Evaluating iterative algebraic algorithms in terms of convergence and image quality for cone beam CT

  • Authors:

  • W. Qiu;T. Pengpan;N. D. Smith;M. Soleimani

  • Affiliations:

  • Engineering Tomography Laboratory (ETL), Department of Electronic and Electrical Engineering, University of Bath, Bath, BA2 7AY, UK;Engineering Tomography Laboratory (ETL), Department of Electronic and Electrical Engineering, University of Bath, Bath, BA2 7AY, UK;Engineering Tomography Laboratory (ETL), Department of Electronic and Electrical Engineering, University of Bath, Bath, BA2 7AY, UK;Engineering Tomography Laboratory (ETL), Department of Electronic and Electrical Engineering, University of Bath, Bath, BA2 7AY, UK

  • Venue:
  • Computer Methods and Programs in Biomedicine
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

Cone beam computed tomography (CBCT) enables a volumetric image reconstruction from a set of 2D projection data. It plays an important role in image guided radiation therapy (IGRT). However, it is desirable to lower the patient radiation dose while maintaining good quality tomographic reconstructions. Hence, methods are needed even when the data is undersampled or there are limited projections. Iterative algorithms such as ART, SART and OS-SART are known to perform well under such circumstances. The performance of ART, SART and OS-SART is here studied based on a range of norm measurements (RMS reconstruction error, RMS projection error, and the L1 norm and L2 norm of the image error). Image quality measurements for uniformity and noise are also introduced. Since image quality often degrades as iterative algorithms converge, a simple function is used to trade off convergence with image quality. Investigations are performed using simulated and experimental data.