A Fast Fluorescence Background Suppression Method for Raman Spectroscopy Based on Stepwise Spectral Reconstruction

• Published in: IEEE access Vol. 6; pp. 67709 - 67717
• Main Authors: Chen, Shuo, Kong, Lingmin, Xu, Wenbin, Cui, Xiaoyu, Liu, Quan
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 2018; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Biological properties; Biomedical signal processing; Computational efficiency; Computing time; Fitting; Fluorescence; fluorescence suppression; Fourier transforms; Image reconstruction; Imaging; Nondestructive testing; Numerical analysis; Numerical methods; Polynomials; Raman imaging; Raman scattering; Raman spectra; Raman spectroscopy; Reconstruction; Spectrum analysis; Wavelet transforms
• ISSN: 2169-3536; 2169-3536
• DOI: 10.1109/ACCESS.2018.2879160

Raman spectroscopy is a rapid and non-destructive technique for detecting unique spectral fingerprints from biological samples. Raw Raman spectra often come with strong fluorescence background that makes spectral interpretation challenging. Although fluorescence background can be suppressed experimentally, this approach requires sophisticated and costly instruments. For convenience and cost-effectiveness, numerical methods have been used frequently to remove fluorescence background. Unfortunately, many of such methods suffer from long computation time. Therefore, a fast numerical method for fluorescence suppression is highly desirable especially in Raman spectroscopic imaging where Raman measurements from many pixels need to be processed rapidly. In response to this demand, we propose a fast numerical method for fluorescence background suppression based on the strategy of stepwise spectral reconstruction that we previously developed. Compared with traditional computational methods, including polynomial fitting, wavelet transform, Fourier transform, and peak detection, our results consistently show significant advantages in both accuracy and computational efficiency when tested on Raman spectra measured from phantoms and cells as well as surfaced enhanced Raman spectra from blood serum samples. In particular, our method yields clean Raman spectra closest to the reference results generated by polynomial fitting while several orders of magnitude faster than others. Therefore, the proposed fast fluorescence suppression method is promising in Raman spectroscopic imaging or related application in which high-computation efficiency is critical and a calibration dataset is available.