
Revision as of 16:13, 1 March 2011
Home < Projects:RegistrationLibrary:RegLib C35Back to ARRA main page
Back to Registration main page
Back to Registration Use-case Inventory
Contents
v3.6.1 
Slicer Registration Library Case #35: Mouse Colon Brain MRI
Input
|
|
|
| fixed image 1/target MRI mouse 1 |
moving image MRI mouse 2 |
Modules
- Slicer 3.6.1 recommended modules: SurfaceRegistration, Transforms (manual registration), Fast Marching Segmentation, Histogram Matching, Model Maker,
Objective / Background
Aim is to register the two MRIs such that the colon aligns.
Download
- DATA
- See detailed pipeline with parameter settings below
Keywords
MRI, colon, mouse, masking, non-human, non-rigid,manual, unsolved
Input Data
- reference/fixed MRI: 0.12 x 0.12 x 0.1 mm, 512 x 512 x 94
- moving MRI: 0.09 x 0.09 x 0.1 mm, 512 x 512 x 94
Registration Challenges
- The image content to be registered is less than 10% of the total image and of low energy (close to background);
- there's large portions of the colon for which no matching/corresponding structures exist, i.e. a straightforward one-one mapping is not avail.
- other landmarks (kidneys etc.) are very different in position and size
Key Strategies
- This is a challenge case and currently unsolved. Best hopes are combinations of manual, fiducial and surface-based registration
- Slicer 3.6.1 recommended modules:
- Surface Registration
- Plastimatch
Registration Results
very rough surface models of colon help visualize misalignment
original, not registered
after surface registration (rigid) + manual correction
