The study of the detailed information of precapillary circulation is very important to examine the control method for a total artificial heart or lung. We developed a new system to analyze visualized capillary image on real-time operation with a microscopic CCD and machine vision system.
On the present system, in the case of narrow capillary, discrete images are extracted by using partial derivatives of a Gaussian smoothing kernel, and line structures are detected according to parameters of the polynominal. The diameter of the capillary was defined from the parameter of them. In the case of relatively bold capillary or arteriola, ROI was defined by using threshold process. The diameter was detected from the smallest surrounding rectangle of the thresholded region. Template matching technique with sub-pixel accuracy was used to determine the velocity profile. Parameters for template matching were defined from the detected diameter and center of line contour profile. The present method was applied to the measurement of mesentery of a bull frog, and the validity of this system was discussed.