Quantitative Susceptibility Mapping for Measuring Super-Paramagnetic Iron Oxide Concentrations in Lymph Nodes

Abstract

Background The sentinel lymph node is the first site to which breast cancer typically spreads. Methods to evaluate whether cancer has spread often entail the use of ionizing radiation. An alternative, non-ionizing method involves injecting the patient with super-paramagnetic iron oxide (SPIO), which accumulates in the sentinel nodes and can be detected using a probe. The accumulation of particles in lymph nodes is also visible on T2*-weighted magnetic resonance imaging (MRI), however none of these approaches provide quantitative information about SPIO concentration. With an image processing technique called quantitative susceptibility mapping (QSM), the content of specific elements can be measured. This is an established method for imaging of the brain. The aim of this study was to investigate the feasibility of using QSM on MR images of the sentinel node to gauge its iron concentration. Method The study was divided into two parts. The first part was creating a QSM pipeline adapted for images of the sentinel node, using already established algorithms developed for imaging of the brain. This was done by a literature review on available methods and trying possible candidates on patient images. The second part was creating a phantom containing vials with varying concentrations of SPIO. The phantom was scanned in an MRI system with vials both parallel and orthogonal to the magnetic field B0. Applying the QSM pipeline on images of the phantom, the relationship between iron concentration and susceptibility value was displayed. The function was then used to calculate the iron concentrations of the patient data. Results The best algorithms that were found were SEGUE for phase unwrapping, VSHARP for background field removal and MEDI for dipole inversion. These produced adequate susceptibility maps, with greater precision at lower susceptibilities and some distortion at higher susceptibilities. The susceptibility values of the phantom showed a relationship linearly increasing with iron concentration for 0-200 μg/ml, but for higher iron concentrations, the susceptibility values did not conform to the trend. The trend was more prominent in the images where the vials were orthogonal to B0. The concentrations of the patients were high and had very large standard deviations. Conclusions The QSM method used in this work is better suited for lower concentrations of iron and could not yield stable results for the high concentrations in the patient data. Further research is needed to find a way to ensure statistical stability for calculating the concentrations in the sentinel node. The phantom worked as intended, in delivering a description of the relationship between iron concentration and susceptibility value. There is much room for improvement regarding the procedure of creating the phantom as well as the method for creating the QSM maps, thereby enabling more precise estimates of SPIO concentrations in patient lymph nodes.