Automatic geometry calibration based on metric optimization in stationary computed tomography baggage scanner with 2π-angle sparsity

Abstract Recently, SSTLabs Co. in Korea has designed and developed a stationary computed tomography (SCT) baggage scanner with 2π-angle sparsity, comprising 20 pairs of monoblock-type X-ray sources and linear array-type dual-layer detectors arranged within a scan angle of 360 ∘ at an equiangular distance. We implemented an SCT reconstruction method for a specific (i.e., very sparse-view) system configuration by adopting an iterative algorithm based on the popular compressed-sensing (CS) mathematical theory. However, the method exhibited poor reconstruction quality owing to the mismatch in scan parameters between the nominal (designed) and actual values used during reconstruction. To overcome this limitation, we propose a pragmatic geometric calibration method based on a metric optimization technique for the mean structural similarity index measure to ensure high-quality CS-reconstructed images from the developed SCT baggage scanner. To verify the efficacy of the proposed method, we conducted a numerical simulation and experiments before its practical implementation and found no significant image distortion and edge blur in the SCT images reconstructed with the proposed geometric calibration, except for artifacts stemming from sparse-view angular sampling. The measured signal-to-noise ratio, contrast-to-noise ratio, and mean square error values between the calibrated and corresponding phantom slice images were improved by approximately 5.2-, 7.4-, and 3.5-fold, respectively, compared to those for the uncalibrated images, demonstrating the efficacy of the proposed geometric calibration method. Furthermore, the proposed method is self-operating and thus does not require a geometric calibration phantom.