Binning approach combined with multivariate curve resolution-alternating least squares for quantitative mass spectrometry imaging

Large and complex data sets generated by modern instruments during the last decade have led to the so-called Big Data era. High resolution mass spectrometry imaging (HRMSI) instrument is one of the most powerful instruments which generates such Big Data and it has a central role in modern analytical chemistry. On this matter, MSI data are difficult to analyze using conventional tools. Chemometric methods have shown potential to fill the gap in the analysis of MSI data especially for quantitative MSI [1]. The aim of this work was quantification of chlordecone as a carcinogenic organochlorinated pesticide (C10Cll0O) in mouse liver using matrix-assisted laser desorption ionization MSI (MALDI-MSI) method [2]. The MSI data sets were corresponded to 1, 5 and 10 days of mouse exposure to the standard of chlordecone in quantity range of 0 to 450 g/g. The raw data size was 70 gigabytes (GB) for one sample. Therefore, binning approach in m/z direction was necessary to group high resolution m/z values and to reduce the data size. To consider the effect of bin size on the quality of the obtained results, three different bin sizes of 0.25, 0.5 and 1.0 were chosen. In this regard, the number of m/z values was reduced from 101205 to 2400, 1200 and 600, respectively. Afterwards, the three-way MSI data arrays (two spatial dimensions and one m/z dimension) for seven standard samples (calibration set) and four unknown samples (test set) were column-wise augmented with m/z values as common mode. Then, these data sets were analyzed using multivariate curve resolution-alternating least squares (MCR-ALS) using proper constraints [3]. Singular value decomposition (SVD) and orthogonal projection approach (OPA) were used to determine the number of components and initial estimates of spectral profiles, respectively. The resolved mass spectrum was used for identification of chlordecone in the presence of complex background and interferences. Additionally, augmented spatial profiles were post-processed to obtain 2D images for each component in different samples. Then, these images were used to set the calibration curve and to obtain analytical figures of merit (AFOM). Finally, the obtained results by MALDI-MSI-MCR (bin size 0.25 as optimum size) were compared with previous results using gas chromatography-mass spectrometry (GC-MS) and MALDI-MSI using msiQuant software. The obtained results using MALDI-MSI-MCR method were higher than MALDI-MSI and lower than GC-MS which confirms the performance improvement of MALDI-MSI using chemometric method. As an instance, the chlordecone quantity for data set II (five days of exposure) was 141.9 g/g by GC-MS, 77.7 g/g by MALDI-MSI-MCR, and 15.0 g/g by MALDI-MSI. All of these results confirmed the potential of chemometrics as an alternative way for quantitative MSI.