Difference between revisions of "2012 Summer Project Week:DifficultRegistration"

From NAMIC Wiki
Jump to: navigation, search
m (Text replacement - "[http://www.na-mic.org/Wiki/images/" to "[https://na-mic.org/w/images/")
 
(8 intermediate revisions by 2 users not shown)
Line 12: Line 12:
  
 
==Registration Result Visualization==
 
==Registration Result Visualization==
 +
 
<b>Slicer:BRAINS<br>[[Image:IntraOp_Slicer-BRAINS_BSpline.gif|400px|lleft|IntraOp via BRAINS]]<br>
 
<b>Slicer:BRAINS<br>[[Image:IntraOp_Slicer-BRAINS_BSpline.gif|400px|lleft|IntraOp via BRAINS]]<br>
 
NiftyReg<br>[[Image:mr-results-nifityreg.gif|400px|lleft|IntraOp via BRAINS]]<br>
 
NiftyReg<br>[[Image:mr-results-nifityreg.gif|400px|lleft|IntraOp via BRAINS]]<br>
Line 49: Line 50:
  
 
<h3>Progress</h3>
 
<h3>Progress</h3>
Three registration approaches evaluated: (NiftyReg), (BrainsFit), (SIFT Landmarks).
+
We held a general discussion of registration challenges in the community. Results were generated for head and neck data (Ivan Kolesov), three registration approaches were evaluated on challenging head and body images: (NiftyReg - Marc Modat), (Slicer:BRAINS - Dominik Meier), (SIFT Landmarks - Matthew Toews).
 
<ul>
 
<ul>
 
<li>Registration of difficult brain images is successful, body images remain challenging.
 
<li>Registration of difficult brain images is successful, body images remain challenging.
 +
<li>Registration direction has an important impact, i.e. which images are designed as fixed or moving. Symmetric or bi-directional registration is helpful.
 
<li>Masking of problematic image regions, e.g. pathology, is an important but time-consuming pre-processing task that could be further investigated or automated. For instance, interactive masking/segmentation of pathology, linking registration and masking.
 
<li>Masking of problematic image regions, e.g. pathology, is an important but time-consuming pre-processing task that could be further investigated or automated. For instance, interactive masking/segmentation of pathology, linking registration and masking.
<li>Registration direction has an important impact, i.e. which images are designed as fixed or moving. Symmetric or bi-directional registration is helpful.
 
 
</ul>
 
</ul>
 
</div>
 
</div>
Line 75: Line 76:
 
Stefan Klein</td><td>Erasmus Medical Center</td><td>General Registration</td><td></td><td>
 
Stefan Klein</td><td>Erasmus Medical Center</td><td>General Registration</td><td></td><td>
 
</td><tr><td>
 
</td><tr><td>
Ivan Kolesov</td><td>Georia Tech</td><td>Articulated, point-based registration</td><td></td><td>
+
Ivan Kolesov</td><td>Georgia Tech</td><td>Articulated, point-based registration</td><td> Plastimatch, NiftyReg </td><td> [http://www.na-mic.org/Wiki/index.php/2012_Summer_Project_Week:Deformable_Registration_for_Head_and_Neck Head Neck Results]
 
</td><tr><td>
 
</td><tr><td>
 
Dominik Meier</td><td>BWH</td><td>General registration</td><td>Slicer: BRAINS</td><td>
 
Dominik Meier</td><td>BWH</td><td>General registration</td><td>Slicer: BRAINS</td><td>
Line 81: Line 82:
 
TBI:[[Image:TBI_Slicer-BRAINS_BSpline.gif|60px|lleft|TBI via BRAINS]]<br>
 
TBI:[[Image:TBI_Slicer-BRAINS_BSpline.gif|60px|lleft|TBI via BRAINS]]<br>
 
CT torso: FAILED</td><tr><td>
 
CT torso: FAILED</td><tr><td>
Marc Modat</td><td>UCL</td><td>neuro-deg. diseases, longitudinal</td><td>NiftiReg</td><td>[http://www.na-mic.org/Wiki/images/d/d5/NiftyReg_projectWeek_test.pdf Some test res]</td><tr><td>
+
Marc Modat</td><td>UCL</td><td>neuro-deg. diseases, longitudinal</td><td>NiftiReg</td><td>[https://na-mic.org/w/images/d/d5/NiftyReg_projectWeek_test.pdf Some test res]</td><tr><td>
 
Albert Motillo</td><td>GE</td><td>Parsing CT, Detection</td><td></td><td>
 
Albert Motillo</td><td>GE</td><td>Parsing CT, Detection</td><td></td><td>
 
</td><tr><td>
 
</td><tr><td>

Latest revision as of 18:27, 10 July 2017

Home < 2012 Summer Project Week:DifficultRegistration

Registration Result Visualization

Slicer:BRAINS
IntraOp via BRAINS
NiftyReg
IntraOp via BRAINS
SIFT Landmark
IntraOp via BRAINS IntraOp via BRAINS

Key Investigators

  • Erasmus Medical Center: Stefan Klein
  • University College London: Marc Modat
  • UNC: Aditya Gupta, Martin Styner
  • BWH: Matthew Toews, Petter Risholm, Dominik Meier, William Wells

Objective

To identify solutions to difficult image registration problems that challenge the limits of current technology. Aspects of difficulty will include:

  • inter-subject registration
  • truncation, missing tissue
  • unknown initialization
  • inter-species registration
  • articulated deformation

Approach, Plan

A set of difficult pair-wise registration problems will be considered. Participants will discuss workable solutions based on their expertise and background, and these solutions will be documented.
Registration cases can be found here.

Progress

We held a general discussion of registration challenges in the community. Results were generated for head and neck data (Ivan Kolesov), three registration approaches were evaluated on challenging head and body images: (NiftyReg - Marc Modat), (Slicer:BRAINS - Dominik Meier), (SIFT Landmarks - Matthew Toews).

  • Registration of difficult brain images is successful, body images remain challenging.
  • Registration direction has an important impact, i.e. which images are designed as fixed or moving. Symmetric or bi-directional registration is helpful.
  • Masking of problematic image regions, e.g. pathology, is an important but time-consuming pre-processing task that could be further investigated or automated. For instance, interactive masking/segmentation of pathology, linking registration and masking.

Delivery Mechanism

Attention Participants: Please log in and update/correct entries in the table below. For bonus points, please provide links or solutions to image registration problems on this page.

ParticipantAffiliationContextTechniquesSolutions/Links
Steven AylwardKitwareCT, US, ressection, brain tumors, changing pathology, sliding organSliding Geometry, Geometric Metamorphosis
Karl DiedrichAZERigid registration, abdomenMulti-resolution registration
James FishbaughSCIShape analysis and registration
Aditya GuptaUNCDTI, enlarged lateral ventricles
Stefan KleinErasmus Medical CenterGeneral Registration
Ivan KolesovGeorgia TechArticulated, point-based registration Plastimatch, NiftyReg Head Neck Results
Dominik MeierBWHGeneral registrationSlicer: BRAINS

IntraOp:IntraOp via BRAINS
TBI:TBI via BRAINS

CT torso: FAILED
Marc ModatUCLneuro-deg. diseases, longitudinalNiftiRegSome test res
Albert MotilloGEParsing CT, Detection
Simrin NagpalQueens UniversityCT/US registration
Samon NuranianUBCUS-guided intervension, spine
Andre RemiUCLALongitudinal changes in TBI, tissue types
Peter RisholmBWHBrain, head and neck, radiation therapyProbabilistic Uncertainty
Samira SojoudiUBCSpine, CT/US registration
Matthew ToewsBWHGeneral registrationSIFT landmark correspondence

MR-brain:lleft
CT-torso:lleft

Bo WangUtahTBI image segmentation
Kevin WangPrincess Margaret HospitalAdaptive radiation therapy, longitudinal
William WellsBWHInterventional applicationsTheory: Segmentation, Registration

Links for Data

Links for Tools & Methods

Links for Papers