Data-Driven Motion Correction Algorithm: Validation in [¹³N]NH₃ Dynamic PET/CT Scans

Area of Science:

• Cardiovascular Imaging
• Nuclear Medicine
• Medical Physics

Background:

• Cardiac PET/CT imaging is challenged by patient motion, impacting diagnostic accuracy.
• Automated data-driven motion correction (DDMC) has shown promise in static cardiac PET/CT.
• This study validates DDMC for dynamic [13N]NH3 PET/CT and assesses its value in evaluating myocardial motion.

Purpose of the Study:

• To validate DDMC in dynamic [13N]NH3 PET/CT acquisitions.
• To compare DDMC with manual in-software motion correction (ISMC).
• To explore the impact of myocardial motion on quantitative myocardial blood flow (MBF) and coronary flow reserve (CFR).

Main Methods:

• Thirty-six normal and forty-three ischemia patient scans underwent processing with no motion correction (NMC), ISMC, and DDMC using QPET software.
• Analysis focused on differences in rest/stress MBF and CFR values and effect sizes.
• DDMC-derived motion vectors were analyzed for inter-phase/axis differences and relation to MBF quantification.

Main Results:

• Motion correction (MC), both DDMC and ISMC, significantly increased stress-MBF and CFR values (p < 0.05).
• DDMC and ISMC showed comparable effect sizes, with minor differences in specific regional MBF values.
• High-intensity motion (>6 mm) occurred predominantly in the Z-axis, was greater during stress than rest, and negatively correlated with stress-MBF.

Conclusions:

• DDMC is feasible for dynamic [13N]NH3 PET/CT, providing results similar to manual ISMC.
• DDMC enhances efficiency by reducing processing time and observer variability.
• Motion correction is crucial in QPET analysis, as Z-axis motion during stress negatively affects stress-MBF quantification.