Representing NIR spectroscopy at SciX 2023

From 8 to 12 October 2023, SciX Conference was successfully held at Nugget Casino Resort in Sparks, Nevada, USA, located near the wonderful site of Lake Tahoe (Figure 1).

Figure 1.

Nugget Casino Resort (Sparks, NV, USA) and Lake Tahoe.

On Wednesday, 11 October, in the morning, an interesting and stimulating session dedicated to NIR Spectroscopy Applications was held, with Christian Huck as a chair and Mika Ishigaki (Shimane University, Japan) as a co-chair. The following five internationally known experts in the field had been invited to present their ongoing research (Figure 2).

Figure 2.

Krzysztof Bec, Marina De Gea Neves, Bayden Wood, Justyna Grabska, Akifumi Ikehata, Heinz Siesler, and Christian Huck.

(SPSJ-04.1) Handheld NIR spectroscopy: A non-destructive, rapid and informative technique for quality control in the materials- and life-sciences

Heinz Siesler, Marina De Gea Neves, Hui Yan, Department of Physical Chemistry, University Duisburg-Essen, Jiangsu University of Science and Technology

Abstract: The presentation considers the rapid development of miniaturized handheld NIR spectrometers over the last decade and provides an overview of current instrumental developments and exemplary applications in the fields of material and food control as well as environmentally relevant investigations. Care is taken, however, not to fall into the exaggerated and sometimes unrealistic narrative of some direct-to-consumer companies, which has raised unrealistic expectations with full-bodied promises, but has harmed the very valuable technology of NIR spectroscopy, rather than promoting its further development. Special attention will also be paid to possible applications that will allow a user group that is not necessarily scientifically trained to solve quality control and authentication problems with this technology in everyday life.

(SPSJ-04.2) Near-infrared spectral pattern classification of glycolytic reactions using oscillating reactions of yeast extracts

Akifumi Ikehata, Miho Sesumi, Food Research Institute

Abstract: To promote bio-applications of NIR spectroscopy, we need to understand the spectral patterns presented by multivariate analyses, such as PLS regression. These patterns can be inferred to originate not only from the objective compound but also from related metabolites; however, this is still unclear. In this study, we will focus on the glycolytic reaction of yeast (Saccharomyces cerevisiae), an experimental microorganism that has been elucidated from the genetic level. Although the yeast extract is not alive, the autocatalytic action of enzymes sustains periodic reactions of carbohydrate metabolism for several hours. The oscillatory reaction with a cycle of 13 min can be monitored by the UV absorbance of NADH at 340 nm. UV and NIR absorption spectra of the same extractant in a cuvette were alternately measured at 90 s intervals for 2 h. From the NIR spectra, we were able to construct a PLS regression to model the UV absorbance of NADH. In other words, the NIR spectra could capture the oscillating reactions of glycolytic metabolism. The problem is to understand the shape of the regression vectors and loadings of the latent variables. Since the PLS regression models only a periodic glycolytic reaction, background metabolisms are automatically excluded. 1H-NMR results performed separately clearly showed periodic time variability of individual metabolites. The results support the phase differences between metabolites predicted in the literature. The PLS regression model was interpreted by comparing it with metabolite kinetics obtained from the 1H-NMR spectra.

(SPSJ-04.3) Hyperspectral image data analytics with deep learning fusion-nets

Bosoon Park, Taesung Shin, U.S. Department of Agriculture, Agricultural Research Service, USDA, ARS

Abstract: Foodborne pathogens cause a serious public health issue every year worldwide. Although various techniques have been used for assessment of bacteria viability, none of them was successful for implementing to identify live/dead bacteria rapidly without incubation process. Recently, hyperspectral microscope imaging (HMI) technology demonstrated the potential for a rapid label-free detection of foodborne bacteria. The HMI with deep learning methods accurately distinguished between live and dead foodborne bacteria. In this study, three deep learning models called Fusion-Net I, II, and III were developed to classify pathogenic bacterial cells of several foodborne bacteria including Escherichia coli, Listeria, Staphylococcus, and Salmonella using their morphological and spectral characteristics of live and dead bacterial cells. Three Fusion-Net models with inputs of spectra and morphological features from 546 nm band images of the cells were employed to classify live/dead bacteria with over 93% accuracy, suggesting that live foodborne bacteria could be accurately identified by HMI with machine learning algorithm effectively prior to causing foodborne outbreaks.

(SPSJ-04.4) Unleashing the potential: Overcoming hurdles to make vibrational spectroscopy a routine diagnostic tool

Bayden Wood, John Adegoke, Karin Jandeleit-Dahm, Diana Bedolla, Phil Heraud, Adele Kinces, Keith Dias, Monash University

Abstract: The potential of vibrational spectroscopy for disease diagnosis has yet to be realized as a routine diagnostic tool due to several reasons. First, the complexity of spectroscopic techniques, such as infrared and Raman spectroscopy, requires specialized expertise and instrumentation, limiting their use in routine clinical settings. Second, the lack of standardized protocols and large-scale validation studies across different diseases and populations has hindered their adoption. Additionally, biological sample variability and the presence of interfering substances pose challenges to accurate measurements. Technological advancements have improved the field, but equipment costs remain high. Finally, integrating vibrational spectroscopy into existing clinical workflows requires restructuring and optimization of healthcare systems. To overcome these challenges, efforts must focus on simplifying the techniques, establishing standardized protocols, accounting for sample variability, reducing costs, and streamlining integration with clinical workflows. The talk will focus on these challenges and how they are being met by the Monash Biospectroscopy Group. The group includes spectroscopists, clinicians, and industry stakeholders whose aim is to realize the potential of spectroscopy as a routine diagnostic tool. In this context, we have extensively investigated biofluids targeting diseases including malaria, HCV, HBV, SARS-CoV-2, and kidney disease. The talk will focus on the spectroscopic approaches to diagnosing these diseases using mid-IR, Near-IR, UV/Vis spectroscopy and multimodal spectroscopy, highlighting the pros and cons of the different modalities.

(SPSJ-04.5) Miniaturized NIR in natural products and food analysis – From mechanistic understanding to framework optimization

Justyna Grabska, Krzysztof Bec, Christian Huck, University of Innsbruck

Abstract: Near-infrared (NIR) spectroscopy has revolutionized analytical chemistry by providing rapid, non-destructive, and cost-effective analysis. NIR spectroscopy is extensively used in various industries, such as agriculture, food analysis, forensics, security, and manufacturing, as a reliable quality control tool. Recent technological and methodological advancements have led to the development of miniaturized and portable instrumentation, as well as new methods for data analysis and interpretation, ushering in a new era in analytical spectroscopy. This talk will explore the potential of the miniaturized NIR spectrometers in analytical applications, with a specific focus on natural products and food analysis. This talk will also address the challenges associated with the use of different miniaturized NIR spectrometers in analytical spectroscopy. These devices operate based on different technological principles, resulting in differences in their performance, sensitivity, and selectivity. Furthermore, this presentation will delve into the mechanistic understanding of NIR spectroscopy and how this understanding can optimize framework development. Firstly, physical interpretation of NIR bands opens new possibilities of profiling the analytical potential of miniaturized sensors, which often can acquire only a specific fragment of the spectral (and thus compositional) information of the sample. Detailed dissection of chemometric models augmented by the interpreted information yields the possibility to assess in detail the sensitivity and selectivity of a particular instrument against a specific compound or chemical moiety. Finally, interpreted information opens the pathway to the understanding of the impact of spectral noise on the analytical application. “Sample-specific” SNR values differ from the nominal SNR values, especially in the lower NIR wavenumber regions. Differences between instruments manifest themselves clearly, particularly in the case of strongly absorbing samples like water.