Rapid identification of Lactobacillus species using near infrared spectral features of bacterial colonies

Abstract

The feasibility of rapid identification of Lactobacillus species using near-infrared spectral features coupled with chemometrics was investigated. First, bacterial colonies of 11 Lactobacillus strains covering four species ( Lactobacillus casei, Lactobacillus reuteri, Lactobacillus brevis, and Lactobacillus fermentum) were cultured using the spread-plate technique. Near-infrared spectra data of the Lactobacillus species were collected directly from the bacterial colonies. Second, 10 wavenumbers were selected from the near-infrared spectra data using uninformative variables elimination and genetic algorithm, and calibration models based on the 10 selected wavenumbers were built using least squares support vector machine. The identification rates for the prediction set and validation set were 89.04 and 85%, respectively. Third, chemical groups of the Lactobacillus cells contributing to the identification of the Lactobacillus strains were identified using mid infrared. The results of mid infrared data analysis indicated that 9 chemical groups could be considered characteristics for categorizing the 11 Lactobacillus strains. The relationship between the 10 selected wavenumbers and identified chemical groups was identified, which supported the satisfactory performance of the least squares support vector machine calibration model. This study demonstrated that near-infrared spectral features of bacterial colonies could be used for Lactobacillus typing at the strain level.

Identification of lactic acid bacteria and rhizobacteria by ultraviolet-visible-near infrared spectroscopy and multivariate classification

“…Overall classification performances from the best models were lower than in previous works on this subject. [14][15][16][17] However, the models of this study were built with significantly more samples and a wider range of bacterial genus and species, leading to more robustness. These results could be further improved by supplementing the model database and prediction set with more strains per genus and species, Table 8.…”

“…Because each spectrometer has an acquisition area of different size, the advantage of confluent growth is to maximize the space occupation of the plate by bacteria to ensure a maximum of bacteriarelated information on acquisitions. 90 mm Petri dishes were prepared with 20 mL of Bennett agar deposited with a motorized pipette (15,249,805 Thermo Fisher Scientific, France). The 226 strains from the lactic flora were revivified from cryotubes by inoculation into Bennett broth and incubation overnight at 30 C. Immediately after incubation, an easySpiral Pro automatic plater (Interscience, Netherlands) was used to deposit 200 mL of each preculture on three different plates in constantdeposition mode.…”

“…After using multiplicative scatter correction (MSC) to correct the spectra, 87% of the strains were correctly classified at the species level by decision-tree prediction. Shi et al 15 grew four strains of Lactobacillus species on MRS agar and preprocessed the acquired spectra by MSC and Savitzky-Golay smoothing, achieving a correct classification rate (CCR) of 85% at the species level by using a support vector machine (SVM) model.…”

Visual detection of microbial community during three bacteria mixed fermentation through hyperspectral imaging technology

“…Hyperspectral imaging technique, as rapid and nondestructive method integrates computer vision and spectral analysis techniques to acquire spectral information of each pixel location in a two-dimensional (2D) objective image, generating a multi-dimensional data (one-dimensional spectral dimension and two-dimensional spatial information) [15,16]. Each strain colony with unique fingerprints spectrum has been identified successfully by taking full advantage of one-dimensional spectral information [17][18][19]. Although hyperspectral imaging technology is very useful for strain identification during mixed fermentation, there are still exist a major issue: rapid and automated enumeration of multiple strains.…”

Identification of bacteria by poly-aromatic hydrocarbon biosensors

“…[20] An additional spectroscopic method that relies on different quantities of bacterial ingredients is near-infrared spectroscopy. [21][22][23][24][25] A major advantage of this method derives from its ability to penetrate glass, which allows to easily measure bacteria in their cultivation vessel without damaging its sterilization. An additional noninvasive method is Terahertz spectroscopy.…”