Accelerated two-dimensional phase-contrast for cardiovascular MRI using deep learning-based reconstruction with...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Healthcare
• Cardiovascular Imaging

Background:

• Accelerated Magnetic Resonance Imaging (MRI) techniques are crucial for reducing scan times, especially in pediatric patients.
• Phase Contrast MRI (PC-MRI) provides essential quantitative flow and velocity data but is often limited by scan duration.
• Developing robust reconstruction methods is vital for maintaining accuracy and precision in highly accelerated 2D PC-MRI.

Purpose of the Study:

• To develop and validate a deep learning (DL) based reconstruction framework for highly accelerated 2D PC-MRI.
• To achieve accurate and precise quantitative measurements of blood flow and velocity using the proposed DL framework.
• To assess the clinical feasibility of the DL reconstruction in pediatric patients.

Main Methods:

• A modified DL-ESPIRiT framework incorporating Complex Difference estimation (CD-DL) was developed.
• CD-DL was trained on 155 fully sampled 2D PC-MRI pediatric datasets and tested on retrospectively undersampled data (6-11x acceleration).
• Performance was compared against Parallel Imaging and Compressed Sensing (PICS), evaluating peak velocity and total flow accuracy and precision.

Main Results:

• Retrospective evaluation showed CD-DL reconstructions at 9x acceleration maintained accuracy and precision within a 5% error margin.
• CD-DL demonstrated superior accuracy and precision over PICS for peak velocity and total flow measurements.
• Prospective feasibility studies in pediatric patients confirmed CD-DL's higher accuracy and precision compared to PICS.

Conclusions:

• The CD-DL framework enables accurate and precise quantitative measurements in highly accelerated 2D PC-MRI (up to 9x acceleration).
• Clinical feasibility was successfully demonstrated with an 8x undersampled acquisition and CD-DL reconstruction in pediatric patients.
• This DL-based approach significantly enhances the utility of accelerated PC-MRI for quantitative cardiovascular assessment.