PDE Software for Digital Image Management

Abstract

In Phase I research, we focused on the research and development of variational level-set methods and their application to the segmentation of the surface of the brain from a three-dimensional MRI dataset and the segmentation of the prostate from a three-dimensional trans-rectal ultrasound dataset. This report contains detailed discussion on curvature-based image processing techniques with application to medical imaging. We begin with a general overview of level-set methods and its application to image segmentation. Implementation of level-set methods requires careful attention to accuracy and robustness as well as computational speed. We discuss the implementation issues we have encountered and solutions to them. Since simple segmentation methods often fail in images where there are indistinct or missing boundaries (as commonly found in medical imagery), we discuss two incremental improvements we used to address this problem. Finally, two new models using region-based features are introduced which provide further robustness to weak or blurred edges.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jul 26, 2000
Accession Number
ADA384519

Entities

People

  • Jill Goldschneider
  • Lydia Ng
  • Vikram Chalana

Tags

Communities of Interest

  • Biomedical
  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Blood Cells
  • Boundaries
  • Computational Science
  • Computer Vision
  • Contracts
  • Curvature
  • Data Sets
  • Detection
  • Digital Images
  • Fluid Dynamics
  • Geometric Forms
  • Image Processing
  • Image Segmentation
  • Images
  • Lines (Geometry)
  • Prostate
  • Three Dimensional

Fields of Study

  • Computer science
  • Medicine

Readers

  • Computer Vision.
  • Finite Element Method (FEM) for solving Partial Differential Equations (PDEs)
  • Medical Imaging.