Improved time series reconstruction for dynamic magnetic resonance imaging

Area of Science:

• Medical Imaging
• Biophysics
• Signal Processing

Background:

• Dynamic magnetic resonance imaging (dMRI) data often shows high temporal correlation.
• Acquiring dMRI at higher temporal resolutions typically requires undersampling data, leading to aliasing artifacts.
• Existing reconstruction methods may struggle with arbitrary k-space trajectories and real-time processing.

Purpose of the Study:

• To develop and evaluate a Kalman filter-based unaliasing reconstruction algorithm for accelerated dynamic MRI.
• To leverage temporal correlations in dMRI data for improved image reconstruction.
• To enable real-time, high-temporal-resolution cardiac MRI.

Main Methods:

• Modeling the dynamic imaging process as a linear dynamical system.
• Implementing a causal Kalman filter for unaliasing undersampled MRI data.
• Handling arbitrary readout trajectories within the reconstruction framework.

Main Results:

• The Kalman filter approach effectively resolves aliasing caused by undersampling.
• The reconstruction is computationally efficient, suitable for real-time applications.
• Successful in vivo application demonstrated for cardiac MRI in healthy volunteers, yielding high temporal resolution.

Conclusions:

• Kalman filter-based unaliasing is a viable and fast method for accelerated dynamic MRI.
• This technique significantly enhances temporal resolution, crucial for dynamic physiological processes like cardiac motion.
• The approach shows promise for real-time cardiac MRI applications.