Performance of Generalized Factor Analysis of Dynamic Sequence (GFADS) in the Automated Characterization of Renal Function and Tissue Enhancement in Dynamic Magnetic Resonance Imaging (MRI)

Ruth Lim¹, Jinsong Ouyang¹, Matthew D. Schmitz¹, Michael S. Gee¹, Ranu Shailam¹, Raul N. Uppot¹, Georges El Fakhri¹

We assessed the performance of a novel generalized factor analysis of dynamic sequences (GFADS) in dynamic, contrast-enhanced renal magnetic resonance imaging (MRI). By detecting unique time-intensity curves for each renal tissue/compartment type, this technique automates the creation of regions of interest (ROIs) around and within the kidneys, and obviates the need for manually-drawn ROIs. These time factor curves are computed from entire factor images and are significantly less affected by noise than time-intensity curves computed within regions of interest that span a few voxels. In this study, we found that GFADS software can successfully, semi-automatically, and rapidly identify the renal cortex, medulla, and collecting system on dynamic contrast-enhanced renal MRI studies while obviating the need to use manually-drawn regions of interest. This enables detailed quantitative assessment of cortical and medullary renal function in normal and abnormal kidneys.