Vibrational spectroscopy combined with chemometric approaches in bacterial studies: a review of recent advances and applications

Abstract

Bacterial infections remain among the leading causes of global mortality and represent significant challenges to public health. Conventional methods such as cultivation procedures, polymerase chain reaction, and instrumental techniques are routinely used in microbiology laboratories. However, these methods are time-consuming and labor-intensive, and require multi-step sample preparation. Moreover, there is a risk of sample contamination. Recently, vibrational spectroscopic techniques, including near-infrared, mid-infrared, and Raman spectroscopy, have gained considerable attention in microbiology research due to their high-throughput evaluation, non-destructive nature, rapid analysis, cost-effectiveness, and simplicity of application without the need for complex sample preparation steps. In combination with vibrational spectroscopy, chemometric analyses are employed to reduce data dimensionality, extract information related to the molecular structure of biological macromolecules, and eliminate irrelevant details. The present review discusses the applications of vibrational spectroscopy methods combined with chemometric approaches in various microbiological studies, including microbial viability assessment, bacterial inactivation evaluation, bacterial species identification, biofilm formation detection, and antibiotic resistance determination. Near-infrared, mid-infrared, and Raman spectroscopy techniques provide valuable information for clinical researchers and microbiologists within a short time by detecting key bacterial components such as proteins, nucleic acids, lipids, and polysaccharides across different spectral ranges. Despite their advantages, supplementary methods are still needed to enhance their precision. In the future, these techniques are expected to advance further to meet the needs of microbiological analyses more efficiently.

---

Maryam Kazemi

Maryam Kazemi is a Ph.D. candidate in the Department of Pharmaceutics and Pharmaceutical Nanotechnology at Shahid Beheshti University of Medical Sciences, Tehran, Iran. She holds B.Sc. in Pure Chemistry from Shahid Chamran University, Ahvaz, Iran, and M.Sc. in Polymer Chemistry from the Institute for Advanced Studies in Basic Sciences, Zanjan, Iran. Her research interests lie in the development of nanoformulations for targeted antimicrobial delivery, characterization of bacterial permeation barriers including biofilms and planktonic cells, and bacterial resistance analysis.

Arash Mahboubi

Arash Mahboubi is a Professor in the Department of Pharmaceutics and Pharmaceutical Nanotechnology at Shahid Beheshti University of Medical Sciences, Tehran, Iran. He earned his Doctor of Pharmacy (Pharm.D.) from Islamic Azad University, Tehran, Iran, and Ph.D. in Pharmaceutics from the Faculty of Pharmacy at Tehran University of Medical Sciences. His research interests are focused particularly on the development of nano-enabled diagnostic platforms for targeted microbial detection, characterization of bacterial microenvironments including biofilm and planktonic structures, and comprehensive analysis of antimicrobial resistance mechanisms.

Reza Jahani

Reza Jahani is an Assistant Professor in the Department of Toxicology and Pharmacology at the School of Pharmacy, Shahid Beheshti University of Medical Sciences, Tehran, Iran. He earned a Doctor of Pharmacy (Pharm.D.) and a Ph.D. in Toxicology, both from Shahid Beheshti University of Medical Sciences. He further pursued a research fellowship at Wageningen University & Research in the Netherlands, focusing on the application of vibrational spectroscopy and chemometrics. His expertise lies in these techniques, including NIR, FTIR, and Raman spectroscopy, with a strong interest in applying them to practical challenges in food science and toxicological evaluation.

Hamid Reza Moghimi

Hamid Reza Moghimi is a Professor of Pharmaceutics in the Department of Pharmaceutics and Pharmaceutical Nanotechnology at the School of Pharmacy, Shahid Beheshti University of Medical Sciences, Tehran, Iran. He earned his Pharm.D. from Tabriz University of Medical Sciences (Tabriz, Iran), and Ph.D. in Pharmaceutics from the University of Bradford, UK. He has completed two fellowships at the University of South Australia, Australia (transdermal drug delivery) and the University of Alberta, Canada (liposomal gene delivery). His research focuses on modeling and characterization of biological barriers including bacteria, lipid nanoparticles, gene delivery and drug permeation mechanisms and enhancement including overcoming bacterial resistance.

---

Introduction

Bacteria are highly adaptable and widespread organisms that inhabit diverse environments, including clinical, environmental, and host-associated settings.¹ While many species play beneficial roles, pathogenic bacteria remain major contributors to the global infectious disease burden, accounting for an estimated 7.7 million deaths per year.² The escalating emergence of antibiotic resistance to commonly used antimicrobials has become a critical international health threat, severely undermining the effectiveness of conventional treatments and contributing to the increasing prevalence of untreatable infections.^3,4

Bacterial typing is essential for distinguishing between strains and identifying pathogenic species. Although traditional culture-based methods are widely used, they are limited by lengthy procedures, contamination risks, and dependence on phenotypic traits.⁵ Molecular techniques such as polymerase chain reaction (PCR) and microarrays offer faster detection and improved sensitivity but are destructive and often insufficient for fully discriminating between bacterial species.⁶ Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) has gained traction due to its high-throughput, protein-based bacterial identification capability. However, its application is hindered by high costs, the need for sample preparation, and the absence of specific protein markers in certain strains.⁷

Assessing microbial viability is crucial for effective infection treatment, particularly for understanding how microorganisms respond to antibiotics. Although classical culture-based techniques and advanced methods such as PCR or next-generation sequencing (NGS) are commonly used, each has its own limitations.^8,9 PCR-based approaches, while rapid, often suffer from low sensitivity in samples with low microbial abundance and may produce false positives due to environmental contamination. Moreover, since PCR detects DNA from both live and dead bacteria, it does not reliably confirm microbial viability, making the interpretation of results challenging in clinical settings.¹⁰

Bacterial biofilms, which are structured microbial communities embedded within a protective extracellular polymeric substance (EPS), complicate matters further. Responsible for approximately 80% of human infections, biofilms exhibit altered gene expression and increased resistance to antimicrobial agents compared to planktonic cells.¹¹ Standard methods for assessing biofilms, such as crystal violet staining, confocal laser scanning microscopy (CLSM), and scanning electron microscopy (SEM), are informative but are limited by labor-intensive procedures, extensive sample preparation requirements, and the potential for altering the native biofilm structure.^12,13 Antimicrobial resistance, recognized by the World Health Organization as a natural yet increasingly alarming phenomenon, currently causes approximately 700 000 deaths annually, a number projected to rise to 10 million by 2050 if no substantial interventions are implemented.¹⁴ The shrinking pipeline of new antibiotics underscores the urgent need to improve diagnostic techniques to ensure the appropriate use of existing treatments and to monitor the evolution of resistance.¹⁵

This review focuses on studies that have employed vibrational spectroscopy and chemometric techniques in various aspects of microbiology, such as microbial viability, bacterial inactivation, bacterial identification, biofilm formation, and antibiotic resistance assessment.

Vibrational spectroscopy

Vibrational spectroscopy, particularly infrared (IR) and Raman spectroscopy, has emerged as a powerful, non-destructive analytical approach for microbial assessment. These methods offer several advantages over conventional techniques, including rapid data acquisition, high throughput, minimal sample preparation, and suitability for in situ analysis using portable instruments.¹⁶ Near-infrared (NIR) spectroscopy and mid-infrared (MIR) spectroscopy are widely used in biological research due to their ability to provide molecular fingerprints of biological samples.¹⁷ Raman spectroscopy, along with its enhanced variant, surface-enhanced Raman spectroscopy (SERS), enables detailed molecular characterization with increased sensitivity and reduced analysis time. SERS leverages plasmonic nanostructures, such as gold or silver nanoparticles, to further minimize laser power requirements and improve reproducibility, making it a particularly attractive option for clinical and microbiological applications.¹⁸

Stable isotope probing for single-cell metabolic analysis

Recent advances in vibrational spectroscopy have enabled label-free, non-invasive biochemical profiling of microorganisms at the single-cell level. Techniques such as Raman spectroscopy, Fourier-transform infrared (FTIR) spectroscopy, and the emerging optical photothermal infrared (O-PTIR) spectroscopy provide complementary insights into the microbial structure, function, and viability.^19–21 O-PTIR, in particular, allows simultaneous acquisition of infrared and Raman spectra from the same spatial location, even in hydrated or optically opaque samples, offering submicron resolution and high chemical specificity.²²

Stable isotope probing (SIP) has significantly expanded the analytical capabilities of vibrational spectroscopy by enabling real-time monitoring of biosynthetic activity through the incorporation of isotopically labeled substrates. Commonly used probes include ¹³C-glucose, heavy water (D₂O), and ¹⁵N-labeled amino acids.²³ These isotopes induce measurable spectral shifts that reflect metabolic activity and cellular viability, facilitating discrimination between active and inactive cells. For instance, ¹³C-glucose incorporation shifts C–C and C=O vibrational modes, while D₂O leads to the emergence of carbon–deuterium (C–D) and O–D bands, indicating metabolic assimilation into cellular macromolecules.²⁴ Similarly, incorporation of ¹⁵N-labeled amino acids and H₂¹⁸O induces detectable shifts in N–H and C=O vibrational bands, enabling single-cell Raman spectroscopy to distinguish metabolically active bacterial populations from inactive ones.²⁵

The integration of vibrational spectroscopy with SIP has found applications across multiple fields. In environmental microbiology, Raman-SIP using ¹³C-labeled substrates has been employed to resolve community-level phylogeny and metabolic function.²⁶ For example, Manzi et al. used ¹³C-labeled bisphenol A (BPA) as a tracer substrate in combination with single-cell Raman microspectroscopy to selectively identify BPA-degrading bacterial populations in activated sludge. By correlating ¹³C-induced spectral shifts in C–H and C–D vibrational bands with isotopic incorporation, the researchers distinguished active degraders from non-utilizing cells and linked metabolic activity to taxonomic identity. This approach enabled functional mapping of BPA biodegradation at single-cell resolution, revealing substrate-specific assimilation patterns and ecological roles of individual bacterial species.²⁷

In a related study, Li et al. utilized ¹³C-labeled methylnaphthalene with single-cell Raman spectroscopy to identify active bacterial degraders in contaminated environments. Their methodology allowed functional mapping of pollutant assimilation and taxonomic resolution at the single-cell level, highlighting the bioremediation potential of complex microbial communities.²⁸

In the food industry, FTIR spectroscopy has been widely adopted for rapid microbial identification and quality control in dairy and meat products.^29,30 Xu et al. applied single-cell Raman spectroscopy combined with deuterium isotope probing (DIP) to identify metabolically active bacteria in high-temperature Daqu, a traditional fermentation starter used in Baijiu production. By tracking C–D vibrational shifts in individual cells, they distinguished active taxa such as Bacillus, Saccharopolyspora, and Streptomyces, which contribute significantly to enzymatic activity and flavor development.³¹ Furthermore, O-PTIR has recently enabled super-resolution chemical imaging of intracellular biochemical alterations in Escherichia coli, revealing metabolic heterogeneity without the need for fluorescent labeling.³²

Chemometric workflows for reliable spectral analysis and model validation

The interpretation of complex spectral data generated by vibrational spectroscopy necessitates advanced analytical techniques, among which chemometric multivariate statistical tools play a pivotal role. These methods enhance spectral analysis by reducing dimensionality, filtering noise, and extracting meaningful biochemical information.³³ Unsupervised classification approaches such as principal component analysis (PCA) and hierarchical cluster analysis (HCA) facilitate pattern recognition and exploratory data analysis. Supervised methods, including linear discriminant analysis (LDA), partial least squares discriminant analysis (PLS-DA), and artificial neural networks (ANN), support predictive modeling and classification. Similarly, regression techniques, ranging from multiple linear regression (MLR) to machine learning algorithms such as support vector machines (SVMs), enable quantitative analysis of spectral datasets.³⁴

Accurate interpretation of microbial vibrational spectra depends on rigorous chemometric workflows that ensure data quality, model reliability, and biological relevance. Each step from spectral pre-processing to validation and model transfer is critical for achieving reproducible classification and prediction across diverse bacterial systems.³⁵

In practice, sample-level data partitioning for microbial vibrational spectroscopy involves assigning all spectra from a single biological replicate exclusively to one subset training, validation, or external test prior to any pre-processing.³⁶ This prevents spectral artifacts or normalization effects from propagating across folds. For instance, when multiple Raman spectra are acquired from one bacterial isolate, all corresponding spectra must remain within the same partition to maintain biological independence.³⁷ A three-way split is commonly used: the training set facilitates model construction, the validation set enables hyperparameter tuning and feature selection via nested cross-validation, and the external test set ideally derived from separate batches or instruments provides an unbiased estimate of model generalizability.³⁸ Automated frameworks such as RamEx facilitate reproducible implementation of microbial Ramanome analysis workflows by enforcing sample-level data partitioning, metadata tracking, and integrated quality control protocols.³⁹ Pre-processing constitutes a foundational stage in chemometric analysis, particularly when handling complex microbial spectra from Raman, FTIR, or NIR platforms. Its objectives include noise reduction, baseline correction, intensity normalization, and enhancement of biochemical signal fidelity.⁴⁰

Effective chemometric modeling of microbial vibrational spectra whether from Raman, FTIR, or NIR sources requires rigorous pre-processing to improve signal fidelity and minimize non-biological variation.⁴¹ Baseline correction techniques, such as asymmetric least squares (ALS) and rubber-band algorithms, are widely used to remove fluorescence and scattering-related offsets, especially in Raman data.⁴² Smoothing and derivative extraction using Savitzky–Golay filters help resolve overlapping peaks and suppress high-frequency noise, thereby improving the detectability of subtle spectral features.⁴³ Normalization methods, including vector and area normalization, correct for intensity variations due to differences in sample thickness or concentration.⁴⁴ Scatter correction techniques, such as standard normal variate (SNV), multiplicative scatter correction (MSC), and extended MSC (EMSC), are commonly applied in FTIR and NIR studies to mitigate light scattering effects and enhance comparability between samples.⁴⁵ In Raman spectroscopy, fluorescence and background removal often achieved through polynomial fitting or adaptive subtraction are essential for isolating genuine vibrational signals.⁴⁶ Finally, wavenumber calibration and drift control using internal standards or reference peaks ensure spectral alignment across instruments and batches, a crucial consideration in longitudinal or multi-instrument studies.⁴⁷ Collectively, these pre-processing steps establish a reliable foundation for multivariate analysis using PCA, PLS-DA, SVM, or ANN in microbial diagnostics.⁴⁸

Robust validation strategies are imperative to ensure the generalizability and reliability of chemometric models derived from spectroscopic data.⁴⁹ In microbial diagnostics, sample-level splitting as opposed to spectrum-level partitioning is essential to prevent overfitting and better reflect real-world biological variability.⁵⁰ Nested cross-validation (CV) is recommended during feature selection to prevent information leakage between training and test folds and to yield unbiased performance estimates. Moreover, the incorporation of external test sets, ideally originating from independent batches or instruments, allows for a realistic evaluation of model transferability.^51,52 Common issues such as data leakage, batch effects, and over-optimistic performance metrics must be carefully mitigated through rigorous experimental design and transparent reporting.⁵³ These validation practices are especially important when applying supervised models like PLS-DA, SVM, and ANN, which are sensitive to subtle spectral variations and susceptible to overfitting in high-dimensional datasets.⁵⁴ Ultimately, reproducibility in microbial spectroscopic diagnostics hinges not only on algorithmic performance but also on the integrity of validation design and the clarity of methodological reporting.⁵⁵

To ensure model robustness and transferability across instruments and experimental batches, it is essential to address batch effects and instrumental variability through systematic calibration strategies.^56,57 Methods such as piecewise direct standardization (PDS) have proven effective in harmonizing spectral discrepancies between devices, thereby facilitating reliable model deployment in practical settings.^58,59 Furthermore, accounting for batch-to-batch variation whether biological or instrumental through external validation sets and transfer learning frameworks significantly improves the generalizability of chemometric models. These measures are particularly critical in microbial spectroscopic diagnostics, where minor spectral shifts can undermine classification accuracy if not adequately controlled.^60,61

To promote reproducibility and transparency in chemometric workflows for bacterial identification using vibrational spectroscopy, the adoption of structured reporting checklists is highly encouraged.³⁹ These checklists should encompass explicit documentation of pre-processing steps (e.g., baseline correction, scatter removal, and normalization), sample-level validation design (e.g., nested cross-validation and external test sets), and model transfer strategies (e.g., calibration transfer and batch correction).⁶² Recent guidelines further stress the importance of reporting data dimensionality, feature selection protocols, hyperparameter tuning procedures, and leakage control mechanisms.⁶³ In addition, reproducibility is bolstered by sharing raw spectra and annotated metadata, which enables independent verification and cross-laboratory benchmarking.⁶⁴ The end-to-end workflow for applying vibrational spectroscopy and chemometric analysis in bacterial studies is illustrated in Fig. 1.

Fig. 1 Workflow for vibrational spectroscopy and chemometric analysis in bacterial studies. NIR: near-infrared; FTIR: Fourier-transform infrared; and SERS: surface-enhanced Raman spectroscopy.

Bacterial identification and typing using vibrational spectroscopy

The cell wall structures of Gram-negative and Gram-positive bacteria are different. The cell wall of Gram-negative bacteria consists of a thin layer of peptidoglycan between two membranes: an inner membrane and an outer membrane. The surface of the outer membrane is covered by lipopolysaccharides (LPS). In contrast, the cell wall of Gram-positive bacteria is composed of a plasma membrane surrounded by a thick layer of peptidoglycan, which is covalently linked to teichoic acids.⁶⁵ However, the main structural distinctions between Gram-negative and Gram-positive bacteria, as well as within bacterial species, are related to the type and composition of polysaccharides. In addition, some bacteria may contain a capsule composed of a gel-like structure of hydrated polysaccharides.⁶⁶ Conventional metabolism-based microbial methods for identifying bacteria and distinguishing their types are manual and time-consuming, hindering rapid diagnosis. Although molecular methods enable the simultaneous identification of one or more bacterial species, these techniques often involve complex and costly procedures.⁶⁷

Following bacterial identification and typing, the bacterial IR spectrum displays a specific fingerprint in the spectral range of 900–1800 cm⁻¹, which is influenced by the major components of the bacterial cell, including proteins, nucleic acids, lipids, and polysaccharides.⁶⁸ The main advantage of the Raman technique lies in its ability to utilize different wavelengths for biological samples, enabling convenient bacterial detection.⁶⁹ The potential of vibrational spectroscopy in differentiating bacterial species is illustrated in Fig. 2. The FTIR spectrometer reveals similarities and differences in the structural components of Gram-negative and Gram-positive bacterial cell walls.

Fig. 2 Identification of bacterial species through vibrational spectroscopy This figure was conceptualized and designed based on the methodology and findings of Kamnev et al.⁶⁸ NIR: near-infrared.

In a study by Treguier et al., NIRS was applied in the 800–2777 nm range to differentiate between Gram-positive bacteria commonly found in raw milk, specifically Lactococcus lactis, Enterococcus durans, E. faecalis, and E. faecium, cultured on agar plates. Utilizing an ANN model, the study achieved discrimination accuracy values of 98.8% at the genus level and 86.3% at the species level between Lactococcus and Enterococcus. PCA loading plots indicated that spectral regions associated with cell-wall polysaccharides played a key role in classification. The findings suggest that NIRS combined with ANN provides a promising alternative to conventional microbiological methods for identifying lactic acid bacteria in dairy products.⁷⁰

Using atomic force microscopy-infrared (AFM-IR) spectroscopy, researchers analyzed the cell wall composition of live Staphylococcus aureus and Escherichia coli at the single-cell level. Kochan et al. were the first to acquire an AFM-IR spectrum from a newly forming bacterial cell wall during division, enabling real-time monitoring of dynamic structural changes. PCA of the spectral data clearly distinguished Gram-positive from Gram-negative bacteria based on unique vibrational signatures primarily associated with peptidoglycans and teichoic acids, which are key constituents of the bacterial envelope. This study underscores the potential of AFM-IR as a powerful, label-free, high-resolution technique for microbial classification.⁷¹

Although MALDI-TOF-MS is widely recognized for its specificity and rapid microbial identification, it faces limitations in differentiating E. coli and Shigella due to their high genetic similarity. To address this, Feng et al. applied both MALDI-TOF-MS and FTIR spectroscopy to analyze 14 strains of E. coli and 9 strains of Shigella, employing a data fusion strategy to enhance classification accuracy. HCA based on MALDI-TOF data alone achieved a 65% correct classification rate, while FTIR data alone yielded 78% accuracy; however, combining both approaches resulted in a 100% correct classification rate. FTIR served as a complementary method to MALDI-TOF by capturing broader biochemical signatures, including variations in polysaccharides and nucleic acids, beyond the ribosomal protein profiles detected by MALDI-TOF. This integrated approach significantly improved bacterial discrimination, particularly between genetically similar species.⁷²

Raman spectroscopy has shown strong potential for rapid bacterial phenotyping, offering high sensitivity and specificity. In this context, Lorenz et al. assessed its effectiveness in distinguishing pathogenic from non-pathogenic E. coli strains. Using Raman microspectroscopy, they analyzed 14 E. coli strains (seven pathogenic and seven non-pathogenic) to develop a predictive model, which was later applied to three additional strains. By combining PCA with SVM, the study achieved 81% sensitivity in training via leave-one-out cross-validation, and an average sensitivity of 77% when identifying the three new strains, correctly classifying 86.3% of pathogenic and 67.1% of non-pathogenic cells. While this sensitivity is lower than those of confirmatory techniques such as PCR or microarrays, the speed and single-cell resolution make Raman spectroscopy a valuable rapid-screening method. However, its inability to elucidate the molecular mechanisms of virulence highlights the need for complementary techniques to characterize bacterial pathogenicity fully.⁷³

The SERS technique integrated with chemometric analysis has also been used to discriminate among six different types of beef pathogenic bacteria (S. typhimurium, E. coli, S. aureus, L. monocytogenes, L. innocua, and L. welshimeri). PCA and HCA of the Raman spectra enhanced the distinction between Gram-positive and Gram-negative bacteria compared to unprocessed data. Bacteria were well separated at the genus level by PCA. Classification at the species level showed favorable differentiation between S. typhimurium and E. coli, while the other four species overlapped. Therefore, PCA was performed for pairs that could be separated with a high degree of species specificity. LDA was employed to validate these models on 540 spectra, achieving an accuracy of 95.65%.⁷⁴

The results of studies on the use of vibrational spectroscopy for detecting bacterial species and strains are summarized in Table 1.

Table 1 A summary of the studies related to the application of vibrational spectroscopy in the detection of bacterial species and strains

Aim of study	Bacterial species	Instrument and spectral range	Important wavelengths	Level of discrimination	Chemometric models	Findings	Ref.
NIR: near infrared; SPA-LDA: successive projection algorithm–linear discriminant analysis; GA-LDA: genetic algorithm-linear discriminant analysis; PCA: principal component analysis; SERS: surfaced enhanced Raman spectroscopy; DFA: discriminant function analyses; and PLS: partial least squares.
The use of NIR spectroscopy and SPA-LDA and GA-LDA algorithms in identifying sensitive and resistant strains of Pseudomonas aeruginosa	P. aeruginosa (63 strains)	Fourier-transform scanning spectrometer: 900–2600 nm	Asymmetric stretching of C–H3 methyl: 1143 nm, C=C alkenes: 1333 nm, stretching and anti-symmetric O–H: 1384 nm, anti-symmetric stretching of N–H2: 1753 nm, and stretching O–H bend: 1871 nm	Strains	SPA-LDA	Sensitive and resistant strains of P. aeruginosa were discriminated	75
					GA-LDA	The SPA-LDA model was able to classify resistant strains of P. aeruginosa with 95% sensitivity using 70 variables
Differentiation of 18 bacterial strains taken from two oak trees by SERS	P. tundrae A10b	Raman spectrometer: 1800–600 cm^−1	Lipids, carbohydrates and tryptophan: 574.2–573.5 cm^−1, guanine and amino acids: 658.4–658.2 cm^−1, adenine and glycosidic ring breathing: 739.1–737.9 cm^−1, proteins (C–N, C–C and C–O stretching): 958.7–958.2 cm^−1, amide III: 1256.9–1247.2 cm^−1, adenine: 1328.4–1327.3 cm^−1, proteins (tryptophan, C–H and C–O deformation): 1366.4–1363.7 cm^−1, and lipids and proteins: 1468.8–1470.4–1468.8 cm^−1	Strains	PCA	8 out of 18 bacterial strains of Paenibacillus sp., Pseudomonas sp., and Pantoea sp. from two oak trees were identified and distinguished with 100% accuracy by PCA and DFA methods	76
	P. agglomerans DSM 3493				DFA
	P. brenneri CFML 97–391
	P. azotoformans
	NBRC 12693
Differentiation of two bacterial species Salmonella enterica and E. coli in 6 culture media by Raman spectroscopy	S. enterica	Raman spectrometer: 1750–350 cm^−1	Carbohydrates (C–O–C): 480 cm^−1, uracil ring: 780 cm^−1, RNA backbone (C–O–P–O–C): 810 cm^−1, amide III (C–N): 1270 cm^−1 C–H, stretching: 1350 cm^−1, C–H2 stretching: 1455 cm^−1 and amide I (C=O): 1665 cm^−1	Strain and species	PCA	The Raman signals were affected by the physicochemical changes of cells and the molecular environment	77
	Paratyphi B (CIP 55.42) E. coli K12 (ATCC 700926)
Identification and differentiation of Salmonella enteritidis among Gram-negative bacteria and live and non-live by SERS	S. enteritidis (BAA-1045)	Raman spectro-microscope: 2000–400 cm^−1	Guanine, tyrosine: 656–654 cm^−1, adenine, glycoside: 730–729 cm^−1, cytosine, uracil: 794–793 cm^−1, proteins: 909–908 cm^−1, C-AgNPs background reference, C–C deformation, and C–N stretching: 959–958 cm^−1, phospholipids, carbohydrates (C–C stretching and C–N stretching): 1023–1020 cm^−1, carbohydrate, lipids, and phospholipids: 1377–1376 cm^−1	Species	PCA	S. enteritidis was detected among Gram-negative bacteria such as E. coli O157: H7 and S. gaminera	78
	S. gaminera (BAA-711)				LDA	Live and dead bacteria were distinguished by citrate-silver nanoparticle-SERS. Non-viable S. enteritidis was discriminated from three live bacteria including S. enteritidis, E. coli and S. gaminera
	E. coli O157: H7 (BAA-43888)				PLS

---

Bacterial inactivation and viability assessment using vibrational spectroscopy

The biochemical profile of metabolic processes during microbial growth is crucial for evaluating cellular viability and metabolic status.⁷⁹ In batch culture systems, planktonic bacteria progress through defined growth stages: lag, exponential, stationary, and death, each characterized by distinct changes in population density, physiological function, and biochemical composition. During the lag phase, cells acclimate to new environmental conditions without active division. The exponential phase is marked by rapid and consistent cellular proliferation. As metabolic by-products accumulate and nutrients become depleted, growth slows, and the rates of cell division and mortality reach equilibrium, defining the stationary phase. Finally, in the death phase, cell viability declines sharply due to sustained stress and nutrient exhaustion.^80,81 Vibrational spectroscopy techniques, particularly IR and Raman spectroscopy, enable precise discrimination between viable and non-viable cells by detecting molecular fingerprints specific to each physiological state.^82,83

Rapid and accurate assessment of bacterial viability and metabolic activity is essential for understanding microbial physiology, monitoring infection progression, and evaluating antimicrobial efficacy.⁸⁴ Although traditional culture-based methods remain widely used, they are time-consuming and often fail to capture the phenotypic heterogeneity inherent in microbial populations.⁸⁵

Raman-deuterium isotope probing (Raman-DIP) enables single-cell resolution detection of metabolically active microbes by identifying C–D vibrational bands in deuterium-labeled biomass, supporting functional profiling and selective isolation within complex communities.⁸⁶ Song et al. applied Raman-DIP to identify metabolically active antimicrobial-resistant bacteria directly in Thames River samples without culturing. By detecting C–D vibrational bands in single-cell Raman spectra after D₂O incubation, they quantified bacterial resistance to carbenicillin, kanamycin, and combined antibiotic exposure. This method facilitated in situ viability assessment and selective isolation of resistant taxa using Raman-activated cell ejection, establishing a link between phenotypic resistance and genotypic identity in unculturable environmental microbes.⁸⁷

Azemtsop Matanfack et al. demonstrated that metabolically active E. coli and S. aureus cells cultured in D₂O exhibit distinct C–D Raman bands in the 2000–2300 cm⁻¹ spectral region, whereas antibiotic-inactivated cells lack these characteristic features.⁸⁸ In contrast, Wang et al. utilized both D₂O and ¹³C-glucose to demonstrate that isotope incorporation into cellular macromolecules produces measurable spectral shifts, enabling reliable differentiation between viable and non-viable bacterial populations.⁸⁹

Weber et al. employed Raman-SIP to investigate metabolic activity in marine microbial communities enriched with ¹³C-labeled substrates. Using single-cell Raman spectroscopy, they detected isotope-induced shifts in vibrational bands corresponding to C–H, N–H, and C=O stretching modes, indicating ¹³C incorporation into cellular macromolecules. These spectral changes enabled functional discrimination between metabolically active and inactive cells without cultivation or fluorescent labeling. Their approach demonstrated the feasibility of in situ metabolic profiling in complex environmental samples, establishing Raman-SIP as a non-invasive method for assessing microbial functionality.⁹⁰

Xiao et al. combined ¹³C-labeled compounds with Raman spectroscopy and liquid chromatography-mass spectrometry (LC-MS) to track isotope assimilation in human and microbial cells. Their experimental design enabled precise detection of ¹³C incorporation into proteins, lipids, and nucleic acids, resulting in measurable spectral shifts in C–C and C=O bands. By comparing labeled and unlabeled populations, the study successfully distinguished active from inactive cells and revealed differential biosynthetic responses to environmental stimuli. These findings underscore the versatility of ¹³C-SIP for functional cell profiling and its potential applications in toxicity assessment and metabolic diagnostics.⁹¹

Cui et al. developed a microfluidic Raman-SIP platform to investigate single-cell carbon and nitrogen fixation in diazotrophic cyanobacteria under nutrient fluctuations. By introducing ¹³CO₂ and ¹⁵N₂ as stable isotope tracers into microfluidic culture chambers, they tracked elemental assimilation using single-cell Raman spectroscopy. Isotope incorporation induced measurable shifts in vibrational bands associated with macromolecular synthesis, enabling quantification of fixation rates and metabolic responses. This approach revealed heterogeneity in nutrient uptake among individual cells and demonstrated interspecies metabolite exchange between cyanobacteria and co-cultured heterotrophs. Thus, SIP served not only as a viability marker but also as a mechanistic tool for resolving nutrient-driven metabolic adaptation at single-cell resolution.⁹²

In a recent study, Lima et al. employed label-free super-resolution optical photothermal infrared (O-PTIR) spectroscopy to investigate metabolic heterogeneity at the single-cell level within Bacillus populations producing poly-3-hydroxybutyrate (PHB). By analyzing vibrational bands associated with PHB biosynthesis, particularly C=O and C–H₂ stretching modes, they identified distinct phenotypic subpopulations with varying metabolic activities. Their findings demonstrate that O-PTIR enables non-invasive, high-resolution chemical imaging of intracellular biopolymer production, offering a powerful approach for assessing microbial viability and metabolic state without fluorescent labeling.⁹³

Fig. 3 schematically illustrates the metabolic activity of active and inactivated bacteria under incubation with D₂O as detected by vibrational spectroscopy.

Fig. 3 Determination of the metabolic activity of bacteria under incubation with deuterium oxide by vibrational spectroscopy (this figure was conceptualized and designed on the basis of Wang and coworker's methodology and findings⁸⁹ and partially created by using https://www.biorender.com/). SERS: surfaced-enhanced Raman spectroscopy and NP: nanoparticle.

Jayan et al. developed a rapid, non-invasive method for evaluating microbial disinfection efficacy in chicken carcass washed water using SERS coupled with SIP. E. coli metabolic activity was monitored under exposure to chlorine and heavy water by incorporating deuterium from D₂O into cellular components. SERS sensitivity was enhanced via in situ synthesis of silver nanoparticles, exploiting the affinity between silver ions and bacterial lipopolysaccharides. The disappearance of the C–D vibrational band in treated samples indicated suppressed metabolic activity. PCA performed on Raman spectra showed two distinct clusters separating disinfectant-treated and untreated bacteria.⁹⁴

Mukherjee et al. investigated the influence of environmental factors on E. coli growth using Raman spectroscopy with excitation wavelengths of 514 nm, 633 nm, and 785 nm. Under 633 nm excitation, a distinctive Raman peak at 740 cm⁻¹ was observed during the exponential phase but was absent in the lag phase. The study examined bacterial cultures incubated at 25 °C, 37 °C, and 45 °C, with the 740 cm⁻¹ peak showing the most incredible intensity at 45 °C, indicating temperature-enhanced metabolic activity. Osmotic stress had minimal impact on the peak. In contrast, treatment with oxidative phosphorylation uncouplers led to significant signal reduction, implicating its link to respiratory activity and chromophores, specifically cytochrome b subunit II. Additional peaks at 1126 cm⁻¹ and 1580 cm⁻¹, associated with protein C–N stretching and nucleic acid content, respectively, were also enhanced under 633 nm excitation. These spectral features reflect the sensitivity of Raman signals to both wavelength selection and environmental growth conditions.⁹⁵

Kumar et al. employed Raman spectroscopy for single-cell biochemical fingerprinting and viability assessment across multiple bacterial strains, including Mycobacterium tuberculosis, M. smegmatis, M. bovis, E. coli, B. subtilis, and Klebsiella pneumoniae, exposed to antibiotics such as isoniazid, ethambutol, rifampicin, and streptomycin. This label-free, non-invasive approach enabled simultaneous bacterial identification and viability classification without using exogenous dyes. PC-LDA achieved strain-level classification with sensitivities of 100% and 99.63% using 514 nm and 633 nm excitation, respectively. Canonical discriminant analysis (CDA) distinguished live, dead, and fixed bacterial cells with 98% accuracy. Fixation resulted in immediate cessation of metabolic activity without macromolecular degradation, whereas antibiotic treatment led to gradual macromolecular breakdown. These differences were evident in the Raman biochemical signatures, particularly in the lipid/protein band ratio at 1450/1659 cm⁻¹, which varied significantly among viability groups. Despite its discrimination capability, the technique could not resolve specific molecular changes induced by antibiotics.⁹⁶

Kochan et al. employed a combined approach using attenuated total reflection Fourier-transform infrared (ATR-FTIR) and Raman spectroscopy to investigate biochemical changes in S. aureus during the lag and exponential growth phases. Spectral analysis revealed phase-dependent variations, with reductions in nucleic acid and protein band intensities observed in the log phase compared to those in the lag phase. In the early lag phase, a marked decline in protein- and lipid-associated peaks (FTIR: 1394 cm⁻¹; Raman: 1453 cm⁻¹) was detected within the first 30 minutes, while increases in nucleic acid-associated peaks (FTIR: 964, 1082, 1215 cm⁻¹; Raman: 785, 1483 cm⁻¹) peaked between 60 and 90 minutes. Multimodal PCA performed simultaneously on ATR and Raman data showed clear, time-dependent discrimination when scans were obtained every 30 minutes during the lag phase. This discrimination was mainly related to nucleic acid content in both spectra.⁹⁷

To address the analytical challenges posed by the high infrared absorption of water and the microscale dimensions of bacterial cells, Meneghel et al. designed two FTIR spectroscopy-based systems for studying cryo-resistance in Lactobacillus bulgaricus strains in aqueous environments. The first system utilized an ATR-inverted microscope coupled to a synchrotron-powered FTIR spectrometer targeting the 1800–1300 cm⁻¹ region to enable single-cell spectral acquisition. The second system employed a thermally powered FTIR microscope with a demountable liquid micro-chamber, operating across the 4000–975 cm⁻¹ range, to probe small bacterial aggregates. Both setups identified spectral markers related to protein secondary structures and cell envelope components associated with differential cryo-resistance. Multivariate spectral analysis revealed greater biochemical heterogeneity in the cryo-sensitive strain CFL1, which exhibited an elevated content of α-helical protein structures (1657–1655 cm⁻¹), whereas the cryo-resistant strain ATCC 11842 was characterized by higher levels of β-sheets and β-turns (1685–1680 cm⁻¹). These findings highlight the utility of advanced FTIR configurations for discriminating phenotypic resilience in live bacterial populations at single-cell and microcolony scales.⁹⁸ The use of vibrational spectroscopy and chemometric techniques for evaluating bacterial viability is summarized in Table 2.

Table 2 A summary of the studies related to the application of vibrational spectroscopy in investigating the metabolic activity of bacteria

Aim of study	Bacterial species	Instrument and spectral range	Important wavelengths	Chemometric models	Findings	Ref.
ATR-FTIR: attenuated total reflection-Fourier transform infrared; HCA: hierarchical cluster analysis; PCA: principal component analysis; SVM: support vector machine; VBNC: viable but non-culturable; and SERS: surfaced-enhanced Raman spectroscopy.
Evaluation of different inactivation methods on Pseudomonas syringae by SERS	P. syringae pv phaseolicola	Raman spectrometer: 1700–450 cm^−1	UV254 treatment: nucleic acids: 678, 730, 960, and 1229 cm^−1	PCA	Three bacteria under different treatments (UV, chlorine, and heat) except for B. subtilis treated with free chlorine were clustered	99
	E. coli K12		UV254 and chlorine treatment: nucleic acids: oxidation guanine: 678 cm^−1, ring breathing of adenine: 730 cm^−1, symmetric stretching vibration of phosphate: 960 cm^−1, guanine: 1333 cm^−1, asymmetric phosphate: 1438 cm^−1, proteins (tryptophan): 619 cm^−1 and C–C twisting: 1365 cm^−1 and carbohydrates: 465 cm^−1	HCA
	B. subtilis ATCC6051		Heat treatment: nucleic acids (ring breathing of guanine): 658 cm^−1, ring breathing of adenine: 733 cm^−1, symmetric stretching vibration of phosphate: 960 cm^−1, guanine: 1319 cm^−1, C–H2/C–H3 wagging mode of guanine, adenine: 1342 cm^−1, deoxyribose: 1455 cm^−1 and proteins: C–C twisting: 619 cm^−1, amide III: 1226 and 1265 cm^−1, and C–H rocking: 1392 cm^−1
Evaluation of the metabolic activity of different bacterial species under incubation with D2O in the presence of various concentrations of UV	Aeromonas sp.	Confocal micro-Raman spectrometer: 3300–500 cm^−1	C–D band: 2300–2040 cm^−1	Without chemometric models	As UV dosages increased the intensity of the C–D band reduced	100
	Pseudomonas sp.				With the increase in UV dosage, the metabolic activity of bacteria decreased in a dose-dependent manner
	E. coli CMCC 44103
	S. aureus ATCC 6538
Examining biomolecular alterations in E. coli W3110 in the VBNC state under stress conditions using ATR-FTIR and PCA	E. coli W3110	ATR-FTIR spectrometer: 4000–650 cm^−1	RNA: 995 cm^−1 (C–O ribose, RNA uracil rings and C–C vibrations)	PCA	Identification of E. coli W3110 in VBNC mode with RNA peak at 995 cm^−1 as a VBNC biomarker	101
Investigating the difference between the live and dead states of bacteria before and after UV radiation	E. coli K-12	Raman spectrometer: 1700–1000 cm^−1	UV radiation: new peaks: lipid: 1450 cm^−1, protein photoproducts: 1410 cm^−1, decreased peak: DNA phosphate backbone: 1100 cm^−1, adenine and guanine: 1571 cm^−1, tryptophan: 1615 cm^−1, and protein amide: 1650 cm^−1, and increased peak: thymine dimer: 1700 cm^−1	PCA	Raman spectra after UV radiation experienced some changes in the intensity of several vibrational regions related to proteins and DNA	102
	E. coli GM1655
	S. marcescens
	M. luteus
	B. thuringiensis
Characterization of metabolic activity in five bacterial species in the form of single cells and biofilm cells	S. aureus DSM20231	Micro-Raman spectroscopy: 1750–650 cm^−1	Single cells in exponential phase: nucleic acids: 671, 725, 780, 1481, and 1578 cm^−1	PCA	Raman spectra of single bacterial cells and single sessile cells in biofilms were different at different growing phases	103
	S. aureus BK31397		Sessile cells in biofilms in stationary phase: DNA and RNA: 671, 725, 780, 1481, and 1578 cm^−1	SVM	Bacteria were classified based on exponential, stationary and biofilm metabolic states with 82.4%, 92.5% and 88.6% accuracy, respectively
	S. epidermidis RP62A
	S. epidermidis VA78203
	E. faecium UK005
	E. faecium BK24498
	E. faecalis DSM20478
	E. faecalis VA245
	P. aeruginosa DSM22644
	P. aeruginosa VA36580

---

Detection of biofilm formation using vibrational spectroscopy

Bacterial biofilms are organized communities of microbial cells enclosed within a self-produced EPS matrix. These bacteria can attach to various living or abiotic surfaces. Complete eradication of biofilms is a challenging process, often leading to persistent contamination. The composition of the biofilm EPS can vary among different species and under varying growth conditions.¹⁰⁴ The main components of the biofilm matrix include lipids, proteins, polysaccharides, and nucleic acids. Crystal violet assay and CLSM are widely used, though complex and time-consuming, methods for measuring biofilm formation. Therefore, there is a strong demand for rapid, on-site methods for biofilm detection. Vibrational spectroscopy can be employed for biofilm identification based on the different functional groups present in the biofilm structure.¹⁰⁵ It provides valuable insights into the mechanisms of biofilm formation.¹⁰⁶ Raman spectroscopy enables characterization of biofilms by obtaining chemical and structural information from bacterial communities and the EPS matrix.^107,108 FTIR spectroscopy can also be used to identify the components of the biofilm matrix, as illustrated in Fig. 4.

Fig. 4 Identification of the molecular components of the biofilm matrix through vibrational spectroscopy. This figure was conceptualized and designed on the basis of Ho and coworker's methodology and findings.¹⁰⁷ eDNA: extracellular DNA.

Bajrami et al. investigated biofilm formation and maturation of Lentilactobacillus parabuchneri, isolated from cheese, using real-time ATR-IR spectroscopy over 72 hours. The spectral region between 1700 and 600 cm⁻¹ proved optimal for biofilm characterization, revealing biochemical contributions from proteins (amides I and II: 1700–1617 and 1577–1464 cm⁻¹), lactic acid/lactate (1465–1293 cm⁻¹), nucleic acids/phospholipids (1280–1180 cm⁻¹), and polysaccharides (1121–997 cm⁻¹). PCA of the IR spectra acquired over four days successfully differentiated biofilm maturation stages and identified a prominent polysaccharide-associated peak at 1081 cm⁻¹ as a key maturation marker. This approach offers a non-invasive, time-resolved framework for monitoring the physicochemical evolution of biofilms and their adhesive properties.¹⁰⁹

Moreover, biofilms composed of two or more different species (dual-species biofilms) can be characterized through vibrational spectroscopy. In this regard, Milho et al. employed ATR-FTIR spectroscopy to elucidate biochemical interactions within dual-species biofilms formed by E. coli and S. enteritidis, two major foodborne pathogens. In contrast to single-species counterparts, dual-species biofilms exhibited reduced overall cell density, suggesting interspecies antagonism or spatial competition during co-development. Spectral characterization of the EPS matrix revealed consistent vibrational features across both configurations, including absorption bands at 3000–2800 cm⁻¹ (lipids), 1700–1500 cm⁻¹ (amide I/II proteins), 1500–1185 cm⁻¹ (phospholipids/nucleic acids), 1185–900 cm⁻¹ (polysaccharides), and 900–600 cm⁻¹ (fingerprint region). The spectral profiles of dual-species matrices essentially represented a superposition of individual EPS signatures; however, compositional dominance by one strain was noted in specific pairings. PCA revealed poor cluster separation in E. coli 434 and S. enteritidis EX2 consortia, indicative of high biochemical overlap, whereas dual-species biofilms composed of E. coli 515 and S. enteritidis 269 aligned more closely with the former. At the same time, the latter formed a distinct cluster.¹¹⁰

Wickramasinghe et al. employed Raman spectroscopy to characterize biofilm composition in four strains of meat spoilage pseudomonads: Pseudomonas fragi (1793, 1832) and P. lundensis (1822, 49968), under storage-relevant temperatures of 10 °C and 25 °C. Biofilms developed at 10 °C exhibited significantly higher total carbohydrate and protein content compared to those grown at 25 °C, suggesting enhanced matrix production under cold stress. No temperature-dependent variation in extracellular DNA (eDNA) was observed. Raman spectra of planktonic cells displayed more intense peaks relative to biofilm-associated cells. PCA of normalized Raman data effectively discriminated between planktonic and biofilm states across all strains. Key spectral markers used in differentiation included bands at 1310 cm⁻¹ (carbohydrates), 1180 and 1600 cm⁻¹ (proteins), 1500 cm⁻¹ (DNA/RNA), and lipid-associated regions. These findings provide detailed biochemical insights into the pseudomonad biofilm architecture and underscore the role of environmental stress in modulating matrix composition during chilled storage.¹¹¹

Kelestemur et al. studied the biofilm formation of P. aeruginosa, S. epidermidis, and Candida albicans on 2D and 3D polymethyl methacrylate (PMMA) surfaces, revealing that 3D fibrous structures enhanced both biomass and biochemical diversity. Using SERS, they monitored metabolic activity over time and found that P. aeruginosa exhibited increasing intensities of protein, carbohydrate, and pyocyanin peaks (1355–1618 cm⁻¹), while a purine-associated signal at 730 cm⁻¹ decreased, indicating maturation. S. epidermidis and C. albicans also showed time- and surface-dependent increases in biomolecular markers, including peaks for carbohydrates, proteins, lipids, and eDNA. PCA and PC-LDA models discriminated biofilms by the species and surface type with up to 100% accuracy, with pyocyanin identified as a key marker for P. aeruginosa identification.¹¹² The results of studies that used vibrational spectroscopy to identify the components of bacterial biofilms are summarized in Table 3.

Table 3 A summary of the studies related to the application of vibrational spectroscopy in identifying the components of bacterial biofilms

Aim of study	Bacterial species (strains)	Instrument and spectral range	Molecular components/associated important wavelengths	Chemometric models	Findings	Ref.
FTIR: Fourier-transform infrared; HCA: hierarchical cluster analysis; PCA-LDA: principal component analysis-linear discriminant analysis; SERS: surfaced-enhanced Raman spectroscopy; and PCA-DFA: principal component analysis-discriminant function analysis.
Investigation of the relationship between V. parahaemolyticus VPS36 biofilm structure and EPS chemical composition	Vibrio parahaemolyticus VPS36	Raman Microscope: 1400–400 cm^−1	Carbohydrates (C–O–C glycosidic ring): 582–561 cm^−1 and 1095–1090 cm^−1, and nucleic acid (O–P–O stretch of DNA): 788 cm^−1	Without chemometric models	A positive correlation between the Raman peak intensity of EPS carbohydrate and Vibrio parahaemolyticus VPS36 biofilm thickness and a negative correlation between the Raman peak intensity of nucleic acid and biofilm porosity were found during the biofilm formation	113
			Carbohydrate:561 and 1095 cm^−1, and nucleic acid: 788 cm^−1
Characterization of the structural components of E. faecalis, A. naeslundii, and two-species biofilm matrices using Raman spectroscopy	E. faecalis	Raman microspectrometer: 1800–600 cm^−1	Polysaccharides: 980–800 cm^−1 and 1125–1000 cm^−1, and proteins: 1335–1270 cm^−1	PCA-DFA	PCA-DFA obtained from Raman spectral profiles of E. faecalis, A. naeslundii, and the dual-species biofilms generated three different clusters based on the different biochemical compositions of extracellular matrices	114
	A. naeslundii
Identification of the chemical profile of Streptococcus biofilms by FTIR and Raman spectroscopy	S. mutans CAPM 6067	FTIR spectrometer: 4000–600 cm^−1	FTIR spectra: carbohydrate: 1200–700 cm^−1, amide I and amide II: 1700–1500 cm^−1, and lipids: 3000–2800 cm^−1	Without chemometric models	The main changes in the FTIR spectra of biofilms were observed in the spectral regions of lipids, proteins and carbohydrates	115
	S. sobrinus DSMZ 20381	Raman microscope: 2000–250 cm^−1	Raman spectra: carbohydrate: 600–475 cm^−1 and 1200–800 cm^−1, lipids: 1500–1175 cm^−1, and proteins:1750–1500 cm^−1
	S. sobrinus CAPM 6070
	S. sobrinus/downei CCUG 21020
	S. sanguis ATCC 10556
Study of S. aureus biofilm formation in a long growth period	S. aureus ATCC 6538	Raman spectrometer: 3300–100 cm^−1	With staining: vibration of the C–C bond and the aromatic ring of crystal violet: 1172 cm^−1, vibration of the n-phenyl moiety of crystal violet: 1384 cm^−1	Without chemometric models	The formation of biofilms in the presence of crystal violet was confirmed	116
			Without staining: bacterial nucleic acids: 640 and 665 cm^−1, bacterial amino compounds and proteins: 1087 and 1590–1575 cm^−1, lipids, proteins, and sugars: 1460–1440 cm^−1		S. aureus biofilm components were determined during a long growth period of 13 days
					The Raman intensity was significantly enhanced by using the gold colloid
					SERS was introduced as a novel, simple, fast, reliable, highly sensitive, on-site, and non-destructive biofilm formation detection technique
Characterization of P. syringae biofilms during development and in the presence of bacteriophage virus Phi6 by SERS to investigate the interaction between virus and bacterial biofilms	P. syringae	SERS: 1600–500 cm^−1	Different biofilm growth steps without Phi6: EPS components: proteins: 620 cm^−1 and 822 cm^−1, guanine: 662 cm^−1, nucleic acids: 804 cm^−1, polysaccharides: 885 cm^−1 and 1125 cm^−1, purine metabolites: 662 cm^−1, and lysogeny broth (LB) medium: 732 cm^−1	PCA-LDA	The PCA score plot of P. syringae biofilms after 5 and 10 hours showed an overlap between biofilms treated with low dose and high dose of phage, while after 20 hours of treatment a clear classification between three groups (uninfected, infected with low and high dose of Phi6) was observed	117
			Different biofilm growth steps with Phi6: nucleic acids: 960 cm^−1 and 1187 cm^−1, membrane phospholipids: 1085 cm^−1, and proteins: 750, 1360, 1621 and 1652 cm^−1	HCA	Uninfected, low dose-infected and high dose-infected biofilms were separated with 91.7%, 93.5% and 99.8% accuracy during 5, 10 and 20 hours, respectively

---

Evaluation of bacterial antibiotic resistance using vibrational spectroscopy

Bacteria exhibit a wide range of phenotypic characteristics.¹¹⁸ The emergence of high cellular diversity within an isogenic population can contribute significantly to increased antibiotic resistance.¹¹⁹ Moreover, selecting the appropriate antibiotic for infection treatment is often challenging because antibiotic susceptibility information is typically unavailable during sampling.¹²⁰ Therefore, there is an urgent need for rapid and high-throughput monitoring techniques to quantify and determine the type and degree of antibiotic resistance. Identification of biochemical changes in bacteria induced by antibiotics can help researchers investigate bacterial responses and susceptibility. Thus, efficient cellular techniques are required to study bacterial metabolism and physiology at the single-cell level.¹²¹ Traditional methods, including disk diffusion and broth or agar dilution assays, require specialized media, laboratories, and skilled personnel. The development of quantitative, multiplex, and real-time PCR techniques has enabled rapid identification of resistance genes. However, these methods are less effective in predicting phenotypic resistance since only a limited number of genes are directly associated with resistance phenotypes. Additionally, methods such as real-time multiplex PCR, isothermal calorimetry, MALDI-TOF-MS, and microarray techniques require expensive and specialized resources, making them more suitable for reference laboratories.^122,123 In this context, vibrational spectroscopy has been successfully used to evaluate phenotypic responses of different bacterial species to various antibiotics.¹²⁴

Single-cell diagnostic platforms have revolutionized antibiotic resistance profiling by enabling direct phenotypic assessment without requiring cultivation. This advancement has significantly reduced diagnostic turnaround time while facilitating rapid detection of metabolic activity and drug response in individual cells.¹²⁵ Such capabilities prove particularly valuable in early-stage or polymicrobial infections, where conventional assays often fail to detect minority resistant subpopulations that can lead to therapeutic failure. Recent research has further demonstrated that integrating vibrational spectroscopy with machine learning enhances detection accuracy in complex clinical samples.^126,127 Building upon these developments, label-free single-cell vibrational spectroscopy has emerged as a powerful approach for culture-independent resistance profiling. Dixneuf et al. investigated Enterococcus faecalis and S. aureus using Raman and O-PTIR spectroscopy, successfully identifying phenotypic heterogeneity and resistant subpopulations without prior cultivation.¹²⁸ Similarly, Yi et al. applied single-cell Raman spectroscopy to detect drug-resistant Escherichia coli, demonstrating a substantial reduction in diagnostic time. These studies collectively underscore the advantages of label-free vibrational techniques in capturing functional resistance signatures at single-cell resolution, particularly in mixed or early-stage infections where traditional culture-based methods may overlook clinically relevant minority populations.¹²⁹

Truong et al. utilized NIRS combined with multivariate analysis to monitor E. coli under varying tetracycline concentrations (0–50 μg mL⁻¹). Distinct NIR absorption peaks at 764 nm (O–H overtones), 830 nm (C–H methylene groups), 846 and 874 nm (C–H aromatic groups), and 910 and 928 nm (C–H₂ bonds) reflected metabolic shifts in bacterial cells during growth. PCA revealed spectral clustering over 24 hours, indicating antibiotic-induced changes, while PLS-DA predicted bacterial behavior with 95% accuracy based on growth time and 80% accuracy across tetracycline concentrations.¹³⁰

Jafari et al. employed ATR-IR spectroscopy to monitor glucose consumption as a proxy for metabolic activity in four bacterial strains Enterobacter aerogenes, Citrobacter freundii, and E. coli (CCUG 11375) following treatment with ampicillin, kanamycin, neomycin, and chloramphenicol. Key absorption bands in the 900–1200 cm⁻¹ fingerprint region (notably at 993, 1035, 1052, 1080, and 1107 cm⁻¹) corresponding to C–O, C–O–C, C–C, and C–C–H vibrations were used to track glucose depletion. Under antibiotic-free conditions, E. coli showed a steady decrease in peak intensity over 24 hours, reflecting glucose consumption. At 1 mg mL⁻¹ ampicillin, metabolic activity persisted for 6 hours, but higher concentrations (10 and 100 mg mL⁻¹) halted glucose uptake, indicating effective inhibition. Similarly, 100 mg mL⁻¹ of kanamycin or neomycin suppressed metabolism within 3 hours, while 10 mg mL⁻¹ had a minimal effect. Chloramphenicol caused a dose-dependent reduction in glucose peaks, suggesting resistance. PCA of ATR-IR spectra over 120 minutes confirmed time-dependent glucose depletion in all strains under sub-inhibitory antibiotic conditions.¹³¹

Ma et al. employed Raman spectroscopy to determine the minimum inhibitory concentrations (MICs) of ampicillin and tetracycline in Campylobacter jejuni strains, using both susceptible (F38011) and resistant strains (1463: ampicillin-resistant; 1143: tetracycline-resistant). Significant spectral changes were observed at 723 cm⁻¹ (adenine) and 778 cm⁻¹ (uracil) in the ampicillin-susceptible strain at low concentrations (above 2 mg L⁻¹), whereas the resistant strain required 256 mg L⁻¹ for similar changes. PCA across the 400–1800 cm⁻¹ Raman region aided in determining MICs, defined as the lowest antibiotic concentration producing a distinct cluster separate from untreated controls. MICs were found to be 1 mg L⁻¹ (ampicillin-susceptible), 256 mg L⁻¹ (ampicillin-resistant), 0.06 mg L⁻¹ (tetracycline-susceptible), and 64 mg L⁻¹ (tetracycline-resistant). These Raman-PCA-derived MICs aligned well with agar dilution reference values for susceptible strains but were less consistent for resistant ones. Additionally, HCA effectively separated antibiotic-treated susceptible strains from untreated controls.¹³²

Ciloglu et al. used SERS to differentiate 19 methicillin-resistant Staphylococcus aureus (MRSA) and 3 methicillin-susceptible S. aureus (MSSA) strains, noting strong spectral similarities but heightened intensity at 732 cm⁻¹ in resistant strains, likely due to peptidoglycan differences. A stacked autoencoder-based deep neural network achieved 97.66% accuracy in classification, outperforming traditional methods such as SVM, neural networks (NN), decision trees (DT), LDA, and KNN, which yielded accuracies ranging from 94.06% to 95.87%.¹³³

Liu et al. investigated biochemical differences between carbapenem-sensitive (CSKP) and carbapenem-resistant (CRKP) Klebsiella pneumoniae strains using SERS. While both groups shared similar spectral peaks at 566, 654, 723, and 1690 cm⁻¹, differences in band intensities and characteristic peaks reflected distinct chemical compositions. Eight supervised machine learning algorithms were applied for classification, with a convolutional neural network (CNN) achieving the highest accuracy of 100% in distinguishing sensitive strains from resistant strains.¹³⁴ The results of studies conducted on the determination of bacterial antibiotic resistance profiles using vibrational spectroscopy are summarized in Table 4.

Table 4 A summary of the studies related to the application of vibrational spectroscopy in determining the profile of antibiotic resistance in bacteria

Aim of study	Bacterial species	Antibiotics	Instrument and spectral range	Important wavelengths or wavenumbers	Chemometric models	Findings	Ref.
LDA: linear discriminant analysis; DA-PC: discriminant analysis of principal component; PCA: principal component analysis; PLSR: partial least squares regression; and SERS: surfaced-enhanced Raman spectroscopy.
Determining the phenotype and metabolism of S. aureus (bacteria) under the influence of antibiotics (erythromycin)	S. aureus ATCC 6538	Erythromycin	Raman-integrated optical mid-infrared photothermal microscope: 1770–1000 cm^−1	Nucleic acid: 1080 cm^−1, amide II: 1550 cm^−1, and amide I: 1650 cm^−1	PCA	Changes in the biochemical composition of bacteria treated with erythromycin were identified	135
						S. aureus treated with erythromycin showed an increase in the intensity of the nucleic acid peak and a decrease in the intensity of the amide II and amide I peaks
						PCA performed on infrared spectra significantly clustered antibiotic-treated and non-treated bacteria
Rapid assessment of antibiotic susceptibility of kanamycin-resistant and sensitive E. coli strains in a microfluidic system by SERS	E. coli ATCC 25922	Kanamycin	Commercial Raman microscope: 1200–400 cm^−1	Adenine and hypoxanthine: 740 cm^−1	Without chemometric methods	The peak intensity of 740 cm^−1 for antibiotic-susceptible E. coli was significantly reduced with the increase of kanamycin concentration compared to untreated E. coli, while the peak at 740 cm^−1 for antibiotic-resistant E. coli had almost the same intensity and was comparable to untreated E. coli	136
	E. coli (DH5-α)
Developing a SERS method to investigate chemical changes in Lb. bulgaricus ATCC 11842 during the initial period of antibiotic treatment to determine antibiotic sensitivity	Lactobacillus delbrueckii subsp. bulgaricus ATCC 11842 (Lb. bulgaricus ATCC 11842)	Penicillin G, ampicillin, and vancomycin	Raman spectro-microscope: 2200–500 cm^−1	735 cm^−1, 1270 cm^−1, 1375 cm^−1, 1455 cm^−1, 1540 cm^−1, amide I: 1612 cm^−1, and tyrosine: 1640 cm^−1	PCA	During the treatment of Lb. bulgaricus with twice MIC of ampicillin, the spectral pattern changed compared to the control group, and new peaks appeared related to amide I and tyrosine	137
					PLSR	The relationship between spectral changes and the ability of Lb. bulgaricus proliferation under penicillin G, ampicillin, and vancomycin showed correlation coefficients of 0.9534, 0.9582, and 0.8811, respectively.
Prediction of antibiotic resistance in the parental E. coli strain and ten antibiotic-resistant laboratory-evolved strains in the absence of antibiotics by Raman spectroscopy	E. coli MDS42	Cefoperazone and cefixime	Raman spectrometer: 1710–600 cm^−1	Skeletal structures of proteins: 853, 936, 972, and 989 cm^−1	DA-PC	Spectral peak intensities were significantly correlated with the expression of some genes contributing to antibiotic resistance	138
						Nucleic acids, proteins, aromatic compounds, and lipids illustrated important differences across strains
The separation of three cefoxitin-resistant Staphylococcus aureus strains and two susceptible S. aureus strains by Raman spectroscopy	S. aureus 44	Cefoxitin (gentamicin, ciprofloxacin, and amoxicillin)	Raman microspectrometer: 3050–650 cm^−1	Nucleic acids: 783 cm^−1, 1240 cm^−1, 1483 cm^−1 and 1574 cm^−1, and proteins 1125 cm^−1 and 1443 cm^−1	LDA	Based on the response of resistant and sensitive strains to cefoxitin, LDA was able to cluster each susceptible strain (SA44 or SA86) compared to the control group while less variation was observed in cefoxitin-resistant strains (SA128 and SA127)	139
	S. aureus 86
	S. aureus 126
	S. aureus 127
	S. aureus 128

---

In summary, lipids, proteins, polysaccharides, and nucleic acids constitute the main molecular components of bacteria in both planktonic and biofilm forms. Any changes in these components can generate significant spectral differences, which can be interpreted using vibrational spectroscopy combined with chemometric techniques. This serves as the basis for microbial viability assessment, bacterial inactivation evaluation, bacterial species identification, biofilm formation detection, and antibiotic resistance assessment. For this purpose, reference libraries are essential for spectral interpretation. In this context, detailed assignments of NIR, FTIR, and Raman bands observed in the spectra of different bacteria in planktonic and biofilm forms are presented in Tables 5, 6 and 7, which can serve as a basis for constructing reference libraries for spectral interpretation in microbiological analyses. Data obtained from spectral peaks and chemometric studies accelerate the early stages of microbial research and provide researchers with an overview to guide further investigations. However, more detailed and complementary analyses are still required.

Table 5 Assignment of NIR bands observed in the spectra of different bacteria

Assignment (compounds)	Band position (cm^−1 or nm)	Assignment (vibration)	Bacterial state	Ref.
Lipids	2924 cm^−1	C–H2 asymmetric stretching in fatty acids	Non-biofilm	140
	1230 nm	C–H stretching	Non-biofilm	70
Proteins	1996 nm	Amide	Non-biofilm	141
Polysaccharides	1315 nm	C–H stretching and deformation	Non-biofilm	70
	1380 nm	deformation of C–H bonds	Non-biofilm	70
	1665, 1705, 1745 nm	C–H stretching of exopolysaccharides	Non-biofilm	70
	1895, and 2025 nm	C=O stretching	Non-biofilm	70
	2329 nm or 4294 cm^−1	C–H stretching or deformation	Non-biofilm	140
	2510, 2560, and 2620 nm	C–H bonds of aromatic rings	Non-biofilm	70

---

Table 6 Assignment of FTIR bands observed in the sspectra of different bacteria

Assignment (compounds)	Band position (cm^−1)	Assignment (vibration)	Biofilm formation	Ref.
Lipids	2967, 2960–2950	C–H3 asymmetric stretching	Non-biofilm and biofilm	115 and 142
	2860–2850, 2859	C–H2 symmetric mode	Biofilm and non-biofilm	115 and 142
	1744, 1740–1730	C=O stretching	Non-biofilm	71, 115 and 142
	1240	PO2^− symmetric vibration	Non-biofilm	97
Proteins	3350–3200	Amide I, stretching N–H	Biofilm	115
	1700–1500	Amide I and amide II	Biofilm	110
	1550, 1548, 1543	Amide II	Non-biofilm	135, 140 and 142
	1240	Amide III	Non-biofilm	140
Nucleic acids	1240	PO2^− symmetric vibration	Non-biofilm	97
	1117	C–C, C–O and P–O–C vibration	Non-biofilm	97
	1080	PO2^− symmetric stretching vibration	Non-biofilm	135
	1034	C–O stretching	Non-biofilm	142
	972	C–C, and C–O vibration	Biofilm	115
Polysaccharides	1117	C–C, C–O and P–O–C vibration	Non-biofilm	97
	1081	P=O vibration of polyphosphate and phosphodiester	Biofilm	109
	999–960	C–O stretching in C–O–C glycosidic linkages	Non-biofilm	135

---

Table 7 Assignment of Raman bands observed in the spectra of different bacteria

Assignment (compounds)	Band position (cm^−1)	Assignment (vibration)	Biofilm formation	Ref.
Lipids	2933	C–H stretching vibrations of fatty acids	Non-biofilm	73 and 143
	1580	C–H3/C–H2 bending	Non-biofilm and biofilm	144 and 145
	1500	C–H2 vibration	Biofilm	114
	1300	C–H3 twisting vibration of fatty acids	Biofilm	114
	1184	C–C and C–H stretching	Non-biofilm	146
	909	C-H2 rocking vibration	Non-biofilm	146
Proteins	1682, 1669, 1667, 1662, 1657	Amide I	Non-biofilm and biofilm	73, 76, 103, 115 and 147
	1599, 1588, 1204, 1174, 853, 818, 801, 655	Tyrosine	Non-biofilm	94, 111, 138 and 146–148
	1545	Amide II	Biofilm	111
	1280, 1241, 1236, 1230	Amide III	Non-biofilm and biofilm	73, 103, 111, 114 and 145
	1049	C–C ring breathing	Non-biofilm	146
	962	C–C deformation, C–N stretching	Non-biofilm	146
Nucleic acids	1588, 1582, 1580, 1574, 1571, 1421, 1361, 1333, 1325, 1031, 964, 746, 735, 733, 730, 728	Ring stretching, ring structure, C–N stretching, and ring breathing in adenine	Non-biofilm and biofilm	73, 76, 102, 111, 114, 133, 134, 144–146 and 148–150
	1333	C–H2/C–H3 wagging in purine bases	Non-biofilm	148
	1220, 786, 746, 676, 566	Thymine	Non-biofilm and biofilm	111, 134, 138, 144 and 145
	1091, 960, 818, 811, 790, 788, 786	PO2^− stretching	Non-biofilm and biofilm	76, 99, 111, 113, 138, 148 and 150
Polysaccharides	1580	C–H3/C–H2 bending	Biofilm	144
	1100	Phosphate backbone	Non-biofilm	102
	868	Ribose, C–O–C stretching	Non-biofilm	146
	790–750	C–O–H, C–C–H, and O–C–H of deformation	Biofilm	114

---

Advanced microfluidic platforms for high-throughput bacterial analysis

Microfluidic systems represent miniaturized analytical platforms that manipulate fluids within microscale channels, typically ranging from tens to hundreds of micrometers in dimensions.¹⁵¹ These devices enable precise control over fluid flow, mixing, separation, and reaction conditions within a compact, integrated format.¹⁵² Their capability to handle nanoliter to picoliter volumes makes them particularly suitable for biological applications where sample availability is limited and rapid analysis is essential.¹⁵³

In bacterial diagnostics, microfluidics offers several advantages over conventional methods, including reduced reagent consumption, minimal sample preparation, faster turnaround time, and compatibility with portable spectroscopic and optical systems.¹⁵⁴ These platforms combine fluidic precision with integrated biosensing modalities, enabling direct analysis of complex samples without extensive preprocessing.¹⁵⁵ Recent innovations have focused on enhancing detection sensitivity and specificity through SERS, magnetic enrichment, and silent spectral encoding.^156–158 Such approaches facilitate simultaneous identification of multiple pathogens at low concentrations, supporting real-time decision-making in decentralized settings.¹⁵⁹

The integration of microfluidic platforms with vibrational spectroscopy techniques, particularly Raman spectroscopy and FTIR, enables label-free, high-throughput biochemical profiling of bacterial cells.¹⁶⁰ This synergistic combination allows real-time monitoring of microbial responses to environmental stimuli or antimicrobial agents, supports discrimination between viable and non-viable cells, and enhances the performance of chemometric models used for classification and prediction.^161,162

Krafft et al. developed a microfluidic device designed to concentrate bacterial cells from drinking water and detect them using SERS. The system employed hydrodynamic focusing to direct bacteria into a detection zone containing plasmonic nanostructures, which amplified Raman signals from bacterial components. The device achieved detection limits below 10³ CFU mL⁻¹ for Escherichia coli and Staphylococcus aureus without requiring culture or DNA extraction. Its compact design and rapid analysis time make it particularly suitable for field deployment in water safety monitoring.¹⁶³

Huo et al. introduced a flexible microfluidic platform integrating magnetic enrichment with silent SERS sensing for multiplex detection of foodborne bacteria. Functionalized magnetic beads captured target pathogens such as Salmonella and Listeria monocytogenes, which were then transferred to SERS-active zones containing engineered Raman reporters with a minimal spectral background. The system enabled simultaneous detection of multiple species with limits of detection below 10² CFU mL⁻¹ and total assay time under 30 minutes. Its compatibility with portable Raman readers highlights its potential for on-site food safety diagnostics.¹⁶⁴

Beyond bulk-level detection, microfluidic platforms increasingly incorporate single-cell resolution to capture bacterial heterogeneity, growth dynamics, and drug susceptibility. Hsieh et al. developed a droplet-based microfluidic system that encapsulates individual bacterial cells in picoliter volumes. Each droplet contains a peptide nucleic acid (PNA) probe targeting 16S rRNA and a growth indicator dye. After thermal lysis and hybridization, fluorescence signals reveal both species identity and antibiotic susceptibility at the single-cell level, with results obtained in under one hour.¹⁶⁵ Le Quellec et al. used microfluidic confinement to expose monoclonal bacterial populations to antibiotics, revealing heterogeneous drug responses among genetically identical cells. This approach highlights the limitations of bulk antimicrobial susceptibility testing and underscores the need for single-cell resolution in resistance profiling.¹⁶⁶ Wang et al. designed a microfluidic device with nutrient flow and waste removal channels to support long-term culture of Mycobacterium smegmatis. Using time-lapse microscopy, they tracked individual cell growth and division over several days, enabling real-time observation of slow-growing bacteria under controlled conditions.¹⁶⁷ Kasahara et al. integrated oxygen-permeable membranes into a microfluidic chip to create spatial oxygen gradients. By culturing bacteria across these gradients and imaging their behavior, they demonstrated how oxygen availability modulates growth, motility, and stress responses at the single-cell level.¹⁶⁸

Cell sorting technologies have also evolved through integration with spectroscopic imaging and machine learning, enabling functional isolation of bacterial subpopulations. Lee et al. employed Raman-activated cell sorting (RACS) to selectively isolate microbial cells based on their spectral signatures. This approach allowed targeted genome retrieval and cultivation of cells with specific metabolic functions.¹⁶⁹ Wang et al. used Raman optical tweezers combined with supervised learning to identify viable but nonculturable Campylobacter jejuni cells, achieving label-free classification at the single-cell level.¹⁷⁰ Dina et al. applied fuzzy logic algorithms to classify bacterial species based on single-cell SERS spectra, demonstrating the potential of spectral fingerprinting for species-level identification.¹⁷¹

Machine learning has become an indispensable component in microfluidic diagnostics, enabling rapid and automated interpretation of complex spectral data. Feizpour et al. implemented convolutional neural networks (CNNs) to perform two-dimensional classification of one-dimensional Raman spectra within SERS-integrated microfluidic chips, achieving high accuracy in bacterial identification.¹⁷² Liu et al. developed a fiber-tip acoustic lysis system that releases intracellular components for Raman analysis, followed by deep learning-based classification of bacterial species. This innovative approach combines mechanical lysis with spectral imaging and artificial intelligence for rapid diagnostics.¹⁷³ Fong et al. integrated SERS nanoprobes with microfluidic flow cytometry to enable multiplex bacterial detection, combining exceptional spectral resolution with high-throughput sorting capabilities.¹⁷⁴

Outlook and future perspectives

To advance the practical implementation of vibrational spectroscopy in microbiology, future research should prioritize the standardization of spectral acquisition protocols, development of robust and generalizable chemometric frameworks, and rigorous validation of these approaches across diverse microbial species, phenotypes, and sample matrices. Techniques including NIR, FTIR, and Raman spectroscopy, and SERS provide rapid, non-destructive insights into bacterial viability, species identification, biofilm composition, and antibiotic resistance profiles. However, selecting the optimal technique for specific microbiological applications necessitates a comprehensive understanding of their respective strengths and limitations.

MIR spectroscopy offers exceptional molecular specificity but is considerably hampered by strong water interference. NIR spectroscopy permits deeper sample penetration and greater tolerance to aqueous environments, yet its utility is constrained by low spectral resolution resulting from overlapping overtone and combination bands. Raman spectroscopy yields detailed molecular fingerprints with minimal sample preparation requirements, though its characteristically weak scattering signals frequently necessitate enhancement strategies. SERS dramatically improves sensitivity but introduces challenges related to substrate-dependent variability and reproducibility. Furthermore, chemometric analysis continues to face obstacles including spectral overlap, baseline drift, and limited model transferability across different instrumental platforms.

Addressing these technical challenges will demand intensified interdisciplinary collaboration spanning spectroscopy, microbiology, and computational modeling. The development of standardized reference databases, implementation of advanced pre-processing algorithms, and establishment of universal calibration protocols represent critical near-term objectives. Additionally, future efforts should focus on creating integrated systems that combine complementary spectroscopic techniques to leverage their individual advantages while mitigating their limitations.

Ultimately, the successful integration of vibrational spectroscopy into routine diagnostic workflows will depend on parallel advancements in scalable instrumentation, regulatory science alignment, and the creation of user-friendly interfaces specifically designed for clinical diagnostics, environmental monitoring, and food safety applications. The creation of validated standard operating procedures and demonstration of cost-effectiveness in real-world settings will be essential for widespread adoption across these diverse application domains.

Conclusions

The use of vibrational spectroscopy and chemometric techniques in the analysis of bacterial microorganisms has been increased in the recent decades due to various advantages of this technique including non-destructiveness, high speed and cost-effectiveness with no need for any complex sample preparation steps. It is worth mentioning that spectroscopic data do not provide us with much information in this field without using chemometric studies. In this review, we have intended to provide a comprehensive insight into the applications of vibrational spectroscopy and chemometric techniques in several important microbiology studies. NIR, MIR, FTIR and Raman spectroscopy techniques are promising approaches in bacterial viability assessment, bacterial inactivation evaluation, bacterial species identification, biofilm formation detection, and antibiotic resistance test. Infrared and Raman spectroscopy can provide valuable information for clinical researchers and microbiologists in a short period of time through detecting the main components of bacteria such as proteins, nucleic acids, lipids, and polysaccharides in different spectral ranges. However, the knowledge of chemometrics is essential to extract relevant information from the vast range of data recorded by these techniques. Despite the advantages noted for vibrational spectroscopy, supplementary methods are required to increase the precision of these techniques. These methods will become more advanced for using in the fields of microbiology in the future and meet the needs of users more efficiently.

Author contributions

M. K. and R. J. wrote the original draft. A. M., R. J. and H. R. M. reviewed and edited the manuscript. All authors have read and agreed to the published version of the manuscript.

Conflicts of interest

The authors declare no competing interests.

Data availability

No primary research results, software or code have been included and no new data were generated or analysed as part of this review.

Acknowledgements

The authors would like to thank the Student Research Committee of Shahid Beheshti University of Medical Science, for their kind financial support (43007688).

References

1. P. Swaminaathan, A. Shaji, A. Saravanan and P. Yaashikaa, Water Conserv. Sci. Eng., 2024, 9, 39.
2. H. Wang, K. M. Abbas, M. Abbasifard, M. Abbasi-Kangevari, H. Abbastabar, F. Abd-Allah, A. Abdelalim, H. Abolhassani, L. G. Abreu and M. R. Abrigo, Lancet, 2020, 396, 1160–1203.
3. M. I. Villalba, E. Rossetti, A. Bonvallat, C. Yvanoff, V. Radonicic, R. G. Willaert and S. Kasas, Proc. Natl. Acad. Sci. U. S. A., 2023, 120, e2221284120.
4. Y. Wang, H. Liu, F. Geng, P. Yang, J. Lü and X. Li, Anal. Bioanal. Chem., 2023, 415, 3515–3523.
5. A. A. Ramadan, Gene Rep., 2022, 29, 101675.
6. L. Wang, W. Liu, J.-W. Tang, J.-J. Wang, Q.-H. Liu, P.-B. Wen, M.-M. Wang, Y.-C. Pan, B. Gu and X. Zhang, Front. Microbiol., 2021, 12, 683580.
7. A. Haider, M. Ringer, Z. Kotroczó, C. Mohácsi-Farkas and T. Kocsis, Microbiol. Res., 2023, 14, 80–90.
8. I. Hernández, C. Sant, R. Martínez and C. Fernández, Front. Microbiol., 2020, 11, 208.
9. A. Kallastu, E. Malv, V. Aro, A. Meikas, M. Vendelin, A. Kattel, R. Nahku and J. Kazantseva, Curr. Res. Food Sci., 2023, 6, 100443.
10. W. Van Holm, J. Ghesquière, N. Boon, T. Verspecht, K. Bernaerts, N. Zayed, I. Chatzigiannidou and W. Teughels, Appl. Environ. Microbiol., 2021, 87, e02653–e02620.
11. L. K. Vestby, T. Grønseth, R. Simm and L. L. Nesse, Antibiotics, 2020, 9, 59.
12. T. George, V. Sivam, M. Vaiyapuri, R. Anandan, G. K. Sivaraman and T. C. Joseph, Arch. Microbiol., 2025, 207, 1–7.
13. D. Mir and M. Rather, Appl. Biochem. Microbiol., 2024, 60, 264–279.
14. J. O’Neill, Tackling drug-resistant infections globally: Final report and recommendations, 2016, https://amr-review.org/sites/default/files/160525_Final%20paper_with%20cover.pdf.
15. G. D. Kaprou, I. Bergšpica, E. A. Alexa, A. Alvarez-Ordóñez and M. Prieto, Antibiotics, 2021, 10, 209.
16. K. V. Hackshaw, J. S. Miller, D. P. Aykas and L. Rodriguez-Saona, Molecules, 2020, 25, 4725.
17. T. Chakkumpulakkal Puthan Veettil, K. Kochan, G. C. Williams, K. Bourke, X. Kostoulias, A. Y. Peleg, D. Lyras, P. A. De Bank, D. Perez-Guaita and B. R. Wood, Anal. Chem., 2024, 96, 18392–18400.
18. W. A. Hassanain, C. L. Johnson, K. Faulds, D. Graham and N. Keegan, Analyst, 2022, 147, 4674–4700.
19. U. Blazhko, D. Byrtusová, V. Shapaval, A. Kohler, C. Sandt and B. Zimmermann, Microb. Cell Fact., 2025, 24, 171.
20. H. Joshi, P. Sinha, D. Bowers and J. P. John, J. Biophotonics, 2024, 17, e202300215.
21. H. R. Khudhur, R. S. Al-Hasnawy, A. Rostaminia, S. H. Abed, S. A. Kadhim, H. Khojasteh and V. Eskandari, BioNanoScience, 2024, 14, 2750–2761.
22. S. Shams, S. Ahmed, D. Smaje, T. Tengsuttiwat, C. Lima, R. Goodacre and H. Muhamadali, Spectrochim. Acta, Part A, 2025, 327, 125374.
23. Y. Wang, Y. Song, Y. Tao, H. Muhamadali, R. Goodacre, N.-Y. Zhou, G. M. Preston, J. Xu and W. E. Huang, Anal. Chem., 2016, 88, 9443–9450.
24. L. Shi, C. Zheng, Y. Shen, Z. Chen, E. S. Silveira, L. Zhang, M. Wei, C. Liu, C. de Sena-Tomas and K. Targoff, Nat. Commun., 2018, 9, 2995.
25. J. Weng, K. Müller, O. Morgaienko, M. Elsner and N. P. Ivleva, Analyst, 2023, 148, 128–136.
26. T. A. Caro, S. Kashyap, G. Brown, C. Chen, S. H. Kopf and A. S. Templeton, FEMS Microbiol. Ecol., 2024, 100, fiae110.
27. H. P. Manzi, D. Qin, K. Yang, H. Li, C. Kiki, J. C. Nizeyimana, L. Cui and Q. Sun, J. Hazard. Mater., 2025, 485, 136862.
28. J. Li, X. Cai, Y. Deng, D. Zhang, B. Li, C. Luo and G. Zhang, Environ. Sci. Technol., 2025, 59, 4709–4720.
29. K. Candoğan, E. G. Altuntas and N. İğci, Food Eng. Rev., 2021, 13, 66–91.
30. O. S. Papadopoulou, A. A. Argyri, V. Kounani, C. C. Tassou and N. Chorianopoulos, Front. Microbiol., 2021, 12, 678356.
31. L. Xu, Z.-M. Lu, X.-J. Zhang, L.-J. Chai, S.-T. Wang, S.-Y. Zhang, C.-H. Shen, B. Li, J.-S. Shi and Z.-H. Xu, Anal. Chem., 2025, 97, 18199–18207.
32. C. Lima, H. Muhamadali and R. Goodacre, Sensors, 2022, 22, 3928.
33. R. Jahani, S. van Ruth, Y. Weesepoel, M. Alewijn, F. Kobarfard, M. Faizi, M. H. S. AliAbadi, A. Mahboubi, A. Nasiri and H. Yazdanpanah, Iran. J. Pharm. Res., 2022, 21, e128372.
34. A. M. Jiménez-Carvelo, A. González-Casado, M. G. Bagur-González and L. Cuadros-Rodríguez, Food Res. Int., 2019, 122, 25–39.
35. A. Kassem, L. Abbas, O. Coutinho, S. Opara, H. Najaf, D. Kasperek, K. Pokhrel, X. Li and S. Tiquia-Arashiro, Front. Microbiol., 2023, 14, 1304081.
36. K. A. Frings, R. Mukherjee, V. Schulze, N. Heine, N. Debener, J. Bahnemann, S. P. Szafrański, M. Stiesch, K. Doll-Nikutta and M. L. Torres-Mapa, Analyst, 2025, 3198–3207.
37. G. Li, Z. Wang, C. Wu, D. Wang, I. Han, J. Lee, D. R. Kaeli, J. G. Dy, K. Q. Weinberger and A. Z. Gu, ISME Commun., 2025, 5, ycaf015.
38. J. Le Galudec, M. Dupoy, L. Duraffourg, V. Rebuffel and P. R. Marcoux, Microb. Biotechnol., 2025, 18, e70093.
39. Y. Zhang, G. Jing, R. Chen, Y. Gong, Y. Li, Y. Wang, X. Wang, J. Zhang, Y. Mao and Y. He, 2025, bioRxiv, Preprint, https://doi.org/10.1101/2025.03.10.642505.
40. B. Dayananda, S. Owen, A. Kolobaric, J. Chapman and D. Cozzolino, Crit. Rev. Anal. Chem., 2024, 54, 2745–2753.
41. S. B. Belhaouari, A. Talbi, M. Elgamal, K. A. Elmagarmid, S. Ghannoum, Y. Yang, Y. Zhao, S. M. Zughaier and H. Bensmail, Heliyon, 2025, 11, e42550.
42. S. Vijayakumar, A. Schwaighofer, G. Ramer and B. Lendl, Spectrochim. Acta, Part A, 2024, 315, 124228.
43. J. Zhang and A. M. Mouazen, Infrared Phys. Technol., 2023, 131, 104720.
44. A. Tsagkaris, K. Bechynska, D. Ntakoulas, I. Pasias, P. Weller, C. Proestos and J. Hajslova, J. Food Compos. Anal., 2023, 119, 105276.
45. S. R. Pushpa, R. K. Sukumaran and S. Savithri, Biomass Convers. Biorefin., 2023, 1–19.
46. C. Ru, W. Wen and Y. Zhong, Spectrochim. Acta, Part A, 2023, 284, 121494.
47. O. Ilchenko, Y. Pilhun, A. Kutsyk, D. Slobodianiuk, Y. Goksel, E. Dumont, L. Vaut, C. Mazzoni, L. Morelli and S. Boisen, Nat. Commun., 2024, 15, 3049.
48. S. Pandian, S. A. Lakshmi, A. Priya, B. Balasubramaniam, J.-L. Z. Zaukuu, R. Durgadevi, V. Abe-Inge and S.-I. Sohn, Processes, 2023, 11, 1191.
49. R. Jahani, A. Nasiri, S. Eslamizad, S. Mokhtari, M. Bayanati, A. Haratian and H. Yazdanpanah, J. Food Compos. Anal., 2025, 139, 107127.
50. S. B. Frempong, M. Salbreiter, S. Mostafapour, A. Pistiki, T. W. Bocklitz, P. Rösch and J. Popp, Molecules, 2024, 29, 1077.
51. D. Lilek, D. Zimmermann, L. Steininger, M. Musso, B. D. Wilts, S. Gamsjaeger, D. R. Hermann, C. Wiesner, A. Grünfelder and B. Herbinger, J. Raman Spectrosc., 2025, 867–877.
52. Q. Tu, M. Li, Z. Sun, H. Niu, L. Zhao, Y. Wang, L. Sun, Y. Liu, Y. Zhu and G. Zhao, Anal. Bioanal. Chem., 2025, 1–12.
53. C. Highmore, N. Hanrahan, Y. Cook, Y. Pritchard, A. Lister, K. Cooper, G. Devitt, A. P. Munro, S. N. Faust and S. Mahajan, npj Antimicrob. Resist., 2025, 3, 74.
54. Z. Hamidi, A. Jamalzadeh, R. Jahani, H. Parastar, F. Kobarfard and H. Yazdanpanah, J. Food Compos. Anal., 2025, 107684.
55. D. P. Dos Santos, M. M. Sena, M. R. Almeida, I. O. Mazali, A. C. Olivieri and J. E. Villa, Anal. Bioanal. Chem., 2023, 415, 3945–3966.
56. Q. Li, Y. Yang, J. Wu, C. Wei, H. Xia, Y. Han, Y. Huang and X. Ma, Vib. Spectrosc., 2024, 133, 103695.
57. B. Zhang, A. K. Nambisan, A. P. Hungund, X. Jones, Q. Yang and J. Huang, RSC Adv., 2025, 15, 4847–4860.
58. J. P. Andries and Y. Vander Heyden, Anal. Bioanal. Chem., 2024, 416, 5351–5364.
59. L. Pétillot, F. Pewny, M. Wolf, C. Sanchez, F. Thomas, J. Sarrazin, K. Fauland, H. Katinger, C. Javalet and C. Bonneville, Eng. Rep., 2020, 2, e12230.
60. D. Bhowmik, J. J. S. Rickard, R. Jelinek and P. G. Oppenheimer, Sustainable Food Technol., 2025, 3, 10–31.
61. H. W. H. Hui, W. Kong and W. W. B. Goh, Briefings Bioinf., 2024, 25, bbae515.
62. M. Li, Q. Cai, T. Zhang, H. Tang and H. Li, J. Chemom., 2025, 39, e70006.
63. P. Király and G. Tóth, J. Chemom., 2025, 39, e70026.
64. E. Rial-Rodríguez, C. O. Kappe and G. Laudadio, Chem.: Methods, 2025, e202500104.
65. S. Raina, Int. J. Mol. Sci., 2023, 24, 7498.
66. M. Winkle, V. M. Hernández-Rocamora, K. Pullela, E. C. Goodall, A. M. Martorana, J. Gray, I. R. Henderson, A. Polissi and W. Vollmer, mBio, 2021, 12, e00836–e00821.
67. F. Zendri, V. Schmidt, N. Mauder, A. Loeffler, R. E. Jepson, C. Isgren, G. Pinchbeck, S. Haldenby and D. Timofte, Front. Microbiol., 2024, 15, 1334268.
68. A. Kamnev and A. Tugarova, J. Anal. Chem., 2023, 78, 1320–1332.
69. M. Hardy and H. O. M. Chu, Analyst, 2025, 150, 1986–2008.
70. S. Treguier, C. Couderc, H. Tormo, D. Kleiber and C. Levasseur-Garcia, J. Microbiol. Methods, 2019, 165, 105693.
71. K. Kochan, D. Perez-Guaita, J. Pissang, J.-H. Jiang, A. Y. Peleg, D. McNaughton, P. Heraud and B. R. Wood, J. R. Soc. Interface, 2018, 15, 20180115.
72. B. Feng, H. Shi, F. Xu, F. Hu, J. He, H. Yang, C. Ding, W. Chen and S. Yu, Anal. Chim. Acta, 2020, 1111, 75–82.
73. B. Lorenz, N. Ali, T. Bocklitz, P. Rösch and J. Popp, Anal. Bioanal. Chem., 2020, 412, 8241–8247.
74. H. Zuo, Y. Sun, M. Huang, S. M. Fowler, J. Liu, Y. Zhang and Y. Mao, Foods, 2024, 13, 3688.
75. A. S. Marques, J. N. Castro, F. J. Costa, R. M. Neto and K. M. Lima, Microchem. J., 2016, 124, 306–310.
76. D. Vaitiekūnaitė and V. Snitka, Microorganisms, 2021, 9, 1969.
77. A. Assaf, E. Grangé, C. Cordella, D. Rutledge, M. Lees, A. Lahmar and G. Thouand, Anal. Bioanal. Chem., 2020, 412, 3595–3604.
78. P. Chuesiang, V. Ryu, U. Siripatrawan, L. McLandsborough and L. He, Lebensm.-Wiss. Technol., 2021, 150, 111962.
79. S. Thapa, H. Li, J. Ohair, S. Bhatti, F.-C. Chen, K. A. Nasr, T. Johnson and S. Zhou, Mol. Biotechnol., 2019, 61, 579–601.
80. V. R. Krishnamurthi, I. I. Niyonshuti, J. Chen and Y. Wang, PLoS One, 2021, 16, e0245205.
81. A. Lo Grasso, A. Fort, F. F. Mahdizadeh, A. Magnani and C. Mocenni, Math. Comput. Model. Dyn. Syst., 2023, 29, 169–185.
82. T. T. Nguyen, N. H. Nguyen, G. T. Pham, J. Vongsvivut, M. H. Brown, V. K. Truong and K. Vasilev, Mater. Adv., 2023, 4, 6342–6352.
83. J. Wang, K. Kong, C. Guo, G. Yin, S. Meng, L. Lan, L. Luo and Y. Song, Front. Microbiol., 2023, 14, 1144607.
84. R. Öberg, T. Dahlberg, D. Malyshev and M. Andersson, Analyst, 2023, 148, 2141–2148.
85. C. Morazzoni, M. Sirel, S. Allesina, M. Veses Garcia, K. Kragh, M. Pane and K. Beilharz, Front. Microbiol., 2024, 15, 1391688.
86. D. Berry, E. Mader, T. K. Lee, D. Woebken, Y. Wang, D. Zhu, M. Palatinszky, A. Schintlmeister, M. C. Schmid and B. T. Hanson, Proc. Natl. Acad. Sci. U. S. A., 2015, 112, E194–E203.
87. Y. Song, L. Cui, J. Á. S. López, J. Xu, Y.-G. Zhu, I. P. Thompson and W. E. Huang, Sci. Rep., 2017, 7, 16648.
88. G. Azemtsop Matanfack, A. Pistiki, P. Rösch and J. Popp, Life, 2021, 11, 1003.
89. C. Wang, R. Chen, J. Xu and L. Jin, Front. Microbiol., 2022, 13, 936726.
90. F. Weber, T. Zaliznyak, V. P. Edgcomb and G. T. Taylor, Appl. Environ. Microbiol., 2021, 87, e01460–e01421.
91. X. Xiao, Y. Zhou, X. Li, J. Jin, J. Durham, Z. Ye, Y. Wang, B. Hennig and P. Deng, Microbiome, 2024, 12, 90.
92. L. Cui, Y. Xin, K. Yang, H. Li, F. Tan, Y. Zhang, X. Li, Z. Zhu, J. Yang and S.-J. Kao, PNAS Nexus, 2023, 2, pgad006.
93. C. Lima, H. Muhamadali and R. Goodacre, Anal. Chem., 2023, 95, 17733–17740.
94. H. Jayan, D.-W. Sun, H. Pu and Q. Wei, Spectrochim. Acta, Part A, 2022, 280, 121549.
95. R. Mukherjee, T. Verma, D. Nandi and S. Umapathy, Anal. Bioanal. Chem., 2020, 412, 5379–5388.
96. S. Kumar, R. Gopinathan, G. K. Chandra, S. Umapathy and D. K. Saini, Anal. Bioanal. Chem., 2020, 412, 2505–2516.
97. K. Kochan, E. Lai, Z. Richardson, C. Nethercott, A. Y. Peleg, P. Heraud and B. R. Wood, Sensors, 2020, 20, 3452.
98. J. Meneghel, S. Passot, F. Jamme, S. Lefrançois, P. Lieben, P. Dumas and F. Fonseca, Anal. Bioanal. Chem., 2020, 412, 7049–7061.
99. W. Wang, A. Rahman, Q. Huang and P. J. Vikesland, Water Res., 2022, 220, 118668.
100. L. Guo, C. Ye, L. Cui, K. Wan, S. Chen, S. Zhang and X. Yu, Environ. Int., 2019, 130, 104883.
101. Ö. K. İzgördü, R. Gurbanov and C. Darcan, World J. Microbiol. Biotechnol., 2024, 40, 1–15.
102. R. Li, D. Dhankhar, J. Chen, A. Krishnamoorthi, T. C. Cesario and P. M. Rentzepis, IEEE Access, 2019, 7, 23549–23559.
103. H. Shen, P. Rösch, L. Thieme, M. W. Pletz and J. Popp, J. Mol. Struct., 2023, 1277, 134831.
104. Y. Liu, S. Long, H. Wang and Y. Wang, Int. Wound J., 2024, 21, e14667.
105. R. Kato, K. Miyazawa, T. Imura and T. Minamikawa, Biosci. Biotechnol. Biochem., 2025, 89, 489–495.
106. M. C. Sportelli, C. Kranz, B. Mizaikoff and N. Cioffi, Anal. Chim. Acta, 2022, 1195, 339433.
107. C.-S. Ho, N. Jean, C. A. Hogan, L. Blackmon, S. S. Jeffrey, M. Holodniy, N. Banaei, A. A. Saleh, S. Ermon and J. Dionne, Nat. Commun., 2019, 10, 4927.
108. R. Weiss, M. Palatinszky, M. Wagner, R. Niessner, M. Elsner, M. Seidel and N. P. Ivleva, Analyst, 2019, 144, 943–953.
109. D. Bajrami, S. Fischer, H. Barth, M. A. Sarquis, V. M. Ladero, M. Fernández, M. C. Sportelli, N. Cioffi, C. Kranz and B. Mizaikoff, npj Biofilms Microbiomes, 2022, 8, 92.
110. C. Milho, M. D. Silva, D. Alves, H. Oliveira, C. Sousa, L. M. Pastrana, J. Azeredo and S. Sillankorva, Sci. Rep., 2019, 9, 18183.
111. N. N. Wickramasinghe, M. M. Hlaing, J. T. Ravensdale, R. Coorey, P. S. Chandry and G. A. Dykes, Sci. Rep., 2020, 10, 1–16.
112. S. Keleştemur, Z. Çobandede and M. Çulha, Colloids Surf., B, 2020, 188, 110765.
113. L. Tan, F. Zhao, Q. Han, A. Zhao, P. K. Malakar, H. Liu, Y. Pan and Y. Zhao, Front. Microbiol., 2018, 9, 1881.
114. T. Ramirez-Mora, C. Dávila-Pérez, F. Torres-Méndez and G. Valle-Bourrouet, J. Spectrosc., 2019, 2019, 1–7.
115. B. Gieroba, M. Krysa, K. Wojtowicz, A. Wiater, M. Pleszczyńska, M. Tomczyk and A. Sroka-Bartnicka, Int. J. Mol. Sci., 2020, 21, 3811.
116. Y. Liu, H. Yu, Y. Cheng, Y. Guo, W. Yao and Y. Xie, Food Anal. Methods, 2020, 13, 1710–1716.
117. A. Garg, W. Nam, W. Wang, P. Vikesland and W. Zhou, ACS Sens., 2023, 8, 1132–1142.
118. H. Nicoloff, K. Hjort, B. R. Levin and D. I. Andersson, Nat. Microbiol., 2019, 4, 504–514.
119. R. Banerjee and R. Patel, JAC Antimicrob. Resist., 2023, 5, dlad018.
120. N. Jackson, C. A. Borges, N. J. Tarlton, A. Resendez, A. K. Milton, T. R. de Boer, C. R. Butcher, N. Murthy and L. W. Riley, J. Microbiol. Methods, 2021, 182, 106160.
121. A. Patnaik, S. K. Rai and R. K. Dhaked, 3 Biotech, 2025, 15, 1–14.
122. T. Hosoda, M. Suzuki, T. Matsuno, K. Matsui, K. Ohyama and Y. Doi, Microbiol. Spectrum, 2025, e01014–e01025.
123. J. Kubiak, A. Morgan, A. Kirmaier, R. Arnaout and S. Riedel, J. Clin. Microbiol., 2023, 61, e00952–e00923.
124. M. Giergiel, T. Chakkumpulakkal Puthan Veettil, A. Rossetti and K. Kochan, Int. J. Mol. Sci., 2024, 25, 8148.
125. G. Reszetnik, K. Hammond, S. Mahshid, T. AbdElFatah, D. Nguyen, R. Corsini, C. Caya, J. Papenburg, M. P. Cheng and C. P. Yansouni, Nat. Commun., 2024, 15, 9719.
126. J.-C. Ding, L.-T. Dai, J.-X. Tang, H.-M. Zhang, L.-L. Zhai, C.-Y. Cai, Z.-R. Xiao, Z.-M. Chen, J.-H. Ling and M.-X. Zou, World J. Microbiol. Biotechnol., 2025, 41, 1–12.
127. W. Lu, H. Li, H. Qiu, L. Wang, J. Feng and Y. V. Fu, Front. Microbiol., 2023, 13, 1076965.
128. S. Dixneuf, A.-C. Chareire-Kleiberg, P. Mahé, M. El Azami, C. Kolytcheff, S. Bellais, C. Guyard, C. Védrine, F. Mallard and Q. Josso, Front. Microbiol., 2023, 14, 1232250.
129. X. Yi, Y. Song, X. Xu, D. Peng, J. Wang, X. Qie, K. Lin, M. Yu, M. Ge and Y. Wang, Anal. Chem., 2021, 93, 5098–5106.
130. V. K. Truong, J. Chapman and D. Cozzolino, Food Anal. Methods, 2021, 14, 1394–1401.
131. M. J. Jafari, M. Golabi and T. Ederth, Spectrochim. Acta, Part A, 2024, 304, 123384.
132. L. Ma, L. Chen, K. C. Chou and X. Lu, Appl. Environ. Microbiol., 2021, 87, e00388–e00321.
133. F. U. Ciloglu, A. Caliskan, A. M. Saridag, I. H. Kilic, M. Tokmakci, M. Kahraman and O. Aydin, Sci. Rep., 2021, 11, 18444.
134. W. Liu, J.-W. Tang, J.-W. Lyu, J.-J. Wang, Y.-C. Pan, X.-Y. Shi, Q.-H. Liu, X. Zhang, B. Gu and L. Wang, Microbiol. Spectrum, 2022, 10, e02409–e02421.
135. J. Xu, X. Li, Z. Guo, W. E. Huang and J.-X. Cheng, Anal. Chem., 2020, 92, 14459–14465.
136. K.-W. Chang, H.-W. Cheng, J. Shiue, J.-K. Wang, Y.-L. Wang and N.-T. Huang, Anal. Chem., 2019, 91, 10988–10995.
137. P. Wang, X. Wang, Y. Sun, G. Gong, M. Fan and L. He, Anal. Methods, 2020, 12, 376–382.
138. A. Germond, T. Ichimura, T. Horinouchi, H. Fujita, C. Furusawa and T. M. Watanabe, Commun. Biol., 2018, 1, 85.
139. A. N. Rousseau, N. Faure, F. Rol, Z. Sedaghat, J. Le Galudec, F. Mallard and Q. Josso, ACS Omega, 2021, 6, 16273–16279.
140. P. Krepelka, A. Bolívar and F. Pérez-Rodríguez, J. Near Infrared Spectrosc., 2021, 29, 63–72.
141. Y.-Z. Feng, G. Downey, D.-W. Sun, D. Walsh and J.-L. Xu, J. Food Eng., 2015, 149, 87–96.
142. A. Salman, U. Sharaha, E. Rodriguez-Diaz, E. Shufan, K. Riesenberg, I. J. Bigio and M. Huleihel, Analyst, 2017, 142, 2136–2144.
143. A. Nakar, A. Pistiki, O. Ryabchykov, T. Bocklitz, P. Rösch and J. Popp, Anal. Bioanal. Chem., 2022, 414, 1481–1492.
144. J. Feng, C. De La Fuente-Núñez, M. J. Trimble, J. Xu, R. E. Hancock and X. Lu, Chem. Commun., 2015, 51, 8966–8969.
145. E. Witkowska, D. Korsak, A. Kowalska, A. Janeczek and A. Kaminska, Anal. Bioanal. Chem., 2018, 410, 5019–5031.
146. S. Bashir, H. Nawaz, M. I. Majeed, M. Mohsin, S. Abdullah, S. Ali, N. Rashid, M. Kashif, F. Batool and M. Abubakar, Photodiagn. Photodyn. Ther., 2021, 34, 102280.
147. W. Ding, W. Jin, S. Cao, X. Zhou, C. Wang, Q. Jiang, H. Huang, R. Tu, S.-F. Han and Q. Wang, Water Res., 2019, 160, 339–349.
148. M. Kashif, M. I. Majeed, H. Nawaz, N. Rashid, M. Abubakar, S. Ahmad, S. Ali, H. Hyat, S. Bashir and F. Batool, Spectrochim. Acta, Part A, 2021, 261, 119989.
149. X. Chen, M. Tang, Y. Liu, J. Huang, Z. Liu, H. Tian, Y. Zheng, M. L. de La Chapelle, Y. Zhang and W. Fu, Mikrochim. Acta, 2019, 186, 1–8.
150. A. V. Gannesen, E. L. Zdorovenko, E. A. Botchkova, J. Hardouin, S. Massier, D. S. Kopitsyn, M. V. Gorbachevskii, A. A. Kadykova, A. S. Shashkov and M. V. Zhurina, Front. Microbiol., 2019, 10, 1284.
151. C.-Y. Ku, Y.-W. Chiang, H.-Y. Hsu, H.-W. Cheng, K.-L. Chen, Y.-Y. Han, J.-K. Wang, Y.-L. Wang and N.-T. Huang, Biosens. Bioelectron., 2025, 117576.
152. Z. Meng, T. Zhang, Z. Zhang, X. Qu, B. Wang, Y. Qiu, P. Wu and P. Ding, Biosens. Bioelectron., 2025, 271, 117012.
153. F. Safir, N. Vu, L. F. Tadesse, K. Firouzi, N. Banaei, S. S. Jeffrey, A. A. Saleh, B. T. Khuri-Yakub and J. A. Dionne, Nano Lett., 2023, 23, 2065–2073.
154. N. Jia, A. Daignault-Bouchard, T. Deng, T. G. Mayerhöfer, A. Bégin-Drolet and J. Greener, Lab Chip, 2023, 23, 3561–3570.
155. Y. Liu, G. Su, W. Wang, H. Wei and L. Dang, Anal. Methods, 2024, 16, 2085–2092.
156. F. Huang, C. Sun, J. Dong, X. Wu, Y. Du, Q. Hu and L. Zhou, Mikrochim. Acta, 2024, 191, 237.
157. N. O'Toole, M. E. Allen, C. Contini and Y. Elani, Chem. Commun., 2025, 61, 12155–12158.
158. Z. Pang, L. Shi, M. Wang and J. Tao, Analyst, 2025, 150, 1398–1408.
159. H. Wu, Y. Gao, Q. Chen, L. Yao, B. Yao, J. Yang and W. Chen, Mikrochim. Acta, 2024, 191, 429.
160. S. Park, K. Kim, A. Go, M.-H. Lee, L. Chen and J. Choo, ACS Sens., 2025, 10, 1217–1227.
161. L. Shang, P. Liang, L. Xu, Y. Xue, K. Liu, Y. Wang, X. Bao, F. Chen, H. Peng and Y. Wang, Anal. Chem., 2023, 96, 248–255.
162. M. Wu, C. Su, S. Cao, Y. Sun, N. Niu, L. Chen and Z. Tang, Biosens. Bioelectron., 2025, 271, 117088.
163. B. Krafft, A. Tycova, R. D. Urban, C. Dusny and D. Belder, Electrophoresis, 2021, 42, 86–94.
164. B. Huo, L. Xia, Y. Hu and G. Li, Biosens. Bioelectron., 2024, 255, 116227.
165. K. Hsieh, K. E. Mach, P. Zhang, J. C. Liao and T.-H. Wang, Acc. Chem. Res., 2021, 55, 123–133.
166. L. Le Quellec, A. Aristov, S. Gutiérrez Ramos, G. Amselem, J. Bos, Z. Baharoglu, D. Mazel and C. N. Baroud, PLoS One, 2024, 19, e0303630.
167. H. Wang, G. M. Conover, S.-I. Han, J. C. Sacchettini and A. Han, Microsyst. Nanoeng., 2021, 7, 37.
168. K. Kasahara, M. Leygeber, J. Seiffarth, K. Ruzaeva, T. Drepper, K. Nöh and D. Kohlheyer, Front. Microbiol., 2023, 14, 1198170.
169. K. S. Lee, F. C. Pereira, M. Palatinszky, L. Behrendt, U. Alcolombri, D. Berry, M. Wagner and R. Stocker, Nat. Protoc., 2021, 16, 634–676.
170. K. Wang, X. Ma, P.-L. Longchamps, K. C. Chou and X. Lu, Anal. Chem., 2025, 97, 2028–2035.
171. N. E. Dina, A. M. R. Gherman, A. Colniţă, D. Marconi and C. Sârbu, Spectrochim. Acta, Part A, 2021, 247, 119149.
172. M. Feizpour, H. Van den Bosche, L. Melikyan, T. Demuyser, P. Cools, H. Thienpont, T. Sarukhanyan and H. Ottevaere, Talanta, 2025, 295, 128325.
173. Y. Liu, M. Ji, X. Ren, Z. Dong, T. Wen, Q. Dong, H. p. Ho, L. Cui, Y. Lu and G. Wang, Adv. Sci., 2025, 12, e07724.
174. Z.-H. Fong, C.-H. Wang, C.-Y. Yang, H.-C. Kan, Y.-J. Lin, Y.-L. Chen, Y.-C. Shen, Y.-C. Yu, L.-K. Chau and C.-W. Io, Microchem. J., 2025, 212, 113436.

---