In-depth analysis of cone-beam CT image reconstruction by ideal observer performance on a detection task

We have recently perform evaluations, of the back-projection filtration (BPF) algorithm for cone-beam CT image reconstruction, based on ideal observer performance for signal detection. The metric used is the efficiency, which is here the squared ratio of the signal detection signal-to-noise ratios (SNRs) in the reconstructed image and in the projection data. We have generally found that efficiencies fall in the range of 75%–85% for minimum projection-data scans. We hypothesize that the efficiency should approach 100% for minimum data scans. In this work, we evaluate the individual steps in the BPF algorithm to identify, and possibly remedy, the loss of SNR due to the algorithm implementation. For the detection task specified here, the efficiency is improved from 82% to 96% by using this detailed analysis.

[1]  Harrison H Barrett,et al.  Validating the use of channels to estimate the ideal linear observer. , 2003, Journal of the Optical Society of America. A, Optics, image science, and vision.

[2]  Xiaochuan Pan,et al.  Theory and algorithms for image reconstruction on chords and within regions of interest. , 2005, Journal of the Optical Society of America. A, Optics, image science, and vision.

[3]  Kyle J. Myers,et al.  Foundations of Image Science , 2003, J. Electronic Imaging.

[4]  Xiaochuan Pan,et al.  Accurate computation of the Hotelling observer for the evaluation of image reconstruction algorithms in helical, cone-beam CT , 2007, 2007 IEEE Nuclear Science Symposium Conference Record.

[5]  Theodore G. Birdsall,et al.  Definitions of d′ and η as Psychophysical Measures , 1958 .

[6]  Eric Clarkson,et al.  Efficiency of the human observer detecting random signals in random backgrounds. , 2005, Journal of the Optical Society of America. A, Optics, image science, and vision.