Infrared vein scanners: Drawing blood gets easier with this new tech

Researchers in Russia over time have developed a mock-up of a medical imaging system that will use neural networks to examine near-infrared images of veins and project a venogram onto a patient’s body.

According to the paper written by researchers at Skoltech in Moscow, this system will help us to make blood draws much easier and less troublesome for patients with difficult access to their veins.

In the paper submitted by those researchers, it has been cited that out of approximately 20 million blood tests performed worldwide every day, almost 45 percent of them are estimated to involve some discomfort for the patient whose veins cannot be accessed easily because of medical conditions such as diabetes, young age or other individual characteristics of the human body. In those cases, where veins are hardly visible and noticeable, even the most experienced medical professionals have to turn towards technical aids otherwise, there is a risk of multiple or inaccurate punctures, which can even have bad consequences on one's health, especially for old age people.

To tackle this problem, Dmitry Dylov – an associate professor at the Center for Computational and Data-Intensive Science and Engineering at Skoltech with the help of his colleagues assembled a near-infrared vein scanner, which can determine vein outlines in an arm or leg rather accurately than a normal human does, automatically, and independently without any help from the user. They did so by using artificial neural networks and reinforcement that will help them to analyze images more effectively and project them back on the patient's body as a visual aid.

He explained that “Infrared vein scanners have become common in clinical practice. However, this is the first of its kind that does everything by entirely depending on modern AI: one neural network will clean and process the infrared signal whereas the second one will detect outlines of the veins, and the third one will do the alignment to assure that the outlines must be projected to the patient’s arm overlapping the actual veins.

Remarkably, all we had to do was to tell the system what will be good and bad during the training period, and the neural networks managed to grasp the rest on their own, consistently finding optimal settings for new patients, even if the system has never experienced them before.

Vito Leli, a lead author of the published work, found out that many factors are obstructing the detection of the veins even in the infrared range where the vein contrast is supposed to be better. They assembled a prototype device which they later tested in experiments with those who volunteered, showing that the instrument was able to detect the pattern of the veins in the near-infrared spectrum and later project it back onto people’s arms as an image.

The patient will have to put their arm under the device, which will immediately search for veins and then project them over the forearm of the patient. If the patient moves the forearm, although it's not recommended during any test that involves needles, nothing to worry about here as the system will immediately start the realigning procedure and that's it now the blood can be drawn easily without any fuss.