Difference between revisions of "Projects:KnowledgeBasedBayesianSegmentation"

From NAMIC Wiki
Jump to: navigation, search
Line 1: Line 1:
  Back to [[NA-MIC_Internal_Collaborations:StructuralImageAnalysis|NA-MIC Collaborations]], [[Algorithm:GATech|Georgia Tech Algorithms]], [[Engineering:Kitware|Kitware Engineering]]
+
  Back to [[NA-MIC_Internal_Collaborations:StructuralImageAnalysis|NA-MIC Collaborations]], [[Algorithm:BU|Boston University Algorithms]], [[Engineering:Kitware|Kitware Engineering]]
 
__NOTOC__
 
__NOTOC__
 
= Knowledge Based Bayesian Segmentation =
 
= Knowledge Based Bayesian Segmentation =

Revision as of 16:16, 12 January 2012

Home < Projects:KnowledgeBasedBayesianSegmentation
Back to NA-MIC Collaborations, Boston University Algorithms, Kitware Engineering

Knowledge Based Bayesian Segmentation

This ITK filter is a segmentation algorithm which utilizes Bayes's Rule along with an affine-invarient anisotropic smoothing filter.

Description

Use Case

I'd like to segment a volume or sub-volume into 'N' classes in a very general manner. I will provide the data and the number of classes that I expect and the algorithm will output a labelmap with 'N' classes.

Data

We have applied this algorithm to 20 normal brain MRI data-sets. We used publicly available data-sets from the Internet Brain Segmentation Repository (IBSR) offered by the Massachusetts General Hospital, Center for Morphometric Analysis. The IBSR data-sets are T1-weighted, 3D coronal brain scans after having been positionally normalized. Manual expert segmentations for these data-sets are publicly available and represent the ground truth used in this work.

Algorithm

This algorithm can be cast in either a static or dynamic framework. In the static framework, the following is the algorithm:

  1. The user sets the number of distinct classes for segmentation: 'N'
  2. Generate 'N' prior images (default, 'N' uniform prior images)
  3. Generate 'N' statistical distributions (default, 'N' normal distributions)
  4. Generate 'N' membership images by applying the statistical distributions to the raw data
  5. Generate 'N' posterior images by applying Bayes' rule to the prior and membership images
  6. Smooth the posterior images for 'm' iterations using an affine-invarient anisotropic smoothing filter and renormalize after each iteration (default, m = 5)
  7. Apply maximum a posteriori rule to apply labeling and finalize segmentation

In the dynamic framework, the following image depicts the adaptation of the static framework to the dynamic formulation:

Dynamic Tissue Tracking Algorithm


The ITK filter design


Flowchart

Some Results

  • White Matter Performance on the 20 ISBR datasets WM Algorithm Comparisons
  • Gray Matter Performance on the 20 ISBR datasets GM Algorithm Comparisons
  • Visual Results Visual Results on ISBR data

Project Status

  • Fully incorporated into itkBayesianClassificationImageFilter and itkBayesianClassificationInitializationImageFilter in ITK CVS.
  • Fully wrapped in VTK for use in Slicer.
  • The working ITK code has been committed to the SandBox

Key Investigators

  • Georgia Tech Algorithms: John Melonakos, Yi Gao, Allen Tannenbaum
  • Kitware Engineering: Luis Ibanez, Karthik Krishnan

Publications

In Print