There is nothing that nuclear spins will not do for you,
as long as you treat them as human beings.

-- Erwin Hahn

    Our research is in the general area of magnetic resonance imaging (MRI), signal processing, and machine learning (ML). Our goal has been to develop advanced data acquisition and processing methods to enable ultrafast and ultrahigh-resolution magnetic resonance imaging and spectroscopy. Our work involves signal processing (e.g., sparse sampling theory, algorithms, subspace modeling, parameter estimation, machine learning, etc.), spin physics (magnetic resonance data acquisition design and implementation), and a bit quantum mechanics (simulation of spin dynamics to generate spectral basis), and biomedical applications (especially, brain mapping, i.e., mapping the structural, functional, and molecular fingerprints of brain function and diseases). Students interested in joining our group, please check out some of our publications to get a better understanding of our past and ongoing research.
SELECTED PRIOR RESEARCH
Mathematical Models for Image Reconstruction from Sparse and/or Limited Data
  • Localized polynomial approximation model
    • "High resolution inversion of limited Fourier transform data using a localized polynomial approximation model,"
    Inverse Problems, vol. 5, pp. 831-847, 1989. (Introduction of sparsifying transform for image reconstruction)
  • Generalized series model
    • "A generalized series approach to MR spectroscopic imaging," IEEE Trans. Med. Imaging, vol. 10, pp. 132-137, 1991.
  • Partial separability model (also known as low-rank models)
    • "Spatiotemporal imaging with partially separable functions," Proc. IEEE Int'l Symp. Biomed. Imaging, pp. 988-991, 2007.
    "Accelerated high-dimensional MR imaging with sparse sampling using low-rank tensors," IEEE Trans. Med. Imaging,
    vol. 35, pp. 2119-2129, 2016.
Data Acquisition Methods
  • "Fast spin-echo imaging with circular k-space sampling," Magn. Reson. Med., vol. 39, pp. 23-27, 1998.
    (Introduction of circular sampling trajectories)
  • "Designing multichannel, multidimensional, arbitrary flip angle RF pulses using an optimal control approach,"
    Magn. Reson. Med., vol. 59, pp. 547-560, 2008. (Use of optimal control approach for RF pulse design)
  • "Reduced field-of-view excitation using second-order gradients and spatial-spectral radiofrequency pulses,"
    Magn. Reson. Med., vol. 69, pp. 503-508, 2013. (Introduction of second-order gradient for reduced FOV excitation)
  • "Design of multidimensional Shinnar-Le Roux radiofrequency pulses,"
    Magn. Reson. Med., vol. 73, pp. 633-645, 2015. (Solution of the multidimensional SLR design problem)
Ultrafast Dynamic Imaging
  • Generalized series / "Keyhole" imaging
    • "Improved temporal/spatial resolution in functional imaging through generalized series reconstruction," Works in Progress,
    Book of Abstracts of SMRM Meeting    , p. 346, 1992.
    "An efficient method for dynamic magnetic resonance imaging," IEEE Trans. Med. Imaging, vol. 13, pp. 677-686, 1994.
  • Fast dynamic imaging with sparse sampling of (k, t)-space
    • "(k,t)-Space sampling considerations for imaging of time-varying functions," Proc. of SMRM Meeting, p. 710, 1993.
    "Dynamic imaging by model estimation," Int'l J. Imaging Syst. Techn., vol. 8, pp. 551-557, 1997.
  • Fast dynamic imaging using temporal eigenbasis; demonstration of 4D real-time
    cardiac imaging without gating
    • "Dynamic imaging by temporal modeling with principal component analysis," Proc. of ISMRM Meeting, p. 10, 2001.
    "Four-dimensional MR cardiovascular imaging: Method and applications," Proc. of IEEE-EMBS Conference, 2011.
Magnetic Resonance Spectroscopic Imaging (MRSI)
Machine Learning-Enabled MRI and MRSI
  • "A neural network approach to NMR spectral estimation," Proc. of ISMRM Meeting, p. 1214, 1994.
  • "Machine learning-enabled high-resolution dynamic deuterium MR spectroscopic imaging,"
    IEEE Trans. Med. Imaging, vol. 40, pp. 3879-3890, 2021.
  • "Reconstructing high-quality sodium MR images from limited noisy k-space data with model-assisted deep learning,"
    Proc. of ISMRM Meeting, p. 4453, 2022.
High-Resolution Metabolic Imaging of the Brain
  • "Ultrafast magnetic resonance spectroscopic imaging using SPICE with learned subspaces,"
    Magn. Reson. Med., vol. 83, pp. 377-390, 2020.
  • "Making SPICE spicier with sparse sampling of (k, t)-space and learned subspaces," Proc. of ISMRM Meeting, p. 1899, 2020.
  • "Rapid high-resolution simultaneous acquisition of metabolites, myelin water fractions, and tissue susceptibility
    of the whole brain using 'SPICY' 1H-MRSI," Proc. of ISMRM Meeting, p. 754, 2019.
  • "Rapid parametric mapping using the unsuppressed water signals in metabolic imaging of the brain,"
    Proc. of ISMRM Meeting, p. 1803, 2021.
  • "Fast high-resolution metabolic imaging of acute stroke with 3D magnetic resonance spectroscopy,"
    Brain, vol. 143, pp. 3225-3233, 2021.
  • "High-resolution label-free molecular imaging of brain tumor," Proc. of IEEE EMBS Conference, pp. 3049-3052, 2021.

ONGOING RESEARCH PROJECTS
Signal Processing and Machine Learning Methods for Ultrafast and Ultrahigh-Resolution
MRI and MRSI

  • ML-enabled image recovery from highly limited, sparse, and noisy data

  • Image prior learning

  • Structure/shape prior learning

  • Multimodal information fusion

  • Performance characterization

    •
Ultrafast and Ultrahigh-Resolution MRSI

  • Ultrafast data acquisition at 3T and 7T

  • Reconstruction of spatiospectral functions

  • Spectral quantification incorporating physics and ML priors

Brain Mapping

  • Simultaneous mapping of brain metabolites and neurotransmitters

  • Multiparametric mapping of multiple molecules

  • Building statistical brain atlases of multiple molecules

  • Clinical applications (e.g., stroke, cancer, Alzheimer’s disease, epilepsy)