Time-Correlated Raman Spectroscopy (TCRS): Enhancing Single-Cell Bacterial Detection through Temporal Variation Analysis

Rapid detection of bacteria is crucial for mitigating the risks associated with bacterial contamination. Raman spectroscopy has emerged as a promising technique for single-cell bacterial identification. However, the Raman signals of single bacterium exhibit significant temporal fluctuations during detection, compromising the stability of spectral features over long integration periods. To tackle these challenges, we propose a novel time-correlated Raman spectroscopy (TCRS) technique that targets temporal variation rate as a key discriminative feature. This approach extracts the temporal evolution patterns of bacteria under laser optical tweezers stimulation directly from the raw Raman signals. By leveraging continuous spectral acquisition coupled with low-rank constrained temporal processing, our method achieves a 5-fold enhancement in signal-to-noise ratio. A two-dimensional convolutional neural network is specifically designed to analyze temporal-spectral images, facilitating automated extraction of both spatial-temporal features and compositional signatures. This method enables real-time detection of single-cell bacteria in approximately ten seconds with an accuracy of 97.7%, offering significant potential for rapid, on-site microbial monitoring in clinical and environmental applications.