Cardiovascular Magnetic Resonance Imaging

Our group develops real-time and self-gated cardiovascular magnetic resonance (CMR) imaging methods that do not depend on a regular heart rhythm or breath-holding, and quantitative flow imaging methods with well-characterized uncertainty.

We are also building low-latency real-time MRI for guiding cardiac catheterization procedures.

Interventional Cardiac MRI

MRI-guided cardiac catheterization avoids ionizing radiation, which is particularly important for children who undergo repeated procedures. Guiding catheters requires images that are both fast and immediately available.

We are developing k-space sampling and spatiotemporal reconstruction strategies that reduce latency in real-time interventional cardiac MRI, in collaboration with Rady Children's Hospital.

Quantitative Flow Imaging and Uncertainty

Phase-contrast MRI measures blood velocity in 2D and 3D (4D flow). We use these measurements to visualize complex flow patterns such as vortices, which can identify locations where flow quantification is unreliable.

We are now characterizing the precision of phase-contrast velocity measurements in vivo, separating the contributions of signal-to-noise ratio and the velocity-encoding setting (v/VENC), and how this uncertainty propagates to clinical flow metrics.

Real-Time Cardiac MRI

We have developed techniques to perform real-time cardiac MRI in normal subjects, clinical patients, and translational animal studies. Parallel imaging, non-Cartesian sampling, and constrained reconstruction overcome the conventional sampling limits of MRI and allow us to image every heartbeat.

This enables evaluation of patients with arrhythmias and of rapid changes in cardiac performance, such as during exercise.

Real-time MRI captures every beat in a patient with premature ventricular complexes

Contijoch et al., Journal of Cardiovascular Magnetic Resonance 2024, Figure 3

Self-gated MR Imaging

Conventional cardiac MRI combines data acquired over several heartbeats to increase spatial and temporal resolution. This is challenging in patients with irregular heart rhythms or changing hemodynamics.

We have developed self-gating methods that sort data by beat morphology, and closed-loop control of k-space sampling driven by physiologic feedback, so that images can be formed for each type of beat.

Self-gating of a patient with trigeminy allows visualization of three contractions