A clinical evaluation of the impact of the Bayesian penalized likelihood reconstruction algorithm on PET FDG metrics

This is the authors' abstract. We don't add key points for this paper.

PURPOSE: The aim of this study was to evaluate the impact of using the Bayesian penalized likelihood (BPL) algorithm on a bismuth germanium oxide positron emission tomography (PET)/computed tomography (CT) system for F-FDG PET/CT exams in case of low injected activity and scan duration. MATERIALS AND METHODS: F-FDG respiratory gated PET/CT performed on 102 cancer patients, injected with ∼2 MBq/kg of F-FDG, were reconstructed using two algorithms: ordered subset expectation maximization (OSEM) and BPL. The signal-to-noise ratio (SNR) was calculated as the ratio of mean standard uptake value (SUV) over the standard deviation in a reference volume defined automatically in the liver. The peak SUV and volumes were also measured in lesions larger than 2 cm thanks to the automated segmentation method. RESULTS: On 85 respiratory gated patients, the median SNR was significantly higher with BPL (P<0.0001) and it is even better when the BMI of the patient increases (odds ratio=1.26).For the 55 lesions, BPL significantly increased the SUVpeak [difference: (-0.5; 1.4), median=0.4, P<0.0001] compared with OSEM in 83.6% of the cases. With BPL, the volume was lower in 61.8% of the cases compared with OSEM, but this was not statistically significant. CONCLUSION: The BPL algorithm improves the image quality and lesion contrast and appears to be particularly appropriate for patients with a high BMI as it improves the SNR. However, it will be important for patient follow-up or multicenter studies to use the same algorithm and preferably BPL.