Comparison of Different Types of Commercial Filtered Backprojection and Ordered-Subset Expectation Maximization SPECT Reconstruction Software

Central (A), radial (B), and tangential (C) FWHMs for FBP reconstruction of projections filtered with order 6 Butterworth filter at 4 cutoff frequencies. cy/px = cycles per pixel.

Percentage difference (Diff) in mean pixel counts between FBP reconstruction of projections filtered with Hanning (Hann) filter or Butterworth (But) filter and FBP reconstruction of unfiltered projections (ramp reconstruction). First digit after “But” indicates order of Butterworth filter. Last 2 digits after filter name abbreviation indicate 100 times cutoff, or cutoff frequency of cycles per pixel (e.g., “35” represents “frequency of 0.35 cycle per pixel”).

Contrast for hot (A–C) and cold (D) rods for FBP reconstruction of projections filtered with Hanning filter at cutoff frequencies of 0.2 (A and D), 0.35 (C), and 0.5 (B) cycles per pixel. For Jetstream, eSoft, Icon, Mirage, and Xeleris, values for contrast were almost identical, and only mean value for these 5 types of software is reported. SD was below 4%; error bars were smaller than square symbols and were omitted for clarity. CC for cutoff frequency of 0.35 or 0.5 cycle per pixel was found to be software independent.

Central (A), radial (B), and tangential (C) FWHMs for OSEM reconstruction with various numbers of subsets (s) and iterations (i). Some combinations were not available for some types of software.

Percentage variation (Diff = difference) of mean pixel counts from value obtained with 1 subset and 32 iterations for all combinations of numbers of subsets (s) and iterations (i) used in this study. Some combinations were not available for some types of software. (A) Jetstream and Xeleris. (B) Mirage and Vision.