Search

Researchers at Nanjing University of Aeronautics and Astronautics in China have developed a new robotic approach for venous blood collection using a near-infrared imaging system. The study focuses on improving vein imaging quality through the use of a U-Net+ResNet18 neural network and optimizing the AD-Census stereo matching algorithm. Results show significant improvements in image segmentation quality, with potential applications for real-time puncture guidance in robotic venous blood collection. This research was funded by various organizations and published in the journal Sensors. [Extracted from the article]