Rutgers University engineers have created a tabletop device that combines a robot, artificial intelligence, and near-infrared and ultrasound imaging to draw blood or insert catheters to deliver fluids and drugs. The device can accurately steer needles and catheters into tiny blood vessels with minimal supervision. It can perform complex visual tasks, including identifying the blood vessels from the surrounding tissue, classifying them, and estimating their depth, followed by motion tracking. Robotic cannulation is driven by predictions from a series of deep convolutional neural networks that encode spatiotemporal information from multimodal image sequences to guide real-time motion control of the robot. This tabletop robotic device can accurately steer needles and catheters into tiny blood vessels with minimal supervision. Courtesy of Martin Yarmush and Alvin Chen. Through imaging and robotic tracking studies in volunteers, the researchers demonstrated the ability of the device to segment, classify, localize, and track peripheral vessels in the presence of anatomical variability and motion. They evaluated robotic performance in phantom and animal models with difficult vascular access and showed that the device could improve success rates and procedure times compared to manual cannulations done by trained operators, particularly in challenging physiological conditions. “Using volunteers, models, and animals, our team showed that the device can accurately pinpoint blood vessels, improving success rates and procedure times compared with expert health care professionals, especially with difficult-to-access blood vessels,” professor Martin L. Yarmush said. The team plans to conduct additional research on the device with a broader range of subjects, including people with normal blood vessels and with vessels that are problematic. “Not only can the device be used for patients, but it can also be modified to draw blood in rodents, a procedure that is extremely important for drug testing in animals in the pharmaceutical and biotech industries,” Yarmush said. The research was published in Nature Machine Intelligence (www.doi.org/10.1038/s42256-020-0148-7).