Difference between revisions of "2011 Winter Project Week:SegEye"

From NAMIC Wiki
Jump to: navigation, search
Line 22: Line 22:
 
<div style="width: 27%; float: left; padding-right: 3%;">
 
<div style="width: 27%; float: left; padding-right: 3%;">
  
<h3>Approach, Plan</h3>
+
<h3>Approach</h3>
  
We plan to  finalize a fully-automatic segmentation approach to identify the blood pool in MRAs. The approach uses the robust statistics segmentation framework developed earlier at Georgia Tech.
+
*Pivotal organ is considered the eye since its segmentation will localize the region of interest when looking for other structures.
 +
*We will reduce the dimensionality of this problem by performing model based segmentation for each structure.
 +
*Once the eye is segmented, we use this knowledge to locate the lens/ initialize optic nerve segmentation.
 +
*We would like to leverage the information provided by a CT scan with additional data from an MRI -- we have to consider this registration problem.
  
 
</div>
 
</div>

Revision as of 17:43, 23 December 2010

Home < 2011 Winter Project Week:SegEye

Key Investigators

  • Georgia Tech: Ivan Kolesov and Allen Tannenbaum
  • MGH: Gregory Sharp

Objective

  • We are interested in segmenting the eye ball, lens, optic nerve, and the optic chiasm.
  • Anatomical structures are highly sensitive to radiation.
  • We are creating a framework to perform these segmentations, which is likely to require a different approach for each structures due to the proximity of multiple structures to each other.

Approach

  • Pivotal organ is considered the eye since its segmentation will localize the region of interest when looking for other structures.
  • We will reduce the dimensionality of this problem by performing model based segmentation for each structure.
  • Once the eye is segmented, we use this knowledge to locate the lens/ initialize optic nerve segmentation.
  • We would like to leverage the information provided by a CT scan with additional data from an MRI -- we have to consider this registration problem.



Progress