Evaluation of a model-based poroelastography algorithm for edema quantification

Abstract

A critical component of end stage renal disease treatment is optimal fluid management. Patients with reduced kidney function risk developing fluid overload, and many dialysis patients exceed recommended levels of fluid retention, leading to frequent clinical intervention and higher mortality. Despite this, current clinical standards of care rely on reactive semi-quantitative tests to grade fluid overload and peripheral edema. Poroelastography has been proposed as a quantitative means of estimating fluid overload in peripheral edema, but clinical studies have met limited success, in part because of measurement noise. We developed a new poroelastography method based on an inverse problem formulation to address the shortcomings of current methods. Our technique iteratively minimizes the error between ultrasound displacement measurements and the solution of a Biot poroelastic model. This formulation does not depend on noise-amplifying derivatives and ratios. We tested our algorithm in a simulation study using synthetic displacement data with Gaussian noise and speckle tracking measurements from simulated sonograms. Our algorithm accurately estimated the Poisson’s ratio at all noise levels tested. Reconstruction errors for all cases were less than 10% and outperformed traditional methods. Although our method required 4-12 hours to run, it is easily parallelized. Future work will focus on benchtop validation.