An effective two-stage spectral library search approach based on lifting wavelet decomposition for complicated mass spectra

Abstract

For rapid and accurate matching of complicated spectra in a standard spectral library with different instruments in various environments, a two-stage spectral library search approach that involves preliminary and main searches based on lifting wavelet decomposition was proposed. In the preliminary search stage, similar spectra were effectively extracted and the number of reference spectra was greatly reduced by using the low frequency component (outlines) of lifting wavelet decomposition to calculate spectral similarity. In the main search stage, search accuracy was improved by using the high-frequency component (details) of lifting wavelet decomposition to calculate spectral similarity. In addition, a self-adaptive method used to confirm the level of lifting wavelet decomposition according to the resolution (half-peak width) of the mass spectrometer was also presented. More than 100 sets of mass spectra as well as measured data of octafluoronaphthalene (OFN) samples analyzed by our instrument were used to compare the proposed method with common methods such as the weighted dot product similarity (WC), weighted dot product composite similarity (WRstC), and wavelet transform (WT) methods in terms of similarity, accuracy, and search time. Relative to the WRstC method, the average improvement of similarity for the proposed method was 15.1%. The results indicate that the proposed method improves the similarity and accuracy of spectral recognition for complicated mass spectra, particularly under conditions of poor spectral quality, owing to the low signal-to-noise ratio at low sample concentrations.