Development of an AI-Assisted, Image-Based System for the Quantitative Analysis of Microbial Loads (µQuant)
The overarching objective is the research and prototype development of a robotic, image-based analysis system for the automated assessment of various microbial loads. This combined approach encompasses the prototype development of hardware for automated macroscopic imaging (digitization) of samples, AI-based image analysis, and the integration of these components into laboratory testing workflows. The analysis system is designed to be independent of specific sample carrier models and labeling methods (e.g., barcodes, handwriting), flexible regarding image acquisition modes (reflected light/transmitted light), and—in the long term—applicable across diverse laboratory domains, all while reducing analysis times. High accuracy and reliability are critical prerequisites for the successful practical deployment of such an automated system. Consequently, the project specifically employs AI and machine learning (ML) algorithms to detect frequently occurring special cases within samples—such as accompanying flora. Initially, the project focuses on the *Legionella* genus (utilizing reflected-light imaging, various colored culture media, and white colonies). Given the adaptability of AI/ML-based approaches, the analysis framework is designed to be modularized and adapted to accommodate different testing protocols, microbial species, and culture media following the project’s completion.
This project is co-financed by the European Union and the SAB.
