Neural-network-integrated electronic nose system for identification of spoiled beef

A commercially available Cyranose-320™ conducting polymer-based electronic nose system was used to analyse the headspace from fresh beef strip loins (M. Longissimus lumborum) stored at 4° and 10 °C. The raw signals obtained from the electronic nose system were pre-processed by various signal-processing techniques to extract area-based features. Principal component analysis was subsequently performed on the processed signals to further reduce the dimensionalities. Classification models using radial basis function neural networks were developed using the extracted features. The performance of the developed models was validated using leave-1-out cross-validation method. The developed models classified meat samples stored at two storage temperatures into two groups, i.e., “unspoiled” (microbial counts<6.0 log₁₀ cfu/g) and “spoiled” (microbial counts ?6.0 log₁₀ cfu/g). Maximum total classification accuracies of 100% were obtained for both the samples stored at 10 and 4 °C. Classification models based on “Area scaled” feature showed higher accuracies than that obtained using “Area unscaled feature.”     

Thermal imaging for detecting fungal infection in stored wheat

The feasibility of using an infrared thermal imaging system to identify the fungal infection in stored wheat was studied. Thermal images of bulk wheat grains infected by Aspergillus glaucus group, Aspergillus niger van Tieghem and Penicillium spp. were obtained using an un-cooled focal planar array type infrared thermal camera after heating grain for 180 s with a plate heater placed 10 mm above the grain and maintained at 90 °C, and then cooling in ambient air for 30 s. In total, twelve temperature features were derived from heated and cooled wheat and four-way and pair-wise classification models were developed by linear and quadratic discriminant analyses (LDA and QDA). Leave-one-out and bootstrapping methods were used to validate the developed classification models. Pair-wise LDA and QDA classification models gave a maximum accuracy of 100% for healthy samples and more than 97% and 96% for infected samples, respectively. Four-way LDA and QDA classification models yielded relatively low classification accuracies for fungus-infected samples due to the non-significant changes in the temperature features between grain samples infected with different species of fungi.     

Investigation of Different Gas Sensor-Based Artificial Olfactory Systems for Screening <Emphasis Type="Italic">Salmonella typhimurium</Emphasis> Contamination in Beef

Two different electronic nose systems (metal oxide and conducting polymer based) were used to identify Salmonella typhimurium contaminated beef strip loin samples (stored at two temperatures). The sensors present in the two systems were ranked based on their Fisher criteria of ranking to evaluate their importance in discriminant analysis. The most informative sensors were then used to develop linear discriminant analysis and quadratic discriminant analysis-based classification models. Further, sensor signals collected from both the sensor systems were combined to improve the classification accuracies. The developed models classified meat samples based on the Salmonella population into “No Salmonella” (microbial counts < 0.7 log₁₀ cfu/g) and “Salmonella inoculated” (microbial counts ≥ 0.7 log₁₀ cfu/g). The performances of the developed models were validated using leave-1-out cross-validation. Classification accuracies of 80% and above were observed for the samples stored at 10 °C using the sensor fusion approach. However, the classification accuracies were relatively low for the meat samples stored at 4 °C when compared to the samples stored at 10 °C. The results indicate that the electronic nose systems could be effectively used as a first stage screening device to identify the meat samples contaminated with S. typhimurium.     

Metal oxide semiconductor sensors for monitoring of oxidative status evolution and sensory analysis of virgin olive oils with different phenolic content

An electronic nose based on an array of six metal oxide semiconductor sensors was used to monitor the oxidative status of virgin olive oils (VOO) during storage. VOO samples, with and without phenolic compounds, were stored at 60 °C for 7 weeks. Once a week, absorbance at 232 and 270 nm, oxidized stability index, electronic nose, and sensory analysis were evaluated. Linear discriminant analysis models were constructed in order to classify samples according to oxidative levels. Based on these models, VOO samples with and without phenolic compounds at different storage times, divided in eight categories, were correctly classified also achieving a good correlation for sensory analysis. The method is a fast and economical tool for on-line monitoring of VOO oxidation status.     

Fumigation with essential oil of mustard retards fungal growth and accumulation of ergosterol and free fatty acid in stored shelled groundnuts

Shelled groundnut (Arachis hypogaea) samples with moisture contents (m.c.) between 7.5 and 10.5% and inoculated with conidia of Aspergillus glaucus and A. parasiticus were stored for 15-90 days at 25 ± 2 °C, and fumigated with synthetic food grade essential oil of mustard (100 μl/l space). Deterioration of the samples was assessed by estimating the percentage of kernels colonized by fungi, the number of colony forming units (CFUs)/kernel, and the accumulation of ergosterol and free fatty acids (FFA). The values of these variables increased with the m.c. and storage period, independent of the fumigation treatment; however, the rate of increase was significantly lower in fumigated samples. After 90 days storage, the proportion of kernels yielding A. glaucus was similar in all samples, but the number of CFUs was 300×, ergosterol content 3.6× and FFA 4× higher in non-fumigated than in fumigated samples. In fumigated samples, no molded kernels were visible, while many were seen in non-fumigated samples after 30 or 60 days storage at 10.5 or 9.3% m.c., respectively. The deterioration retardation featured a reduced inoculum on kernel surfaces. Aspergillus parasiticus did not colonize kernels independently of m.c. and fumigation treatments. There was a strong positive correlation between CFUs and ergosterol or FFA content when the data of fumigated and non-fumigated samples were analyzed separately. However, this relationship was absent when data were pooled to disregard the fumigation effect. The correlation between ergosterol and FFA content remained high regardless of the fumigation treatment (r = 0.99). The ergosterol or FFA content of stored groundnuts can be used interchangeably, as a sensitive indicator, to assess deterioration caused by xeric storage fungi. However, the latter was preferable because it was simpler to assess and provided a direct indication of economic losses due to reduced oil yield.     

电子鼻判别不同储藏条件下糙米品质的研究

采用德国Airsense 公司生产的PEN3 型电子鼻系统对不同储藏条件下的糙米进行分析检测。通过对传感器响应值进行PCA、LDA 方法的分析,发现PCA 和LDA 均能准确判别出不同水分含量的糙米;PCA、LDA 方法均可判别不同温度储藏的糙米样品,LDA 方法呈现出良好的集中性和单向趋势;LDA 可以很好的区分不同氧气体积分数储藏的糙米样品,并根据氧气体积分数的不同呈现出明显的规律,但总贡献率要低于PCA 方法。另外,通过方差分析发现不同储藏条件对电子鼻响应值的影响大小有所差异,顺序为水分条件＞温度＞氧气体积分数,水分和温度存在交互作用。另外,采用Loadings 分析方法可以得知传感器W5C、W1S 在检测中起到的作用最大,可以对电子鼻的传感器进行优化与选择,根据不同的具体条件选择适当的传感器阵列组合。     

Volatiles for mycological quality grading of barley grains: determinations using gas chromatography-mass spectrometry and electronic nose

The possibility of using an electronic nose or gas chromatography combined with mass spectrometry (GC-MS) to quantify ergosterol and colony forming units (CFU) of naturally contaminated barley samples was investigated. Each sample was split into three parts for (i) ergosterol and CFU analysis, (ii) measurements with the electronic nose and (iii) identification of volatiles collected on an adsorbent with a GC-MS system. Forty samples were selected after sensory analysis to obtain 10 samples with normal odour and 30 with some degree of off-odour. The data set of volatile compounds and the data collected from the electronic nose were evaluated by multivariate analyse techniques. SIMCA classification (soft independent modelling of class analogy) was used for objective evaluation of the usefulness of the data from the GC-MS or electronic nose measurements for classification of grain samples as normal or with off-odour. The main volatile compounds of grain with normal odour were 2-hexenal, benzaldehyde and nonanal, while 3-octanone, methylheptanone and trimethylbenzene were the main volatile compounds of grain with off-odours. Using data from the electronic nose three samples of 40 were misclassified, while data analysis of the volatile compounds detected with the GC-MS, led to six misclassified samples. Regression models (partial least-squares, PLS) were built to predict ergosterol- and CFU-levels with data from the GC-MS or electronic nose measurements. PLS models based on both GC-MS and electronic nose data could be used to predict the ergosterol levels with high accuracy and with low root mean square error of prediction (RMSEP). CFU values from naturally infected grain could not be predicted with the same degree of confidence.     

Distribution of fungi and aflatoxins in a stored peanut variety

The objective of the present study was to evaluate the mycoflora and occurrence of aflatoxins in stored peanut samples (hulls and kernels) from Tupã, State of São Paulo, Brazil. The samples were analyzed monthly over a period of one year. The results showed a predominance of Fusarium spp. (67.7% in hulls and 25.8% in kernels) and Aspergillus spp. (10.3% in hulls and 21.8% in kernels), and the presence of five other genera. The growth of Aspergillus flavus was mainly influenced by temperature and relative humidity. Analysis of hulls showed that 6.7% of the samples were contaminated with AFB₁ (mean levels = 15–23.9 μg/kg) and AFB₂ (mean levels = 3.3–5.6 μg/kg); in kernels, 33.3% of the samples were contaminated with AFB₁ (mean levels = 7.0–116 μg/kg) and 28.3% were contaminated with AFB₂ (mean levels = 3.3–45.5 μg/kg). Analysis of the toxigenic potential revealed that 93.8% of the A. flavus strains isolated were producers of AFB₁ and AFB₂.     

Monitoring beer during storage by fluorescence spectroscopy

The present study demonstrates the use of fluorescence spectroscopy and multivariate data analysis for monitoring changes in beer during storage. Total luminescence spectra and synchronous scanning fluorescence spectra were recorded for fresh beer and beer samples stored in clear glass vials for three weeks in darkness at 4 °C and at 22 °C, and exposed to light at 22 °C, respectively. A pronounced decrease of fluorescence features ascribed to riboflavin was observed in samples exposed to light as compared to those kept in the dark. Principal component analysis of synchronous scanning fluorescence spectra revealed clear clustering of samples according to storage conditions. Successful classification of differently stored samples was accomplished using both the nearest neighbour method (kNN) and linear discriminant analysis (LDA).     

Monitoring water uptake in whole barley (Hordeum vulgare L.) grain during steeping using near infrared reflectance spectroscopy

The aim of this study was to evaluate the use of near infrared reflectance (NIR) spectroscopy to monitor water uptake and steeping time in whole barley samples as a rapid and easy to use technique. Whole barley grain samples were steeped in water, and subsamples were analyzed for water uptake (gravimetric method) and using NIR spectroscopy. The spectra and the analytical data were used to develop partial least squares (PLS) calibrations to predict water uptake and steeping time. Cross validation models for water uptake and steeping time gave a coefficient of determination in cross validation (R²) and the standard error of prediction (SEP) of 0.90 (SEP = 5.36 g/100 g fw) and 0.92 (SEP = 3.93 h), respectively. This study showed that NIR spectroscopy combined with PLS regression showed promise as a rapid, non-destructive method to monitor and measure water uptake and steeping time in whole barley during soaking.     

Development of a low cost optimized handheld embedded odor sensing system (HE-Nose) to assess ripeness of oranges

Presented work addresses the development and application of artificial olfactory (e-nose) as an efficient nondestructive handheld system to extract quality attributes of variety of orange cultivars at various cultivated lands. An ARM-9 (S3C2440 controller) based advance embedded electronic nose system has been developed for on-site odor acquisition, processing and ripeness level prediction for various orange cultivars. Developed handheld electronic nose system is light weight, low power, and easy operable to every consumer. Various statistical multivariate data analysis techniques (PCA, LDA, QDA, and KNN) implementation on electronic nose measurements helped estimating optimal harvest dates for various orange cultivars. Developed handy system has been optimized in the sense of selectivity, sensitivity of gas sensors using the implementation of temperature control for heating element of the gas sensors. Various temperature, humidity models also have been developed to improve the performance of developed system in different environmental conditions. Data acquisition process has been performed using developed system for 100 orange samples have two different cultivars (mandarin and sweet orange). Extracted parameters have been subjected towards supervised Levenberg–Marquardt back-propagation algorithm to training and testing of developed handheld electronic nose system to predict quality parameters of oranges. A good correlation has been found between developed handheld electronic nose system signals and quality attributes indicators and it shows that system can successfully detect and predict various quality parameters of orange samples. Results of developed handheld electronic nose system have been validated with commercial standard electronic nose Alpha Fox 3000 system and it has been observed that efficiency varies around 94 ± 0.3 %.     

Sterol and vitamin D2 concentrations in cultivated and wild grown mushrooms: Effects of UV irradiation

Raw and processed mushroom samples including wild grown (chanterelles and king bolete) and cultivated samples (white and brown button, portabella, shiitake, oyster) were bought from the food market and analysed for sterols and vitamin D₂ contents. The different commercial mushrooms selected are abundant in almost every Swedish and European supermarket or outdoor market. Ergosterol was the most abundant sterol found in mushrooms present in somewhat higher concentrations in cultivated mushrooms (4.0-5.0 mg/g dry matter (d.m.)) than those found in wild mushrooms (1.7-3.5 mg/g d.m.). In addition, three closely related minor sterols, including ergosta-7,22-dienol, ergosta-5,7-dienol, and ergosta-7-enol were identified. Chanterelles and king bolete were found to be good sources of vitamin D₂ (0.7-2.2 μg/g d.m.) compared with cultivated mushrooms that had a low content (< 0.1 μg/g d.m.). Canned samples of Agaricus bisporus/white were slightly lower in ergosterol and vitamin D₂ compared to fresh samples. Irradiation with UV light in the A region (366 nm) only slightly affected ergosterol and vitamin D₂ content. In contrast, irradiation with UV light conducted in the C region (254 nm, 0-2 h, 20 cm distance) for fresh white button mushrooms and freeze-dried chanterelles resulted in nonsignificant decrease in ergosterol content, whereas vitamin D₂ increased up to 9-fold (Cantharellus tubaeformis) and 14-fold (A. bisporus/white), respectively.     

Polycyclic aromatic hydrocarbons (PAHs) in two Spanish traditional smoked sausage varieties: “Androlla” and “Botillo”

The polycyclic aromatic hydrocarbon (PAH) content in two traditional smoked sausages from Spain was determined. Determination and quantification of PAHs in smoked sausages were performed by HPLC with fluorescence detection. Results showed, that total mean levels of PAHs found were higher in “Androlla” (36.45 µg/kg) than in “Botillo” (29.39 µg/kg) although no significant differences (P < 0.05) were observed. In all examined samples content of phenanthrene was the highest in the two traditional sausage varieties. The maximum level for benzo[a]pyrene (BaP) of 5 µg/kg in smoked meat products was not exceeded in any samples. BaP represented 1.3% and 1.2% in “Androlla” and “Botillo” samples, respectively of the total sum of the 15 PAHs investigated in both sausages. Correlation statistic analysis (P < 0.01) showed that BaP was a good marker for 6 IARC possible and probable carcinogenic PAHs in “Androlla” samples (R_BaP/6IARC = 0.63) and in “Botillo” samples (R_BaP/6IARC = 0.96).     

Norharman and harman in instant coffee and coffee substitutes

Norharman and harman are two neuroactive β-carbolines present in several plants and thermally processed foods, including roasted coffee. The objective of this study was to evaluate their amounts in commercial instant coffee-based and coffee substitute beverages (n = 48), with variable amounts of coffee, chicory, barley, malt, and rye. All samples contained variable amounts of both β-carbolines, always with a higher proportion of norharman than harman. The highest levels (p < 0.01) were found in 100% instant coffees, with mean amounts of 3.8 μg/g and 1.5 μg/g for NH and H, respectively, followed by plain chicory. The lowest amounts were found in 100% barley, with 1.1 μg/g for NH and 0.3 μg/g for H. The NH and H content was statistically higher (p < 0.01) in all mixtures with coffee when compared to those without it. Nevertheless, and except for 100% barley, the amounts provided by beverages based on coffee substitutes are within those reported for standard coffee brews.     

Use of a MS-electronic nose for prediction of early fungal spoilage of bakery products

A MS-based electronic nose was used to detect fungal spoilage (measured as ergosterol concentration) in samples of bakery products. Bakery products were inoculated with different Eurotium, Aspergillus and Penicillium species, incubated in sealed vials and their headspace sampled after 2, 4 and 7 days. Once the headspace was sampled, ergosterol content was determined in each sample. Different electronic nose signals were recorded depending on incubation time. Both the e-nose signals and ergosterol levels were used to build models for prediction of ergosterol content using e-nose measurements. Accuracy on prediction of those models was between 87 and 96%, except for samples inoculated with Penicillium corylophilum where the best predictions only reached 46%.     

Survival of human derived Lactobacillus plantarum in fermented cereal extracts during refrigerated storage

The aim of this study was to investigate the survival of a potentially probiotic Lactobacillus plantarum strain in barley, wheat and barley malt extracts. The extracts were produced from three flour/water suspensions, i.e., 5%, 20%, 30% w/w. After inoculation, the cultures were incubated for 24 h at 37 °C, and were subsequently stored at 4 °C for up to seventy days. The lactic acid and reducing sugar concentrations at the beginning of storage were significantly different between the fermented media, ranging from 0.5 g/L to 17 g/L and from 0.8 g/L to 6.5 g/L respectively, while the pH ranged between 2.9 and 3.4. It was observed that the cells survived much better in the malt extracts compared to barley and wheat extracts during refrigerated storage. Based on the results from a study using model media and supplemented cereal extracts it was derived that this was most likely due to their higher sugar concentration and the presence of protective unidentified compounds, albeit the fact that the malt extracts contained higher amounts of lactic acid.     

RELATIONSHIP BETWEEN FUSARIUM INFESTATION OF BARLEY AND THE GUSHING POTENTIAL OF MALT

Fifty barley samples, displaying a range of 0 to 100% kernels infested with Fusarium, were collected in North Dakota, South Dakota, and Minnesota during the harvest of 1994. Samples were micromalted, and the levels of the fungal metabolites, deoxynivalenol and ergosterol, were determined. Fusarium infestation and the levels of fungal metabolites were evaluated as predictors of gushing in laboratory trials. Malt samples which were infested with Fusarium or contaminated with the fungal metabolites exhibited a propensity to gush. However, only the levels of deoxynivalenol and ergosterol were found to be strongly correlated with the actual amount of gushing observed. This suggests that their production may parallel that of the component which actual causes gushing, and that screening barley and malt for these metabolites, may offer a means of reducing gushing problems in the brewery. Determination of deoxynivalenol is rapid, when it is present, and necessary because of food safety and malt quality concerns.     

Mid infrared spectroscopy and multivariate analysis: A tool to discriminate between organic and non-organic wines grown in Australia

The combination of mid infrared (MIR) spectroscopy and multivariate analysis was explored as a tool to classify commercial wines sourced from organic (ORG) and non-organic (NORG) production systems. Commercial ORG (n = 57) and NORG (n = 115) red and white wine samples from 13 growing regions in Australia were analysed using a MIR spectrophotometer. Discriminant models based on MIR spectra were developed using principal component analysis (PCA), discriminant partial least squares (DPLS) regression and linear discriminant analysis (LDA). Overall, the LDA models based on the PCA scores correctly classified on average, more than 75% of the wine samples while the DPLS models correctly classified more than 85% of the wines belonging to ORG and NORG production systems, respectively. These results showed that MIR combined with discriminant techniques might be a suitable method that can be easily implemented by the wine industry to classify wines produced under organic systems.     

Traceability of olive oil based on volatiles pattern and multivariate analysis

An automated head-space solid-phase microextraction (HS-SPME)-based sampling procedure, coupled to gas chromatography–ion trap mass spectrometry (GC–ITMS), was developed and employed for fast characterisation of olive oil volatiles. In total, 914 samples were collected, over three production seasons, in north-western Italy—Liguria (n = 210) and other regions—in addition to the rest of Italy, Spain, France, Greece, Cyprus, and Turkey (n = 704) with the aim to distinguish, based on analytical (profiling) data, the olive oils labelled as “Ligurian” (protected denomination of origin region, PDO) from all the others (“non-Ligurian”). For the chemometric analysis, linear discriminant analysis (LDA) and artificial neural networks with multilayer perceptrons (ANN-MLP) were tested. Employing LDA, somewhat lower recognition (81.4%) and prediction (61.7%) abilities were obtained. The classification model was significantly improved using ANN-MLP. Under these conditions, the recognition (90.1%) and prediction (81.1%) abilities were achieved. The diagnostic value of the data obtained by one-dimensional GC–ITMS were compared with those generated by two-dimensional gas chromatography–time-of-flight mass spectrometry (GC × GC–TOFMS), allowing a comprehensive analysis of olive oil volatiles.