Hierarchical Region Splitting Detects Iron-Oxide Labeled Neural Stem Cells in Susceptibility Weighted Magnetic Resonance Imaging

Human neural stem cells are multipotent cells that aid in recovery from hypoxic ischemic injury (HII). The work proposes that an automated and computational method, Hierarchical Region Splitting (HRS) of Susceptibility Weighted Images (SWI) can identify iron oxide labeled human neural stem cells (hNSCs) within the hypoxic-ischemic brain. Using a unilateral carotid artery occlusion model in neonatal rat pups, we implanted hNSCs that were labeled with Feridex (iron-oxide agent). The rats were then non-invasively imaged (1-90 days post implantation) using SWI magnetic resonance imaging (MRI), which enhances visualization of hypointensity areas that contain iron-oxide labeled hNSCs. SWI images were further enhanced by creating minimum intensity projections (MIP's) from raw magnitude images that were combined with phase images. SWI, due to its sensitivity to paramagnetic compounds, such as iron oxide, resulted in an 8-fold enhancement of small regions of iron oxide labeled stem cells. HRS analysis allowed for semi-automatic detection of hNSCs from SWI images. We found hNSCs in the striatum and regions adjacent to the lesion that corresponded to histological Prussian blue staining, confirming the presence of hNSCs in the areas identified by HRS. hNSC volumes were increased at 30 days and then plateaued until 90 days post implantation. Phase values (radians) were shown to be increased in MR images containing stem cells at later time points consistent with a reduction in iron oxide as hNSCs replicate and migrate from the implantation site. Quantitative SWI derived hNSC volumes were similar to that previously reported. Our results confirm that HRS in conjunction with the high sensitivity of SWI could be used to monitor iron-labeled hNSCs in a clinically relevant manner.