ACCELERATING MAGNETIC RESONANCE IMAGING VIA DEEP LEARNING.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Deep Learning

Background:

• Magnetic Resonance Imaging (MRI) is crucial for diagnostics but often requires long acquisition times.
• Accelerating MRI scans without compromising image quality is a significant challenge in medical imaging.

Purpose of the Study:

• To develop a deep learning-based approach for accelerating MRI acquisition.
• To train a convolutional neural network (CNN) to reconstruct high-quality MR images from undersampled k-space data.

Main Methods:

• A deep learning model, specifically an off-line convolutional neural network (CNN), was designed and trained.
• The CNN learns the mapping between zero-filled and fully-sampled k-space data to reconstruct MR images.
• The method is designed for compatibility with online constrained reconstruction techniques.

Main Results:

• The proposed deep learning approach effectively accelerates MRI acquisition.
• The CNN successfully restores fine structures and details in the reconstructed MR images.
• Experimental results on real MR data demonstrate encouraging performance for efficient imaging.

Conclusions:

• Deep learning offers a promising solution for accelerating MRI scans.
• The developed CNN method enables efficient and effective MRI acquisition while maintaining image quality.
• This approach has the potential to improve patient comfort and throughput in MRI examinations.