Enhanced spectral resolution and reduced acquisition time in fiber-based wavelength-swept source Raman spectroscopy.

Abstract

Significance: We introduce a fast Raman spectroscopy (SSRS) system that reduces acquisition time and enhances data quality, providing a breakthrough in SSRS for real-time applications. We demonstrate its utility in differentiating brain tissue regions based on lipid and protein content.

Aim: Our primary goal was to develop a fast SSRS system that enables rapid data acquisition for in vivo applications. We aimed to investigate its effectiveness in differentiating brain tissue types by analyzing lipid and protein content, ultimately enhancing classification accuracy and supporting advancements in medical diagnostics.

Approach: We implemented an optimized circuit and signal processing technique to reduce high-frequency noise and improve signal-to-noise ratio. Brain tissue measurements were validated against staining models, and classification accuracy was tested with principal component analysis (PCA) and support vector machine (SVM).

Results: Our SSRS system captures spectra in 1 s which is significantly faster than similar systems. This rapid method enables real-time monitoring and accurate classification of brain regions based on lipid-protein content, confirmed by neurofilament and Nissl staining correlations ( $R^2=0.75$ and 0.55, respectively). Tissue classification showed 80.20% accuracy using spectral intensity at the wavenumbers associated with C-H, ${\mathrm{CH}}_3$ , and ${\mathrm{CH}}_2$ vibrations and 81.23% accuracy using PCA-derived features (PC1, PC2, and PC3).

Conclusions: The fast-SSRS system marks a significant advance in Raman spectroscopy, improving speed and data quality. Our setup captures finer spectral details, facilitating reliable differentiation of tissue types, as verified by staining methods and PCA. This method shows promise for real-time tissue analysis and medical diagnostics, outperforming traditional Raman techniques in speed and data throughput.