Difference between revisions of "2008 Winter Project Week:Prostate Segmentation"
| Line 32: | Line 32: | ||
====Jan 2008 Project Half Week==== | ====Jan 2008 Project Half Week==== | ||
| − | + | 1. The grow cut algorithm, implemented from the paper: "Vezhnevets V., Konouchine V., 'Grow-Cut' - Interactive Multi-Label N-D Image Segmentation. Graphicon-2005", has been written as an ITK filter and submitted into the NA-MIC Sandbox(http://www.na-mic.org/ViewVC/index.cgi/trunk/itkGrowCutImageFilter/?root=NAMICSandBox). | |
| + | The filter deals with N-Dimensional scalar value image. | ||
| + | |||
| + | 2. The Grow Cut filter needs a rough initial manually labeling. This is solved by using an initial label map image which can be used through the whole data set. | ||
| + | |||
| + | 3. The goal is to put the filter into Slicer where the manual labeling problem won't be a problem anymore. | ||
</div> | </div> | ||
| Line 39: | Line 44: | ||
</div> | </div> | ||
| − | |||
===References=== | ===References=== | ||
Revision as of 21:05, 7 February 2008
Home < 2008 Winter Project Week:Prostate Segmentation Return to 2008_Winter_Project_Week |
Key Investigators
- Georgia Tech: Yi Gao, Ponnappan Arumuganainar, John Melonakos, Allen Tannenbaum
- JHU: Gabor Fichtinger
- GE: Xiaodong Tao
Objective
In this work, we develop tools for the segmentation of the prostate from MRI and ultrasound imagery.
Approach, Plan
Our approach includes grow cut and spherical wavelet based segmentation tools. We will build a command-line tool using ITK which will then be integrated into Slicer3.
Progress
Jan 2008 Project Half Week
1. The grow cut algorithm, implemented from the paper: "Vezhnevets V., Konouchine V., 'Grow-Cut' - Interactive Multi-Label N-D Image Segmentation. Graphicon-2005", has been written as an ITK filter and submitted into the NA-MIC Sandbox(http://www.na-mic.org/ViewVC/index.cgi/trunk/itkGrowCutImageFilter/?root=NAMICSandBox). The filter deals with N-Dimensional scalar value image.
2. The Grow Cut filter needs a rough initial manually labeling. This is solved by using an initial label map image which can be used through the whole data set.
3. The goal is to put the filter into Slicer where the manual labeling problem won't be a problem anymore.
