Program Information
Uncertainty Analysis of a FDG-PET Imaging Based Tumor Voxel Dose Response Marker
S Chen*, D Yan , William Beaumont Hospital, Royal Oak, MI
Presentations
SU-E-605-6 (Sunday, July 30, 2017) 1:00 PM - 1:55 PM Room: 605
Purpose: Tumor voxel dose response marker (SF2) determined from multiple FDG-PET imaging has shown a great value on predicting local tumor recurrence (LR) and ability to be used as a novel objective for dose-painting-by-number. However, uncertainties included in the SF2 construction can introduce bias in its application. This study aims to quantify the uncertainties and their impact.
Methods: Tumor voxel dose response marker, SF2(v), has been constructed using pretreatment and weekly FDG-PET/CTs using the deformable image registration (DIR). SF2 of tumor voxel v was derived based on the changes of weekly SUV, within which a k-factor represents a linear relationship between the logarithm of the ratio of SUV change and the logarithm of the tumor cell survival fraction. The k-factor was determined assuming the average tumor cell size and cell density saturation in a voxel with the SUVmax at pretreatment. The SF2 uncertainty was formulated as ∆SF2(v)=f(v,∆SUV,∆k), where ∆SUV was induced by PET-imaging partial-volume-effect (PVE), PET-imaging time-interval after FDG injection (TI) and weekly-to-pretreatment PET/CT DIR; ∆k was induced by assumptions of tumor cell size, density saturation and heterogeneous FDG absorption. Perturbation analysis was performed including different time-phase PETs, DIR methods, PVE and k values to construct SF2. The SF2 uncertainty was quantified and its influence on the predictability of LR was evaluated by receiver-operating-characteristic analysis.
Results: SUV variability induced by PVE (assuming 15% SUV variability), DIR (inconsistency-vector-difference 1.24±1.07mm) and TI variation (interval 0.96±0.11hours) and different k values assuming tumor cell diameter range 10μm~20μm result in voxel SF2 variation of 0.06±0.13. Without perturbation, area under the curves (AUC) of tumor sub-volume with SF2>0.9 or 1.0 is 0.98~1 (p<0.003); with perturbation induced by uncertainties results in AUC drop of 0.04±0.02.
Conclusion: FDG-PET imaging based tumor dose response marker (SF2) is robust respect to the uncertainties. It is reliable for early response assessment.
Contact Email: