A computer vision system for automated measurement of leukocyte rolling velocities in vivo.

摘要

White blood cells, or leukocytes, are an important component of the mammalian immune system. They are responsible for detecting foreign material in the body, producing the antibodies to destroy the invaders, and creating the conditions necessary to fight the invasion and repair tissue. When these conditions arise, a cascade of reactions occurs that activates the leukocytes. These activated cells stick to the cells lining the small veins found in the body, causing the leukocytes to roll slowly along the vessel wall and eventually stop. Subsequently, they move through the vessel wall into the surrounding tissue to attack the invader or alter the environment in the tissue. As the degree of the immune response increases, so does the attraction between the leukocytes and the blood vessel wall, resulting in slower moving cells and more sticking cells along the wall. Measurement of this rolling behavior provides a means to quantify the degree of the ongoing immune system response in a living animal.;It is a relatively simple task for the human vision system to detect the rolling white blood cells in videotaped microscopic images. This project was designed to mimic these abilities using digital image analysis of the same sequence of images. Various methods of motion estimation were attempted to determine those that were most effective. Also, several different ways of segmenting the cells from the images were tested. Once the moving cells had been found, the computer was able to measure the distance they traveled and collect statistical data about the behavior they exhibited. The results of the automated analysis were compared with manually collected data to ensure that they were statistically similar. Additionally, the system was able to compute the velocity data in real time.