1222 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 21, NO. 10, OCTOBER 2002 Tracking Leukocytes In Vivo With Shape and Size Constrained Active Contours Nilanjan Ray, Student Member, IEEE, Scott T. Acton , Senior Member, IEEE, and Klaus Ley Abstract—Inflammatory disease is initiated by leukocytes (white blood cells) rolling along the inner surface lining of small blood ves- sels called postcapillary venules. Studying the number and velocity of rolling leukocytes is essential to understanding and successfully treating inflammatory diseases. Potential inhibitors of leukocyte recruitment can be screened by leukocyte rolling assays and suc- cessful inhibitors validated by intravital microscopy. In this paper, we present an active contour or snake-based technique to automat- ically track the movement of the leukocytes. The novelty of the pro- posed method lies in the energy functional that constrains the shape and size of the active contour. This paper introduces a significant enhancement over existing gradient-based snakes in the form of a modified gradient vector flow. Using the gradient vector flow, we can track leukocytes rolling at high speeds that are not amenable to tracking with the existing edge-based techniques. We also pro- pose a new energy-based implicit sampling method of the points on the active contour that replaces the computationally expensive explicit method. To enhance the performance of this shape and size constrained snake model, we have coupled it with Kalman filter so that during coasting (when the leukocytes are completely occluded or obscured), the tracker may infer the location of the center of the leukocyte. Finally, we have compared the performance of the pro- posed snake tracker with that of the correlation and centroid-based trackers. The proposed snake tracker results in superior perfor- mance measures, such as reduced error in locating the leukocyte under tracking and improvements in the percentage of frames suc- cessfully tracked. For screening and drug validation, the tracker shows promise as an automated data collection tool. Index Terms—Active contours, cell tracking, inflammatory dis- ease, leukocytes, video microscopy. I. INTRODUCTION T RACKING leukocytes in vivo is becoming increasingly important among medical research groups that are studying inflammatory disease [1], [2]. Leukocyte rolling is largely mediated by the selectin family of adhesion molecules with contributions from integrins and integrins [1]. Analysis of leukocyte rolling is an important tool in dis- covering potential novel anti-inflammatory treatments. For example, E-selectin inhibitors have been shown to reduce the number and increase the velocity of rolling leukocytes in a model of inflammation in living animals [3]. Increased rolling Manuscript received October 25, 2001; revised July 14, 2002. This material is based upon work supported in part by the Whitaker Foundation. Asterisk in- dicates corresponding author. N. Ray is with the Department of Electrical and Computer Engineering, the University of Virginia, Charlottesville, VA 22904 USA. S. T. Acton is with the Department of Electrical and Computer Engineering, the University of Virginia, 351 McCormick Road, Charlottesville, VA 22904 USA (e-mail: acton@virginia.edu). K. Ley is with the Department of Biomedical Engineering and Cardiovascular Research Center, the University of Virginia, Charlottesville, VA 22904 USA. Digital Object Identifier 10.1109/TMI.2002.806291 velocity under otherwise identical hemodynamic conditions is indicative of weaker, fewer or shorter-lived bonds between the rolling cell and the endothelial lining of the inflamed blood vessel. Currently the analysis of rolling velocities is laborious and requires tens of hours of user-interactive image processing work after each experiment. Rolling velocity is a key predictor of inflammatory cell recruitment [4]. The most powerful description of leukocyte rolling velocities is a velocity distribution, preferably for hundreds of cells [5]. In addition to its use in intravital microscopy, a robust and automatic tracking algorithm would also expand the scope of flow chamber assays. A flow chamber [6] consists of a trans- parent parallel-plate apparatus perfused at low Reynolds num- bers to match wall shear stresses observed in blood vessels in vivo. The vessel wall is modeled as an isolated protein sup- porting leukocyte rolling in a planar lipid bilayer [7] or directly immobilized on glass or plastic [8] or by endothelial cells grown on the lower plate of the flow chamber [9]. Centroid trackers are successful at tracking leukocytes rolling on transparent substrata like protein-coated plastic [8], but when rolled over endothelial cells the tracking becomes difficult [10]. This difficulty is due to the structural clutter and obstructions introduced by the optical properties of the endothelial cells. Flow chamber experiments are widely used to screen for compounds that may inhibit leukocyte interaction with in- flamed blood vessels. Glycotech, Inc. (Rockville, MD) offers a single-channel flow chamber for such uses in drug screening. More recently high-throughput approaches are being developed by using hydrodynamic focusing (CelTor, Inc., Santa Clara, CA). In these systems, cells are visualized using phase contrast microscopy, a technique that can yield a “bright” or “dark” image of the cell, dependent on the position of the focus of the objective relative to the rolling cell. These and other approaches would benefit from a robust tracking algorithm that can track leukocytes even in the presence of clutter, obstruction and change of focus. The most challenging application is intravital microscopy where rolling cells are observed in living microvessels (in vivo) under conditions of inflammation. These experiments add motion artifacts to the challenge of image processing, and no currently existing algorithm is successful at tracking rolling leukocytes in vivo. In this paper, we present an active contour or snake [11] based tracking of the leukocytes from video sequences. As an ex- ample, a portion of a video sequence is shown in Fig. 1. The imaging technology and the in vivo experimental setup have been described in [12]. As a result of the mismatch of refrac- tory indices of the rolling cell and the surrounding plasma, con- trast/intensity change/reversal occurs quite often in such video 0278-0062/02$17.00 © 2002 IEEE