Quantification of kidney volume from MRI without constrast medium injection in ADPKD patients

QUANTIFICATION OF KIDNEY VOLUME FROM MRI WITHOUT CONTRAST MEDIUM INJECTION IN ADPKD PATIENTS INTRODUCTION Autosomal dominant polycystic kidney disease (ADPKD) is the most common renal genetic disorder with a prevalence greater than 1:1000 individuals. ADPKD progresses to end-stage renal disease (ESRD), requiring renal replacement therapy (RRT) in 45% of patients by the age of 60 (1). However, the progression rate of ADPKD to ESRD is quite variable and the age at onset of chronic renal failure (CRF) ranges from 2 to 80 years. The mechanism of CRF in ADPKD is not yet defined. It has been proposed that growth of renal cysts leads to renal failure by compressing adjacent normal parenchyma. As the cysts become more numerous and increase in size expanding the total cyst volume (TCV), the kidney become completely replaced by cysts with increasing of total renal volume (TRV). Imaging techniques, such as Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) well document enlarged bilateral kidneys with multiple cysts in ADPKD and they have been employed to calculate both TCV and TRV. The latter is considered a predictive marker of the CRF development. TRV in ADPKD has been mostly evaluated using CT or MRI data with contrast medium (2). However, contrast medium may cause renal failure and shouldn’t be administered to patients with impaired renal function. In addition, to obtain TRV and TCV, kidney and cyst segmentation is based on manual tracing or relies on a heavy manual intervention. In this study we developed a method based on T2-weighted MRI acquisition without the use of contrast medium, in order to quantify TRV rapidly, automatically and safely. METHODS To test the accuracy and reliability of our imaging method and data analysis a phantom was designed to simulate a kidney and cysts within it. The phantom was prepared following Mitchell et al. (3). The volume of the agarose solution simulating kidney parenchyma (590 ml) as well as the volume of the grapes and tomatoes simulating cysts (150 ml) were measured with a graduated cylinder. The phantom was placed in an oil bath, and then in an external water bath before MR scanning. The in-vivo validation of the developed technique included twenty-six ADPKD patients (age range: 24÷68 yrs; mean±SD: 42±11 yrs; 11 males) with a wide range of kidney volumes (127÷3132 ml). ADPKD patients with normal renal function and CRF were recruited and in all of them the diagnosis of ADPKD was made with echography investigation and based on Ravine criteria. MRI data were obtained using a 1.5T Philips scanner. SPIR sequences were used to acquire axial images. The MRI datasets were analyzed using custom software, which provides kidney contour detection on each acquired slice. The procedure to detect kidney contours required several steps. First, a fully automated 3D reconstruction of the volumetric data was performed from the short-axis slices. In the mid slice of the volume the operator selected two points inside the right and left renal cavities. Subsequently, automatic kidney segmentation was performed separately for each kidney, starting from the corresponding manually selected point, applying a region growing technique. Then the detected rough contour was optimized by applying (1) a flood-fill operation on background pixels, followed by (2) a curvature motion in the areas in which the contour curvature is negative and (3) morphological opening operations (fig. 2c). On each slice the kidney area in mm2 was calculated as the number of pixels inside the detected contour whose spacing is known. The kidney volume was then computed by applying the method of discs. To obtain reference values for areas and volumes, images were also analyzed using a commercial software (ImageJ, http://rsbweb.nih.gov/ij/). applying the stereology method to manually traced contours. RESULTS AND DISCUSSION The analysis of a single dataset using the reference technique required 20 min for each ki...