We report anatomically correct 3D‐printed mouse phantoms that can be used to plan experiments and evaluate analysis protocols for magnetic particle imaging (MPI) studies. The 3D‐printed phantoms were based on the Digimouse 3D whole body mouse atlas and incorporate cavities representative of a liver, brain tumor, and orthotopic breast cancer tumor placed in anatomically correct locations, allowing evaluation of the effect of precise doses of MPI tracer. To illustrate their use, a constant tracer iron mass was present in the liver for the breast (200 μgFe) and brain tumor (10 μgFe) model, respectively, while a series of decreasing tracer iron mass was placed in the tumor region. MPI scans were acquired in 2D and 3D high sensitivity and high sensitivity/high resolution (HSHR) modes using a MOMENTUM imager. A thresholding algorithm was used to define regions of interest (ROIs) in the scans and the tracer mass in the liver and tumors was calculated by comparison of the signal in their respective ROI against that of known mass fiducials that were included in each scan. The results demonstrate that this approach to image analysis provides accurate estimates of tracer mass. Additionally, the results show how the limit of detection in MPI is sensitive to the details of tracer distribution in the subject, as we found that a greater tracer mass in the liver cavity resulted in poorer sensitivity in tumor regions. These experiments illustrate the utility of the reported 3D‐printed anatomically correct mouse phantoms in evaluating methods to analyze MPI scans and plan in vivo experiments.

Purpose: Magnetic particle imaging (MPI) is being explored in biological contexts that require accurate and reproducible quantification of superparamagnetic iron oxide nanoparticles (SPIONs). While many groups have focused on improving imager and SPION design to improve resolution and sensitivity, few have focused on improving quantification and reproducibility of MPI. The aim of this study was to compare MPI quantification results by two different systems and the accuracy of SPION quantification performed by multiple users at two institutions. Procedures: Six users (3 from each institute) imaged a known amount of Vivotrax+ (10 micrograms Fe), diluted in a small (10 microliter) or large (500 microliter) volume. These samples were imaged with or without calibration standards in the field of view, to create a total of 72 images (6 users x triplicate samples x 2 sample volumes x 2 calibration methods). These images were analyzed by the respective user with two region of interest (ROI) selection methods. Image intensities, Vivotrax+ quantification, and ROI selection was compared across users, within and across institutions. Results: MPI imagers at two different institutes produce significantly different signal intensities, that differ by over 3 times for the same concentration of Vivotrax+. Overall quantification yielded measurements that were within +/- 20% from ground truth, however SPION quantification values obtained at each laboratory were significantly different. Results suggest that the use of different imagers had a stronger influence on SPION quantification compared to differences arising from user error. Lastly, calibration conducted from samples in the imaging field of view gave the same quantification results as separately imaged samples. Conclusions: This study highlights that there are many factors that contribute to the accuracy and reproducibility of MPI quantification, including variation between MPI imagers and users, despite pre-defined experimental set up, image acquisition parameters, and ROI selection analysis.

No abstract

No abstract