The Need

Current cardiac MRI workflows rely heavily on breath-holds, ECG gating, and retrospective binning, which break down in patients with arrhythmias, irregular breathing, or limited ability to cooperate. Existing 3D approaches often average away beat-to-beat variability or suffer from motion artifacts, while 2D real-time imaging sacrifices volumetric coverage. There is a significant unmet need for a practical method that can capture the full heart volume in real time, under free-breathing and irregular rhythm conditions, without extending scan times or requiring large training datasets.

The Technology

OSU engineers have developed a scan-specific, learning-based reconstruction framework for real-time 3D cardiac MRI. It reconstructs dynamic volumetric image series directly from highly undersampled measurements by jointly modeling spatial content and temporal changes. The approach adapts itself to each individual scan rather than relying on pre-trained models, enabling recovery of complex motion patterns and temporal variability. Importantly, it avoids conventional gating, binning, or cycle averaging, allowing true real-time volumetric imaging from a single continuous acquisition.

Commercial Applications

• Clinical cardiac MRI in patients with arrhythmias or poor breath-hold capability
• Advanced cardiac functional assessment during stress or exercise imaging
• Research imaging platforms for studying beat-to-beat cardiac dynamics

Benefits/Advantages

• Preserves true beat-to-beat dynamics
• Robust to motion and rhythm irregularities, including free breathing and arrhythmia
• No reliance on large external training datasets
• Enables highly accelerated 3D imaging