Advanced Image Reconstruction and Sampling Pattern Optimization in Silent MRI

PASSCODE: argmin

Silent MRI is a technology that allows for MRI scans to be conducted with less noise than traditional MRI machines. This technology is important for a few reasons: First, the loud noises generated by traditional MRI machines can be uncomfortable for some patients, particularly those with anxiety disorders, dementia, or sensory sensitivities. Second, silent MRI can be useful in auditory and speaking studies and pediatric studies. Third, the noise generated by traditional MRI machines can interfere with speech communication, making it difficult for healthcare providers to communicate with patients during the scan.

Model-based image reconstruction is a technique that uses mathematical models to suppress image noise, reduce acquisition time, and improve image quality, especially in dynamic and quantitative MRI. We proposed to use MBIR, learning-based optimized sampling patterns, and an optimized excitation module to improve the image quality of silent MRI can in terms of spatial resolution, temporal resolution, noise, and artifacts. These findings suggest that by carefully designing reconstruction algorithms, sampling patterns, and excitation modules, the image quality of silent MRI can be improved for broader use in both research and clinical settings.

CO-CHAIRS: Professors Jeffrey Fessler & Douglas Noll