Quantitative Single-Cell Analysis Method for Stimulated Raman Scattering Images

Label-free imaging such as stimulated Raman scattering (SRS) microscopy provides important chemical information of single cells to study changes in cell response to various stimuli. Single-cell analysis is a valuable tool to obtain quantitative and statistical information from cell images. Single-cell analysis can provide insight regarding cell distribution and individual cell features that are not available with ensemble analysis. This work explains the development of a single-cell analysis methodology that combines machine learning for cell segmentation and quantitative single cell analysis for SRS microscopy. Machine learning was used for cell segmentation and then followed by analysis using CellProfiler. The implementation of deep learning in this procedure made cell segmentation efficient despite the lack of nucleus staining. Using CellProfiler enabled measurement and data collection for a wide variety of cell characteristics that could be adjusted depending on experimental objectives. Both the machine learning and CellProfiler portions of this pipeline are adaptable and can thus be applied to cell images with varying cell sizes, shapes, intensities, concentrations, and so forth. Using this approach, we have systematically studied how different stimuli induce apoptosis and necrosis that can be quantified by label-free SRS microscopy.