Position estimation of veins under fossa cubitalis from near-infrared images

This study aims to develop a system for automating or assisting the intravenous injection operation. For the purpose, in this paper, position estimation of veins under fossa cubitalis from near-infrared camera vision is considered. In general, near-infrared images are used for veins detection, and veins are captured on the images as black area because reduced hemoglobin in veins absorbs near-infrared light. The accuracy of position estimation from images is affected by light wavelength, irradiation intensity, lens distortion, quantization error and detecting algorithm. This study counts such causes of estimation error, and considers corresponding correction methods. In parallel, allowable limits of estimation error is derived from relationship between the size of acus and veins. An arm phantom is developed to evaluate the proposed position estimation for veins and to verify its effectiveness.