Page 23 - Quantitative Imaging of Small Tumours with Positron Emission Tomography

Chapter 1 Most PVC methods developed to date are post-reconstruction methods, which can be applied to PET images that are acquired clinically. The region-based methods require exact definitions of anatomical or PET-based regions, which often requires co-registered high-definition images (80). In oncological PET, this is problematic since low-dose CT does not allow for exact tumour delineation. The region-based PVC methods are mainly used in functional neuroimaging, where grey and white matter regions are more easily segmented (80,81). For tumour imaging, most studies have used a simple recovery coefficient correction using phantom data (82-84). More advanced or parametric PVC methods might be more suitable for oncological PET since these do not rely on segmentation of lesion boundaries. However, it is yet unclear how the net accuracy is affected by tumour delineation accuracy (85,86). In recent years, use of reconstruction-based methods has become more common, especially since vendors started to provide such algorithms with novel PET systems (87-89). These reconstructions have the advantage of higher detection rates through improved spatial resolution, but could have lower quantitative accuracy due to increased image noise characteristics and Gibbs ringing artefacts (90,91). As these image reconstructions with corrections for PVE are being implemented worldwide, the impact of PVC needs to be taken into account in quantitative PET studies. This might be of particular importance for test-retest and response monitoring studies, as the lower precision caused by PVC methods might affect the minimal detectable change in tumour uptake. 22