Quantitative comparison between regularized time-of-flight and OSEM PET reconstructions for small ¹⁸F-FDG-avid lesions

Objective We seek to quantitatively assess the relationship between lesion size and maximum standard uptake value (SUV_max) differences for small fluorine-18-fluorodeoxyglucose-avid lesions through the comparison of regularized time-of-flight (R-ToF) PET and standard ordered subset expectation maximization (OSEM) PET reconstructions. Patients and methods Data from 30 patients (male to female ratio 8: 7, 15 different primary malignancies) who underwent noncontrast PET-CT in October 2015 through February 2016 were evaluated. For each patient, multiple reconstructions of the PET data, including R-ToF, time-of-flight (ToF), and non-ToF, were constructed using OSEM of three iterations and 24 subsets. Volumes of interest were contoured for lesions up to 3 cm in size, and SUV_max measurements and lesion size were recorded for a total of 64 lesions. Liver measurements were also recorded for background analysis. A comparative statistical analysis was subsequently carried out. Results R-ToF showed the highest SUV_max for all 64 lesions and the highest SUV_max mean (P<0.0001) among all lesions, followed by ToF and non-ToF. R-ToF also showed a statistically significant negative correlation (P<0.001) between lesion size and SUV_max versus ToF and non-ToF PET; the smaller the lesion, the greater the disparity. R-ToF also showed the largest mean background signal to noise ratio (P<0.0001), mean lesion signal to noise ratio (P<0.001), and mean lesion signal to background ratio (P<0.0001) versus ToF and non-ToF PET. Conclusion R-ToF shows higher SUV_max values than standard OSEM reconstructions for all fluorine-18-fluorodeoxyglucose -avid lesions, with greater differences for increasingly smaller lesions. R-ToF offers improved suppression of background noise versus OSEM and therefore increased signal to noise and background ratios. These findings offer potential for improved small lesion visibility and characterization.