Comparison of different motion correction techniques for dynamic FDG-PET/CT studies in breast cancer patients.

Abstract

BACKGROUND\nOur aim was evaluate interchangeability of different motion correction methods in the assessment of dynamic FDG-PET/CT studies in breast cancer patients as well as to assess the interrater reliability of these methods.\n\n\nMETHODS\nIn our prospective study we included patients with malignant breast tumours. Dynamic PET acquisition lasted for 60 minutes after tracer (FDG) injection. Every study was assessed by the same two experienced observers. We assessed plasma activity noninvasively. In case of the primary tumour VOIs we applied two different approaches to correct motion artefacts: Method I) frame-by-frame manual motion correction; Method II) frame-by-frame semi-automatic software-based motion correction. FDG two-compartment kinetic modelling was applied to assess K1, k2, k3 rate coefficients and to calculate Ki (tracer flux constant) and MRFDG (FDG metabolic rate).\n\n\nRESULTS\n35 lesions detected during 34 dynamic studies were included in this current analysis. Interrater reliability of both applied motion correction methods proved to be excellent (ICC= 0.89-0.99), except Ki measured by Method I (ICC=0.66). Bland-Altman analysis revealed that Method II resulted in significantly lower values than Method I regarding k3 and Ki in case of both observers, and regarding MRFDG in one of the observers. In case of K1 and k2 the two methods were in good agreement.\n\n\nCONCLUSIONS\nBoth applied methods proved to be reproducible and reliable, especially Method II, where every measured kinetic parameter showed excellent interrater reliability. Different approaches of motion correction could have a significant effect on the results of the kinetic modelling; therefore careful selection of the most reliable method is advised.