Quantitative capabilities of current commercial SPECT-CT systems

Aim: To assess the quantitative capabilities of commercial SPECT-CT systems with their full 3D iterative reconstruction which includes resolution recovery, attenuation and scatter corrections.
Materials and methods: A NEMA NU2-1994 attenuation and scatter correction accuracy phantom, a cylindrical phantom with uniform part, cold and hot rods (diameter 4-25 mm), three uniform cylindrical phantoms of different sizes were scanned on a Siemens Symbia T6, a Philips BrightView XCT and a GE Infinia Hawkeye 4. For each phantom, activity concentration was obtained from the technetium-99m activity measured with an activimeter and its water volume. Emission projections were collected for 128 angles over 360 in 128*128 matrix (pixel size of 2.7 mm) using the 140 keV system usual energy window, and standard CT attenuation protocol was executed. Full 3D iterative reconstructions (Siemens Flash3D, Philips Astonish, GE Evolution) were performed. Residual fraction (RF) in the cold inserts of the NEMA phantom, contrast recovery coefficient (CRC) in the cold and hot rods were computed from the images. Regions of interest (ROI) of different sizes (60-100% of the phantom physical size) were drawn on the three uniform phantoms to obtain a calibration factor (CF).
Results: RF dropped below 4% and CRC almost stabilized after 18 iterations. Hot rod CRC were above 90% for the rods with a diameter larger than 16 mm rising above 100% in some cases. Accurate quantification was shown feasible to within 5% or even less on some systems. CF was found to depend on both the phantom and the ROI sizes used for its determination, with variation of a few percents to more than 10%. The best phantom/ROI combination was also dependent on the system.
Conclusion: With a careful design of the calibration procedure, quantification with these SPECT-CT systems and 3D iterative reconstruction seems to be feasible. However partial volume effect will remain a concern when object size drops below 2 cm.