Conditional anisotropic diffusion for leukocyte identification

Leukocyte identification is pursued in order to measure important parameters of inflammation including rolling leukocyte flux and rolling leukocyte volume fraction. The automated identification of leukocytes also serves as a precursor to leukocyte tracking. In this study, a conditional anisotropic diffusion model is used to enhance edges within video microscopy depicting rolling leukocytes within small venules. The detected edge chains are pruned according to the geometric information given in the leukocyte models, and candidate edge chains representing potential leukocytes are approximated by circular contours. The performance of the proposed approach is evaluated using images of TNF-α treated venules within the mouse cremaster.