Detection and tracking of rolling leukocytes from intravital microscopy

We present a method for detecting and tracking rolling leukocytes from intravital video microscopy. The paper emphasizes a new method for detecting cells that is based on finding ellipses that have maximal outward normal gradient and minimal gradient variance around the boundary. A fast convolution-based implementation of the detector is discussed. For tracking, we present a radial active model that exploits a gradient vector flow field tailored to tracking objects in the direction of blood flow. Results are presented that compare the detection performance and tracking performance to that of existing methods.