Purpose: Intratumoral treatment response can be detected using a pre-treatment and serial FDG-PET/CT imaging feedback during the treatment and used to guide dose painting. However, to reliably implement this technique, the effect of uncertainties in Â¹â?¸F-FDG PET/CT imaging/processing on response assessment and dose painting needs to be determined.
Methods: Three major uncertainties, (1) PET image acquisition time variation (TI), (2) PET image partial volume effect (PVE) and (3) discrepancy of deformable image registration (DIR) for tumor, were determined using images obtained from a cohort of 18 head and neck cancer patients treated with chemo-radiotherapy. Perturbation analysis was performed to quantify the effect of the three uncertainties on the tumor voxel baseline/pretreatment standard uptake value (SUV0), dose response matrix (DRM) and dose prescription function (DPF), where the uncertainty of TI was determined using the linear interpretation from the patient dual-point measured PET images, PVE was determined using the Richardson-Lucy iterative deconvolution method and DIR was determined using the two different registration processes, i.e. target weekly PET/CT image to the pretreatment PET/CT image directly or target weekly PET/CT image to its early adjacent weekly PET/CT image iteratively until the pretreatment PET/CT image.
Results: TI and PVE induced (-7 Â± 6.78)% and (1.78 Â± 8.01)% of tumor voxel SUV0 discrepancies respectively. TI, PVE and DIR induced (5.26 Â± 7.29)%, (0.62 Â± 8.5)% and (7.16 Â± 8.04)% of tumor voxel DRM discrepancies. As consequence, they caused tumor voxel DPF discrepancies to be -1.95 Â± 3.89, 2.72 Â± 4.72 and 2.09 Â± 3.35 Gy of EQD2 respectively.
Conclusion: Effect of uncertainties in FDG-PET/CT imaging/processing on intratumoral treatment response assessment was not negligible. They should be properly compensated for in the treatment planning of dose painting.