Abstract
Raman spectroscopy is a powerful non-destructive technique for qualitatively and quantitatively characterizing materials. However, noise often obscures interesting Raman peaks due to the inherently weak Raman signal, especially in biological samples. In this study, we develop a method based on spectral reconstruction to recover Raman spectra with low signal-to-noise ratio (SNR). The synthesis of narrow-band measurements from low-SNR Raman spectra eliminates the effect of noise by integrating the Raman signal along the wavenumber dimension, which is followed by spectral reconstruction based on Wiener estimation to recover the Raman spectrum with high spectral resolution. Non-negative principal components based filters are used in the synthesis to ensure that most variance contained in the original Raman measurements are retained. A total of 25 agar phantoms and 20 bacteria samples were measured and data were used to validate our method. Four commonly used de-noising methods in Raman spectroscopy, i.e. Savitzky-Golay (SG) algorithm, finite impulse response (FIR) filtration, wavelet transform and factor analysis, were also evaluated on the same set of data in addition to the proposed method for comparison. The proposed method showed the superior accuracy in the recovery of Raman spectra from measurements with extremely low SNR, compared with the four commonly used de-noising methods.
Highlights
Raman spectroscopy is a laser-based spectroscopic technique that exploits Raman scattering for qualitative or quantitative biological material characterization [1]
It should be noted that these are raw spectra without going through background removal. For both phantoms and bacteria samples, the Raman spectra with low signal-to-noise ratio (SNR) are much noisier than reference Raman spectra and Raman peaks are overwhelmed by noise
25 phantoms and 20 bacteria samples were used and the mean relative root mean square error (RMSE) was used to evaluate the overall performance of those techniques
Summary
Raman spectroscopy is a laser-based spectroscopic technique that exploits Raman scattering for qualitative or quantitative biological material characterization [1]. Raman spectra or peaks inside could be employed to differentiate biological components This method has shown great potential in many biomedical applications [3, 4]. Such applications are often hampered by inherently weak Raman signals from biological molecules [5]. In this case, measurement noises obscure Raman peaks of interest rendering a low signal-to-noise ratio (SNR) [6]. It is common to solve this problem by increasing the power of the excitation laser and/or exposure time These methods cannot be used when measuring unstable materials or observing fast changing phenomena. It is important to develop a method to quickly recover Raman spectra with low SNR without increasing laser power
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
