CONVOLUTIONAL FRAMEWORK FOR ACCELERATED MAGNETIC RESONANCE IMAGING.

Area of Science:

• Medical Imaging
• Biophysics
• Computer Vision

Background:

• Magnetic Resonance Imaging (MRI) offers excellent soft-tissue contrast without ionizing radiation.
• Long MRI data acquisition times limit clinical utility.
• Accelerated MRI reconstruction from undersampled k-space data is crucial.

Purpose of the Study:

• To present a simple and scalable Convolutional Framework (CF) for MR image reconstruction.
• To address the challenges of non-convex optimization in existing reconstruction methods.
• To demonstrate the versatility of CF across various MRI applications.

Main Methods:

• Exploiting Hankel data matrix rank deficiency for k-space recovery.
• Developing a novel Convolutional Framework (CF) for image reconstruction.
• Testing CF with measured data from 2D, 3D, and dynamic MRI.

Main Results:

• The proposed Convolutional Framework (CF) provides a feasible and versatile approach to MR image reconstruction.
• CF demonstrates effectiveness in reconstructing images from highly undersampled k-space data.
• The scalability of CF is validated across diverse MRI datasets.

Conclusions:

• Convolutional Framework (CF) presents a promising, scalable method for accelerated MRI.
• CF can potentially reduce MRI scan times, enhancing clinical workflow.
• The approach is adaptable for 2D, 3D, and dynamic imaging scenarios.