Single‐cell Raman spectrum extraction from clinic biosamples
Xinxin Han, Yihui Wu, Mingbo Chi, Sujun Gao, Quan Wang
Abstract
Abstract Raman spectra of clinical samples are often affected by both the substrate and background fluorescence. The overlay of the substrate spectrum, biological Raman spectrum, and the fluorescence spectrum significantly affects the identification of biometrics. In this paper, we propose a specific‐scale analysis algorithm and a zero‐order Savitzky–Golay filtering algorithm combining local minima to separate the substrate and background fluorescent, respectively. The specific‐scale analysis algorithm based on wavelet transform can realize the linear separation of the substrate spectrum through multiresolution analysis. The zero‐order Savitzky–Golay filter algorithm combining local minima is an empirical algorithm. It estimates the background fluorescence by building a smooth curve that passes through local minima. We tested our algorithms with simulated spectra and with the Raman spectra of clinical biosamples recorded on glass. The analysis results of three gastric cancer cell lines indicated that the classification error of the processed spectrum decreased significantly.