Reproducibility of DCE-MRI time-intensity curve-shape analysis in patients with knee arthritis: A comparison with qualitative and pharmacokinetic analyses

To compare the between-session reproducibility of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) combined with time-intensity curve (TIC)-shape analysis in arthritis patients, within one scanner and between two different scanners, and to compare this method with qualitative analysis and pharmacokinetic modeling (PKM). Fifteen knee joint arthritis patients were included and scanned twice on a closed-bore 1.5T scanner (n = 9, group 1), or on a closed-bore 1.5T and on an open-bore 1.0T scanner (n = 6, group 2). DCE-MRI data were postprocessed using in-house developed software ("Dynamo"). Disease activity was assessed. Disease activity was comparable between the two visits. In group 1 qualitative analysis showed the highest reproducibility with intraclass correlation coefficients (ICCs) between 0.78 and 0.98 and root mean square-coefficients of variation (RMS-CoV) of 8.0%-14.9%. TIC-shape analysis showed a slightly lower reproducibility with similar ICCs (0.78-0.97) but higher RMS-CoV (18.3%-42.9%). The PKM analysis showed the lowest reproducibility with ICCs between 0.39 and 0.64 (RMS-CoV 21.5%-51.9%). In group 2 TIC-shape analysis of the two most important TIC-shape types showed the highest reproducibility with ICCs of 0.78 and 0.71 (RMS-CoV 29.8% and 59.4%) and outperformed the reproducibility of the most important qualitative parameter (ICC 0.31, RMS-CoV 45.1%) and the within-scanner reproducibility of PKM analysis. TIC-shape analysis is a robust postprocessing method within one scanner, almost as reproducible as the qualitative analysis. Between scanners, the reproducibility of the most important TIC-shapes outperform that of the most important qualitative parameter and the within-scanner reproducibility of PKM analysis