TY - JOUR T1 - Partial volume correction strategies for quantitative FDG PET in oncology JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 1494 LP - 1494 VL - 50 IS - supplement 2 AU - Nikie Hoetjes AU - Adriaan Lammertsma AU - Otto Hoekstra AU - Corneline Hoekstra AU - Nanda Krak AU - Ronald Boellaard Y1 - 2009/05/01 UR - http://jnm.snmjournals.org/content/50/supplement_2/1494.abstract N2 - 1494 Objectives Quantification of FDG PET using SUV is affected by partial volume effects. Therefore different partial volume correction (PVC) methods were evaluated using simulations, together with phantom and clinical data. Methods 3 PVC methods were tested: (1) inclusion of the point spread function in the reconstruction; (2) iterative deconvolution of reconstructed images; (3) calculation of spill-in and spill-out factors based on tumour masks. Simulations were based on a mathematical phantom including variously sized and shaped tumors. Phantom data were obtained using the NEMA NU2 image quality phantom. Clinical evaluation included a test-retest study (n=14, lung cancer) and a response monitoring study (n=16, breast cancer, 2 to 4 scans per patient). In all studies volumes of interest were generated using an adaptive relative thresholds technique. Results Simulations and phantom data revealed similar results. All methods were able to recover true SUV within 10% for spheres larger than 1 ml. Reconstruction based recovery, however, provided up to 2-fold better precision than image based methods. Clinical studies showed that PVC increased SUV with 5 to 80%. Test-retest variability slightly worsened from 9.8±6.5% without to 10.8±7.6% with PVC. Finally, PVC resulted in a smaller SUV response, i.e. from -30% without to -26% with PVC after 1 cycle of treatment (p < 0.001). Conclusions PVC improved accuracy of SUV without decreasing test-retest variability significantly. It had a small, but significant, effect on observed tumour response. Reconstruction based PVC outperformed image based methods, although differences were small for clinical data. ER -