Segmentation of renal tubules and automatic biomarker quantification in ciliopathy preclinical models.

Abstract

We develop a method for multi-structure segmentation of the mice kidney, which enables segmenting the renal cortex, the renal tubules, and tubular lumens. This facilitates quantitative exploration of kidney anatomy and the definition of image-based biomarkers. We apply this method to nephronophthisis (NPHP), a rare genetic renal disease from the family of ciliopathies, characterized by the appearance of large, dilated tubules within the cortex. We combine tubule detection with geometric criteria to select dilated structures. This enables the automatic computation of quantitative NPHP biomarkers with high correlation with manually collected measures, proving the reliability of the proposed method.