Difference between revisions of "2010 Winter Project Week RegistrationCaseLibrary"
Line 49: | Line 49: | ||
*'''Segmentation''': | *'''Segmentation''': | ||
**[[2010_Winter_Project_Week_SegmentationWizard|High Level Wizard for Segmentation of Images]] (Mark Scully): uses register images to align atlas; functionality and compatibility considerations for RegisterImages redesign (both algorithm & GUI) | **[[2010_Winter_Project_Week_SegmentationWizard|High Level Wizard for Segmentation of Images]] (Mark Scully): uses register images to align atlas; functionality and compatibility considerations for RegisterImages redesign (both algorithm & GUI) | ||
− | **[[2010_Winter_Project_Week_LongitudinalLupusAnalyses|Longitudinal Analyses of Lesions in Lupus]] (Mark Scully): uses ChangeTracker | + | **[[2010_Winter_Project_Week_LongitudinalLupusAnalyses|Longitudinal Analyses of Lesions in Lupus]] (Mark Scully): |
+ | ***uses ChangeTracker as current prototype, although input data differs: T1,T2 +FLAIR are standard channels, currently T1 is used for registration. As with all change tracking and intra-subject registration: precision is key, robustness requirements less than others. | ||
+ | ***will provide example dataset to library | ||
**[[2010_Winter_Project_Week_ProstateSeg|Prostate segmentation using shape-based method]] (Andras Lasso): Does this involve co-registration of an atlas? | **[[2010_Winter_Project_Week_ProstateSeg|Prostate segmentation using shape-based method]] (Andras Lasso): Does this involve co-registration of an atlas? | ||
**[[2010_Winter_Project_Week_SegmentationEpicardialWall|Epicardial Wall Segmentation]] (Behnood Gholami): employs its own shape-registration scheme, uses ITK registration algorithm, does not go through slicer interface -> N/A | **[[2010_Winter_Project_Week_SegmentationEpicardialWall|Epicardial Wall Segmentation]] (Behnood Gholami): employs its own shape-registration scheme, uses ITK registration algorithm, does not go through slicer interface -> N/A | ||
Line 72: | Line 74: | ||
*'''NA-MIC Kit Internals''' | *'''NA-MIC Kit Internals''' | ||
**[[2010_Winter_Project_Week_SPECTRE_3DSlicer_Integration|Integration of SPECTRE Java module into 3D Slicer]] (Nicole Aucoin): SPECTRE: skull stripping is an important element for registration preprocessing, documentation and possibly GUI will link to this module. | **[[2010_Winter_Project_Week_SPECTRE_3DSlicer_Integration|Integration of SPECTRE Java module into 3D Slicer]] (Nicole Aucoin): SPECTRE: skull stripping is an important element for registration preprocessing, documentation and possibly GUI will link to this module. | ||
− | *'''MIDAS''' | + | === Progress: Notes === |
− | + | *consider a ''priority matrix'' for each case that lists the requirements for robustness, precision, speed. E.g. change tracking applications rank high in precision but low in the other two, IGT applications rank high in speed and robustness etc. | |
− | + | *'''MIDAS''': discussed with Steve & Casey the use of MIDAS server to host the case library; this would enable additional use of library cases for automated testing | |
==References== | ==References== |
Revision as of 04:15, 5 January 2010
Home < 2010 Winter Project Week RegistrationCaseLibraryThe 3DSlicer Registration Case Library Project
Key Investigators
- BWH: Dominik Meier, Ron Kikinis
- Kitware: Casey Goodlett
Objective
This work is part of a 2-year effort to supplement the 3DSlicer Registration Module with extensive set of documentation, tutorials and use case scenarios. We seek to build a comprehensive library of use cases for every form of registration challenge using 3DSlicer. Each library entry will contain a dataset along with Registration Parameter Presets and a guided tutorial on how to successfully register the images.
We also have a Registration Breakout Session on Tuesday 1-2pm (Amethyst Room). If registration is on your processing list, please drop in, have a look at what's new, and share your thoughts.
Approach, Plan
The use case library relies mainly on user contributions for content. There is an official call for datasets to all slicer users to consider a unique time-limited service for any dataset/registration problem that is not yet in the library. Users who agree to share one anonymized dataset for the library will have that dataset registered by us and the solution and strategies posted. Look here for an example of a library entry.
Our plan for the project week is to first get a survey of all projects using registration. We hope to catch most of these projects from reviewing these project description pages, so it is important that if you use registration in your procedure, please make sure to mention explicitly. We will collect as many example cases as possible, and select a few focus cases with particularly vexing registration challenges and begin to develop the appropriate tailored registration strategies.
Progress
- List of contacts for new use cases: see below
- insights/comments on registration settings/setting changes that made a difference go here
- Notes re. user insights/comments/wishes on registration go here
Progress: Target Projects
The projects below have been tentatively identified as having registration ties. We will seek discussion with these projects during the week. If your project is listed here, I will try to discuss the registration aspects at some point during the week. Type and length of discussion will vary greatly. Text next to each project below identifies the registration link.
- Segmentation:
- High Level Wizard for Segmentation of Images (Mark Scully): uses register images to align atlas; functionality and compatibility considerations for RegisterImages redesign (both algorithm & GUI)
- Longitudinal Analyses of Lesions in Lupus (Mark Scully):
- uses ChangeTracker as current prototype, although input data differs: T1,T2 +FLAIR are standard channels, currently T1 is used for registration. As with all change tracking and intra-subject registration: precision is key, robustness requirements less than others.
- will provide example dataset to library
- Prostate segmentation using shape-based method (Andras Lasso): Does this involve co-registration of an atlas?
- Epicardial Wall Segmentation (Behnood Gholami): employs its own shape-registration scheme, uses ITK registration algorithm, does not go through slicer interface -> N/A
- Segmentation of TBI (Traumatic Brain Injury) Subjects from Multimodal MRI (Marcel Prastawa): includes registration in new (ABC) module: category: intra-subject multi-contrast reg. Currrently also uses its own registration algorithm.
- Cadiac Ablation Scar Segmentation (Michal Depa): employs a label fusion segmentation algorithm which first registered all of the training images to the test one
- Registration
- Fusion of Anatomy,MRI and Electrophysiology in Parkinson's (Andrzej Przybyszewski): co-registration of subject MRI with anatomical atlas; involves moving image with drastically clipped FOV, inter-subject registration
- electrophysiology data avail. as second/validation channel
- final atlas needs to move into patient-specific stereotactic space
- Registration with Varying Elastic Parameters (Peter Risholm)
- Fusion of Anatomy,MRI and Electrophysiology in Parkinson's (Andrzej Przybyszewski): co-registration of subject MRI with anatomical atlas; involves moving image with drastically clipped FOV, inter-subject registration
- IGT
- MRI-guided Robotic Prostate Intervention (Andras Lasso and Junichi Tokuda): Transperineal device: includes registration, where accuracy should be improved; currently the registration is based on a single-slice image; it could possible to improve the accuracy by increasing the number of slices used for the registration.
- Radiotherapy
- Adaptive Radiotherapy for Head, Neck, and Thorax (Ivan Kolesov): Once segmentation is available, the goal is to register patient scans to account for anatomical changes between visits
- Analysis
- Vervet MRI Longitudinal Analysis (Andriy Fedorov): Register follow-up to baseline
- Informatics
- XNAT : discussion with XNAT group re. feasibility of using XNAT central as repository for use-case library.
- Processing and loading images in NRRD format: NRRD and NIFTI are the default formats for the image data in the registration case library.
- Diffusion
- Atlas-Based White Matter Segmentation (Lauren O'Donnell)
- NA-MIC Kit Internals
- Integration of SPECTRE Java module into 3D Slicer (Nicole Aucoin): SPECTRE: skull stripping is an important element for registration preprocessing, documentation and possibly GUI will link to this module.
Progress: Notes
- consider a priority matrix for each case that lists the requirements for robustness, precision, speed. E.g. change tracking applications rank high in precision but low in the other two, IGT applications rank high in speed and robustness etc.
- MIDAS: discussed with Steve & Casey the use of MIDAS server to host the case library; this would enable additional use of library cases for automated testing