Improving best-phase image quality in cardiac CT by motion correction with MAM optimization

Area of Science:

• Medical Imaging
• Cardiovascular Imaging
• Image Reconstruction

Background:

• Cardiac CT angiography requires motion correction for coronary artery image quality.
• Current motion estimation relies on 3-D/3-D registration, necessitating extended scan ranges and increasing radiation dose.
• Existing methods offer limited improvements in image quality compared to the best-phase image.

Purpose of the Study:

• To propose a novel motion estimation algorithm for cardiac CT.
• To improve best-phase image quality using a short scan data interval.
• To avoid the need for extended scan ranges and reduce patient radiation dose.

Main Methods:

• Motion artifact metrics (MAM), including entropy and positivity, were defined to quantify motion artifacts.
• Motion field parameters were optimized using a gradient descent procedure to minimize artifacts.
• Analytical methods were employed for efficient motion estimation and compensation.
• The proposed MAM-optimization was compared to a 3-D/3-D registration-based algorithm using simulated data and clinical cases.

Main Results:

• The MAM-approach significantly improved best-phase image quality, with up to 100% increase in normalized cross-correlation (NCC) and 81% in root-mean-square deviation (RMSD).
• The registration-based approach showed maximum improvements of 20% in NCC and 40% in RMSD.
• The MAM algorithm outperformed the registration-based method in phases with less motion, while the latter was superior in rapid motion phases.
• Visual confirmation of image quality improvement was observed in clinical cases.

Conclusions:

• The proposed software-based method enhances best-phase image quality in coronary CT angiography.
• A short scan data interval is sufficient, eliminating the need for additional data from other cardiac phases.
• The algorithm is compatible with standard cardiac CT acquisition protocols.