Background: ¹⁸F-fluorodeoxyglucose positron emission tomography has proven relevance in oncological diagnosis, staging and follow-up. The standardized uptake value (SUV) is one of the most widely used semi-quantitative criteria in PET imaging. However, factors such as noise and image resolution affect the measurement of the SUV. We reported earlier that a device-dedicated projector [attenuation-weighted ordered-subsets expectation maximization detector response (AW-OSEM DR), based on point-source measurements] introduces less noise than a geometrical model (AW-OSEM).
Objective: The aim of this study was to investigate the AW-OSEM DR method's impact on SUV measurements under clinical conditions.
Methods: We first performed a bias analysis to assess the accuracy of the quantitation for the two reconstruction methods as a function of target size and the number of iterations, with 14 acquisitions of the NEMA IEC/2001 phantom. We then used each method to calculate the maximum and average SUVs, respectively for 32 lesions.
Results: For all spheres and all iterations, the bias was significantly lower with AW-OSEM DR than with AW-OSEM (P=0.012). Moreover, a paired Student's t-test showed significant intermethod differences for maximum SUV and average SUV (both P<0.001) in cancer patients. Conversely, the two methods did not differ significantly in terms of the mean SUV and signal-to-noise ratio calculated in the liver for each patient (P=0.5 and 0.08, respectively).
Conclusion: Phantom and patient studies were performed to quantify the effects of AW-OSEM DR on PET images. The phantom study highlighted the fact that our method produces more accurate images in terms of the SUV, which is an essential quality for ensuring correct diagnosis, follow-up and treatment planning.