Food Analytical Methods最新文献

The specialty coffee industry relies heavily on manual grading to maintain the ultimate cupping quality of the specialty coffee. This is a subjective and costly process, as this grading is performed by skilled labor. Therefore, this study aims to evaluate the potential of computer vision and machine learning approaches to classify green coffee beans into specialty and defective categories. In this regard, two traditional machine learning models, including random forest (RF) and support vector classifier (SVC), and three deep learning models, including a custom lightweight Convolutional Neural Network (CNN), MobileNetV2, and MobileNetV3, were evaluated on this task. Model performances were assessed using accuracy, precision, recall, F1-score, learning curves, Grad-CAM visualization, and precision-recall analysis. According to the results, the traditional machine-learning models achieved classification accuracies of 98% with RF and 95% with SVC. Similarly, the deep-learning models achieved accuracy values of 99.6% with the lightweight custom CNN, 99.6% with MobileNetV2, and 98.7% with MobileNetV3. Moreover, inference time was tested on a Raspberry Pi 5 to assess the feasibility of real-time deployment capabilities of the models on low-cost edge devices. The results demonstrated ultra-fast inference time of 0.155 ms with SVC compared to RF (1.226 ms). Similarly, average inference time for deep learning models demonstrated 94.811 ms for CNN with custom architecture, 125.144 ms with MobileNetV2, and 115.86 ms with MobileNetV3. Furthermore, this inference time was reduced significantly after the conversion of the models to a TFLite model. Based on overall evaluations of the models, the lightweight CNN with a custom architecture outperformed, maintaining consistent inference time and strong feature interpretability with generalized performance.

Surfactant-induced aqueous two-phase systems (ATPS) offer an environmentally friendly alternative for the separation and preconcentration of analytes, minimizing toxic waste generation and operational costs. In this work, we report for the first time the application of a surfactant-driven ATPS to the simultaneous extraction and preconcentration of cobalt and nickel from food matrices. The system was composed of Triton X-100 + Na₂SO₄ + H₂O in the presence of 4-(2-Pyridylazo)resorcinol (PAR) as the complexing agent, followed by detection via flame atomic absorption spectrometry. Key parameters, including pH, PAR concentration, centrifugation time, and incubation time, were optimized through multivariate analysis based on a desirability function approach. Optimal conditions were pH 9.2, centrifugation time 10 min, thermostatic bath time 11 h, and PAR concentration 0.0750% w/w. Under these conditions, the limits of detection and quantification were 0.330 and 1.10 µg·kg⁻¹ for Co, and 0.0370 and 0.890 µg·kg⁻¹ for Ni, respectively, with enrichment factors of 20.2 and 16.7. The method showed good precision, with RSDs of 6.3% for Co and 7.4% for Ni, and accuracy verified using the certified reference material NIST 1515 (apple leaves), yielding recoveries of 97.6 ± 2.7% for Co and 97.9 ± 2.9% for Ni. In real food samples, recoveries ranged from 96 to 106%, further confirming the reliability of the approach. This novel methodology, by combining micellar extraction with the principles of green chemistry, provides a reliable, cost-effective, and sustainable strategy for trace metal monitoring in food safety applications.

This study presents a voltammetric electronic tongue for the classification and electrochemical characterization of honey from Brazilian native stingless bees, an underexplored and chemically complex food matrix. Ten samples from different species and regions were analyzed using unmodified commercial screen-printed electrodes: carbon (C110), gold cured at high temperature (220AT), gold cured at low temperature (220BT), and platinum (550), under four pH conditions (pure, 2.0, 7.0, and 12.0). Au-AT and Pt were selected for their superior voltammetric definition and classification performance, assessed via principal component analysis (PCA), hierarchical cluster analysis (HCA), and artificial neural networks (ANNs). The most distinctive redox profiles emerged under neutral and alkaline conditions, with peak attribution restricted to the class level (sugars, flavonoids, and phenolic acids). NN models using combined Au-AT and Pt data at pH 7.0 and pH 12.0 achieved 91.7% accuracy in training and 100% in validation, successfully discriminating samples by both geographical origin and bee species. Overall, the minimalist bare-electrode e-tongue combined with AI enabled robust and interpretable classification, offering a powerful tool for the authentication and valorization of native stingless bee honeys.

Graphical Abstract

In this study, the Cu₇Mn₂ bimetallic cluster was synthesized via a solvothermal method using copper (II) chloride dihydrate (CuCl₂·2H₂O) and manganese (II) chloride tetrahydrate (MnCl₂·4H₂O) as the metal centers, and 1,2-cyclohexanediamine-N,N′-bis-(3-carboxylsalicylide) as the ligand. Subsequently, the Cu₇Mn₂ bimetallic cluster fiber coating was prepared using a physical coating method and applied for the solid phase microextraction (SPME) of 7 polycyclic aromatic hydrocarbons (PAHs). The coating was characterized by Fourier transform infrared spectroscopy, scanning electron microscopy, energy dispersive X-ray spectroscopy, thermogravimetric analysis, and contact angle measurement. Based on adsorption isotherms and density functional theory (DFT) calculation, the adsorption mechanism was speculated to be hydrophobic interaction, π-π stacking, and N − H···π interaction. The coating exhibited higher extraction efficiency for 7 PAHs than the Cu cluster, Mn cluster, commercial PDMS, and PDMS/CAR fiber coatings. Under the optimized extraction conditions, an SPME-HPLC method was developed for the determination of 7 PAHs in corn oil, sesame oil, blended oil, and rapeseed oil. The linear ranges of the method were 1.0 − 1000 µg·kg^–1 for naphthalene, 0.1 − 1000 µg·kg^–1 for anthracene, phenanthrene, and pyrene, 3.0 − 1000 µg·kg^–1 for fluorene, 0.3 − 1000 µg·kg^–1 for fluoranthene, and 1.5 − 1000 µg·kg^–1 for 1-methylnaphthalene (R² > 0.99), respectively. The limits of detection (LODs, S/N = 3) and the limits of quantitation (LOQs, S/N = 10) were in the ranges of 0.030 − 1.0 µg·kg^–1 and 0.10 − 3.0 µg·kg^–1, respectively. The precisions (RSDs) were less than 10.0% and the recoveries were in the range of 82.9 − 115.2%. These results demonstrate that the Cu₇Mn₂ bimetallic cluster coating is an effective SPME adsorbent for separation and sensitive determination of PAHs in edible oils.

Graphical Abstract

Sudan dyes, a class of azo dyes, are genotoxic and carcinogenic to humans. However, they are still used illegally to color food products and spices due to their color fastness. In this study, multi-functionalized magnetic silica nanoparticles (C₁₈/DVB/NVP(4:1)-Fe₃O₄@SiO₂ MNPs) were synthesized, and solid–liquid extraction (SLE) coupled with magnetic solid phase extraction (MSPE) was employed for the extraction and preconcentration of five Sudan dyes (Sudan I − IV, Sudan Red 7B), followed by high-performance liquid chromatography/ultraviolet spectroscopy (HPLC–UV) for their detection in powdered chili pepper samples. The SLE conditions for the analytes in powdered chili pepper samples were optimized. Under optimal conditions, the extraction recovery ranged between 78.0% and 106.9%. We investigated and optimized the potential factors that may affect the efficiency of MSPE. Under these optimal conditions, calibration curves showed good linearity (R² > 0.9989) across the tested concentration range of 0.98 to 125 µg/g. The limits of detection for the five Sudan dyes were in the range of 0.06–0.26 µg/g. The intra-day and inter-day accuracy ranged from 93.5% to 106.3% with a relative standard deviation (RSD) of not more than 9.1%. The developed method was successfully applied to analyze ten different commercial brands of powdered chili pepper samples purchased from markets in Wuhan, confirming its feasibility for routine analysis of illegal Sudan dyes in these samples.

Daqu serves as a key saccharifying and fermenting agent in Baijiu production, and its reducing sugar content has a significant impact on the aroma and quality of the liquor. However, traditional methods for measuring reducing sugar content in Daqu are complex and suffer from time lag, which makes monitoring and analyzing reducing sugar levels during fermentation difficult. Given the influence of environmental variables on the reducing sugar content in Daqu, this study aimed to propose a time-series prediction method by combining Daqu microenvironment parameters with a stacked ensemble learning approach. The microenvironment parameters were recorded and analyzed, and RF-Relief was used to identify the most important monitoring points from various environmental variables. Single models, including support vector regression (SVR), eXtreme Gradient Boosting (XGBoost), random forest (RF), and ridge regression (RR), as well as a stacking ensemble model (Stacking), were established, and their prediction accuracies for reducing sugar content were compared. The results showed that the stacking model significantly outperformed the single models, achieving R² values of 0.9610, 0.9661, and 0.9501 for the upper, middle, and lower layers, respectively. This represented an improvement by 17.08%, 11.58%, and 16.47% over SVR; by 20.09%, 15.23%, and 14.86% over XGBoost; by 10.12%, 6.05%, and 9.09% over RF; and by 7.80%, 10.60%, and 15.81% over RR. The proposed model enabled accurate real-time prediction of reducing sugar content, thereby overcoming the limitations of traditional methods and providing reliable data support for the intelligent regulation of the Daqu fermentation process.

A novel approach combining a cold probe with dispersive liquid-liquid microextraction based on solidification of floating organic drop (DLLME-SFO) was developed to simplify the retrieval of organic solvents from aqueous sample surfaces. Custom-designed cooled probes containing a 1:1 water-ethylene glycol mixture enabled efficient and rapid solvent collection without the need for cooling the sample. The method was applied to quantify propargite in fruit samples using the gas chromatography-mass spectroscopy technique. Extraction parameters were optimized using both one-variable-at-a-time and simplex methods, yielding optimal conditions (1 mL dispersing solvent, 200 µL extraction solvent, 2 min extraction time, 5 min centrifugation, 50 °C, no salt addition). The method achieved a limit of detection of 0.051 µg mL⁻¹, a relative standard deviation of 8.6%, and an average extraction recovery of 103.9% across six replicate analyses. The technique demonstrated robust applicability for determining propargite in real fruit samples, offering a simple and effective approach for trace-level pesticide analysis.

Ziziphi Spinosae Semen (ZSS) is the seed of Ziziphus jujuba Mill. var. spinosa (Bunge) Hu ex H. F. Chou. As a food and medicine homologous herbal medicine, it was extensively employed to treat insomnia for the safety and significant therapeutic effects. To establish a comprehensive quality evaluation method for ZSS, this study proposed an integrated strategy combining chemometrics and weighted TOPSIS-GRA fusion model. Ultra-high performance liquid chromatography tandem quadrupole time-of-flight mass spectrometry (UHPLC-Q-TOF–MS) was employed to identify 96 chemical components in ZSS. Thereafter, a high-performance liquid chromatography-triple quadrupole tandem mass spectrometry (HPLC-QQQ-MS/MS) method was developed for the simultaneous quantification of 22 compounds. Multivariate statistical analysis was applied to compare different batches of ZSS samples, and a multi-index comprehensive evaluation system was constructed by the technique for order preference by similarity to ideal solution (TOPSIS) and gray correlation analysis (GRA) fusion model analysis to achieve quantitative ranking and grading of ZSS quality. The results demonstrated that this strategy could effectively evaluate the quality differences of ZSS and identify six key differential markers, including magnoflorine, 6‴-feruloylspinosin, jujuboside A, jujuboside B, spinosin, and frangufoline. The comprehensive evaluation model developed in this study can provide reference for the quality control and standardization of ZSS.

In many countries, palm jaggery is a widely used traditional sweetener. The adulteration has damaged its reputation and negatively impacted the livelihoods of those directly involved in its production. Beyond its traditional medicinal uses, adulterated jaggery can pose serious health risks. Among the available analytical techniques, Raman spectroscopy stands out as one of the most effective methods for detecting adulteration and verifying the authenticity of jaggery in the market. An attempt was made to use portable Raman spectroscopy to study the authentication of jaggery and the identification of the adulterant. Palm jaggery was bought on a laboratory scale as pure, adulterated with sugar at different concentrations. Thus, 3 samples of authentic palm jaggery (smoked, non-smoked, and dried) and 9 samples of adulterated palm jaggery were analyzed. The spectral peak at 636 cm⁻¹ is the deforming vibration peak of C–C–O and C–C–C bonds, which indicates the presence of fructose. The peak at 846 cm⁻¹ represents the strong deformation peak of C–C–C and C–C–O, which indicates the presence of sucrose with higher intensity. The peak at 398 cm⁻¹ Raman shift indicates the strong stretching of the C–O bond, which represents the glucose. SNR values take the investigation of adulteration to the next level. The novel idea of the work was to identify palm jaggery adulteration using a portable Raman spectrometer (785 nm).

Graphical Abstract

Ensuring the authenticity of grapevine cultivars is crucial for traceability, fraud prevention, and consumer protection. In the Republic of North Macedonia, viticulture and wine production have a long-standing tradition and remain an important pillar of the country´s agricultural sector. Recently, the Macedonian government adopted a ten-year National Strategy for the Development of Viticulture and Winemaking to ensure authenticity and sustainability of wine production. Traditional identification of grape cultivars relied on ampelographic methods which are strongly affected by environmental conditions. The analysis of simple sequence repeat (SSR) markers by capillary electrophoresis has become the standard for grape cultivar identification, but is time-consuming and cost-intensive. In this study, we assessed nine SSR markers recommended by the Organisation Internationale de la Vigne et du Vin for their suitability to differentiate eleven grape cultivars commonly grown in North Macedonia using PCR combined with high-resolution melting (HRM) analysis. Compared to capillary electrophoresis, HRM reduces time and cost, as it is carried out directly after PCR in the same instrument without requiring any pretreatment steps. Among the markers tested, assays targeting VrZAG62 and VrZAG79 showed the highest discriminatory power, distinguishing 50 and 53 of 55 cultivar pairs, respectively. The complex melting curves indicate that the amplified regions differ not only in repeat number but also in base composition. Preliminary experiments demonstrate that the discriminatory power can be even enhanced by performing multiplex PCR-HRM. Our findings indicate that PCR-HRM is a reliable screening method for grape cultivar authentication, supporting food authenticity testing and protection of local viticulture.