View source for 2011 Summer Project Week Non-rigid inter-patient registration of bone masks derived from CT for Head and Neck Cancer Radiation Therapy

__NOTOC__
<gallery>
Image:PW-SLC2011.png|[[2011_Winter_Project_Week#Projects|Projects List]]
Image:HeadStrucs.png
Image:NeckStrucs.png
Image:ik_0624_1.png
Image:ik_0624_2.png
Image:ik_0624_3.png
Image:ik_0624_4.png
</gallery>

==Key Investigators==
* Georgia Tech: Ivan Kolesov, Yi Gao, and Allen Tannenbaum
* MGH: Gregory Sharp

<div style="margin: 20px;">
<div style="width: 27%; float: left; padding-right: 3%;">

<h3>Objective</h3>
*Obtain segmentations of a number of organs in the head, neck region.



</div>

<div style="width: 27%; float: left; padding-right: 3%;">

<h3>Approach, Plan</h3>
*Segmentation based on intensity alone is insufficient => does not lead to robust performance
*We require shape information to constrain the problem
*An atlas of organs must be created 
*We will explore deformable registration techniques applied to our data
*Also, we will consider possible sparse representations of our imagery to simplify the problem

</div>

<div style="width: 40%; float: left;">

<h3>Progress</h3>
*Tested pipeline to perform atlas based segmentation
*Used BRAINSFIT to perform Bspline registration of patient bone masks
*First results are promising
*Next Steps:
**Create patient masks to remove fixtures
**Automate the segmentation pipeline and use multiple atlases
**Use the atlases as initializations as opposed to final segmentations
</div>
</div>

<div style="width: 97%; float: left;">

==Delivery Mechanism==

This work will be delivered to the NA-MIC Kit as a command line module.

==References==