Dynamic Mode Decomposition (DMD) for Low-Latency Real-Time Cardiac MRI

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging
• Image Reconstruction

Background:

• Real-time cardiac MRI requires high spatiotemporal resolution for accurate diagnosis.
• Traditional reconstruction methods often face latency challenges, limiting online applications.
• Dynamic Mode Decomposition (DMD) offers a potential solution for rapid image reconstruction.

Purpose of the Study:

• To demonstrate the application of Dynamic Mode Decomposition (DMD) for high spatiotemporal, low-latency online reconstruction in 2D real-time cardiac MRI.
• To evaluate the efficacy of DMD in de-aliasing and preserving physiological motion during cardiac MRI acquisition.

Main Methods:

• DMD was applied to 2D spiral balanced steady state free precession (bSSFP) cardiac MRI data from adult and fetal subjects.
• Reconstruction performance was assessed against spatiotemporally constrained reconstruction (STCR) as a reference.
• Low-latency online reconstruction was evaluated at temporal resolutions of 21 ms/frame and 42 ms/frame.

Main Results:

• DMD accurately represented cardiac dynamics with a normalized root-mean-square error (NRMSE) below 7% when all modes were retained.
• The low-latency DMD-based online reconstruction successfully performed de-aliasing.
• Physiological motion was preserved, supporting high framerates essential for cardiac imaging.

Conclusions:

• The DMD framework is applicable to 2D real-time cardiac MRI.
• DMD facilitates low-latency de-aliasing, significantly enhancing online reconstruction quality.
• This technique holds promise for improving real-time cardiac MRI diagnostics.