DEISM: Deep Reconstruction Framework With Self-Calibration Mechanisms for Accelerated Chemical Exchange Saturation...

Area of Science:

• Magnetic Resonance Imaging
• Medical Imaging
• Biomedical Engineering

Background:

• Chemical Exchange Saturation Transfer (CEST) imaging offers valuable molecular information but suffers from prolonged scan times due to multiple data acquisitions.
• Accelerated imaging techniques are crucial to overcome the limitations of long acquisition durations in CEST MRI.
• Exploiting artifact information as a prior for image reconstruction is an underexplored area in accelerated CEST imaging.

Purpose of the Study:

• To develop and validate a novel deep reconstruction framework with self-calibration mechanisms (DEISM) for highly accelerated CEST imaging.
• To investigate the utility of artifact information, specifically spatial-frequential redundancy, as a prior for CEST image reconstruction.
• To enhance the quality of CEST images, molecular maps, and spectra in accelerated acquisitions.

Main Methods:

• Proposed a novel deep reconstruction framework (DEISM) integrating a model-based network for initial reconstruction and a data-driven artifact suppression (AS) network.
• Developed a unique encoder-decoder architecture with multi-scale feature fusion for robust artifact estimation and correction.
• Trained the DEISM framework end-to-end using simulated data and evaluated performance on healthy volunteers and brain tumor patients with varying acceleration factors.

Main Results:

• Demonstrated the feasibility of the data-driven artifact suppression (AS) concept in accelerated CEST imaging.
• Showcased the effectiveness of exploiting spatial-frequential correlations within the artifact field for improved reconstruction.
• DEISM provided high-quality source images, molecular maps, and CEST spectra, outperforming conventional and state-of-the-art methods.

Conclusions:

• The DEISM framework successfully accelerates CEST imaging by leveraging artifact information as a powerful prior.
• Integrating image artifact priors into learning-based reconstruction significantly enhances CEST imaging quality and diagnostic potential.
• DEISM offers a promising solution for reducing scan times in CEST MRI without compromising image fidelity.