Correlation imaging with arbitrary sampling trajectories.

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging (MRI)
• Image Reconstruction

Background:

• High-speed Magnetic Resonance Imaging (MRI) is crucial for reducing scan times and improving patient comfort.
• Conventional MRI reconstruction methods face limitations with arbitrary sampling trajectories.
• Correlation imaging offers a promising framework for advanced MRI reconstruction.

Purpose of the Study:

• To develop a generalized linear approach for image reconstruction using arbitrary sampling trajectories in MRI.
• To implement and validate a hybrid-space correlation imaging framework for accelerated MRI.
• To enhance the performance of MRI by leveraging multiple reconstruction mechanisms.

Main Methods:

• Developed a generalized linear approach based on the "correlation imaging" framework.
• Implemented correlation imaging in a multi-dimensional hybrid space combining physical and virtual spaces.
• Introduced an undersampling trajectory with uniformity and randomness within the hybrid space.
• Demonstrated the approach in multi-slice 2D, multi-scan, and radial dynamic imaging protocols.

Main Results:

• Hybrid-space correlation imaging effectively utilizes multiple reconstruction mechanisms: coil sensitivity encoding, data sparsity, and information sharing.
• The method demonstrated superior performance compared to several conventional MRI reconstruction techniques.
• Successful implementation across various imaging scenarios, including multi-slice, multi-scan, and dynamic imaging.

Conclusions:

• The proposed hybrid-space correlation imaging provides a robust and generalized linear approach for MRI reconstruction with arbitrary sampling trajectories.
• This method significantly accelerates MRI acquisition by effectively using undersampled data.
• The approach holds potential for accelerating multi-scan clinical MRI protocols, especially those requiring diverse sampling strategies.