Artificial Intelligence Facilitates Empty Well Detection

Published on: 
In the Lab eNewsletter, Pharmaceutical Technology's In the Lab eNewsletter, August 2021, Volume 16, Issue 08

Through a partnership with the University of Hertfordshire, Ziath discovers means of using AI to detect empty wells in sample tube racks.

Ziath, a UK-based instrumentation control and information management company focused on laboratory automation, in conjunction with the University of Hertfordshire, has discovered a means of using artificial intelligence (AI) to enable discrimination between empty wells in sample tube racks from wells that may have an obscured or poorly rendered barcode.

Ziath utilizes DataMatrix barcodes to track biological and compound samples. These barcodes are laser-etched onto the underside of sample tubes, which are then stored in racks. Tube identification is then done via a barcode reader that scans the bottom of the rack and decodes it.

A problem had emerged, however, where traditional barcode readers were sometimes unable to determine the difference between an empty well and those that have hard-to-read labels. This error can happen for various reasons, such as ambient lighting, background image lasering, or simple variation in barcode lasering and material.

Alexander Beasley, PhD, from the University of Hertfordshire, used his expertise in embedded systems design to develop a scanner that circumvents these issues. By utilizing a convolutional neural network (CNN) for feature extraction of images, Beasley has been able to successfully discriminate between empty and full wells.

The new technology is expected to launch in Ziath barcoded tube scanners by the end of 2021. In the meantime, Ziath has also implemented the technology in the latest version of its DP5 control software, offering customers the opportunity to try the new technology immediately.


“The CNN I have chosen is designed to be very lightweight, allowing for quick execution,” said Beasley, in July 8, 2021 press release. “When compared to the pre-existing heuristic methods, the CNN approach was almost ten times faster to execute with virtually 100% accuracy.”

“This is just the first deliverable from our collaboration with Alexander and the University of Hertfordshire team,” said Ziath Managing Director Neil Benn in the press release. “We are expecting this project to revolutionize the way we decode DataMatrix tubes and help us produce the next generation of faster, lighter, go-anywhere tube rack readers. It’s an exciting development that will, very soon, improve sample tracking and tracing for scientists everywhere.”

Source: Ziath