Improving the Quality of Brain PET Images Using MR-Guided PET Reconstruction Technique With Integrated PET/MR

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Radiology

Background:

• Integrated PET/MR systems offer combined anatomical and functional imaging capabilities.
• Optimizing image quality in brain PET imaging is crucial for accurate diagnosis and treatment monitoring.
• Conventional PET reconstruction methods may have limitations in complex imaging scenarios.

Purpose of the Study:

• To evaluate an MR-guided PET reconstruction (MRg) algorithm for enhancing brain PET image quality.
• To compare the performance of MRg against the routine Ordered Subsets Expectation Maximization (OSEM) algorithm.
• To assess the clinical utility of MRg in various neurological and oncological conditions.

Main Methods:

• Implementation of an MR-guided PET reconstruction algorithm on an integrated PET/MR system.
• Phantom studies using a Hoffman phantom to assess image quality metrics under varying dose and iteration parameters.
• Clinical studies involving patients with diverse conditions (no brain lesions, DLBCL, PD, cavernous angioma) to evaluate MRg and OSEM reconstructions.

Main Results:

• Phantom studies showed MRg improved image quality metrics (contrast, variability), especially at lower doses, with optimal performance at penalization factor β=2.
• Clinical studies demonstrated MRg significantly outperformed OSEM in image quality for healthy subjects.
• MRg significantly improved lesion detection and quantification in DLBCL patients and enhanced signal-to-noise ratios in PD patients.

Conclusions:

• The MR-guided PET reconstruction method offers significant advantages over conventional OSEM for brain PET imaging.
• MRg demonstrates substantial improvements in image quality, contrast, and noise reduction in clinical brain PET/MR examinations.
• This technique holds considerable potential for routine clinical application in brain PET/MR imaging.