Co-Registration of Bioluminescence Tomography, Computed Tomography, and Magnetic Resonance Imaging for Multimodal In Vivo Stem Cell Tracking

Abstract

We present a practical approach for co-registration of bioluminescence tomography (BLT), computed tomography (CT), and magnetic resonance (MR) images. To this end, we developed a customized animal shuttle composed of non-fluorescent, MR-compatible Delrin plastic that fits a commercially available MR surface coil. Mouse embryonic stem cells (mESCs) were transfected with the luciferase gene and labeled with superparamagnetic iron oxide (SPIO) nanoparticles. Cells were stereotaxically implanted in mouse brain and imaged weekly for 4 weeks with BLI (IVIS Spectrum CT scanner) and MRI (11.7T horizontal bore scanner). Without the use of software co-registration, in vitro phantom studies yielded root-mean-square errors (RMSE) of 7.6×10^-3, 0.93 mm, and 0.78 mm along the medial-lateral (ML), dorsal-ventral (DV), and anterior-posterior (AP) axes, respectively. Rotation errors were negligible. Software co-registration by translation along the DV and AP axes resulted in consistent agreement between the CT and MR images, without the need for rotation or warping. In vivo co-registered BLT/MRI mouse brain data sets demonstrated a single, diffuse region of BLI photon signal and MRI hypointensity. Over time, the transplanted cells formed tumors as validated by histopathology. Disagreement between BLT and MRI tumor location was greatest along the DV axis (1.4±0.2 mm) compared to the ML (0.5±0.3 mm) and AP axis (0.6 mm) due to the uncertainty of the depth of origin of the BLT signal. Combining the high spatial anatomical information of MRI with the cell viability/proliferation data from BLT should facilitate pre-clinical evaluation of novel therapeutic candidate stem cells.

Keywords: Cell tracking; Computed tomography; Multimodal imaging; Stem cells; bioluminescence imaging; co-registration; magnetic resonance imaging.

Figure 1.

Unmodified commercial mouse imaging shuttle (A). Custom modification (indicated in red) to accommodate the radiofrequency (RF) magnetic resonance imaging (MRI) surface coil with cutouts to hold clear animal restraint straps and a respiration sensor lead (green) (B). Drawing showing the animal holder assembled with the RF MRI coil (white), placed directly above the mouse brain during magnetic resonance (MR) imaging (C).

Figure 2.

Transaxial images of coregistered air–water phantom images from computed tomography (CT) (A) and MRI (B), and showing excellent agreement between the sample overlay of the 2 modalities (C).

Figure 3.

Coronal (A) and sagittal (B) in vivo mouse brain images of coregistered CT (gray scale) and T1-weighted fast low-angle shot (FLASH) MR images (hot color scale).

Figure 4.

In vivo coronal images 4 weeks post cell transplantation (A–C). The bioluminescence tomography (BLT) (hot color scale)-reconstructed luciferase (Luc)-mouse embryonic stem cells (mESC) location is superimposed on the T2-weighted rapid acquisition with relaxation enhancement (RARE) MR images for all 3 mice. T2-weighted MR volume-rendered mouse brain from panel (A) at 4 weeks post cell transplantation, showing an overlay of the BLT-reconstructed Luc-mESC location (orange) and the segmented MRI tumor volume (green) (D).

Figure 5.

Hematoxylin and eosin (H&E)-stained coronal section showing tumor near implantation site (A, B). Prussian blue-stained section with nuclear fast red counterstain (C, D). Superparamagnetic iron oxide (SPIO) appears as blue deposits in the stain. Immunohistological stain for Luc (green) against DAPI (4′,6-diamidino-2-phenylindole) nuclear counterstain, showing Luc-expressing cells at both the original transplantation site and superficial lesion (E, F).

Figure 6.

Coronal MR images at 1, 2, and 4 weeks (left to right) after transplantation. The SPIO-labeled cell hypointensity induces a blooming effect, masking an initial visualization of tumor growth at the 1- and 2-week time points. By week 4, the tumor has considerably expanded, with fragmented pockets of the originally hypointense cells located within the center.

Figure 7.

Bioluminescent imaging (BLI) signal and segmented tumor volume values at different time points following transplantation, normalized to the initial values at day 1 (n = 3) (A). Comparison of the total BLI signal, normalized to the initial values at day 1, against the BLT-reconstructed source power (n = 3) (B). Results are presented as mean values ± SD; asterisks denote significance level versus day 1 (*P < .10, **P < .05).