Magnetic properties of iron-filled hydrogel clusters: a model system for quantitative susceptibility mapping with MRI

Abstract

Quantitative approaches in clinical Magnetic Resonance Imaging (MRI) benefit from the availability of adequate phantoms. Ideally, the phantom material should reflect the complexity of signals encountered in vivo. In the present study we validate and characterize clusters consisting of sodium-polyacrylate embedded in an alginate matrix that are unloaded or loaded with iron for Quantitative Susceptibility Mapping (QSM), yielding a non-uniform iron distribution and tissue-mimicking MRI properties. Vibrating sample magnetometry (VSM) was used to characterize the phantom material and verify the accuracy of previous MRI-based observations of the QSM based molar susceptibility (χM). MRI at 14.1 T with high resolution acquisitions was used to determine the size of hydrogel clusters and to further investigate the suitability of the phantom material as a model system for QSM at high field. VSM demonstrated that the iron-solution used for manufacturing the phantoms consisted of ferric iron. The χM of clusters with a constant iron-to-polyacrylate-ratio (8.3 μg/mg) observed with VSM was 50.7 ± 8.0 ppb mM−1 but showed a tendency towards saturation at total iron concentrations >1 mM. On unwrapped and background corrected phase-images obtained with gradient-echo MRI and an isotropic voxel size of 37 μm at 14.1T, the iron-free clusters had a roundish shape and blurry border with an equivalent sphere diameter of 276 ± 230 µm and a QSM of 7 ± 7 ppb. Iron-loading led to strong phase wrapping, necessitating the use of short echo times, or short inter-echo delays below 10 ms at 14.1 T. The equivalent sphere diameter of the iron-loaded clusters was estimated to 400–500 µm as verified using different MRI modalities (spin-echo, inversion recovery, and gradient echo MRI). With a constant iron-to-polyacrylate ratio, the cluster density was 10 mm−3 mM−1 iron. In agreement with previous observations, χM of samples with a constant amount of polyacrylate was 50.6 ± 11.4 ppb mM−1 at 3 T while samples containing clusters with a constant iron-to polyacrylate-ratio yielded χM = 56.1 ± 6.3 ppb mM−1 at 3T and 55.6 ± 0.7 ppb mM−1 at 14.1 T. In conclusion we found that the molar susceptibility of the proposed model system corresponds to that predicted for ferritin in vivo loaded with 3000 iron atoms. The reproducibility was within 12% across MR scanners, batches, and phantom types and compared well with results obtained with vibrating sample magnetometry.