abstract | dataset | people

Abstract  

Intravital microscopy has been used to visualize the microcirculation by imaging fluorescent labeled red blood cells (RBCs). Traditionally, microcirculation has been modeled by computing the mean velocity of a few, randomly selected, manually tracked red blood cells. However, this protocol is tedious, time-consuming, and subjective with technician related bias. We present a new method for analyzing the microcirculation by modeling the red blood cell motion through automatic tracking. The tracking of red blood cells is challenging as in each image, as many as 200 cells move through a complex network of vessels at a wide range of speeds while deforming in shape. To reliably detect RBCs traveling at a wide range of speeds, a window of temporal template matching is applied to detect the cells. Then, cells appearing in successive frames are corresponded based on the motion behavior constraints in terms of the direction,magnitude, and path. The performance evaluation against a ground truth indicates the detection accuracy up to 84% TP at 6% FP and a correspondence accuracy of 89%. We include an in-depth discussion on comparison of the microcirculation based on motion modeling from the proposed automated method against a mean velocity from manual analysis protocol in terms of precision, objectivity, and sensitivity.

Dataset

The dataset for evaluating the tracking  was composed of 3 experimental conditions:  Mild remote trauma (Femur Fracture: ffx) injected with saline, a severe polymicrobial sepsis induced by cecal ligation and puncture followed by a femur fracture (sequential stress, SS) injected with apc drug, and SS injected with saline. 9 video sequences of 100 frames each captured at 30 fps were used to evaluate the tracking of the algorithm. A mask image labeling the central vein in white, the lobule area in gray, and the non-lobule area in black is used as input for the algorithm along with the input frames. Red blood cells only appearing in the lobule are are tracked.

The ground truth is formatted in a csv file. Below is the format of the ground truth file. An area of the lobule is described as a bounding box (top left point and and bottom right point), and all rbc positions appearing in that bounding box for two consecutive frames are recorded. Each rbc detected in a frame is given an index and a correspondence index that the rbc corresponds to in the neighboring frame. An rbc can have no correspondence. In that case, it is given a correspondence index of 0.

The rbctrack output is composed of a movie in the mp4 format. The movie is created at 5 fps so tracking output can be more easily seen. Quicktime is required to play. Right click on the link and select "Save Link As " or "Save Target As" to download. The central vein is outlined in magenta. Red blood cells detected are drawn in a bounding box. A cyan bounding box means the algorithm detected no correspondence for that cell. A red bounding box is the current position of the detected rbc and the blue line drawn from the red bounding box to the blue bounding box shows the path of the rbc between the current frame and the next frame.

People

• Walid S. Kamoun   Email: wskamoun at uncc dot edu
• Stephen J. Schmugge  Email: sjschmug at uncc dot edu
• Jerrod Kraftchick Email: jpkraftc at uncc dot edu
• Mark G. Clemens
• Min C. Shin   Email:  mcshin at uncc dot edu (contact person)