A new vessel segmentation algorithm for robust blood flow quantification from two-dimensional phase-contrast magnetic...

Area of Science:

• Cardiovascular imaging
• Medical image analysis
• Biomedical engineering

Background:

• Accurate time-resolved vessel segmentation of the aorta and pulmonary artery is crucial for phase-contrast magnetic resonance imaging (PC-MRI) blood flow measurements.
• Current semi-automatic methods often require extensive manual correction, limiting efficiency and relying heavily on user expertise.

Purpose of the Study:

• To develop and evaluate a novel semi-automatic vessel segmentation algorithm incorporating shape constraints for the aorta and pulmonary artery.
• To assess the algorithm's robustness in healthy volunteers and patients with cardiovascular conditions.
• To validate the method using a pulsatile flow phantom and make it publicly available for research.

Main Methods:

• Developed a semi-automatic algorithm utilizing shape constraints derived from manual delineations of the aorta and pulmonary artery.
• Required only one manual delineation per vessel, significantly reducing user interaction.
• Validated the algorithm's performance against manual segmentations and phantom flow measurements.

Main Results:

• The algorithm demonstrated low bias and variability for flow volume measurements in both the aorta (0.0 ± 1.9 ml) and pulmonary artery (-1.7 ± 2.9 ml).
• Interobserver variability was significantly lower for the proposed semi-automatic method compared to manual delineations.
• Phantom validation confirmed good agreement with timer-and-beaker flow measurements (0.4 ± 2.7 ml).

Conclusions:

• The developed semi-automatic vessel segmentation algorithm offers an efficient and robust solution for analyzing blood flow and shunt volumes.
• This method enhances the accuracy and reduces the time required for cardiovascular image analysis.
• The algorithm's availability promotes further research in cardiac MRI blood flow quantification.