Food Analytical Methods最新文献

Garlic (Allium sativum L.) is a functional food of considerable therapeutic interest. However, the clinical utility of its primary organosulfur compound, allicin, is limited by its rapid degradation. To overcome this instability, the synthesis of allithiamine, a stable lipid-soluble derivative, has been proposed as a promising therapeutic strategy. Nevertheless, the lack of robust analytical methods for monitoring the formation of this complex remains a major challenge. The primary objective of this study was to develop and validate a specific, sensitive, and precise high-performance liquid chromatography method with photodiode array detection (HPLC-PDA) for the simultaneous quantification of allithiamine, thiamine, and allicin in a garlic-brewer’s yeast complex, in strict accordance with International Council for Harmonisation (ICH) Q2 (R2) guidelines. The method exhibited excellent selectivity and linearity, with coefficients of determination (R²) exceeding 0.999 for all analytes and achieved a low limit of quantification (LOQ) for allithiamine of 0.95 µg/mL. Accuracy was confirmed through recovery studies (95.01%–104.49%), and precision assessments yielded coefficients of variation (CVs) below 1.6%. In conclusion, the proposed HPLC-PDA method represents the first reported ICH-compliant protocol for the simultaneous analysis of allithiamine and its precursors in a functional food matrix. This method provides a robust tool for quality control and supports the standardization of next-generation garlic supplements.

Fish spoilage caused by microbial activity and enzymatic degradation remains a major challenge for quality control in the seafood industry. This study developed a chitosan/polyvinyl alcohol (Cs/PVA)-based colorimetric film incorporating butterfly pea flower anthocyanin extract (BPF) as a smart indicator for real-time, non-destructive monitoring of fish freshness. The total anthocyanin content of the extract was 140.60 mg/L, expressed as cyanidin-3-glucoside equivalents. Films containing 1–5 mL of anthocyanin extract were prepared and characterized using liquid chromatography–high-resolution mass spectrometry (LC-HRMS), ultraviolet–visible (UV–Vis) spectroscopy, field emission scanning electron microscopy (FESEM), and Fourier transform infrared (FTIR) spectroscopy. The films were evaluated for pH sensitivity and ammonia responsiveness under refrigerated (6 ± 1 °C) and ambient (25 ± 1 °C) storage conditions. Among the formulations, Cs/PVA-BPF2 (2 mL anthocyanin extract) exhibited the best overall performance, showing the most pronounced visual color transitions, a maximum color difference (ΔE) of 54.78 ± 0.30, and the highest colorimetric sensitivity index (S%) of 51.3% toward ammonia exposure. During refrigerated storage of Oreochromis mossambicus, a distinct color transition became visible at approximately pH 7.0, corresponding to days 5–6 of storage, indicating storage-associated chemical changes. Under ambient conditions, all formulations developed a noticeable green coloration after 15 h, corresponding to pH values of 7.0–8.0. Integration with a custom card-reader system enabled quantitative RGB/CIELAB analysis, improving objectivity and detection reliability. These results demonstrate the potential of Cs/PVA-BPF2 as a low-cost, bio-based polymers for monitoring the freshness of protein-rich perishable foods.

Palm oil adulteration in butter fat represents an important authenticity concern, requiring rapid analytical approaches capable of distinguishing chemically similar lipid matrices. This study evaluated attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy combined with principal component analysis (PCA) and partial least squares regression (PLSR) for the quantitative determination of palm oil in butter fat. A total of 48 samples, including pure butter fat, pure palm oil and mixtures covering a broad adulteration range, were investigated. PCA was used as a data-driven exploratory tool to identify spectral regions contributing to differentiation between the lipid matrices, highlighting the 630–930 and 1060–1310 cm⁻¹ intervals as potentially informative regions. PLSR models were subsequently developed using 36 calibration and 12 prediction samples and evaluated under different spectral preprocessing conditions. The 630–930 cm⁻¹ region with first-derivative preprocessing provided the most promising quantitative performance, with R²cal = 0.999, RMSEC = 0.637% (w/w), R²pred = 0.936 and RMSEP = 7.485% (w/w). Examination of individual predictions, however, revealed concentration-dependent variability, with larger deviations for some low-to-intermediate adulteration levels and generally closer agreement at higher levels. Approximate LOD and LOQ values of 0.526 and 1.75% (w/w), respectively, were estimated using the conventional σ/slope approach and should therefore be interpreted cautiously because this calculation does not fully account for multivariate prediction uncertainty. GC-FID analysis of pure butter fat and palm oil provided complementary fatty acid compositional information supporting their chemical differentiation but was not used as direct validation of the FTIR-PLSR predictions. Overall, the PCA-guided ATR-FTIR-PLSR approach shows potential as a rapid and nondestructive screening strategy for palm oil adulteration in butter fat; however, further validation using larger, independently sourced sample sets, particularly at low adulteration levels, is required before routine quantitative application.

Food contamination by heavy metals is a major threat to public safety, as these metals tend to bioaccumulate and can cause gastrointestinal, systemic, and neurological disorders. Consequently, demand is increasing for precise, accessible, and low-cost detection of these hazardous metals in food. Electrochemical nanosensors have been developed into highly sensitive analytical devices for detecting trace amounts of heavy metals in complex food systems. In recent years, engineering nanostructured materials has become important, as they enhance electrochemical performance by increasing active surface area and improving electron-transfer kinetics and analyte adsorption at the electrode interface. This review highlights recent developments in nanomaterial-based electrochemical sensors for detecting heavy metals across a wide range of food matrices and water, including dairy products, seafood, fruits and vegetables. It begins by introducing the fundamentals of electrochemical sensors and the techniques used in electrochemical sensing, especially sensing materials, such as carbon nanomaterials, metal and metal oxide nanostructures, and hybrid nanocomposites. These materials have unique physicochemical properties that improve analytical sensitivity. Furthermore, the review compares reported limits of detection and shows that nanostructured sensing platforms are well suited for analyzing trace levels of heavy metals in food systems. This review also examines emerging technologies, including flexible and wearable sensors, AI-assisted sensing, microfluidic approaches, paper-based sensing, and smartphone-integrated sensing.

Graphical Abstract

Laser-induced breakdown spectroscopy (LIBS) and Raman spectroscopy provide complementary fingerprints for geographical classification. This study examined 210 intact Astragali Radix slices from six Chinese producing regions: 180 original and 30 added slices. Raman preceded LIBS at nine corresponding points per slice. A dual-stream convolutional neural network (CNN), support vector machine (SVM), random forest (RF), and partial least-squares discriminant analysis (PLS-DA) were assessed with grouped partitions and inner model selection. Development used 150 original and 30 added slices; 30 original slices formed an internal holdout. Fused CNN accuracy was 95.56% out of fold (OOF), compared with 93.89%, 89.44%, and 91.67% for expanded SVM, expanded RF, and PLS-DA. CNN achieved 96.67% on both the internal holdout and added-batch OOF subset; expanded RF achieved 100.00% and 96.67%, respectively. CNN OOF accuracy was 87.22% for LIBS alone and 73.89% for Raman alone. In source-only analysis, direct added-batch CNN accuracy was 43.33%, rising to 76.67% after unlabeled mean alignment; aligned RF reached 86.67%. Fusion improved classification within represented acquisition conditions, whereas transfer performance and model ranking depended on inference conditions. Added-batch OOF assessment is retrospective and does not establish independent external validity.

Bisphenol A (BPA) has been widely used in epoxy coatings and polycarbonate food-contact materials, whereas bisphenol F (BPF) and bisphenol S (BPS) are common structural alternatives. An 8-min liquid chromatography-Orbitrap high-resolution mass spectrometry method was developed and validated for their simultaneous determination in canned foods and beverages. The method combines parallel reaction monitoring with optimized hydrophilic-lipophilic balanced solid-phase extraction across five representative matrices. Negative heated electrospray ionization and an Accucore RP-MS column were selected. Validation followed the 2025 Eurachem guide. Statistically estimated limits of detection and quantification were 0.39–0.89 and 1.31–2.98 µg/kg for foods or µg/L for beverages, respectively. Apparent recoveries at 5, 25, and 75 µg/kg or µg/L ranged from 70.8% to 103.5%, with relative standard deviations ≤ 8.2%; intermediate-precision relative standard deviations were 3.9% to 6.6%. In 80 Egyptian retail products, BPA was quantified in 77 samples (96.3%), BPF in 16 (20.0%), and BPS in 71 (88.8%). The highest category-mean BPS concentrations were observed in hot dogs and tuna, although several categories had few samples. Current European Union legislation prohibits BPA in relevant food-contact materials, subject to derogations and transitional provisions; therefore, the former 0.05 mg/kg migration limit was not used as a compliance benchmark. The purposive survey design precludes national prevalence estimates or causal attribution to packaging. The method supports rapid multi-bisphenol surveillance, including monitoring of BPA substitutes.

Microplastics (MPs) are ubiquitous environmental pollutants increasingly detected in food matrices, raising concerns about human exposure. Rapid, sensitive, and matrix-tolerant detection methods are needed to monitor MPs in consumables. This study reports the synthesis and characterization of Nile Red carbon dots and their application in paper-based detection strips for identifying low-density polyethylene (LDPE) and polypropylene (PP) microplastics in food. LDPE and PP standards were prepared and confirmed using fluorescence microscopy and Fourier-transform infrared spectroscopy. Selected food matrices, including milk, juice, and cheese, were artificially spiked with MPs, and detection was performed via Nile Red carbon dot–coated strips. Nile Red carbon dots exhibited strong red fluorescence under green-light excitation, excitation-independent photoluminescence, and uniform nanoscale morphology as confirmed by UV–visible spectroscopy, photoluminescence, and transmission electron microscopy (TEM). The paper-based strips demonstrated sensitive, rapid visual screening method of MPs at 100 ppm across all food matrices, with clear differentiation between spiked and blank samples. The platform offers a low-cost, rapid, and portable approach for microplastic screening in diverse foods and may support future standardization of MP detection methods.

Fruit juices are an important part of a balanced diet because they are rich in vitamins, minerals, phenolic compounds, and other bioactive components. Common practices include water dilution, sugar addition, and mixing with lower-cost fruit juices, which may alter product composition and mislead consumers. Apple juice is particularly relevant to authenticity studies because its composition can be modified by these practices. Water, sugar, and orange juice were selected as representative adulterants because they introduce different compositional changes to the juice matrix. Various analytical techniques, including chromatography, spectroscopy, mass spectrometry, and high-field proton nuclear magnetic resonance spectroscopy, have been applied to fruit juice authentication. In this study, time-domain nuclear magnetic resonance (TD-NMR) was used to detect adulterations in apple juice created by the addition of water, orange juice, and sugar. For this purpose, the spin–lattice (T₁) and spin–spin (T₂) relaxation times of the samples were measured, and structural changes associated with adulteration were revealed by evaluating the T₂ distributions through inverse Laplace transform (ILT) analysis applied to Carr-Purcell-Meiboom-Gill (CPMG) data. ILT analysis facilitated the separation of distinct water populations and relaxation components, thereby enabling a more detailed characterization of adulteration-induced structural modifications. The results demonstrated that the relaxation behavior of T₁ and T₂ varied significantly depending on the type of adulteration: water addition increased both T₁ and T₂ times, orange juice addition increased T₁ but decreased T₂, while sugar addition reduced both relaxation times. The occurrence of multiple peaks in the ILT analysis further confirmed that different adulteration types could be differentiated under the tested conditions. The novelty of this study lies in the combined application of T₁, T₂, and ILT analysis to characterize water, orange juice, and sugar adulteration in apple juice. This approach provides complementary relaxation-based information for differentiating the tested adulteration conditions under the investigated experimental settings. As a result, the combined assessment of T₁ and T₂ relaxation measurements with ILT analysis demonstrated that TD-NMR is a rapid, non-destructive, reliable, and highly accurate technique for detecting apple juice adulteration.

A multivariate sensory analytical methodology integrating response surface methodology, descriptive sensory analysis and principal component analysis was developed and validated for optimizing carob-enriched jelly candy formulations. Preliminary experiments compared three carob incorporation strategies in a pectin-based gel matrix, demonstrating that coating with carob flour provided the most suitable technological approach, whereas direct flour incorporation and syrup concentrations above 1% adversely affected product texture. Based on these findings, a central composite design was applied to optimize carob syrup (0.5–1.5%) and coating carob flour (30–70%) levels. Response surface methodology identified odor and texture as the hedonic sensory responses most significantly affected by formulation, with maximum overall acceptability predicted at 0.52% carob syrup and 31.4% coating carob flour. Descriptive sensory analysis revealed brown color intensity and hardness as the principal discriminating attributes among formulations, while carob odor and flavor differentiated only a subset of samples with high panel reproducibility. Principal component analysis further confirmed distinct sensory profiles associated with the combined effects of coating and syrup concentrations. The optimized formulation (desirability = 0.63) exhibited a balanced sensory profile characterized by moderate brown color, good elasticity, and low carob aroma intensity, while maintaining physicochemical and sensory stability during 15 days of storage, with moisture loss below 2% and no perceptible quality deterioration. The study revealed meaningful relationships between formulation parameters and the sensory and hedonic characteristics of carob-based gummy candies, demonstrating the value of a sensory-driven approach to formulation assessment.

Dry matter (DM) content is a crucial quality indicator for vegetables and fruits, yet its conventional determination methods are destructive and time-consuming. Visible and near-infrared (Vis–NIR) spectroscopy offers a rapid, non-destructive alternative, but full-spectrum modeling often suffers from data redundancy and poor interpretability. In this study, a joint wavelength screening strategy combining latent variable magnitude (LVM) and variable importance in projection (VIP) was employed to enhance partial least squares (PLS) prediction of DM content. A total of 250 leek samples were analyzed over the 470–900-nm range, and the same strategy was further applied to broccoli and apple to evaluate its applicability across different crop species. The LVM-VIP approach reduced the number of wavelengths by 85.9% for leek, 75.3% for broccoli, and 71.3% for apple. At the same time, the predictive performance improved dramatically, prediction coefficients of determination (R²_P) increased from 0.6845 to 0.9360 for leek, from 0.7119 to 0.9305 for broccoli, and from 0.6647 to 0.9207 for apple, with corresponding root mean square error of prediction (RMSE_P) reductions to 0.33%, 0.44%, and 0.59%, respectively. The selected wavelengths consistently clustered in the 500–550 nm and 750–850 nm regions across all three crops, corresponding to pigment reflectance and internal light scattering that are indirectly associated with DM content. The proposed LVM-VIP wavelength screening strategy not only significantly enhances model parsimony and predictive accuracy but also proves effective across different crop species, providing a practical strategy framework for developing low-cost, multispectral sensors for real-time quality assessment in the fresh produce industry.