Data-driven and coarse-to-fine baseline correction for signals of analytical instruments

Abstract

Baseline correction is an indispensable step in the signal processing of chemical analysis instruments. With the increasing demand for on-site applications, a variety of analytical instruments require a more friendly, rapid and adaptive baseline correction method. In this paper, a data-driven and coarse-to-fine (DD-CF) baseline correction scheme mainly based on the empirical mode decomposition (EMD) algorithm is proposed. For eliminating the mode-mixing effect of the original EMD, the proposed method firstly obtains a coarse baseline estimation using automatic peak detection, elimination and interpolation; and the EMD is applied on the coarse baseline to get a fine baseline finally. We have compared this method with the adaptive iteratively reweighted Penalized Least Squares algorithm (airPLS) and the sparse representation baseline correction methods using simulated signals and experimental signals from different analytical instruments. Results indicate that the proposed DD-CF scheme can effectively estimate the baseline more accurate than the comparing methods for varies of analytical signals such as mass spectrometer, ion mobility spectrometer, gas chromatograph, etc. Furthermore, with signals of different length, different peak distributions and even from totally different instruments, the proposed method requires minimal user intervention, in which the parameters of the comparing methods should be adjusted for a wide range.