Wavelet-Based Adaptive Denoising and Baseline Correction for MALDI TOF MS

Abstract

Proteomic profiling by MALDI TOF mass spectrometry (MS) is an effective method for identifying biomarkers from human serum/plasma, but the process is complicated by the presence of noise in the spectra. In MALDI TOF MS, the major noise source is chemical noise, which is defined as the interference from matrix material and its clusters. Because chemical noise is nonstationary and nonwhite, wavelet-based denoising is more effective than conventional noise reduction schemes based on Fourier analysis. However, current wavelet-based denoising methods for mass spectrometry do not fully consider the characteristics of chemical noise. In this article, we propose new wavelet-based high-frequency noise reduction and baseline correction methods that were designed based on the discrete stationary wavelet transform. The high-frequency noise reduction algorithm adaptively estimates the time-varying threshold for each frequency subband from multiple realizations of chemical noise and removes noise from mass spectra of samples using the estimated thresholds. The baseline correction algorithm computes the monotonically decreasing baseline in the highest approximation of the wavelet domain. The experimental results demonstrate that our algorithms effectively remove artifacts in mass spectra that are due to chemical noise while preserving informative features as compared to commonly used denoising methods.