Next-Generation Cardiac MRI
Free-breathing Multi-Parametric Mapping
3D B1+ corrected simultaneous myocardial T1 and T1 mapping with subject-specific respiratory motion correction and water-fat separation
Our Free-Breathing 3D Whole-Heart Simultaneous T1 and T1 Quantification technique integrates IR and T1 pulses, dual-echo Dixon acquisition, and diaphragm navigation for motion tracking. Combined with our SubNdMoCo non-rigid motion correction and dictionary-based multicontrast matching, this approach delivers precise T1, T1, and B1+ maps with enhanced efficiency and robustness for diagnosis of cardiomyopathies.
Click to see the zoomed-in pipeline.
Free-breathing simultaneous native myocardial T1, T2 and T1 mapping with Cartesian acquisition and dictionary matching
We developed FB-MultiMap, a free-breathing technique that simultaneously quantifies T1, T2, and T1 parameters in a single scan. Compared with conventional separate beath-hold mapping techniques, this technique can greatly simplify the acquisition process and reduce the scan time from ~5 minutes to 1 minute. FB-MultiMap has shown great potential for clinical translation of diagnosing complex cardiomyopathies without contrast agent.
Click to see the full pipeline of FB-MultiMap.
Free-breathing non-contrast T1 dispersion magnetic resonance imaging of myocardial interstitial fibrosis in comparison with extracellular volume fraction
We explored the feasibility of the non-contrast parameters including T1, T2, T1 and myocardial fibrosis index for diagnosing diffuse myocardial fibrosis without gadolinium-based contrast agents. The results suggest that FB-MultiMap is a promising alternative to contrast enhanced techniques.
Click to see detailed results.
AI Empowered Cardiac MRI Analysis and Diagnosis
SRMapping: A super-resolution network with multi-level fine-grained feature fusion for cardiac MR T1 mapping
We developed SRMapping: a transformer-based network that simultaneously performs super-resolution and quantitative T1 mapping for low-resolution Cardiac MR T1 acquisitions. This approach addresses challenges such as motion blurring, compromised spatial resolution, and noise sensitivity in traditional methods.
Click to see the full pipeline of SRMapping.
Predicting Late Gadolinium Enhancement of Acute Myocardial Infarction in Contrast-Free Cardiac Cine MRI Using Deep Generative Learning
We proposed cine generated enhancement (CGE), a deep learning framework that transforms widely available non-contrast cardiac cine images into LGE (late gadolinium enhancement)-like images for myocardial infarction diagnosis. CGE achieved superior image quality to LGE and accurate scar delineation in patients with acute MI of both internal and external datasets. CGE can significantly simplify the CMR examination, reducing scan times and risks associatedwith gadolinium-based contrasts, which are crucial for acute patients.
Click to see the full graphic abstract of CGE.
Pushing MRI Speed Limits
A new 3D non-Cartesian trajectory.
Non-cartesian Trajectory
By carefully planning the sampling trajectory, non-Cartesian imaging acquires more data every excitation compared to conventional Cartesian sampling. This technique can be applied to various MRI scenario, including efficient volumetric imaging, large-coverage quantitative imaging and 3D real-time MRI.
One spiral (6ms) per frame with NUFFT (left) and our proposed method (right).
Accelerated Reconstruction
We developed advanced deep learning reconstruction methods to reconstruct high-quality images from highly undersampled data.
Representative Publications
Free-breathing simultaneous native myocardial T1, T2 and T1 mapping with Cartesian acquisition and dictionary matching
Published in JCMR, 2023
Free-breathing non-contrast T1 dispersion magnetic resonance imaging of myocardial interstitial fibrosis in comparison with extracellular volume fraction
Published in JCMR, 2024
Predicting Late Gadolinium Enhancement of Acute Myocardial Infarction in Contrast-Free Cardiac Cine MRI Using Deep Generative Learning
Published in CCI, 2024
Join Us
Interested in our research? We welcome passionate students/researchers to join our team.