Image quality and quantification in99mTcmyocardial SPECT-CT with a 3D CZT camera: what to expect from BSREM reconstruction methods?

Objective. The 3D CZT SPECT-CT offers betterγray detection, particularly improved energy resolution, which could be useful for more accurately accounting for scattered radiation. These new cameras also benefit from advanced reconstruction methods. Notably, the BSREM algorithm controls noise during the iterative process. This study aims to investigate the influence of using a narrow energy window for scatter correction on image quality and quantification performance. It will also examine the regularisation methods (and their parameters) implemented by the manufacturer.Approach. Acquisitions were performed on two99mTc-filled phantoms (a thorax-mimicking phantom including a myocardium insert and a cylindrical tank). Scatter was taken into account using the Jaszczak subtraction method (DEW) and a narrow photopeak window (SEW). The images were reconstructed using the Median Root Prior and Relative Difference Prior regularisation methods. The following were evaluated: sensitivity, sharpness index (a spatial resolution indicator), contrast-to-noise ratio, signal-to-noise ratio, and concentration recovery.Main results.Sensitivity was independent of the regularisation method and was measured at (942.4±2.1) and (1074.4±12.8) counts.(MBq.s)-1, respectively, for DEW and SEW. The sharpness index ranged from (0.138±0.014) to (0.033±0.003) mm-1, the contrast-to-noise ratio ranged from (4.0±0.2) to (2.1±0.1), and the signal-to-noise ratio ranged from (11.2±0.4) to (4.1±0.1). Concentration recovery decreased with theβparameter for both MRP and RDP, as well as for both scatter corrections.Significance. Using a narrower spectrometric window makes the system more sensitive, providing equivalent image and quantification quality to that achieved using the Jaszczak subtraction method. When the BSREM reconstruction method is used, the quality of the images and the quantification in myocardial SPECT-CT are highly dependent on the regularisation methods and their parameters.