Visual vein segmentation for an intravenous injection support system

In the medical field, one of the most frequently performed medical practice is intravenous injection and it is also a significant factor to increase the workload of medical physicians. Recently, many support systems have been developed to give better injection and to reduce the workload. However, most of them are large-scaled and expensive. To deal with this problem we have been engaged in developing a low-cost injection support system by using a USB camera and robot. This paper presents a vein segmentation method for this system. In this method, an emphasizing process is firstly described to accentuate visual veins by focusing on the red component of the color image. Then, after thresholding a number of randomly selected sub-images with different sizes, the visual veins are obtained by aggregating those thresholded sub-images into a continuous segmented image by majority vote. The performance of the proposed method is demonstrated by making experiments with 15 subjects' vein images.