Effects of pre‐processing of Raman spectra on in vivo classification of nutrient status of microalgal cells

Abstract

Raman spectra were obtained from cells of the chlorophyte unicellular eukaryotic alga Dunaliella tertiolecta, which had been grown either under nutrient-replete conditions or starved of nitrogen for 4 days. Spectra were rich in bands which could all be attributed to either chlorophyll a or bcarotene. A cursory examination of the differences between the spectra of replete and starved cells indicated a decline in chlorophyll a and an increase in b-carotene in chlorophytes. Unprocessed spectra showed pronounced baseline effects. A variety of pre-processing techniques were used in an attempt to visualise the spectral, and hence chemical, differences in the transformed data and perform classification based upon these differences. Six types of spectral pre-processing were compared: baseline correction with vector normalisation; Multiplicative Scatter Correction (MSC), Extended Multiplicative Signal Correction (EMSC); Standard Normal Variate (SNV); and vector normalised 1st and 2nd derivative spectra. Results for Soft Independent Modelling of Class Analogy (SIMCA) and Partial Least Squares (PLS) discriminant analysis were compared. All pre-processing methods allowed spectral differences between N-replete and N-starved spectra to be visualised, with derivatives and EMSC scoring the lowest RMSEC and RMSEV values with PLS and also the best overall classification results. SIMCA was not suited to classifying the nutrient classes under any of the pre-treatments, due to the small model distances involved.

Micro‐Raman spectroscopy of algae: Composition analysis and fluorescence background behavior

“…Most spectra collected contained a pronounced background, which is likely caused by fluorescence of the pigments in algal cells. Since the spectral backgrounds do not contain any chemical-specific information, they were usually omitted in previous studies by baseline correction using different mathematical procedures, for example, Heraud et al (2006). To separate the fluorescence background, the raw spectra obtained were subjected to a RCF, which is a high-pass filter that can provide efficient background subtraction without introducing significant distortion to the high frequency components (Brandt et al, 2006).…”

Review of multidimensional data processing approaches for Raman and infrared spectroscopy

“…The differentiating Raman features are enhanced by preprocessing the data, which reduces the variability among the spectra from the same group (SNR) of the data used in the training phase (PC modeling phase). In spite of this, SIMCA has been applied successfully to spectroscopic data to solve many classification problems [72][73][74][75].…”

“…Otherwise, FCMCA is very similar to KMCA except that these membership values have to be included in the objective function. In vibrational spectroscopy FCMCA is often used for solving the soft clustering problem [73,84,87].…”

Spectral pre-processing for biomedical vibrational spectroscopy and microspectroscopic imaging

“…In vibrational spectroscopy this is useful to compensate for the differences in sample quantity or a different optical pathlength. In data mining tasks involving spectral distance measurements normalization has been shown to improve the accuracy and efficiency of the models [1][2][3]. 3.…”

“…Pre-processing has been shown to be of crucial importance for subsequent data mining tasks. In fact, it is now widely recognized that quantitative and classification models developed on the basis of pre-processed data generally perform better than models that solely use raw data [1][2][3]. With this review it is intended to explore the concepts and techniques of pre-processing methods and to discuss the applicability of distinct pre-processing techniques in the field of biomedical IR and Raman spectroscopy.…”