Purpose: Molecular imaging studies reveal tumor heterogeneity which may explain therapeutic resistance. Alternatively, variations observed in the SUV uptake may not be due to intrinsic heterogeneity of the tumor, but due to effects of position resolution (volume dependence) or tumor motion. The objective of this study is to evaluate motion artifacts on PET SUV uptake within tumors for different volumes. Methods: We conducted a phantom study: spheres of homogeneous18F‐FDG activity of motion amplitudes (0, 5, 10, 15, 20, 25, 30 mm) and volumes (internal diameter 10, 13, 17, 22, 28, 37 mm). Any variation observed in SUV uptake distribution with homogeneous activity spheres must be due to motion and volume effects. Therefore, we used this data to quantify uptake variations as a function of motion and volume. PET SUV uptake values for each tumor were fit with a Woods‐Saxon model with 2 parameters: radius and skin depth. When the ratio of radius to skin depth becomes small, this simplifies to a Gaussian. By using this parameterization for stationary phantoms, using the probability density function for motion, and using an image convolution technique we were able to simulate the SUV uptake distribution of moving tumor in the central 2D plane. Taking voxel by voxel SUV uptake distribution in 2D we quantified the accuracy of our simulation for motion affected, tumor volume affected SUV uptake heterogeneity within the phantom. Results: Phantom study for stationary spheres show motion has a high impact on SUV uptake distributions. Maximum percentage uptake variation across the phantoms was 22%. Convolution results confirm that motion is the largest contributor to the SUV heterogeneity of the tumors. Average difference between convolution and measurement is 6.7%. Conclusion: These results will impact the studies that utilize tumor heterogeneity such as tumor detection, planning with dose painting, and treatment response.
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging