| Optimization of recovery coefficients for partial volume correction in PET Imaging using different reconstruction methods |
| Paper ID : 1027-ISCH |
| Authors |
|
Hanan Khaled Metwally * Medical Biophysics department, Faculty of Science, Helwan University, Cairo, Egypt. |
| Abstract |
| Purpose. The aim is to find out a universal formulation for recovery coefficient of partial volume correction using different positron emission tomography (PET) lesions and various contrast conditions. Methods. NEMA body phantom is used for evaluation of the quality of reconstructed images and simulation of whole-body imaging and PET data were acquired using (Siemens Healthcare, Biograph), The phantom was filled with water containing 18F-fuorodeoxyglucose (FDG) to make the sphere versus background activity concentration ratio at significantly wide range of lesion to background ratios (i.e. from 2:1 up to 20:1). For every reconstructed PET image of the 12 contrast ratios, an exponential fit was then generated of the resulting sphere volume and different thresholds applied. The reconstructed PET images of the 10 contrast ratios 2:1, 3:1, 5:1, 6:1, 8:1, 10:1, 12:1, 16:1, 18:1 and 20:1 was used to generate formulation for correction of SUV mean and SUV max. Then for Validation were used 2 another contrast ratio 4:1 and 14:1. Results. Iterative+TOF performed best for SUVmean and SUVmax predictions, with lower variability and quicker stabilization. TrueX+TOF performed well for higher contrast ratios but exhibited more variability. SUVmean and SUVmax errors were negative, while higher contrast ratios stabilized quicker. Increasing iterations improved recovery coefficient predictions accuracy. Conclusion. The study highlights the importance of method-specific optimization and higher iteration numbers for accurate PET image reconstructions, especially in lower contrast ratios. It recommends the Iterative+TOF method for consistency and lower error margins. |
| Keywords |
| PET, RC, SUVmean, SUVmax. Iterative+TOF, TrueX+TOF, TrueX |
| Status: Abstract Accepted (Poster Presentation) |