An improved peak detection algorithm in mass spectra combining wavelet transform and image segmentation

Peak detection is a crucial step in the analysis of mass spectrometry data. However, the measured spectrum inevitably contains random noise and altering baseline, which directly impact peak detection. Although many methods were developed to deal with these issues, how to identify weak peaks and overlapped peaks while reducing false peaks is still a challenge. In this study, an improved peak detection algorithm combing continuous wavelet transform and image segmentation was proposed. The ridges in wavelet space that correspond to peaks positions were completely identified by a new searching method named stair scanning. And false ridges outside peaks regions were removed, which are segmented from wavelet space by an image threshold segmentation method. The peaks are recognized by the information of final ridges, valleys and original spectrum. This method was applied to the peak detection of simulated spectra and real spectra. The results show that the proposed method has better performance on peak detection than previous methods. Specifically, the proposed method makes progress on detection of weak and overlapped peaks as well as removal of false peaks. Besides, this method shows good reliability and practicability on processing of real spectra. (c) 2021 Elsevier B.V. All rights reserved.

Automated summary

Peak detection is a crucial step in mass spectrometry data analysis. An improved algorithm combining continuous wavelet transform and image segmentation has been proposed, showing better performance in identifying weak and overlapped peaks while reducing false peaks. The method has been applied to simulated and real spectra, demonstrating good reliability and practicability.