Computational Methods for Enhancing Sensitivity to MRI Cell-Tracking Agents

New methods for programming cells to perform desired functions in vivo promise to enable new diagnostic tools and therapies. To develop, confidently deploy, and routinely use these emerging cellular therapies, it is necessary to have the ability to non-invasively detect and monitor transplanted cells, or those that have been genetically-modified in situ. MRI offers non-invasive high-resolution imaging within deep tissues without the use of ionizing radiation. In order for cells of interest to appear in MR images, they are labeled with iron oxide contrast agent (CA), genetic instructions to produce their own CA, or with fluorine-based tracer agents. Regardless of the type of label used, it is a challenge to achieve sufficient MR image signal and contrast in order to differentiate labeled cells from background tissue or image noise, especially when they reside in inhomogeneous tissue, or when scan time is limited. Improvements in sensitivity to labeled cells are needed for their ready detection and quantification.

Cells labeled with iron oxide CA appear in conventional proton (¹H) MR images as hypo-intense spots or regions within the organ or anatomy being imaged. To facilitate cell-tracking that employs iron-oxides, we present three methods: The first, called the Two-Compartment T2 Contrast Model (T2CM), is a model for predicting the relationship between iron oxide CA concentration and expected image contrast. The second method, called Phase Slope Magnitude Imaging (PSM), highlights arbitrary distributions of iron oxide CA in tissue. The third method, called Phase Map Cross-Correlation Detection and Quantification (PDQ), detects isolated magnetic dipoles that indicate the presence of an iron oxide-labeled cell or cell cluster. PDQ then measures the magnetic moment of each dipole and registers its location for the purpose of cell-tracking and 3D visualization.

Cells labeled with fluorine-based tracer agents appear in fluorine (¹⁹F) MR images as hyper- intense spots or regions against a background of only image noise. The background is devoid of anatomical features, since tissue fluorine concentration is insignificant relative to that within labeled cells -- distinguishing fluorine tracer from anatomical tissue features is not an issue. However, fluorine-based tracer agents often have a sparse spatial distribution and produce low levels of MRI signal, so it is often difficult to distinguish labeled cells from background image noise, especially when scan times are limited during in vivo experiments. To facilitate cell-tracking that employs fluorine-based tracers, we implemented and evaluated compressed sensing acquisition and reconstruction. This method generates 3D images with higher signal-to-noise ratios than conventional methods, allowing for 3D fluorine acquisitions with higher resolutions or shortened scan times.

Overall these methods for enhancing sensitivity to cells labeled with iron oxide CAs and fluorine tracer agents will help enable MRI as a platform for detecting and tracking cells in living subjects. Improved MRI cell monitoring will help researchers understand how normal and diseased cells behave and migrate inside living systems, and will help to determine the efficacy of new cellular therapies.