Automatic Peak Integration and Baseline Correction for Micro-scale Gas Chromatographs Using Continuous Wavelet Transform

Abstract

Chromatograms generated by micro-scale gas chromatographs are complex sensor signals comprised of multiple chemical retention peaks superimposed upon a non-monotonic baseline. Depending on the chemical properties, the peaks may have Gaussian shapes with different widths, heights, skew, and overlap, all of which pose challenges to automated recognition and quantification of chemicals. This paper presents an automatic algorithm based on continuous wavelet transforms for peak integration and baseline correction of chromatograms generated by micro-scale gas chromatographs. This algorithm identifies peaks using a dynamic filter based on the inherent and known relationship between the peak widths and peak locations. The width of each peak is determined from continuous wavelet transform coefficients by locating the pair of local minima that straddle the apex. This approach provides peak detection and width estimation with low false positive rates, even for skewed peaks with tailing or fronting and for non-monotonic baselines.