Microelectrode-array-based sensing of bacterial biofilm antibiotic susceptibility using impedance spectroscopy and convolutional neural networks

• Published in: Biomedical Circuits and Systems Conference pp. 1 - 5
• Main Authors: Haeverbeke, Maxime Van, Cums, Charlotte, Vackier, Thijs, Braeken, Dries, Stock, Michiel, Steenackers, Hans, Baets, Bernard De
• Format: Conference Proceeding
• Language: English
• Published: IEEE 24.10.2024
• Subjects: Antibiotics; antimicrobial susceptibility; Bacterial biofilm; Convolutional neural networks; electrochemical impedance spectroscopy; equivalent electrical circuit; Impedance; Impedance measurement; Integrated circuit modeling; machine learning; Microelectrodes; Point of care; Precision medicine; Sensors; Testing
• ISSN: 2766-4465
• DOI: 10.1109/BioCAS61083.2024.10798403

In light of the worldwide emergence of antibioticresistant bacteria, appropriate and targeted approaches to antibiotic treatments are crucial. Such targeted or precision medicine approaches rely on precisely and quickly determining antibiotic susceptibility. The current routinely applied antibiotic susceptibility testing assays operate within a timeline of one to two days, hindering timely treatment decisions. Furthermore, point-of-care testing would further improve the efficiency of the tests. Recent research demonstrates that electrochemical impedance spectroscopy and related techniques are promising solutions because they can be developed as miniaturised point-of-care devices and high-throughput installations in healthcare facilities. This paper proposes a label-free impedimetric sensing approach to antibiotic susceptibility testing. The proposed novel method uses unmodified microelectrode arrays and convolutional neural network models leveraging impedance spectroscopy measurements and their derived equivalent electrical circuit features to accurately determine bacterial biofilms' antibiotic susceptibility.