Related Experiment Video
Updated: Oct 14, 2025
10:35
Simultaneous PET/MRI Imaging During Mouse Cerebral Hypoxia-ischemia
Published on: September 20, 2015
12.4K
Harmonization of PET image reconstruction parameters in simultaneous PET/MRI
Richard Laforest1, Mehdi Khalighi2, Yutaka Natsuaki3
1Mallinckrodt Institute of Radiology, Washington University, St. Louis, MO, USA. laforestr@wustl.edu.
EJNMMI Physics
|November 5, 2021
Summary
Researchers identified specific image reconstruction parameters for GE Signa and Siemens mMR PET/MRI scanners to achieve comparable quantitative PET performance. This harmonization enables more reliable data comparison across different systems, improving accuracy in positron emission tomography/magnetic resonance imaging studies.
Area of Science:
- Medical Imaging Physics
- Quantitative PET Imaging
- Multimodal Imaging Systems
Background:
- Simultaneous PET/MRI scanners exhibit variability in quantitative PET performance due to system design and reconstruction differences.
- Inconsistent quantitative accuracy hinders meaningful data combination and comparison across diverse PET/MRI scanners.
- Harmonization of image reconstruction is crucial for reliable multicenter studies and clinical trial data.
Purpose of the Study:
- To define specific image reconstruction parameters that yield comparable contrast recovery curves (CRCs) across different simultaneous PET/MRI systems.
- To establish a standardized methodology for harmonizing quantitative PET data acquired on GE Signa and Siemens mMR scanners.
- To optimize reconstruction parameters for achieving both high contrast recovery and minimal discrepancy between scanners.
Main Methods:
- The NEMA NU-2 image quality phantom was imaged on GE Signa and Siemens mMR PET/MRI scanners.
- Images were reconstructed using manufacturer-provided iterative algorithms with and without point spread function (PSF) modeling, and varying post-reconstruction Gaussian filter sizes (3-7 mm).
- Contrast recovery coefficients (CRCs) for SUVmean, SUVmax, and SUVpeak were calculated. A hybrid metric combining root-mean-squared discrepancy (RMSD) and CRC values was used to identify optimal harmonization parameters.
Main Results:
- Significant variability in quantitative performance was observed across different reconstruction parameters and scanners.
- Optimal parameter sets for harmonized CRCs varied by metric (CRCmean, CRCmax, CRCpeak) and scanner.
- Specific parameter combinations, such as 2 iterations-3 mm filter (GE) and 4 iterations-3 mm filter (Siemens), both with PSF, achieved simultaneous harmonization with low RMSD values (2.8% for CRCmean, 5.8% for CRCmax, 3.2% for CRCpeak).
Conclusions:
- User-selectable parameters in iterative reconstruction, smoothing, and PSF modeling allow for harmonization of PET/MRI quantitative performance.
- Nearly identical CRC curves can be achieved on different commercial PET/MRI scanners by carefully selecting image reconstruction parameters.
- This harmonization strategy is essential for improving the reliability and comparability of quantitative PET data across multicenter research and clinical applications.
