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

From NAMIC Wiki
Jump to: navigation, search
 
(21 intermediate revisions by the same user not shown)
Line 2: Line 2:
 
<gallery>
 
<gallery>
 
Image:PW-MIT2012.png|[[2012_Summer_Project_Week#Projects|Projects List]]
 
Image:PW-MIT2012.png|[[2012_Summer_Project_Week#Projects|Projects List]]
 +
Image:CTUSDiagram.png|Sequence diagram
 +
Image:CTUSModule.png|Module interface
 +
Image:CTUSTools.png|Slicer plug-ins
 +
Image:CTUSReg.png|Registered phantom(L3)
 
</gallery>
 
</gallery>
 
 
==Key Investigators==
 
==Key Investigators==
* University of British Columbia, Robotics & Control Laboratory
+
* University of British Columbia, Robotics & Control Laboratory: Saman Nouranian, Samira Sojoudi
 
* Queen's University
 
* Queen's University
 
 
<div style="margin: 20px;">
 
<div style="margin: 20px;">
 
<div style="width: 27%; float: left; padding-right: 3%;">
 
<div style="width: 27%; float: left; padding-right: 3%;">
 
 
<h3>Objective</h3>
 
<h3>Objective</h3>
We are developing a new module as a puzzle block for a spine injection image-guided intervention.<br />
+
To bring the realtime needle navigation for spine injections into 3D-slicer framework, a chain of tools are required.
The ultimate goal is to plan for injection based on prior CT-image and perform the treatment using ultrasound-based needle guidance.
+
Ultrasound machine is one of the most applicable devices for realtime guidance due to its non-poisonous physics behind and low cost availability.
A volumetric representation of the patient's lumbar section is being reconstructed using tracked frames.
+
Out of the patient's lumbar section of the spine, a 3D volume is reconstructed using tracked frames acquired from ultrasound machine. This volume needs to be registered to prior CT-image of the same area of the patient. At the end a realtime guidance is performed based on outputs generated.
This volume should go through a registration algorithm to rigidly align with the model generated from CT image.
+
<p>
 
+
In this phase, we are focused on developing a new module for slicer that performs a rigid registration between segmented CT (model) and ultrasound volumetric representation (reconstructed volume). Current algorithm is based on single vertebra inputs and will be extended to include mechanical characteristics of spine for a multi-vertebrae case.
 
+
</p>
 
 
 
 
 
 
  
 
</div>
 
</div>
Line 27: Line 25:
  
 
<h3>Approach, Plan</h3>
 
<h3>Approach, Plan</h3>
A bone probability volume is generated from the original ultrasound volume.  
+
We assume a single vertebra model and volume are provided using existing slicer modules, like crop-volume and segmentation (editor).
From the CT image, a subset of visible points is extracted.
+
From the CT model, a subset of visible points is extracted regarding position of the probe in ultrasound data acquisition step.
A guassian mixture model method is performed to solve for this surface to volume registration problem.
+
3D ultrasound volume (reconstructed before) is processed to generate a bone probability volume. <br/>
 
+
An iterative optimization algorithm is performed using Guassian Mixture Model method to solve for this surface to volume registration problem.
 +
Results of each iteration step is visualaized in slicer to show the convergence of the algorithm.<br/>
 +
Output from each step of the algorithm and all transforms from each iteration are saved to disk and scene.
 
</div>
 
</div>
  
Line 36: Line 36:
  
 
<h3>Progress</h3>
 
<h3>Progress</h3>
All implementations are based on a single vertebra registration for now.
+
* A loadable module is created which accepts a CT model (polydata) and an ultrasound volume (scalar) as inputs. All inputs need to be limited to region of interest (i.e. a single vertebra: L3)
A loadable module is created which accepts a CT model (polydata) and an ultrasound volume (scalar) as inputs.  
+
* Core implementation of the algorithm is MATLAB-based (The algorithm is quite fast and speed is not an issue). Re-distributable MATLAB Compiler Runtime (MCR) is a prerequisite for the slicer module to run on any machine.
All inputs need to be limited to region of interest (i.e. a single vertebra: L3).
+
* A spine phantom data (L1-5) is used for the experiment.
Core implementation of the algorithm is MATLAB-based, since the algorithm is quite fast and speed is not an issue.
+
<h3>Progress during project week</h3>
A gaussian mixture model is used to register surface to volume. output of the module is the rigid transformation matrix obtained in this way.
+
* MATLAB core is compiled and exported as shared libraries to be used in external projects.
A spine phantom data is used for validation. Patient recruitment is an ongoing task for this project.
+
* A wrapper project is added to the slicer solution to provide an interface with compiled MATALB dlls.
 
+
* Entire procedure for single vertebra registration is implemented using slicer interface and current plugin modules:
 +
** loading complete ultrasound volume of the lumbar phantom
 +
** cropping ultrasound volume using "Crop Volume" module
 +
** Segmentation of the cropped CT volume using "Editor" module
 +
** Passing model and volume nodes to "SpineCTUSRegistration" module and calling the registration commands
 +
<h3>Works to do</h3>
 +
* Still there is a need to initialize model and volume inputs regarding LPS/RAS orientation conflicts
 +
* Improving animated visualization of the transform evolution on each iteration
 +
* Algorithm modification by using original MHA files instead of reconstructed VTK files
 +
* Performing entire procedure for a real patient data.
  
 
</div>
 
</div>
Line 55: Line 64:
 
##Built-in
 
##Built-in
 
##Extension -- commandline
 
##Extension -- commandline
##Extension -- loadable
+
##Extension -- loadable --> Yes
 
#Other (Please specify)
 
#Other (Please specify)
 
==References==
 
*Fletcher P, Tao R, Jeong W, Whitaker R. [http://www.na-mic.org/publications/item/view/634 A volumetric approach to quantifying region-to-region white matter connectivity in diffusion tensor MRI.] Inf Process Med Imaging. 2007;20:346-358. PMID: 17633712.
 
* Corouge I, Fletcher P, Joshi S, Gouttard S, Gerig G. [http://www.na-mic.org/publications/item/view/292 Fiber tract-oriented statistics for quantitative diffusion tensor MRI analysis.] Med Image Anal. 2006 Oct;10(5):786-98. PMID: 16926104.
 
* Corouge I, Fletcher P, Joshi S, Gilmore J, Gerig G. [http://www.na-mic.org/publications/item/view/1122 Fiber tract-oriented statistics for quantitative diffusion tensor MRI analysis.] Int Conf Med Image Comput Comput Assist Interv. 2005;8(Pt 1):131-9. PMID: 16685838.
 
* Goodlett C, Corouge I, Jomier M, Gerig G, A Quantitative DTI Fiber Tract Analysis Suite, The Insight Journal, vol. ISC/NAMIC/ MICCAI Workshop on Open-Source Software, 2005, Online publication: http://hdl.handle.net/1926/39 .
 

Latest revision as of 15:18, 22 June 2012

Home < 2012 Summer Project Week:VertebraCTUSReg

Key Investigators

  • University of British Columbia, Robotics & Control Laboratory: Saman Nouranian, Samira Sojoudi
  • Queen's University

Objective

To bring the realtime needle navigation for spine injections into 3D-slicer framework, a chain of tools are required. Ultrasound machine is one of the most applicable devices for realtime guidance due to its non-poisonous physics behind and low cost availability. Out of the patient's lumbar section of the spine, a 3D volume is reconstructed using tracked frames acquired from ultrasound machine. This volume needs to be registered to prior CT-image of the same area of the patient. At the end a realtime guidance is performed based on outputs generated.

In this phase, we are focused on developing a new module for slicer that performs a rigid registration between segmented CT (model) and ultrasound volumetric representation (reconstructed volume). Current algorithm is based on single vertebra inputs and will be extended to include mechanical characteristics of spine for a multi-vertebrae case.

Approach, Plan

We assume a single vertebra model and volume are provided using existing slicer modules, like crop-volume and segmentation (editor). From the CT model, a subset of visible points is extracted regarding position of the probe in ultrasound data acquisition step. 3D ultrasound volume (reconstructed before) is processed to generate a bone probability volume.
An iterative optimization algorithm is performed using Guassian Mixture Model method to solve for this surface to volume registration problem. Results of each iteration step is visualaized in slicer to show the convergence of the algorithm.
Output from each step of the algorithm and all transforms from each iteration are saved to disk and scene.

Progress

  • A loadable module is created which accepts a CT model (polydata) and an ultrasound volume (scalar) as inputs. All inputs need to be limited to region of interest (i.e. a single vertebra: L3)
  • Core implementation of the algorithm is MATLAB-based (The algorithm is quite fast and speed is not an issue). Re-distributable MATLAB Compiler Runtime (MCR) is a prerequisite for the slicer module to run on any machine.
  • A spine phantom data (L1-5) is used for the experiment.

Progress during project week

  • MATLAB core is compiled and exported as shared libraries to be used in external projects.
  • A wrapper project is added to the slicer solution to provide an interface with compiled MATALB dlls.
  • Entire procedure for single vertebra registration is implemented using slicer interface and current plugin modules:
    • loading complete ultrasound volume of the lumbar phantom
    • cropping ultrasound volume using "Crop Volume" module
    • Segmentation of the cropped CT volume using "Editor" module
    • Passing model and volume nodes to "SpineCTUSRegistration" module and calling the registration commands

Works to do

  • Still there is a need to initialize model and volume inputs regarding LPS/RAS orientation conflicts
  • Improving animated visualization of the transform evolution on each iteration
  • Algorithm modification by using original MHA files instead of reconstructed VTK files
  • Performing entire procedure for a real patient data.

Delivery Mechanism

This work will be delivered to the NA-MIC Kit as a (please select the appropriate options by noting YES against them below)

  1. ITK Module
  2. Slicer Module
    1. Built-in
    2. Extension -- commandline
    3. Extension -- loadable --> Yes
  3. Other (Please specify)