Effects of the collimator magnification factor in the geometrical calibration of SPECT systems

In compact systems, precise measurement in the projection space may be compromised due to minification. The objective of this work is to investigate the impact of the magnification factor in a model-based calibration procedure. This has direct relevance to the geometrical calibration of the clinical INSERT camera. Projection data from three point sources were simulated for a single pinhole collimator with magnification and single pinhole and slit-slat collimators with minification, for 100 noise realizations and 3 count levels. Model-based calibration was performed to estimate geometric parameters and data corresponding to a Derenzo phantom were simulated and reconstructed with true and worst-case estimates for each collimator. Experimental projection data were acquired with an INSERT prototype camera and four
^99m
Tc line sources in different locations within the FoV. The collimator-CoR distance was varied in order to obtain different minifications and model-based calibration was performed. The results from the simulations suggest that calibration is less robust when minification is present, with higher biased calibration parameters, which result in activity underestimation. For experimental data, estimated parameters improved with a higher magnification factor, in line with the simulation results. However, some inconsistencies in the results suggest that there is still room for improvements. To conclude, geometric calibration of SPECT systems is more sensitive to minification than magnification, which will impact image quality.