Artificial intelligence facilitates empty well detection

6-Aug-2021

DataMatrix barcodes play a key role in tracking and tracing samples. They’re usually laser-etched onto the underside of sample tubes, and the tubes are then stored in racks

Ziath is working with the University of Hertfordshire (UK) on a program of development for the next generation of Ziath barcoded tube scanners due to launch around the end of 2021. Part of the program is the ground-breaking use of artificial intelligence to enable discrimination between empty wells in sample tube racks from wells with a tube present which may have an obscured or poorly rendered barcode.

DataMatrix barcodes play a key role in tracking and tracing samples. They’re usually laser-etched onto the underside of sample tubes, and the tubes are then stored in racks. Tube identification using a barcode reader to scan the rack and decode the barcodes in one go is subject to recognised issues in identifying which location has a tube and which is empty.

Ambient lighting, background image noise, variation in barcode lasering and material quality all contribute to detection difficulties.

Dr Alexander Beasley, from the University of Hertfordshire, is an expert embedded systems design engineer. He has used a convolutional neural network (CNN) technique for feature extraction of images from a Ziath camera-based barcode reader. In this developmental project he has taken the notion of the CNN and applied it specifically to discriminating empty wells from full wells.

Dr Beasley said: “The CNN I have chosen is designed to be very lightweight allowing for quick execution. When compared to the pre-existing heuristic methods, the CNN approach was almost ten times faster to execute with virtually 100% accuracy.”

Sign up for your free email newsletter

Ziath has already implemented the feature in the latest version of its DP5 control software.

Featured Companies