Volume 11 Supplement 1

Abstracts of the 12th Annual SCMR Scientific Sessions – 2009

Open Access

Real-time free-breathing strategy for tracking labeled cells with in-vivo vardiac MRI

  • Yijen L Wu1,
  • Cornelius Brinegar2,
  • Zhi-Pei Liang2,
  • Qing Ye1,
  • T Kevin Hitchens1,
  • Lesley M Foley1 and
  • Chien Ho1
Journal of Cardiovascular Magnetic Resonance200911(Suppl 1):P31


Published: 28 January 2009


Non-invasive in vivo cellular imaging with MRI has greatly broadened our understanding in many cellular and molecular processes, as well as in monitoring stem cell therapy. The target cells can be made MRI-detectable by labeling them with iron-oxide contrast agents, such as micrometer-sized-iron-oxide (MPIO) particles. With sufficient iron loading per cell, it is possible to detect single cells in vivo with MRI. However, motion artifacts and the need for gated acquisition have made imaging cells in the heart more challenging and time-consuming. The goal of this study is to develop a strategy to enable real-time imaging of individual labeled cells without the need for a gated or breath-hold protocol. To this end, we have implemented a new model-based imaging scheme for high-resolution free-breathing imaging, using the Partially Separable Function (PSF) model.


1. Model system

We use a rodent abdominal heterotopic working heart and lung transplantation model with DA to BN rat pair, using BN-to-BN transplantation as isograft controls.

2. Iron oxide particle labeling

Immune cells, mostly macrophages, were labeled in vivo by direct intravenous injection of 3 mg micrometer-sized-iron-oxide (MPIO) particles 1 day prior to MRI scans.

3. MRI protocols

Real-time free-breathing imaging, and ECG and respiration gated T2* and cine imaging were performed using a Bruker AVANCE 4.7-T system with a home-built 5.5 cm surface coil and an in-plane resolution of 195 μm.

4. Partially Separable Function (PSF) method and theory

The PSF model approximates the hypersurface, d(k, t), formed by the time varying data as a summation of M simpler functions that are each the product of one-dimensional functions α m (k) and ϕ m (t), and recent results have shown that cardiac imaging with breathing exists in a low rank space where M = 16. Data collection consists of interleaving two scans: (i) a high temporal resolution data set for estimating {ϕ m (t)} M m = 1and (ii) a high spatial resolution scan used to estimate {α m (k)} M m = 1. The low rank of the model allows for temporal interpolation over time intervals failing the Nyquist criterion thereby avoiding the need to time data collection with the cardiac and respiratory motions.


We have used a rodent heterotopic transplantation model for tracking labeled immune cells in vivo. In this model, immune cells, mostly macrophages, can be labeled in vivo by direct intravenous administration of MPIO particles; labeled immune cells then migrate to the rejecting transplanted hearts and are detected by T2*-weighted imaging. Low-dosage MPIO particles are used in this study for sparse labeling of macrophages to promote the visualization of single cells, even though the images lack the necessary spatial resolution by at least a factor of 4.

Without gating, real-time free-breathing MRI results in blurring due to motion artifacts (Fig. 1A). The model-based data collection and reconstruction scheme (Fig. 1B) yields a high-quality image comparable to ECG-and-respiration-gated cine imaging (Fig 1C), however in a much shorter acquisition time. The model-based method enables identification of MPIO-labeled cells in vivo (Fig. 1D) in real-time data acquisition. The model-based method acquires data continuously whereas the gated scheme collects data intermittently, depending upon the relationship of the ECG and respiratory signals. In addition to a shorter acquisition time, the continuous model-based scheme preserves all information in the k-t space, which is especially important in functional evaluation of the diseased conditions with arrhythmia or compromised breath-holding.
Figure 1

(A) Fourier reconstruction of free-breathing MRI. (B) Model-based reconstruction of the same free-breathing MRI. (C) ECG-and-respiration-gated cine imaging for comparison. (D) Enlarged view of the heart showing the infiltrated MPIO-labeled cells.


The model-based imaging scheme allows real-time free-breathing imaging with resolution and image quality sufficient for tracking labeled cells in vivo in beating hearts.

Authors’ Affiliations

Pittsburgh NMR Center for Biomedical Research, Carnegie Mellon University
Electric and Computer Engineering, University of Illinois at Urbana-Champaign


  1. Magn Reson Med. 2006, 55: 242-9. 10.1002/mrm.20718.Google Scholar
  2. IEEE International Symposium on Bioimaging. 2007, 988-991.Google Scholar
  3. Proc Natl Acad Sci. 2006, 103: 1852-7. 10.1073/pnas.0507198103.Google Scholar
  4. Brinegar, et al.: Engineering in Medicine and Biology Conference. 2008Google Scholar


© Wu et al; licensee BioMed Central Ltd. 2009

This article is published under license to BioMed Central Ltd.