Baseline correction of high resolution spectral profile data based on exponential smoothing

• Apply exponential smoothing to baseline correction of spectral profile firstly
• Invoke exponential smoothing iteratively in ATEB to fit background progressively
• ATEB is an automatic algorithm that hardly needn't user intervention and prior information.
• Obviously predominance in speed and accuracy of processing large spectral data

Extraction of qualitative and quantitative information from large amounts of analytical signals is difficult with drifted baselines, especially in multivariate analysis. Baseline drift obscures, “fuzzy” signals, and even deteriorates analytical results. In order to obtain accurate and clear results, some effective methods should be proposed and implemented to perform baseline correction before further data analysis. However, most of the classic methods require user's intervention or are prone to variability, especially with low signal-to-noise signals in large data. In this study, a novel baseline correction algorithm based on two-side exponential smoothing algorithm and iterative fitting strategy is proposed. In addition, the iteratively smoothing strategies were creatively implemented in progressively smoothing the residuals between fitted baseline and original signals. This method, named Automatic Two-side Exponential Baseline correction algorithm (ATEB), does hardly require user intervention and prior information, such as peak detection. It's worth noting that the innovative ATEB algorithm has some obvious advantages, especially, when it comes to the processing speed and corrected accuracy of high resolution spectral data with large scale dataset. After a series of benchmarks with high resolution spectral datasets and comparisons with several other popular methods, using various kinds of analytical signals (including hepatocellular carcinoma, MALDI-TOF mass spectrometry, coronary heart disease serum, NMR spectrum and GC–TOF-MS data), the proposed method is found to be accurate, fast, flexible and easy to use on real datasets.