{"title":"Molecular, structural, and functional effects of corn fiber gum hydrolysate containing arabinoxylooligosaccharides in pasta: Implications for digestibility and probiotic growth","authors":"Adieh Anvar, Amin Karimi, Mahshad Davoudi, Fataneh Hashempour-Baltork, Hassan Ahmadi Gavlighi","doi":"10.1016/j.lwt.2026.120128","DOIUrl":"10.1016/j.lwt.2026.120128","url":null,"abstract":"<div><div>Although corn fiber gum is a dietary fiber with functional properties, the application of enzymatically hydrolyzed corn fiber gum containing arabinoxylooligosaccharides (AXOS) in pasta remains underexplored, particularly regarding processing performance and potential prebiotic activity. This study evaluated hydrolyzed corn fiber gum (HCFG) containing AXOS in pasta at 2, 4, and 6 g/100 g and assessed its effects on cooking quality, color, texture, microstructure, <em>in vitro</em> digestibility, and probiotic growth. HCFG increased total dietary fiber from 2.08% to 5.73% at 6 g/100 g. Increasing HCFG levels decreased water absorption and swelling indices by 10.9–22.0% and 10.5–17.8%, respectively, while cooking loss increased by 7.5–16.7%. Texture analysis showed increased hardness from 0.40 to 0.50 N and decreased gumminess from 0.29 to 0.28 N, cohesiveness from 0.72 to 0.57, and chewiness from 0.98 to 0.83 mJ. Fourier-transform infrared spectroscopy showed decreased β-sheet content from 36.9% to 22.2% and increased random coil content from 8.6% to 12.2%, indicating partial disruption of gluten organization. At 120 min, <em>in vitro</em> starch digestibility decreased by 16.5%, 27.9%, and 32.7% in HCFG2, HCFG4, and HCFG6, respectively, while protein hydrolysis increased from 48.2% to 62.1%. Pasta containing 6 g/100 g HCFG supported probiotic growth, reaching 9.8 and 10.8 log<sub>10</sub> CFU/mL for <em>Lactobacillus casei</em> and <em>Lactobacillus plantarum</em>, respectively. Overall, HCFG containing AXOS shows promise for developing fiber-fortified pasta while valorizing corn processing by-products.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120128"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364788","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Xinyao Yan, Xi Li, Le Du, Huihua Fan, Ge Fang, Zhanat Bauyrzhankyzy Duisembi, Shaohui Ouyang
{"title":"Structural and functional enhancement of oat bran dietary fibre by sequential extrusion–alkaline–enzymatic pretreatment","authors":"Xinyao Yan, Xi Li, Le Du, Huihua Fan, Ge Fang, Zhanat Bauyrzhankyzy Duisembi, Shaohui Ouyang","doi":"10.1016/j.lwt.2026.120162","DOIUrl":"10.1016/j.lwt.2026.120162","url":null,"abstract":"<div><div>This study investigated extrusion, alkaline, enzymatic, and sequential combined pretreatments of oat bran dietary fibre (ODF) using a three-stage physical-chemical-biological modification strategy. Compared with untreated ODF, the combined pretreatment markedly increased the proportion of soluble dietary fibre, with SDF increasing from 6.29% to 13.47%. A multi-scale characterisation framework further demonstrated reduced cellulose, hemicellulose, and lignin contents, disruption of the compact fibre matrix, decreased particle size, increased specific surface area, and enhanced surface roughness and porosity, together with altered hydrogen bonding, preserved polysaccharide backbone, and increased relative crystallinity and higher thermal transition temperatures. Functionally, the combined pretreatment enhanced oil-holding capacity, flowability, glucose adsorption, glucose dialysis retardation index, α-amylase inhibition, cholesterol and bile salt binding, and antioxidant activities. Multivariate analyses further explored structure-function associations, indicating that increased SDF, reduced particle size, enlarged specific surface area, and increased relative crystallinity were collectively associated with the observed functional changes. Overall, the sequential combined pretreatment provides an effective strategy for producing modified ODF for potential use as a functional food ingredient.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120162"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364789","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Cynthia D. Garambas, Michael Karl M. Barnuevo, Melchor S. Lumiked, Maravilla M. Senado, Jenelyn A. Martin
{"title":"Sensory, safety, and physicochemical characterization of Pandan (Pandanus amaryllifolius) - Flavored cocoa tablets","authors":"Cynthia D. Garambas, Michael Karl M. Barnuevo, Melchor S. Lumiked, Maravilla M. Senado, Jenelyn A. Martin","doi":"10.1016/j.lwt.2026.120087","DOIUrl":"10.1016/j.lwt.2026.120087","url":null,"abstract":"<div><div>The Cordillera Administrative Region (CAR) is experiencing significant growth in its cacao industry, prompting a need for value-added innovation. This study focused on developing raw cocoa flavored with pandan into tablets, aiming to diversify the product offerings and enhance the industry's economic potential. The developed pandan-flavored cocoa tablets underwent rigorous quality assessments. Organoleptic characterization confirmed desirable sensory properties, indicating strong consumer appeal. Physicochemical composition (crude fat, moisture, and water activity) analyses demonstrated high repeatability and reproducibility, affirming the product's consistent nutritional profile and suitability for consumption. Crucially, comprehensive microbiological testing verified the product's safety, with <em>E. coli</em>, aerobic plate, yeast, and mold counts all within acceptable limits (<10 CFU/g), and the complete absence of Salmonella in 25 g samples. These findings collectively validate the quality and safety of the innovative pandan-flavored cocoa tablets. This research underscores the significant potential for such value-added products to bolster CAR's burgeoning cacao industry, effectively integrating local agricultural resources with market-ready food innovation.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120087"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364793","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Understanding the physicochemical properties, multi-scale structure, and in vitro digestibility of sweet potato starch noodles supplemented with fucoidan","authors":"Guanghui Li, Jialu Gao, Hanmeng Jiang, Shengli Zhao, Qianran Hu, Lining Ge, Junjie Yuan, Yonghui Wang, Shenghua He, Xueli Gao, Weiyun Guo, Jihong Huang","doi":"10.1016/j.lwt.2026.120180","DOIUrl":"10.1016/j.lwt.2026.120180","url":null,"abstract":"<div><div>This study aimed to explore the effects of fucoidan on the physicochemical properties, multi-scale structure, and in vitro digestibility of sweet potato starch noodles (SPSNs). The underlying interaction mechanisms between SPS and fucoidan were also investigated. Results showed that fucoidan decreased the rehydration time (18.75-15.25 min), rehydration ratio (365.38-341.09%), and expansion ratio (165.75-137.70%) of SPSNs, whereas the broken rate (0-10%) and cooking loss (0.11-0.18%) increased. In addition, the hardness, chewiness, springiness, and gumminess of SPSNs were improved with the addition of 100-200 g/kg fucoidan. Moreover, Fourier transform infrared spectroscopy and X-ray diffraction analyses revealed that fucoidan elevated the ratio of 1047 to 1022 cm<sup>−1</sup> (1.79-2.92) and 995 to 1022 cm<sup>−1</sup> (3.41-5.52), and the crystallinity (4.41-6.58%) of SPSNs. Scanning electron microscope demonstrated that SPSNs with fucoidan exhibited a porous structure with larger pore size and greater pore density. These structural alterations led to an increase in slowly digestible starch (SDS) (14.65-28.87%) and SDS + RS (68.58-81.32%), and a decrease in rapidly digestible starch (31.28-18.68%), without affecting the resistant starch (RS) content. The enthalpy, peak temperature, pasting temperature, final viscosity, and setback of SPSNs increased upon fucoidan addition, whereas the peak viscosity, trough viscosity, and breakdown decreased. In addition, principal component analysis and Pearson correlation analysis confirmed that the cooking properties and in vitro digestibility were significantly correlated with the starch structure. Molecular dynamics simulations further revealed that fucoidan bound to sweet potato starch via hydrogen bonding and electrostatic interactions. This study provides new insights into the potential of fucoidan as a functional ingredient for developing low-glycemic starch-based foods.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120180"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364265","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nalinpat Chalermchotiwong, Yajie Ma, Ziyue Han, Han Zhang, Yiru Yan, Xiang Luo, Stefano Cucurachi, Justin Z. Lian, Renshi Li, Bin Dong, Boyang Yu
{"title":"Water-based extraction of garlic phenolic compounds: Integrating life cycle assessment and machine learning for quality and sustainability optimization","authors":"Nalinpat Chalermchotiwong, Yajie Ma, Ziyue Han, Han Zhang, Yiru Yan, Xiang Luo, Stefano Cucurachi, Justin Z. Lian, Renshi Li, Bin Dong, Boyang Yu","doi":"10.1016/j.lwt.2026.120144","DOIUrl":"10.1016/j.lwt.2026.120144","url":null,"abstract":"<div><div>Garlic polyphenols are high-value bioactive compounds widely sought for functional food and nutraceutical applications, yet their production via water-based extraction involves trade-offs between extract quality, process efficiency, and environmental burden. This study presents an integrated gate-to-gate life cycle assessment (LCA) and machine learning (ML) framework to simultaneously evaluate and optimize both food-quality performance and environmental sustainability across three water-based extraction technologies: hot water extraction (HWE), ultrasound-assisted extraction (UAE), and subcritical water extraction (SWE). Under the experimental conditions tested, SWE produced the highest total phenolic content (31.7 mg GAE/g extract) and strongest antioxidant activity (SC<sub>50</sub> = 0.48 mg/mL). Gate-to-gate LCIA confirmed that SWE also carries the lowest environmental impacts across 15 midpoint categories, with electricity consumption driving more than 80% of climate change impacts across all methods. ML regression with SHAP analysis identified temperature (HWE) and solid-to-liquid ratio and time (UAE) as the key parameters governing both phenolic recovery and energy use. ML optimization reduced environmental impacts for HWE and UAE by 66% and 73%, respectively, while preserving extract quality, narrowing the environmental gap between optimized UAE and SWE from 11–13-fold to 2.5–3-fold. These findings provide data-driven guidance for garlic polyphenols extraction process selection, with the acknowledgment that all results are based on laboratory-scale data and require validation at pilot or industrial scale before commercial application.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120144"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The level of methyl esterification of soluble pea polysaccharide affects the interactions with caseins at low pH","authors":"Himari Ogura, Norifumi Adachi, Hiroko Ashida, Sayuri Sakurada, Masayuki Shibata, Munehiko Asayama, Takashi Yarita, Akihiro Nakamura","doi":"10.1016/j.lwt.2026.120129","DOIUrl":"10.1016/j.lwt.2026.120129","url":null,"abstract":"<div><div>Interactions between casein particles and pea soluble polysaccharides with different degrees of esterification (HM-SPPS, DE 71%; LM-SPPS, DE 25%) were investigated during GDL-induced acidification and compared with high-methoxyl pectin (HMP, DE 75%). HMP effectively stabilized acidified milk only after acidification, whereas its addition before acidification promoted viscosity increase and curd formation. HM-SPPS showed little effect on acid-induced casein aggregation. In contrast, LM-SPPS adsorbed onto positively charged casein particles during acidification, suppressing particle growth and viscosity development. SEM observations revealed a dense network of finely dispersed casein particles in LM-SPPS-containing systems. In acidified milk beverages, LM-SPPS addition before acidification produced smaller particles (0.73 μm) than addition after acidification (1.5 μm). The apparent thickness of the adsorbed LM-SPPS layer (28–37 nm) was comparable to its molecular size (32 ± 6 nm), suggesting monolayer adsorption on casein particle surfaces. These results demonstrate that LM-SPPS stabilizes acidified milk through a mechanism distinct from that of HMP and is a promising stabilizer for acidified dairy beverages.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120129"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364419","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Sara Razaghi, Kooshan Nayebzadeh, Kiana Hasibi, Seyede Marzieh Hosseini, Leila Mirmoghtadaie, Amir Mohammad Mortazavian
{"title":"Fine-tuning the properties of whey protein isolate-inulin binary hydrogel-based bigels: Exploring the effects of protein-polysaccharide ratio and pH","authors":"Sara Razaghi, Kooshan Nayebzadeh, Kiana Hasibi, Seyede Marzieh Hosseini, Leila Mirmoghtadaie, Amir Mohammad Mortazavian","doi":"10.1016/j.lwt.2026.120161","DOIUrl":"10.1016/j.lwt.2026.120161","url":null,"abstract":"<div><div>Bigels structured with protein-polysaccharide binary hydrogels are promising for developing fat substitutes with tunable properties. However, the application of whey protein isolate (WPI)-inulin (INU) binary hydrogels in bigels remains unexplored. This study developed and characterized WPI-INU/beeswax bigels by integrating WPI-INU binary hydrogels with a beeswax oleogel. A Box-Behnken design was employed to systematically investigate the linear and combined effects of the WPI:INU mass ratio (20:80–80:20), pH (3.5–9.5), and oleogel-to-hydrogel (OG:HG) mass ratio (40:60–70:30) on oil binding capacity (OBC), water holding capacity (WHC), elastic modulus in the linear viscoelastic region (G'<sub>LVR</sub>), gel strength (parameter A), and work of shear. Regression models highlighted that increasing the WPI:INU mass ratio improved OBC but reduced G'<sub>LVR</sub>, which could be attributed to intensified WPI aggregation at higher protein concentrations. INU served as a water-binding and viscosity-enhancing component in bigels, while the effect of the WPI:INU mass ratio on WHC was dependent on pH. Acidic pH enhanced OBC, whereas alkaline pH strengthened bigel structures. An elevated OG:HG mass ratio enhanced G'<sub>LVR</sub>, parameter A, and work of shear. The optimal formulation (BG-OPT) showed an OBC of 91.62 ± 0.46%, a WHC of 99.12 ± 1.15%, and a parameter A of 78580 ± 3200 Pa s<sup>1/z</sup>, along with enhanced viscoelasticity and thermal characteristics compared with the WPI-based bigel. These findings provide insights into tailoring WPI-INU/beeswax bigel properties through the systematic variation and optimization of formulation variables for the development of fat-reduced food products.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120161"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364421","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Xiaomin Zang, Binghong Gao, Ying Yang, Xiao Chuan Huang, Chuanyan Rao, Yanlin Liu, Li Li, Jian Sun, Dongqing Ye
{"title":"Expanding the oenological potential of ‘Lingfeng’ (Vitis quinquangularis Rehd hybrid) wine: Product- and dose-dependent quality modulation by yeast derivative products","authors":"Xiaomin Zang, Binghong Gao, Ying Yang, Xiao Chuan Huang, Chuanyan Rao, Yanlin Liu, Li Li, Jian Sun, Dongqing Ye","doi":"10.1016/j.lwt.2026.120171","DOIUrl":"10.1016/j.lwt.2026.120171","url":null,"abstract":"<div><div>‘Lingfeng’, an interspecific <em>Vitis</em> hybrid cultivar, produces deeply colored and fruity red wine, but high acidity and weak aroma expression constrain its sensory balance. Post-fermentation application of yeast derivative products (YDPs) provides winemakers with an additional opportunity to address specific quality limitations that remain in the resulting base wine. Accordingly, four compositionally distinct commercial YDPs (MP, SV, OS, and SF) were applied at 0.1 and 0.3 g/L and evaluated after 15 d for physicochemical properties, color, electronic-tongue responses, apparent viscosity, volatile compounds, and sensory attributes. Relative to the control, all treatments increased aroma intensity by 14%–41% and reduced bitterness by 16%–35% while the basic physicochemical profile remained broadly stable. MP increased apparent viscosity by 18%–19% without a statistically significant increase in perceived mellowness. Additionally, 0.3 g/L SV reduced bitterness by 35% but presented 14% lower aroma intensity than 0.1 g/L. SF at 0.3 g/L reduced sourness, astringency, and total phenolics (by 26%), but increased higher alcohols to 405 mg/L β-Damascenone increased by 30%–67% across treatments. Among the tested combinations, OS at 0.3 g/L showed the most balanced profile across the wine characteristics evaluated, increasing aroma intensity (by 41%) and total phenolics (by 23%), while reducing sourness and astringency. These findings indicate that matching YDP formulation and dose enables targeted post-fermentation modulation of quality parameters and sensory intensity, providing a practical strategy to adjust the sensory balance and expand the oenological potential of ‘Lingfeng’.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120171"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364792","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Multi-scale protective effects of static magnetic field-assisted freezing on chicken breast during repeated freeze-thaw cycles: From quality attributes to gel properties","authors":"Xin Chen, Jichao Huang, Zimu Li, Jinxue Hou, Jingxin Sun, Ming Huang","doi":"10.1016/j.lwt.2026.120194","DOIUrl":"10.1016/j.lwt.2026.120194","url":null,"abstract":"<div><div>The repeated freeze-thaw (F-T) cycles in cold chain logistics are the key factor leading to the deterioration of frozen poultry meat quality. However, conventional freezing methods are limited in mitigating this damage. Herein, we employed static magnetic field-assisted freezing (SMFF) as a physical strategy to address this challenge. This study investigated the effects of SMFF at different intensities (3, 6, 9 mT) on the physicochemical quality, myofibrillar protein (MP) stability, and gelation properties of chicken breast after multiple F-T cycles. Results showed that SMFF effectively alleviated the cumulative damage caused by F-T cycles, with 3 mT showing the best protective effects among the tested intensities (3, 6, and 9 mT). At the tissue level, 3 mT SMFF preserved microstructural integrity, thereby improving water-holding capacity, as evidenced by a 41% reduction in thawing loss after the final F-T cycle. At the molecular level, the treatment inhibited MP oxidation and stabilized the protein conformation, as evidenced by a reduced random coil ratio and enhanced endogenous fluorescence during repeated F-T cycles. Consequently, MP from the 3 mT treatment group exhibited superior processing potential, with gel hardness and water-holding capacity significantly increased (<em>P</em> < 0.05). Overall, SMFF effectively alleviates the quality decline caused by F-T cycles and maintains the processing potential of chicken breast, providing a promising strategy and theoretical reference for quality preservation in the frozen chicken industry.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120194"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364399","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Jointing thermal imaging and one-dimensional convolutional neural network for detecting different adulterations in cow ghee","authors":"Mohammad Hossein Nargesi, Kamran Kheiralipour","doi":"10.1016/j.lwt.2026.120166","DOIUrl":"10.1016/j.lwt.2026.120166","url":null,"abstract":"<div><div>In this study, the feasibility of using thermal imaging combined with machine learning and deep learning algorithms to detect and quantify adulteration in cow ghee was investigated. Pure cow ghee and adulterated samples containing different proportions of margarine, sheep tail fat, and vegetable oil were prepared under controlled laboratory conditions. Thermal images were acquired before and after cooling, and color and textural features were extracted for classification. Multilayer perceptron, support vector machine, random forest, and convolutional neural network models were evaluated to compare their performance in identifying adulterant types and levels. The convolutional neural network model achieved the highest classification accuracies, reaching 99.65%, 99.65%, and 99.46% for sheep tail fat, vegetable oil, and margarine adulteration, respectively. support vector machine achieved accuracies of 98.81%, 98.52%, and 97.28%, whereas multilayer perceptron yielded 96.49%, 97.33%, and 95.60%, respectively. Random forest showed comparatively lower accuracies, ranging from 80.59% to 86.47%. Analysis of the confusion matrices showed that most classification errors occurred between adjacent adulteration levels. These findings demonstrate the potential of thermal image-based analysis for distinguishing compositional differences associated with multiple adulterant types and concentrations in cow ghee. By comparing conventional machine learning models with a deep learning approach within the same experimental framework, this study provides a comparative assessment of their suitability for ghee adulteration analysis. The proposed approach may offer a rapid and non-destructive screening strategy for food authenticity assessment, although further validation under broader sample and operating conditions is required before practical implementation.</div></div>","PeriodicalId":382,"journal":{"name":"LWT - Food Science and Technology","volume":"259 ","pages":"Article 120166"},"PeriodicalIF":7.1,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149364266","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}