Band target entropy minimization for retrieving the information of individual components from overlapping chromatographic data

Band target entropy minimization (BTEM) is a self-modeling curve resolution (SMCR) approach relying on non-negative criterion and minimization of Shannon entropy. In this study, BTEM algorithm was applied to retrieving the information of individual components from overlapping gas chromatography-mass spectrometry (GC-MS) data. The algorithm starts with dividing the whole data into bands along the retention time. In each band, singular value decomposition (SVD) is used to decompose the data into scores and loadings. Because the pure chromatographic signal possesses the lowest Shannon entropy, the chromatographic signal of each component can be constructed by optimizing the combination of the loadings with minimal Shannon entropy under non-negative criterion. To show the efficiency of the algorithm, a simulated four-component overlapping GC-MS data and an experimental GC-MS data of 18 organophosphorus pesticide mixture are investigated. The results show that both the chromatographic profiles and mass spectra of the components can be successfully extracted from the overlapping signals.