Effect of block-iterative acceleration on Ga-67 tumor detection in thoracic SPECT

Through a combination of human localization ROC (LROC) and channelized Hotelling observer (CHO) ROC psychophysical studies, the authors investigated how accelerated OSEM and rescaled block-iterative EM (RBI) reconstruction affected tumor detection in simulated Ga-67 SPECT images. The tumors were 1-cm spheres within the chest region of the 3D mathematical cardiac-torso phantom. Previous work with iterative detector resolution compensation had shown that 8 iterations of the OSEM algorithm with a subset size of 8 (16 subsets) offered optimal observer performance. For the LROC study in this paper, the OSEM and RBI algorithms were implemented using subset sizes P and iterations K that satisfied the relation P=K for P=1,2,4, and 8. The CHO investigated reconstruction strategies that deviated from this relation. The authors´ results show that the strategies with P⩽2 penalized observer performance compared to strategies with larger subset sizes. Other researchers have reported on the more stable convergence and noise properties of the RBI algorithm, and in a similar vein, the authors found that an RBI strategy with a subset size of P produced the same performance as an OSEM strategy with subset size 2P. They conclude that the RBI algorithm with P=4 and K=3 should give human LROC performance on par with OSEM with P=8 and K=8