CineVN: Variational network reconstruction for rapid functional cardiac cine MRI.

Area of Science:

• Cardiovascular Imaging
• Medical Image Reconstruction
• Deep Learning in Radiology

Background:

• Accelerated cardiac cine MRI is crucial for assessing heart function.
• High spatiotemporal resolution and low temporal blurring are desired for accurate analysis.
• Current reconstruction methods face limitations in speed and accuracy.

Purpose of the Study:

• To develop a novel reconstruction method for highly accelerated cardiac cine MRI.
• To achieve high spatiotemporal resolution and minimize temporal blurring.
• To accurately estimate ventricular volumes and myocardial strain in various subject groups.

Main Methods:

• The CineVN method utilizes a spatiotemporal Variational Network with conjugate gradient descent.
• Evaluation involved retrospectively undersampled data for image quality assessment.
• Prospectively accelerated data were acquired in healthy subjects and patients for functional parameter analysis.

Main Results:

• CineVN demonstrated superior image quality compared to compressed sensing on retrospectively undersampled data.
• Functional parameters and myocardial strain measurements by CineVN were more accurate than state-of-the-art compressed sensing methods.
• The method was validated in a cohort of 46 patients.

Conclusions:

• Deep learning-based reconstruction with CineVN allows accurate real-time cardiac function evaluation.
• The method achieves high spatiotemporal resolution in cine MRI.
• CineVN has the potential to enhance cardiac imaging for patients with arrhythmias or breath-holding difficulties.