A critical phantom study of the energy window used for 99mTc quantitative explorations with a ring CZT SPECT system.

Background: This study investigated, mainly for quantitative ^99mTc explorations with a ring CZT SPECT system (GE HealthCare Starguide), the use of a narrow symmetric or a fully asymmetric energy window to reject scattered photons. The results were compared with the manufacturer's post-acquisition dual energy window approach.

Methods: Two uniform and two cold and hot rod contrast cylindrical phantoms of various sizes were scanned with the Starguide system to acquire a very high number of counts. After rebinning the list-mode files for different energy windows, data were reconstructed with manufacturer's iterative algorithm including attenuation correction, resolution recovery and eventually scatter correction, but without any regularization technique. Cold rod residual scatter fraction, hot and cold rod contrast recovery coefficient, coefficient of variation in phantom uniform areas and quantification accuracy using calibration with one of the homogeneous phantoms were, among others, computed.

Results: Narrow symmetric photopeak-centred windows or fully asymmetric (≥ 140 keV) window led, on one hand, to decreased scatter residual fraction and sensitivity and, on the other hand, to increased noise, cold and hot recovery coefficients when compared to a standard 15-20% wide symmetric window. With a 6-7% wide symmetric window we obtained very comparable results to the dual energy window scatter correction used by the manufacturer for all measured parameters, but larger recovery coefficients especially for small hot objects in a cold background. Similar results were obtained with the fully asymmetric window at the cost of a higher noise level resulting from a drastic reduction of the sensitivity.

Conclusions: Narrow symmetric or asymmetric energy windows were found an interesting alternative to the standard dual energy window method to reject ^99mTc scattered photons. As a key feature, they allowed to avoid the erasing of small hot objects in a null background that was observed with the standard dual energy window scatter correction.