Numerical observer studies comparing FORE+AWOSEM, FORE+NECOSEM and NEC based fully 3-D OSEM for 3-D whole-body PET imaging

In this work, we use the NEC (Noise Equivalent Count) transformation method of Nuyts in the OSEM reconstruction technique as a procedure to partially restore the Poisson characteristics of 3-D whole-body PET pre-corrected data. We compare the use of FORE followed by NEC-weighted OSEM (FORE+NECOSEM) with FORE+AWOSEM was well as with fully 3-D implementations of NEC weighted OSEM, in terms of lesion detectability, using a 3-D version of the non-prewhitening matched filter numerical observer. In FORE+NECOSEM, only a multiplicative weight was applied (NEC-scaling) that accounts for the effects of both the quantitative corrections and FORE itself. For the fully 3-D reconstruction, we used both the shifted-Poisson approach (SPOSEM3D) and a 3-D implementation of NEC-scaling OSEM. In addition to NECOSEM, FORE+AWOSEM (attenuation weighted) and ANWOSEM3D (attenuation and normalization weighted) were also tested. We simulated 3-D whole-body PET exams with 1 cm diameter lesions. The results demonstrate an overall improved performance of FORE+NECOSEM and fully 3-D algorithms compared to FORE+AWOSEM. We also found that the behavior of the fully 3-D NECOSEM reconstruction algorithm was sensitive to the manner of calculating the NEC weights.