Vein Tracking Using 880nm Near Infrared and CMOS Sensor with Maximum Curvature Points Segmentation

Vein Tracking Using 880nm Near Infrared and CMOS Sensor with Maximum Curvature Points Segmentation

Author		: ANGGUNMEKA LUHUR PRASASTI; Richard Karel Willem Mengko
Published on	: 7th WACBE World Congress on Bioengineering

Abstract

Injection error usually occurs in obese patients and children (especially dehydrated children). More than one injection errors may cause uncomfortable condition for patient and can lead to hematoma. The tools which is used to obtain vein image tends to be expensive because it uses high intensity infrared camera with CCD sensor along with highly sensitive to infrared so that vein image can be seen very clearly. CMOS sensors can be an alternative to CCD because the price is more affordable and readily available. However, vein image from CMOS sensor is not as good as CCD sensor so it needs advanced image processing to get better vein image quality. The 880nm near-infrared light, CMOS sensor and 880nm optical filters, help clarify the veins existence up to 3mm below the skin surface. Image acquisition results have low contrast and spreading noise resembles info (vein) that thresholding can not be conducted to separate the info and background. Adaptive smoothing and retinex is used as preprocessing to eliminate noise and to enhance information and background contrast. Maximum curvature points methode is quite good in segmentation for vein location tracking. It can be used in phlebhotomy robot and calculation of many things related to vein changes, for example in detecting vein collapse which is very dangerous and important to be known. Testing on white skin subjects have higher accuracy and precision than the dark and thick skin subjects. This system can track hand vein with 83.92% accuracy and 94.28% precision rate.