Vivoquant imaging software

In vitro phantoms comprising of different numbers of VivoTrax labeled islet organoids (0, 25, 50, 100, 200, and 400) in PBS were imaged using an MPI scanner (MOMENTUM MPI, Magnetic Insight Inc., Alameda, CA, United States) with the fiducial markers. Each 2D MPI images were acquired with parameters of a field-of-view (FOV) of 6 cm × 12 cm, a 5.7 T/m selection field gradient, a drive field strength of 20 mT peak amplitude and a 45.0 kHz drive frequency. Images were reconstructed using x-space reconstruction. Quantification of the islet organoids phantoms was performed using the 2D MPI image intensity calibrated against a fiducial marker of known iron content (2.2 μg of iron) using VivoQuant imaging software (Invicro, Boston, MA, United States).

Under anesthesia with 2% isoflurane, mice were imaged using 3D MPI at 1, 7, and 28-day post-transplantation (n = 3) with operating parameters, a FOV of 6 cm × 6 cm × 12 cm, acquisition time of 10 s per projection (total 55 projections) plus 30 s for automatic set up of the magnets, with an approximate total time of 35 min including for image reconstruction. Anatomic CT reference images were acquired by the whole-body model of Perkin Elmer QuantumGX microCT. MPI images were co-registered to CT with fiducial markers using VivoQuant Imaging Software (Invicro, Boston, MA, United States). Control animals did not receive islet organoids (n = 2).

In order to measure accuracy of the algorithm and determine its validity, its TIV prediction was compared to that of a manual rater, in this case a board-certified radiologist. The rater segmented the ROI manually on 15 images using the VivoQuant Imaging Software (Invicro, Boston, MA, United States), and total pixel sum values were extracted for TIV analysis from the ROIs using a ratio method with reference to a singular fiducial marker for TIV prediction (Makela et al., 2020 (link)), calibrated against a 40% fiducial marker of known iron concentration (2.2 μg of iron). For statistical analysis, SPSS statistical software (IBM, Armonk, NY, United States) was used to calculate intraclass correlation coefficient (ICC), which provides a measure of the inter-rater reliability between the rater and algorithm. A two-way mixed model with a confidence interval of 95% was selected, and a measure of absolute agreement was calculated with the ICC. A higher ICC score indicates greater reliability of algorithm performance due to the high degree of agreement between the rater and algorithm.