Food Analytical Methods最新文献

Coconut oil (CO) is commonly known to have health benefits and thus is a commonly used functional oil in numerous consumer products. Nevertheless, CO is rather expensive and highly demanded, which leads to economic profit through the adulteration with cheaper, low-quality vegetable oils. This adulteration does not only affect the quality of the oil but also creates great health and safety hazards for consumers. Conventional ways of identifying such adulteration tend to be time-consuming, are toxic, as well as complex in the preparation of the sample. Consequently, this necessitates the need to have faster, precise, and environmentally friendly methods of conducting analytical procedures to identify CO adulteration. This study developed an advanced method for processing CO using HSI camera and deep learning method for prediction of adulteration in CO. The proposed model named TKRnet, which is composed of a Transformer for feature enhancing in data, KBest for feature selection and Random Forest for prediction. Raw spectral data is transformed by the transformer and spectral indices with statistical features are integrated to give a rich feature set. This proposed architecture has several levels of feature engineering in which there is the calculation of normalized difference indices and statistical descriptors, and dimensionality reduction with SelectKBest to provide optimal feature selection. The designed features are fed into the RF model and this guarantees a powerful and effective performance in the regression and classification processes. The method of preprocessing can enhance the predictive accuracy and interpretability of the models because it retains meaningful spectral information at a low dimensionality. The performance appraisals show little variation with the 8 chosen features explaining 99.65 maximum performance of the model. The findings indicate that the TKRnet is better than the traditional models in their R² score, root mean squared error (RMSE), and mean absolute error (MAE), and the proposed model has the highest predictive power. The experiment underscores the significance of integrating high-level feature selection and transformation with the state-of-the-art ensemble-based learning algorithms such as the Random Forest to get high predictive accuracy. We find that Random Forest models that are enhanced using transformers work well on tasks with high performance and robustness, especially on complex and high-dimensional data.

The intelligent management and reasonable nutrition of food depend on the continuous improvement of cultivar discrimination technology. A model for food classification and detection based on image deep learning is proposed, which is used to discriminate hundreds of commonly used foods such as fruits, vegetables, and meat. We use smartphones to collect food images and form a dataset under uncontrolled lighting and camera parameters such as focal length and camera stability. It also discriminates partially occluded objects with similar features and achieves good results. Testing on a detection dataset yielded a mAP value of 94.35% and an inference time of 27.53 ms, with a slight increase in parameter scale. Further enhancements in feature fusion and loss function testing showed that IDNet, integrating both BiFPN and WIOU, achieved the best overall performance with a mAP value of 96.24% and an inference time of only 29.86 ms. Experiments are conducted using diverse foods placed in the refrigerator, confirming its superiority.

Lycopene, a lipophilic carotenoid, is widely recognized for its health-promoting bioactivities and broad industrial applications. Tomatoes are the richest source of lycopene, and their processing by-products are also abundant in this valuable pigment. This study proposes a green extraction approach for recovering lycopene from tomato peel by-products and is the first to evaluate acetophenone as a novel solvent for eco-friendly lycopene extraction. Intermittent irradiation microwave-assisted performance was applied to intensify lycopene extraction by acetophenone. In comparison with conventional lycopene extraction, microwave-assisted extraction (MAE) exhibited a higher extraction yield (95.37%) and a reduced extraction time (10 min). The single-factor test was employed to determine the optimal conditions for the lycopene extraction process using the acetophenone-MAE method. Response surface methodology (RSM) was then applied to optimize the extraction process. A four-factor Box-Behnken design (BBD) evaluated the impact of solid-liquid ratio (2–2.5%, w/v), microwave power (300–600 W), extraction time (15–25 min), and On/Off pulsed ratio (30–50 s). Extraction yield was significantly (p < 0.05) influenced by microwave irradiation. The yield increased to 99.04% with a moderate extension of the extraction time to 20 min and a significant reduction in microwave power and duty cycle to minimum values. The novelty of this work lies in the simultaneous study of the use of acetophenone and the effect and interdependencies of MAE parameters to improve lycopene extraction efficiency. The results of this research provide into the development of innovative and environmentally sustainable methods for lycopene extraction, with potential applications in various industrial sectors.

Accurate and reproducible determination of sodium chloride (NaCl) in processed meats is critical for product quality, safety, and regulatory compliance. This study compared the classical Mohr titration method with the Dicromat-II conductivity-based analyser for quantifying NaCl in dry-fermented sausages. Validation was performed across three matrices: aqueous NaCl standards, “added salt” pepperoni homogenates, and commercial pepperoni products. Both methods exhibited excellent linearity (R² > 0.99) in standard and “added salt” matrices. However, the Dicromat-II demonstrated consistently higher precision (%RSD) and superior matrix tolerance, maintaining robust accuracy across all validation tiers. In commercial products, Mohr titration showed substantial variability, with weak correlation and poor agreement in most samples. The Dicromat-II, by contrast, offered reliable performance, reduced chemical handling, and strong repeatability, supporting its use in industrial meat analysis. This study represents the first direct validation-based comparison of these two methods in fermented meat matrices, highlighting the Dicromat-II as a practical and scalable alternative to traditional titration.

Fish consumption is rapidly increasing worldwide, generating significant by-products and waste that offer opportunities for reuse as food, cosmetic, or pharmaceutical ingredients aimed at a sustainable economy. This study evaluated the effect of simulated gastrointestinal digestion (GID) on the generation of antioxidant peptides from blue whiting muscle as well as from by-products hydrolysates obtained by alcalase hydrolysis and ultrasound-assisted alcalase hydrolysis. A comprehensive peptide profile was obtained by nano- or ultra-high-performance liquid chromatography coupled to mass spectrometry, identifying 1246 medium- and 415 short-sized peptides. In this context, GID led to a significant reduction in the number of identified medium-sized peptides in all samples. Concurrently, the quantity of identified short peptides increased in the digested muscle, whereas it remained largely unchanged in the by-product hydrolysates. Peptides containing proline, aspartate, histidine, tryptophan, or glycine residues showed high resistance to GID, and in silico analyses predicted 36 short-sized sequences as bioactives. Results of in vitro assays showed that antioxidant activity generally increased after GID when measured by ABTS radical-scavenging capacity and oxygen radical absorbance capacity assays, whereas it decreased in DPPH free radical–scavenging and ferric-reducing antioxidant power assays. These findings underline the antioxidant potential of peptides derived from both blue whiting muscle and by-products, which could be valorised as a sustainable source of functional ingredients, promoting circular economy and reducing waste in aquaculture.

The existence of residues of aminoglycoside antibiotics, including gentamicin and streptomycin, in meat products is a significant risk to consumer health and a serious food safety concern. In this study, we developed a novel, fast, simple, and sensitive microplate immunoassay based on dual-color fluorescent carbon dot-linked antibody (MIA-DFCDs) for simultaneous detection of aminoglycoside antibiotics in meat samples. Blue and green fluorescent carbon dots were covalently labeled to gentamicin and streptomycin antibodies (B-FCDs_Gen and G-FCDs_Strep), respectively, and utilized as fluorescent probes in a competitive assay format. The assay achieved high sensitivity with very low detection limits (LOD, 0.32 ng/mL for gentamicin and 0.13 ng/mL for streptomycin) and exhibited excellent precision and selectivity. In meat sample analysis (chicken, pork, and beef), the analytical performances including recovery rate (88.9–118.1%) and relative standard daviation (2.9–9.6% RSD) were comparable to a standard ultra-high-performance liquid chromatography tandem mass spectrometry (UHPLC-MS/MS), demonstrating the accuracy and reliability of the proposed assay. In addition, the assay is faster (dual detection within 35 min), simpler, and more cost-effective than UHPLC-MS/MS. These advantages make it a promising tool for the routine monitoring of aminoglycoside antibiotic residues to ensure meat product safety.

Graphical Abstract

Phenolic compounds from elderberry (Sambucus nigra L.) were extracted using a green, water-based method assisted by β-cyclodextrin (β-CD). Central composite design was applied to evaluate the effects of three key parameters: β-CD concentration (1–3% w/v), liquid-to-solid (L/S) ratio (10–30% w/v), and extraction time (30–90 min). Multiple responses were modeled, including total phenolic content (TPC), flavonoids (TFC), total monomeric anthocyanin and total proanthocyanin contents (TMAC, TPAC), and antioxidant activity (DPPH, FRAP, CUPRAC). TPC followed a quadratic trend, while TFC peaked around 2% β-CD. TMAC and TPAC increased linearly across the tested ranges. Among antioxidant assays, DPPH showed the highest sensitivity to process conditions, whereas FRAP and CUPRAC were less affected. Time had a relatively minor influence within the tested interval. Optimal extraction was achieved at moderate L/S ratios and ~ 2% β-CD, balancing efficiency while minimizing dilution and mass transfer limitations. Overall, β-CD-assisted extraction is an effective, low-solvent approach for recovering elderberry phenolics. These findings support future studies on inclusion mechanisms, stability, and scale-up through techno-economic and life-cycle assessments.

The concept of blending oils has emerged as a rational approach to formulate oils with optimized fatty acid ratios, superior stability, and improved nutritional and functional properties. This study aimed to develop a stable oil blend of eri pupal oil (EPO), a sustainable source of bioactive lipids with significant therapeutic benefits and flaxseed oil (FO), and to characterize it using Fourier-transform infrared spectroscopy (FTIR), gas chromatography–mass spectroscopy (GC-MS), and antimicrobial assays. FTIR analysis confirmed the presence of key functional groups such as carbonyl (C=O), alkene (=CH), and methylene (CH₂), providing insights into the structural composition of the blend. GC-MS analysis demonstrated a reduction of α-linolenic acid (ALA) from 83.89% to 76.92%, with increases in palmitic acid from 7.76% to 16.62% and oleic acid from 0.93% to 2.49%, reflecting significant alterations in the fatty acid profile that may influence nutritional quality, oxidative stability, and functional properties. The antimicrobial efficacy was assessed against Staphylococcus aureus, Salmonella typhi, Candida albicans, and Aspergillus niger via the disc diffusion assay, while the minimum inhibitory concentration was determined by the broth dilution method. Antimicrobial assays revealed the strongest inhibitory activity against A. niger, with a zone of inhibition measuring 9.33 ± 2.31 mm and a minimum inhibitory concentration of 20% (v/v). These findings highlight the potential of insect oil–based blends as functional ingredients for future nutraceutical and pharmaceutical applications.

Graphical Abstract

A decanol–tetrahydrofuran supramolecular solvent (SUPRAS)–based vortex-assisted salt-saturated microextraction method coupled with UV–Vis spectrophotometry was developed for the determination of quercetin in real samples. The extraction efficiency was significantly enhanced by salt saturation, resulting in a 45.5% increase in analytical signal compared with the same SUPRAS-based procedure without salt addition. The effects of extraction solvent volume, pH, tetrahydrofuran volume, centrifugation speed and time, and vortex time were optimized using response surface methodology and one-variable-at-a-time approaches. Under optimal conditions, the method exhibited a linear dynamic range of 0.001–5.0 mg L⁻¹, with limits of detection and quantification of 0.28 µg L⁻¹ and 0.97 µg L⁻¹, respectively. An enrichment factor of approximately 136 was achieved. The proposed method showed good precision (RSD < 3.7%) and accuracy. Applicability was confirmed by analysis of fish, apple and tomato samples, yielding recoveries in the range of 96.9 to 100.0%. Owing to its simplicity, low solvent consumption, and satisfactory analytical performance, this method provides an efficient and environmentally friendly approach for routine spectrophotometric determination of quercetin.

The aroma characteristics of Jiangxiang-type baijiu are primarily formed during the third to fifth production rounds. Given the complexity of its aroma compounds, quality assessment currently depends mainly on manual sensory evaluation, with limited objective and scientific methodologies available. This study presents a rapid classification approach using atmospheric pressure solid analysis probe-time of flight (ASAP-TOF) mass spectrometry that requires no sample pretreatment. The method enables real-time discrimination of three aroma types (Jiangxiang-type, Chuntian-type, and Jiaodi-type) and two quality grades (grade I/II), with the entire workflow from sample introduction to classification completed within minutes. A total of 233 baijiu samples, validated by sensory evaluation from the production enterprise, were divided into seven representative groups. Mass-to-charge ratio (m/z) data were acquired using ASAP-TOF mass spectrometry to identify characteristic ions. A recognition model for aroma type and quality grade classification was then developed using LiveID software with principal component analysis and linear discriminant analysis, and its robustness was assessed through fivefold cross-validation. The model achieved a classification accuracy of 90.37% for grade I Jiangxiang-type baijiu from rounds 3 to 5, and demonstrated applicability to unknown samples. Analysis of the m/z data revealed several differential peaks. Based on comparison with literature data and gas chromatography-mass spectrometry (GC-MS) results, guaiacol, 2-methylfuran, 2,3-dimethylpyrazine, and isovaleric acid were identified as potential key flavor compounds distinguishing grade I Jiangxiang-type baijiu across different production rounds. In summary, this study establishes a real-time classification model for grade I Jiangxiang-type baijiu, providing a chemometric tool applicable to quality control and process optimization.