Deep Learning Image Recognition-Assisted Atomic Force Microscopy for Single-Cell Efficient Mechanics in Co-culture...

Area of Science:

• Mechanobiology
• Biophysics
• Cellular Mechanics
• Deep Learning Applications in Life Sciences

Background:

• Atomic force microscopy (AFM) is crucial for single-cell mechanical property characterization.
• Current AFM methods suffer from low throughput and require manual operation.
• Label-free identification of co-cultured cells is a significant challenge for AFM applications.

Purpose of the Study:

• To develop a deep learning-assisted AFM system for fluorescence-independent cell recognition in co-cultures.
• To enable high-throughput and automated mechanical measurements of identified single cells.
• To facilitate the study of cell-cell interactions and mechanical cues in native cellular environments.

Main Methods:

• Utilized a deep learning-based image recognition model for analyzing bright-field microscopy images of co-cultured cells.
• Integrated image recognition with AFM for automated probe positioning and force measurements.
• Applied AFM indentation assays (Young's modulus) and single-cell force spectroscopy (adhesion forces) on identified cells.

Main Results:

• Successfully identified cell types and viability in co-culture environments using only bright-field images, confirmed by fluorescent labeling.
• Demonstrated automated, precise AFM probe targeting and force measurements based on deep learning recognition.
• Validated the method's applicability for measuring Young's modulus and cell adhesion forces using different AFM probe types.

Conclusions:

• Deep learning-assisted AFM provides a label-free, high-throughput approach for single-cell mechanics under co-culture conditions.
• This technology overcomes limitations of manual operation and enhances the utility of AFM in life sciences.
• The method holds promise for advancing mechanobiology by enabling detailed analysis of cell-cell mechanical interactions.