Conditional Generative Adversarial Network for Spectral Recovery to Accelerate Single-Cell Raman Spectroscopic Analysis

Raman spectroscopy is a powerful tool to investigate cellular heterogeneity. However, Raman spectra for single-cell analysis are hindered by a low signal-to-noise ratio (SNR). Here, we demonstrate a simple and reliable spectral recovery conditional generative adversarial network (SRGAN). SRGAN reduced the data acquisition time by 1 order of magnitude (i.e., 30 vs 3 s) by improving the SNR by a factor of similar to 6. We classified five major foodborne bacteria based on single-cell Raman spectra to further evaluate the performance of SRGAN. Spectra processed using SRGAN achieved an identification accuracy of 94.9%, compared to 60.5% using unprocessed Raman spectra. SRGAN can accelerate spectral collection to improve the throughput of Raman spectroscopy and enable real-time monitoring of single living cells.

Automated summary

Raman spectroscopy is a powerful tool for studying cellular heterogeneity, but its application in single-cell analysis is limited by low signal-to-noise ratio. In this study, a simple and reliable spectral recovery conditional generative adversarial network (SRGAN) is developed to improve the SNR and reduce data acquisition time. The performance of SRGAN is tested on the classification of five major foodborne bacteria based on single-cell Raman spectra, achieving a significantly higher identification accuracy compared to unprocessed spectra. SRGAN has the potential to enhance the throughput of Raman spectroscopy and enable real-time monitoring of single living cells.