Objective assessment of renal DCE-MRI image segmentation

Abstract

In dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) of renal perfusion with injection of a contrast agent, the segmentation of kidney in regions of interest like cortex, medulla and pelvo-caliceal cavities is necessary for accurate functional evaluation. Several semiautomatic segmentation methods using time-intensity curves of renal voxels have been recently developed. Most of the time, quantitative result validation consists in comparisons with a manual segmentation by an expert. However it can be questionable to consider such a segmentation as a ground truth, especially because of intra- and inter-operator variability. Moreover it makes comparisons between results published by different authors delicate. We propose a method to built synthetic DCE-MRI sequences from typical time-intensity curves and an anatomical model that can be used for objective assessment of renal internal structures.