Deep learning-powered high-efficient atomic force microscopy single-cell nanomechanical analysis on diverse biointerfaces

Haodong Huang, Zhihui Zhang, Lianqing Liu, Mi Li,
Deep learning-powered high-efficient atomic force microscopy single-cell nanomechanical analysis on diverse biointerfaces,
Biochemical and Biophysical Research Communications,
Volume 786,
2025,
152761,
ISSN 0006-291X,
https://doi.org/10.1016/j.bbrc.2025.152761.
(https://www.sciencedirect.com/science/article/pii/S0006291X25014779)
Abstract: The extracellular matrix (ECM) is crucial in tuning cellular behavior, and quantifying cellular mechanical changes in response to ECM stimuli can help reveal the underlying physical mechanisms of cell-ECM interactions for a comprehensive understanding of physiological and pathological processes. Particularly, atomic force microscopy (AFM) has become a key tool for single-cell force measurements, but its throughput and automation level still need to be improved for better application. Here, we present a method that combines AFM-based single-cell indentation assay with vision foundation model-enabled image recognition, enabling reliable and laborsaving AFM force measurements of numerous cells on diverse biointerfaces. With the use of the pre-trained deep learning model, cells were accurately recognized in the optical bright-field images in real time, which was leveraged to achieve autonomous high-efficient AFM single-cell indentation assay. The effectiveness of the proposed method was verified on a variety of commonly used substrates, including regular cell culture dishes, hydrogels, microgrooves, and micropillars. The study vividly demonstrates a promising way based on deep learning to enhance the capability of the AFM-based force spectroscopy toolbox for probing cell-ECM interactions, which is significant for the advancement of the field of mechanobiology.
Keywords: Atomic force microscopy; Deep learning image recognition; Extracellular matrix; Hydrogel; Microgroove; Micropillar