A Raman peak recognition method based automated fluorescence subtraction algorithm for retrieval of Raman spectra of highly fluorescent samples

Abstract

Intense fluorescence background is a major problem in the application of Raman spectroscopy. An appropriate algorithm which can faithfully retrieve weak tissue Raman signals is required. In this article, we propose a new algorithm for automated and artifact-free recovery of Raman spectra which combines a novel Raman peak recognition method (RPR method) with an improved iterative smoothing method (SG-SR method). The SG-SR method, based on the modified Savitzky–Golay iterative process, substantially improves its convergence speed. By applying a novel negative relaxation factor to the successive relaxation iterative method, automatic recognition of Raman peaks is realized. In the proposed algorithm (RIA-SG-RPR algorithm), a real Raman peak position is first detected by the RPR method to serve as the intrinsic criterion of convergence for the SG-SR method to avoid human interference. Then, real Raman signals are recovered from the iterative procedure of the SG-SR method. This algorithm has been optimized and validated with mathematically simulated Raman spectra as well as experimentally recorded Raman spectra from various fluorescent samples, resulting in a significant improvement in the rejection of both high fluorescence background and direct human intervention. This algorithm drastically avoids false Raman features to benefit the utilization of Raman spectroscopy to characterize molecular specifics in more challenging Raman applications.