Evaluation of the clinical performances of a large NaI(Tl) crystal 3D PET scanner.

Aim: This study was aimed at assessing the clinical performances of a NaI(Tl) crystal 3D PET scanner, C-PET (ADAC-UGM), using a multi-ring 2D BGO PET scanner (multi-ring PET), as a reference.

Methods: Thirty-seven oncological patients were studied in sequence with multi-ring PET and C-PET, within 30 days of a CT study. In order to assess the behaviour of C-PET in relation to acquisition count rate, patients were divided into 3 groups according to the count rate at the time of the C-PET scan acquisition. Group A (n=21): 3000-5000 kcounts/sec (recommended count rate range); Group B (n=8): <3000 Kcounts/sec and Group C (n=8): >5000 Kcounts/sec.

Results: The number of lesions detected by multi-ring PET and C-PET, classified according to size, was compared. For Group A and Group B there was a good agreement between C-PET and multi-ring PET in terms of lesion detectability (relative sensitivity: 99.9% and 96.0%, respectively), while for Group C the relative sensitivity of C-PET was 61.9%.

Conclusion: Optimal performances of the C-PET scanner can thus be obtained at a count rate within or below the recommended range. Despite a lower lesion/background contrast resulting from a high scatter and random noise, the sensitivity of C-PET in detecting hypermetabolic lesions is comparable to that of multi-ring PET. These findings are discussed in relation to the physical performance of the two scanners and particularly in relation to the 3D vs 2D acquisition modality.