View source for 2014 Summer Project Week:An ITK implementation of Physics-Based Non-Rigid Registration method for Brain Shift

__NOTOC__
<gallery>
Image:PBNRRSlicer.jpg | The PBNRR CLI module. Top Left: moving image, top right: moving image volume render, bottom left: fixed image, bottom right: warped moving image. 
Image:PBNRR_UI.jpg | PBNRR's UI.
</gallery>

==Key Investigators==
* Fotis Drakopoulos (CRTC)
* Yixun Liu (CRTC)
* Andriy Kot (CRTC)
* Andrey Fedorov (BWH/SPL)
* Olivier Clatz (Asclepios,INRIA)
* Ron Kikinis (BWH/SPL)
* Nikos Chrisochoides (CRTC)

[[Image:PBNRR_extension.jpg|800px]]

==Project Description==
This project implements ITK's Physics-Based Non-Rigid Registration (PBNRR) method. The PBNRR compensates for the brain shifts during the Image-Guided Neurosurgery (IGNS). The method uses a linear homogeneous bio-mechanical model to compute a dense deformation field that defines a transformation for every point in the fixed image to the moving image. The PBNRR includes the following three components combine together to provide a user-friendly interface. 
*Feature Point Selection. 
Selects highly discriminant features (blocks of voxels) in the moving image. We use the variance of the image intensity within the block region to measure its relevance and only select a fraction of all potential blocks based on a predefined parameter of the algorithm.
*Block Matching. 
Matches the selected blocks in the moving image with blocks in the fixed image using a predefined similarity measure (NCC).
*Finite Element Solver.
Builds a Finite Element (FE) bio-mechanical model, applies the block matching displacements to the model, and estimates the mesh deformations iteratively by first using an approximation method. Once many of the outliers from block matching are rejected, we use an interpolation formulation to compute the image deformation field that maps positions from the fixed to the moving coordinate frame.

<div style="margin: 20px;">
<div style="width: 27%; float: left; padding-right: 3%;">
<h3>Objective</h3>
* Develop an extension that encapsulates a CLI module for the '''PBNRR''' method.
</div>
<div style="width: 27%; float: left; padding-right: 3%;">
<h3>Approach, Plan</h3>
* The '''CBC_3D_I2MConversion''' extension can be used to generate suitable meshes for the PBNRR method.
</div>
<div style="width: 27%; float: left; padding-right: 3%;">
<h3>Progress</h3>
* The extension is available in the nightly build version of Slicer with name: '''PBNRR'''.
* The documentation page for the extension is available here: https://www.slicer.org/wiki/Documentation/Nightly/Extensions/PBNRR.
* Two brain MRI cases from SPL's public data-set are provided for testing.
</div>
</div>

==References==
* An ITK implementation of a Physics-Based Non-Rigid Registration method for Brain Deformation in Image-Guided Neurosurgery. 
Liu Y, Kot A, Drakopoulos F, Yao C, Fedorov A, Enquobahrie A, Clatz O and Chrisochoides NP (2014), Front. Neuroinform. 8:33. doi: 10.3389/fninf.2014.00033

* Non-Rigid Registration for Brain MRI: Faster and Cheaper.
Yixun Liu, Andrey Fedorov, Ron Kikinis and Nikos Chrisochoides.
International Journal of Functional Informatics and Personalized Medicine (IJFIPM), Publisher Inderscience Enterprises Ltd., Volume 3, No. 1, pages 48 -- 57, 2010

* Robust Non-Rigid Registration to Capture Brain Shift from Intra-Operative MRI.
Olivier Clatz, Hervé Delingette, Ion-Florin Talos, Alexandra J. Golby, Ron Kikinis, Ferenc Jolesz, Nicholas Ayache, and Simon Warfield, IEEE Transactions on Medical Imaging, 24(11):1417-1427, Nov. 2005