Comparison of Inline R2*MRI versus FerriScan for liver iron quantification in patients on chelation therapy for iron overload: preliminary results

Objectives MRI quantification of liver iron concentration (LIC) using R-2 or R-2* relaxometry requires offline post-processing causing reporting delays, administrative overhead, and added costs. A prototype 3D multi-gradient-echo pulse sequence, with inline post-processing, allows immediate calculation of LIC from an R-2* map (inline R-2*-LIC) without offline processing. We compared inline R-2*-LIC to FerriScan and offline R-2* calibration methods. Methods Forty patients (25 women, 15 men; age 18-82 years), prospectively underwent FerriScan and the prototype sequence, which produces two R-2* maps, with and without fat modeling, as well as an inline R-2*-LIC map derived from the R-2* map with fat modeling, with informed consent. For each map, the following contours were drawn: ROIs, whole-axial-liver contour, and an exact copy of contour utilized by FerriScan. LIC values from the FerriScan report and those calculated using an alternative R-2 calibration were the reference standards. Results were compared using Pearson and interclass correlation coefficients (PCC, ICC), linear regression, Bland-Altman analysis, and estimation of area under the receiver operator curve (ROC-AUC). Results Inline R-2*-LIC demonstrated good agreement with the reference standards. Compared to FerriScan, inline R-2*-LIC with whole-axial-liver contour, ROIs, and FerriScan contour demonstrated PCC of 94.8%, 94.8%, and 92%; ICC 93%, 92.7%, and 90.2%; regression slopes 1.004, 0.974, and 1.031; mean bias 5.54%, 10.91%, and 0.36%; and ROC-AUC estimates 0.903, 0.906, and 0.890 respectively. Agreement was maintained when adjusted for sex, age, diagnosis, liver fat content, and fat-water swap. Conclusion Inline R-2*-LIC provides robust and comparable quantification of LIC compared to FerriScan, without the need for offline post-processing.