Accurate zone of inhibition measurement for rapid antimicrobial susceptibility testing

Abstract

Antimicrobial Susceptibility Testing (AST) is a critical tool in combating bacterial infections and guiding effective antibiotic treatments. This paper introduces an automated algorithm leveraging YOLOv5 for accurately measuring the diameter of Zones of Inhibition (ZOIs) in AST plates is to address the limitations of manual measurement. The proposed system employs image processing techniques and object detection to identify antibiotics and ZOIs, enabling precise classification can be made as resistant, intermediate, or susceptible based on radius measured using Clinical and Laboratory Standards Institute (CLSI) guidelines. Special cases, such as overlapping and scattered zones, are managed through enhanced Harris-Stephens corner detection methods. The system was evaluated using 300 annotated images of Escherichia coli and Klebsiella pneumonia, achieving high accuracy in ZOI measurement and susceptibility classification. Results demonstrate the algorithm’s potential to enhance the reliability and efficiency of AST, offering a robust solution for clinical decision-making in the fight against antimicrobial resistance.