#118: Coupled edge profile geodesic active contours for red blood cell flow analysis


Abstract

Physiological properties of blood flow at the microvasculature scale can be measured by tracking the movement and density of red blood cells (RBCs). In this paper we propose a method for individual RBC segmentation to enable tracking and capturing dynamically varying bulk transport properties. RBCs have varying annular and disk like morphologies, and are often clustered into clumps that are difficult to segment using watershed-based methods. Edge profile active con- tours in combination with graph coloring based coupling (C-EPAC) are introduced as a robust approach to prevent merges between ad- jacent, clumped RBCs by modifying the active contour energy func- tion to be sensitive to a specific edge profile and not just the mag- nitude as in the traditional methods. Explicit coupling is combined with graph coloring to efficiently compute the contour evolution us- ing the fewest number of level sets to support high-throughput stud- ies of RBC flow characterization under varying physiological condi- tions.