Abstract
Phlebotomy may cause unnecessary injuries to a patient whose veins are not easily visible to a healthcare professional. To mitigate this problem we designed a new system to image subcutaneous veins. Multispectral images were obtained using a microprocessor, an IR (infrared) camera, different wavelengths of NIR (near-infrared) sources, and an IR band-pass filter. Raw vein images were enhanced, colored, and displayed on a monitor using an easy-to-use interface. The mean dice similarity index (DSI) between the vein border specified by a doctor on the raw images manually and the automated segmented by the proposed system is determined as 0.92 ± 2.1 for 20 subjects. Also, the average peak signal-to-noise ratio (PSNR) obtained a high value of 68.37 ± 1.56 from the enhanced image. Phlebotomists can easily observe the subcutaneous veins in real-time with the three different options using the proposed device. As a result, this study advances the vein imaging field which has the potential to reduce injury to the patient during venipuncture.