Wroclaw company automates microbiological analysis with artificial intelligence

NeuroSYS, a Wroclaw-based IT company, is implementing an innovative project related to the use of artificial intelligence, or more specifically Deep Learning, to automate the detection and identification of bacterial colonies in industrial laboratories. The solution may find application in the medical, pharmaceutical, food, cosmetic and veterinary industries, among others. The innovative idea of NeuroSYS was recognized by the National Center for Research and Development and received funding under the Fast Track competition.

Today, the industry is using półautomatic, and often even fully manual, microbiological analysis – someóre its stages, e.g. preparation of próbki, labeling or image acquisitionów, can be done with robotic devices. At the same time, the evaluation of the resultsóin takes place in the mainównie manually and requires tedious, repetitive work performed by teams ofó³ expertów. Time and accuracy are crucial for analysis in industrial laboratories, where microbiological testing is a legally required part of the production process. Software, on whichóThe team ofół R&D NeuroSYS, will automate and streamline laboratory processes and contribute to the elimination of potential human errorsów.

The ongoing research has already achieved significant success – compositeół has developed a system thatóD NeuroSYS will automate and streamline laboratory processes and contribute to the elimination of potential human errors. with efficiency to assess whether a petri dish is sterile or whether microorganisms have appeared on it. Despite appearances, this is a difficult task, with whichówhich standard algorithms cannot handle (on the petri dishes of theócz bacteria can be found rótainted dirt and cracks). The system is used in production and has been implemented by one of the five largest corporations in the worldóin pharmaceuticals around the world, where it functions in a machine that automates the operation of a microbiology laboratory. However, the tool created by the teamóR&D NeuroSYS is not reserved exclusively for high-tech companies, it can be used róAlso by means of a camera placed on a tripod.

We jointlyóIn cooperation with experts from the University of Wroclaw, further research is underway, whichóre will allow identification of bacterial strains and precise determination of the number of bacterial colonies. A training database is currently being developed, whichóra will collect a total of 20,000 images próbek – microbiologists at the Institute of Genetics and Microbiology prepare subsequent bacterial culture cultures with detailed descriptions of Petri dish images, and R&D NeuroSYS use them in the process of supervised learning (ang. Supervised Learning).

The work is carried out by an interdisciplinary teamół, in the composition of whichóThe system is composed of specialists and scientists from the field of microbiology and the team ofóR&D NeuroSYS. Their fascinating work is led by Tomasz Golan, a theoretical physicist with many years of experience in developing software for numerical calculations and Monte Carlo simulations for particle physics. He has a postdoctoral fellowship at Fermilab (Fermi National Accelerator Laboratory), one of the most important centers ofóin research around the world. Współ has developed international scientific projects, published in Philadelphia-listed journals and authored lecturesów for studentsóIn the Department of Physics and Astronomy at the University of Wroclaw, including a course in Python programming and a lectureóin monographic “Introduction to Machine Learning”.