The effect of different reconstruction algorithms in FDG PET/CT imaging on quantitative evaluation in lymphoma patients

Introduction: Lymphoma is a heterogeneous group of malignancies that respond to curative treatments. Factors such as treatment and survival, histological subtype, clinical stage, and response to treatment affect its prognosis. Fluor-18 Fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) is frequently used in lymphoma, and semi-quantitative parameters such as standardized uptake value (SUV) and the Deauville scoring system obtained with these parameters have an important place in the evaluation of the disease. Among the many factors affecting radiopharmaceutical uptake and SUV measurements ​​in FDG PET/CT imaging are reconstruction parameters. Therefore, different reconstruction methods used in PET/CT imaging may change Deauville scores in the evaluation of lymphoma patients. The aim of our study is to investigate the effect of different reconstruction methods on the quantitative evaluation of FDG PET/CT in lymphoma patients. Material and Methods: Raw PET data of 255 patients who underwent FDG PET/CT imaging with the diagnosis of lymphoma between November 1, 2020 and April 30, 2021 were reconstructed separately with Q.Clear(GE Healthcare), a Bayesian Penalized Likelihood(BPL) algorithm, and SharpIR(GE Healthcare), an Ordered Subset Expectation Maximization(OSEM) algorithm. In both images, each patient's liver, mediastinal blood pool and SUV measurements of a total of 487 lesions selected from the patients were performed and Deauville scores were calculated. The data were evaluated statistically. Results: In our study, we determined that BPL and Deauville scores increased significantly (p<0.001), and the scoring increased to 4 in thirty patients who were evaluated as 3 with OSEM. It was found that the quantitative values of the lymph nodes and the signal background ratios increased statistically with BPL (p<0.001), and the quantitative values of the liver from the reference regions were significantly decreased (p<0.001). In addition, this difference in lymph node quantification was independently associated with the size of the lesion and was significantly more pronounced in small (<13 mm) lesions (p<0.001). The effects of the BPL algorithm were more pronounced in the SUVmax measurements than in the SUVmean and SUVpeak measurements. Deauville-mean and Deauville-peak scores, which are Deauville scores calculated by SUVmean and SUVpeak measurements, were less changed by BPL than Deauville-max scores calculated by SUVmax measurements. Conclusion: Different reconstruction algorithms in FDG PET/CT affect the quantitative evaluation. This may affect the change in Deauville scores in lymphoma patients, thus affecting patient management.