Reference-based MRI.

Area of Science:

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

Background:

• Existing Magnetic Resonance Imaging (MRI) techniques can be enhanced using prior imaging data to reduce scan times or improve Signal to Noise Ratio (SNR).
• Clinical MRI scenarios often present inherent similarities between reference and acquired images, such as adjacent slices, different contrasts, or serial scans of the same patient.

Purpose of the Study:

• To present FASTMER, a framework designed for fast MRI acquisition by effectively exploiting reference image information.
• To develop a method that can accelerate MRI scans and enhance image quality in diverse clinical settings.

Main Methods:

• The FASTMER framework utilizes the similarity between a reference image and the acquired image.
• An iterative weighted reconstruction approach is employed, adapting weights based on the degree of similarity to accommodate variations.
• This method is designed to handle scenarios with low similarity between the reference and acquired images.

Main Results:

• Demonstrated SNR improvement in high-resolution brain MRI.
• Enabled fast Fluid-Attenuated Inversion Recovery (FLAIR) scanning by exploiting similarity with T2-weighted images.
• Facilitated rapid follow-up MRI acquisition by utilizing baseline scan similarity.
• FASTMER outperformed existing state-of-the-art image reconstruction methods in experimental evaluations.

Conclusions:

• FASTMER provides a versatile framework for accelerating MRI by exploiting reference images.
• The iterative algorithm supports varying degrees of similarity, broadening its applicability across different MRI scenarios.
• The framework has the potential to significantly improve reconstruction in numerous MR applications due to the prevalence of reference images in clinical tasks.