Cardiac motion compensation and resolution modeling in simultaneous PET-MR: a cardiac lesion detection study

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Cardiovascular Imaging

Background:

• Cardiac PET imaging is degraded by motion and partial volume effects (PVE), leading to artifacts and inaccurate measurements.
• Simultaneous PET-MR scanners enable concurrent assessment of myocardial contractility and PET perfusion data.
• Existing methods struggle with accurate quantification and defect characterization in cardiac PET.

Purpose of the Study:

• To develop a list-mode iterative reconstruction framework for cardiac PET-MR that incorporates tagged-MR derived motion and detector point spread function (PSF) modeling.
• To evaluate the effectiveness of this framework in improving image quality, contrast recovery, and lesion detectability using a beating cardiac phantom.
• To assess the impact of MR data acquisition time (full vs. half k-space) on motion correction accuracy.

Main Methods:

• Developed a list-mode iterative reconstruction framework integrating non-rigid myocardial motion from tagged-MR and position-dependent detector PSF.
• Utilized non-rigid B-spline registration to derive wall motion information from tagged MR volumes acquired with full or half k-space data.
• Evaluated the method in a beating cardiac phantom with transmural and non-transmural cold defects.

Main Results:

• MR-based motion correction improved defect/myocardium contrast recovery by 34-206% and lesion detectability by 115-235% compared to uncorrected studies.
• The method enabled differentiation between transmural and non-transmural defects, crucial for assessing residual ischemia.
• Incorporating PSF modeling alongside MR-based motion compensation further enhanced contrast recovery (5.1-8.5%) and lesion detectability (39-56%).
• No significant difference in PET contrast or lesion detectability was observed between full and half k-space tagged MR data.

Conclusions:

• The proposed PET-MR reconstruction framework effectively corrects for cardiac motion and PVE, significantly improving image quality and diagnostic accuracy.
• This integrated approach enhances the ability to detect and characterize myocardial defects, offering clinical advantages over current single-modality imaging.
• Acquisition of half k-space tagged MR data is sufficient for generating accurate motion fields, potentially accelerating imaging time without compromising results.