Limited view CT reconstruction and segmentation via constrained metric labeling

  • Authors:

  • Vikas Singh;Lopamudra Mukherjee;Petru M. Dinu;Jinhui Xu;Kenneth R. Hoffmann

  • Affiliations:

  • Biostatistics & Medical Informatics and Computer Sciences, University of Wisconsin-Madison, Madison, WI 53706, USA;Mathematical and Computer Sciences, University of Wisconsin-Whitewater, Whitewater, WI 53190, USA;Computer Science & Engineering, State University of New York at Buffalo, Buffalo, NY 14260, USA;Mathematical and Computer Sciences, University of Wisconsin-Whitewater, Whitewater, WI 53190, USA;Toshiba Stroke Research Center, State University of New York at Buffalo, Buffalo, NY 14214, USA

  • Venue:
  • Computer Vision and Image Understanding
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper proposes a new discrete optimization framework for tomographic reconstruction and segmentation of CT volumes when only a few projection views are available. The problem has important clinical applications in coronary angiographic imaging. We first show that the limited view reconstruction and segmentation problem can be formulated as a 'constrained' version of the metric labeling problem. This lays the groundwork for a linear programming framework that brings metric labeling classification and classical algebraic tomographic reconstruction (ART) together in a unified model. If the imaged volume is known to be comprised of a finite set of attenuation coefficients (a realistic assumption), given a regular limited view reconstruction, we view it as a task of voxels reassignment subject to maximally maintaining consistency with the input reconstruction and the objective of ART simultaneously. The approach can reliably reconstruct (or segment) volumes with several multiple contrast objects. We present evaluations using experiments on cone beam computed tomography.