/* ---- Google Analytics Code Below */
Showing posts with label Veins. Show all posts
Showing posts with label Veins. Show all posts

Wednesday, April 07, 2021

Finding Veins Using AI

Having been on the receiving end of this, like the claim.  

Skoltech Scientists Use ML to Help Doctors Find Veins for No-Fuss Blood Draws

Skolkovo Institute of Science and Technology (Russia)

March 29, 2021

A prototype medical imager developed by scientists at Russia's Skolkovo Institute of Science and Technology (Skoltech) employs machine learning to analyze near-infrared images of veins and project their patterns onto a patient's body, to make the drawing of blood samples easier. Skoltech’s Dmitry Dylov said, "This is the first [infrared vein scanner] that does everything entirely by virtue of modern AI [artificial intelligence]: one neural network cleans and processes the infrared signal, the second one detects contours of the veins, and the third one continuously 'worries' about alignment to assure the contours projected to a patient's arm overlap with the actual veins."  ... ' 

Wednesday, February 17, 2021

More Biometrics for Identification

Handprints are a much older biometric ID capability, and iris scanning is also well known.  We tested the methods in our innovation centers. 

MIT News,Rachel Gordon,February 8, 2021

AI Can Use the Veins on Your Hand Like Fingerprints to Identify You  By New Scientist

February 17, 2021

Researchers at the University of New South Wales have developed a technique that identifies individuals using the unique pattern of veins on the back of their hands.

The researchers used 500 photos of the hands of 35 people to train a neural network to connect the pattern of veins to a particular subject. The model identified the test subjects with an accuracy rate of 99.8%, then identified four new subjects not included in the original dataset with a 96% accuracy rate.

Researcher Syed Shah at the university said vein detection is reliable for people of all ethnicities and is less vulnerable to attacks than existing biometric tests using fingerprints or face recognition. The technique potentially could be adapted for use with smartphones and CCTV cameras, Shah said. 

From New Scientist  ..