Human Sensory Evaluation Research Articles

Research on the Evaluation of Baijiu Flavor Quality Based on Intelligent Sensory Technology Combined with Machine Learning

Baijiu, one of the world’s six major distilled spirits, has an extremely rich flavor profile, which increases the complexity of its flavor quality evaluation. This study employed an electronic nose (E-nose) and electronic tongue (E-tongue) to detect 42 types of strong-aroma Baijiu. Linear discriminant analysis (LDA) was performed based on the different production origins, alcohol content, and grades. Twelve trained Baijiu evaluators participated in the quantitative descriptive analysis (QDA) of the Baijiu samples. By integrating characteristic values from the intelligent sensory detection data and combining them with the human sensory evaluation results, machine learning was used to establish a multi-submodel-based flavor quality prediction model and classification model for Baijiu. The results showed that different Baijiu samples could be well distinguished, with a prediction model R2 of 0.9994 and classification model accuracy of 100%. This study provides support for the establishment of a flavor quality evaluation system for Baijiu.

Identification of key aroma‐active compounds in long shelf‐life yoghurt and milk produced by different pre‐sterilisation processes of raw milk using SBSE‐GC × GC–O–MS

SummaryAroma is a vital attribute of dairy products. This study employed a combination of human sensory evaluation and instrumental analysis to study the effect of two different pre‐sterilisation processes (A and B) on the odour characteristics of milk and long shelf‐life yoghurt. The results from the sensory evaluation and electronic nose analysis indicated that the pre‐sterilisation process significantly influences the aroma profile and compounds of the samples. Stir bar sorptive extraction coupled with comprehensive two‐dimensional gas chromatography−olfactometry−mass spectrometry (SBSE‐GC × GC−O−MS) was applied to identify the aroma‐active compounds present. Aldehydes and ketones emerged as the most abundant aroma‐active compounds. Quantitative analysis showed variations in the concentrations of aroma‐active compounds. For example, higher concentrations of hexanal, butanoic acid and hexanoic acid were observed in long shelf‐life yoghurt and milk produced by process B than process A, while higher concentrations of nonanal, 2,3‐butanedione and 3‐hydroxy‐2‐butanone were detected in products from process A than process B. Initial odour activity values provided insight into the contribution of aroma‐active compounds to the overall odour perception. Based on the results of aroma recombination, omission and spiking experiments, 13 and 11 key aroma‐active compounds were identified in long shelf‐life yoghurt and milk, respectively. These compounds, including 2,3‐butanedione, 2,3‐pentanedione, 2‐heptanone, 2‐undecanone, hexanal, octanal, nonanal, decanal, furfural, butanoic acid, hexanoic acid, 3‐hydroxy‐2‐butanone and styrene, can serve as potential indicators for determining odour differences between long shelf‐life yoghurt (milk) samples. Consequently, they offer a chemical basis for manufacturers to make informed decisions regarding production processes.

Grade identification of rice eating quality via a novel flow-injection voltammetric electronic tongue combined with SFFS-BO-SVM

Given the importance and complexity of grade identification for rice eating quality, this study aimed to use a self-established flow-injection voltammetric electronic tongue (FIE-tongue) combined with SFFS-BO-SVM to replace human sensory evaluation. A novel FIE-tongue was firstly designed and established in this work, and the optimal rice pretreatment conditions were determined by fractional factorial design (FFD) which used for FIE-tongue detection. Based on optimized detection data, the optimized features and hyperparameters of 16 samples in three eating quality grades for SVM model were obtained by Sequential Floating Forward Selection (SFFS) algorithm and Bayesian Optimization (BO) algorithm, respectively. Compared to the original SVM model established by all features, the accuracy of SFFS-BO-SVM model for different grades were all higher than 92%, and have the increasement of 5.5%, 18.8%, and 11.4%, respectively. Therefore, the FIE-tongue combined with SFFS-BO-SVM method should be promising for rice eating quality grading and other food quality control.

Comparative assessment of Riesling wine fault development by the electronic tongue and a sensory panel.

Wine faults threaten brand recognition and consumer brand loyalty. The objective of this study was to compare the acuteness of e-tongue and human sensory evaluation of wine fault development in Riesling wine over 42days of storage. Riesling wines uninoculated (control) or inoculated with 104CFU/mL cultures of Wickerhamomyces anomalus, Acetobacter aceti, Lactobacillus brevis, or Pediococcus parvulus were assessed every 7days with the e-tongue and a rate-all-that-apply (RATA) sensory panel. After 7days of storage, the e-tongue detected differences in all four wine spoilage microorganism treatments, compared to control wine, with discrimination indices over 86%. The RATA sensory panel detected significant differences beginning on day 35 of storage, 28days after the e-tongue detected differences. This study showed that the e-tongue was more sensitive than the human panel as a detection tool, without sensory fatigue. PRACTICAL APPLICATION: This research is useful for winemakers seeking additional instrumental methods in the early detection of wine faults. Given the results of this study, the e-tongue can be a useful tool for detecting early chemical changes in white wines that have undergone microbial spoilage, providing winemakers with time to mitigate faults before they surpass sensory thresholds.

Characterization of flavor quality deterioration of postharvest Chinese bayberry (Myrica rubra cv. Dongkui) at different storage temperatures

The present study characterized the flavor quality deterioration of ‘Dongkui’ Chinese bayberry (Myrica rubra) stored at different temperatures. The fruit were stored at 0 °C, 4 °C, 10 °C and 20 °C for 12 d. The quality deterioration was characterized by combining human sensory evaluation and artificial sensory evaluation, which included e-nose and e-tongue. The sensory quality deterioration was characterized in the aspects of taste strength, bitterness, and odor, in which the odor was found to be most related to the sensory quality deterioration. In the artificial sensory evaluation, 12 e-nose sensors showed differential responses with the elongation of storage time and temperature. Based on the human sensory evaluation and the response values of the 12 e-nose sensors, the stored fruit were classified, with those stored at 20 °C for 2 d or longer and those stored at 10 °C for 7 d or longer classified as “off-flavored” fruit. The artificial neural network (ANN) model verified the classification. A total of 27 volatile compounds were identified by HS-SPME–GC–MS. Eight differential volatiles were screened out with p < 0.05 and variable importance in projection (VIP) > 1 by orthogonal partial least squares discriminant analysis (OPLS-DA). Among these volatiles, (Z)-3-hexenoic acid methyl ester and (E,E)-2,4-hexadienal decreased, while octanoic acid, acetaldehyde, ethanol, phenylethyl alcohol, 2-pentylfuran and benzaldehyde increased in the deteriorated fruit.

Colloidal study on the texture of sous vide (SV)-processed Tibetan chicken breast with human-machine collaborated electromyography analysis

Sous Vide (SV) is a potential meat thermal processing technology. However, the texture evolution of SV processed meat is not easy to understand due to the complex hierarchical structure of meat. In our research, a two-level meat colloidal model was carried out to analyze the texture of SV-processed Tibetan chicken breast. Moreover, electromyography analysis was used to collaborate the texture results between human sensory evaluation and machine detection. Although the denaturation and dehydration of sub-continuous phase cytoskeletal proteins are unavoidable even under 55 °C at 1 h or 65 °C at 0.5 h heating, the main-continuous phase collagen can achieve slow denaturation and sufficient conversion to gelatin if SV is applied. Different from collagen under high-temperature processing, gelatin does not exert shrinkage stress to the sub-continuous phase upon SV. Moreover, insignificant lateral contraction was revealed by differential scanning calorimetry (DSC) and scanning electron microscopy (SEM). Thus, limited water migration and minimized cooking loss were achieved because of the integrity of the colloid system upon SV. Furthermore, physical properties proved the tenderization effect of SV. Thus, our results showed that the colloidal model combined with an electromyography study was useful in revealing the textual detail of pure meat or imitation meat.

Pre-processing technique of Aquilaria species from Malaysia for four different qualities

The paper interprets data distribution by using boxplot pre-processing in classify the quality of Agarwood oil for eleven chemical substances into four different qualities. The varieties usage of Agarwood oil makes it considered as an expensive and valuable product on the essential oil market. Perfumes, fragrances, incense, aromatherapy, and traditional medicine are the most popular Agarwood oil applications. However, the classification of Agarwood oil grades does not yet have standard grading method. This because it has been graded manually into different qualities by using human sensory evaluation. Boxplot analysis involving eleven chemical subtances that will be focusing in this study by concerned the quality for low, medium low, medium high and high. ɤ-eudesmol, ar-curcumene, β-dihydro agarofuran, ϒ-cadinene, α-agarofuran, allo aromadendrene epoxide, valerianol, α-guaiene, 10-epi-ɤ-eudesmol, β-agarofuran, and dihydrocollumellarin compounds are the selected significant compounds that represent the input for boxplot. Agarwood oil consist 660 data samples from low, medium low, medium high, and high quality. The result in this study showed that the four selected significant compounds (ɤ-eudesmol, 10-epi-ɤ-eudesmol, β-agarofuran, and dihydrocollumellarin) are important as a marker for Agarwood oil quality classification. The identification of chemical substances on high quality done as reference for future research studies.

Machine learning based classification of yogurt aroma types with flavoromics

Traditional sensory evaluation, relying on human assessors, is vulnerable to subjective error and lacks automation. Nonetheless, the complexity of human sensation makes it challenging to develop a computational method in place of human sensory evaluation. To tackle this challenge, this study constructed logistic regression classification models that could predict yogurt aroma types based on aroma-active compound concentrations with high classification accuracy (AUC ROC > 0.8). Furthermore, indicator compounds discovered from feature importance analysis of classification models led to the derivation of classification criteria of yogurt aroma types. Through constructing and analyzing machine learning models on yogurt aroma types, this study provides an automated pipeline to monitor sensory properties of yogurts.

Formation and conversion of flavour related compounds of thermal cooked Ruditapes philippinarum soup

SummaryThermal reactions affect flavour, but how different thermal reactions influence the flavour of cooked soup has rarely been investigated. Herein, the effect of heat treatment (0–50 min) at different temperatures (120, 140 and 160 °C) on Ruditapes philippinarum soup was studied. The acceptable heating of Ruditapes philippinarum soup was assigned at 140 °C for 30 min by human sensory evaluation and compound analysis. In addition, the detailed relationship between the formation of key volatile compounds and heating temperature was revealed by partial least squares discriminant analysis (PLS‐DA) and the odour coding method. A total of 28 key volatile compounds were selected according to the results of variables important in the projection (VIP) and odour activity value (OAV). Whereas 3‐ethyl‐2,5‐dimethyl pyrazine and 2‐heptylfuran were the most important volatiles produced only in the thermal reaction at 140 °C and associated with cocoa and roasted nut odours. These results indicated the relationship between thermal reactions and flavour formation. The odour coding obtained from OAV could be as a reference benchmark for heating seafood soup products.

Establishment of a new cell-based assay to quantitatively evaluate the sweetness of sugar and sugar alcohol

Sweetness is a crucial indicator for identifying sweeteners used for flavor regulation and nutritional matching of foods. This study established a new quantitative sweetness evaluation method based on a combination of transiently transfected sweet taste receptor cells (STRs) and human sensory evaluation. Here, nine different sweeteners were evaluated and 10 sensory difference intensities and their corresponding concentrations were obtained through sensory evaluation. Using the same concentration solution as in the sensory evaluation, the dosage-response curve and EC50 of the nine sweeteners were obtained by analyzing the intracellular calcium signal fluctuation of STRs. Data analysis revealed that the calcium signal intensity and sweetness intensity under the same concentration of sweetener presented a strong linear correlation (R2 > 0.9). Additionally, we found a favorable correlation (R2 = 0.8746) between the relative sweetness and EC50. These findings indicate that the quantitative evaluation of sweetness based on STRs could be a potential alternative to human sensory evaluation.

Synthesis of N'-[(2,4-dimethoxyphenyl)methyl]-N-(2-pyridin-2-ylethyl) oxamide and evaluation of its taste-enhancing effect

The development of efficient and safe taste-enhancing compound is important to reduce sodium intake, at the same time maintains food palatability. Hence, in this study, an excellent and safe taste-enhancing compound (N'-[(2,4-dimethoxyphenyl)methyl]-N-(2-pyridin-2-ylethyl) oxamide (DE)) was synthesized in 96% yield under optimized conditions. The chemical structure of DE was clearly characterized by 1H and 13C NMR spectrum. Human sensory evaluation showed that the addition of DE (0.9 ppm) significantly enhanced the kokumi, umami, and salt tastes, the degree of enhancement by 66.4%, 56.6%, and 26.2%, respectively. The result of electronic tongue was consistent with the human sensory evaluation, which was again illustrated that DE is a promising taste-enhancer. Time-intensity (TI) results revealed that the sample (with addition of 0.9 ppm DE) umami duration (210.0 ± 0 s) were significantly prolonged that of control group (150.0 ± 0 s). Moreover, the DE could bind and activate calcium-sensing and T1R1/T1R3 receptors, inducing stronger umami and kokumi tastes. This work revealed the potential of DE as an excellent taste enhancer.

TastePeptides-EEG: An Ensemble Model for Umami Taste Evaluation Based on Electroencephalogram and Machine Learning.

In the field of food, the sensory evaluation of food still relies on the results of manual sensory evaluation, but the results of human sensory evaluation are not universal, and there is a problem of speech fraud. This work proposed an electroencephalography (EEG)-based analysis method that effectively enables the identification of umami/non-umami substances. First, the key features were extracted using percentage conversion, standardization, and significance screening, and based on these features, the top four models were selected from 19 common binary classification algorithms as submodels. Then, the support vector machine (SVM) algorithm was used to fit the outputs of these four submodels to establish TastePeptides-EEG. The validation set of the model achieved a judgment accuracy of 90.2%, and the test set achieved a judgment accuracy of 77.8%. This study discovered the frequency change of α wave in umami taste perception and found the frequency response delay phenomenon of the F/RT/C area under umami taste stimulation for the first time. The model is published at www.tastepeptides-meta.com/TastePeptides-EEG, which is convenient for relevant researchers to speed up the analysis of umami perception and provide help for the development of the next generation of brain-computer interfaces for flavor perception.

The Integration of Metabolomics, Electronic Tongue, and Chromatic Difference Reveals the Correlations between the Critical Compounds and Flavor Characteristics of Two Grades of High-Quality Dianhong Congou Black Tea.

Tea's biochemical compounds and flavor quality vary depending on its grade ranking. Dianhong Congou black tea (DCT) is a unique tea category produced using the large-leaf tea varieties from Yunnan, China. To date, the flavor characteristics and critical components of two grades of high-quality DCT, single-bud-grade DCT (BDCT), and special-grade DCT (SDCT) manufactured mainly with single buds and buds with one leaf, respectively, are far from clear. Herein, comparisons of two grades were performed by the integration of human sensory evaluation, an electronic tongue, chromatic differences, the quantification of major components, and metabolomics. The BDCT possessed a brisk, umami taste and a brighter infusion color, while the SDCT presented a comprehensive taste and redder liquor color. Quantification analysis showed that the levels of total polyphenols, catechins, and theaflavins (TFs) were significantly higher in the BDCT. Fifty-six different key compounds were screened by metabolomics, including catechins, flavone/flavonol glycosides, amino acids, phenolic acids, etc. Correlation analysis revealed that the sensory features of the BDCT and SDCT were attributed to their higher contents of catechins, TFs, theogallin, digalloylglucose, and accumulations of thearubigins (TRs), flavone/flavonol glycosides, and soluble sugars, respectively. This report is the first to focus on the comprehensive evaluation of the biochemical compositions and sensory characteristics of two grades of high-quality DCT, advancing the understanding of DCT from a multi-dimensional perspective.

Prediction of the intramuscular fat content of pork cuts by improved U2-Net model and clustering algorithm

The intramuscular fat content is an essential indicator of pork quality, directly affecting sensory quality and consumers' willingness to buy. Traditional testing methods are subjective and destructive, their assessment is scored by trained assessors according to a marbling scale, but the human sensory evaluation has great subjectivity and randomness. Nowadays, there are more methods to predict the intramuscular fat content of livestock meat using computer vision techniques. However, the complex background of the image makes it difficult to segment the target and background. This study proposes a method based on semantic segmentation networks and machine learning algorithms to detect the intramuscular fat content of multi-part pork cuts. The images of five different pork cuts (belly, loin, collar, ham, and hock) are used as experimental data in the study, the results show that the method proposed in this paper can objectively detect the intramuscular fat content of pork and the average accuracy of prediction can reach 93.28%. This new information method can be used to assess the quality of pork in markets and food processing plants, enabling processors and consumers to distinguish pork cuts with different fat content. It improves the quality of pork to meet the needs of the food processing industry.

Standardization of strawberry sourness and sweetness intensities using a taste sensor system and an invariant taste standard solution.

Taste is an essential factor for evaluating the quality of agricultural products. However, it is usually difficult to compare data acquired at different times or by different people because there is no invariant reference and because the evaluation methods are largely subjective. Here, we addressed these problems by developing a method for standardizing strawberry sourness and sweetness intensities using a taste sensor approach with a taste standard solution composed of sour and sweet compounds. This standard solution allows highly efficient sensor measurements because it contains the standard compounds citric acid and sucrose. In addition, we found that polyphenol destabilized the sensor response for strawberry sweetness, and its removal from the sample by appropriate treatment with polyvinylpolypyrrolidone allowed stable evaluation of the sweetness intensity. The taste sensor data obtained using this method were in good agreement with the chemical analysis values related to human sensory evaluation.

Analyzing the influence of withering degree on the dynamic changes in non-volatile metabolites and sensory quality of Longjing green tea by non-targeted metabolomics.

Withering is an important processing stage in green tea, which contributes to the tea flavor quality. The aim of this work was to comprehensively investigate the changes of chemical features and flavor attributes in Longjing green teas produced with five different withering degrees (moisture content of 75.05, 72.53, 70.07, 68.00, and 64.78%, w.b.). Combined with human sensory evaluation, electronic tongue and chromatic differences analysis, an assessment of the relationship between the withering degree and the sensory quality of Longjing tea was obtained. By using a non-targeted metabolomics approach, 69 significantly differential metabolites were screened. As the withering degree increased, most free amino acids and catechin dimers were increased, largely attributed to the hydrolysis of proteins and catechin oxidative polymerization, respectively. The contents of organic acids as well as phenolic acids and derivatives were reduced. Interestingly, flavone C-glycosides decreased overall while flavonol O-glycosides increased. The correlation analysis revealed that metabolites such as theasinensin F, theasinensin B, theaflavin, theaflavin-3,3'-gallate, theaflavin-3'-gallate, malic acid, succinic acid, quinic acid, theanine glucoside and galloylglucose had a greater influence on the taste and color of tea infusion (|r| > 0.6, p < 0.05). Overall, an appropriate withering degree at a moisture content of around 70% is more favorable to enhance the Longjing tea quality. These results may enhance the understanding of green tea flavor chemistry associated with withering and provide a theoretical basis for green tea processing.

Preliminary study on agarwood essential oil and its classification techniques using machine learning

Using essential oils derived from trees for pharmaceutical purposes, incense, aromatherapy, and other areas has expanded its popularity on the international market. However, since human sensory evaluation is still the primary technique used to grade essential oils in Malaysia, the classification technique for determining their grade is still below standard. Nonetheless, prior studies established new approaches for classifying the grade of essential oils by studying their chemical compounds. Therefore, agarwood essential oil was selected for the suggested model due to the increasing demand and the high cost of the world's natural raw materials. The support vector machine (SVM) using one versus all (OVA) approach was selected as the classifier for agarwood essential oil. This study provides an overview of essential oils and their prior research techniques. In addition, a review of SVM is conducted to demonstrate that the technique is appropriate for future studies.

A machine learning method for juice human sensory hedonic prediction using electronic sensory features

This study proposed a method that combines fused electronic sensory analysis technology with artificial neural network to predict the human sensory hedonic of fruit juice. Quantitative descriptive analysis (QDA) and the scoring test method were utilized for human sensory evaluation. The first step involved modeling the fused e-sensory features with human sensory attributes, followed by establishing a fitting model of human sensory attributes and acceptance. The R2 and RMSE values obtained were 0.77 and 0.42 (QDA method), and 0.63 and 0.63 (scoring test method). Finally, the relationship between the fusion e-sensory features and the human sensory hedonic was established. Model-1 achieved an R2 of 0.95 and an RMSE of 0.04, while model-2 achieved an R2 value of 0.88 and an RMSE value of 0.21. This study demonstrates the potential of fusing e-sensory technologies to replace human senses, which may lead to the development of devices with simultaneous multiple senses.

An overview of bitter compounds in foodstuffs: Classifications, evaluation methods for sensory contribution, separation and identification techniques, and mechanism of bitter taste transduction.

The bitter taste is generally considered an undesirable sensory attribute. However, bitter-tasting compounds can significantly affect the overall flavor of many foods and beverages and endow them with various beneficial effects on human health. To better understand the relationship between chemical structure and bitterness, this paper has summarized the bitter compounds in foodstuffs and classified them based on the basic skeletons. Only those bitter compounds that are confirmed by human sensory evaluation have been included in this paper. To develop food products that satisfy consumer preferences, correctly ranking the key bitter compounds in foodstuffs according to their contributions to the overall bitterness intensity is the precondition. Generally, three methods were applied to screen out the key bitter compounds in foods and beverages and evaluate their sensory contributions, including dose-over-threshold factors, taste dilution analysis, and spectrum descriptive analysis method. This paper has discussed in detail the mechanisms and applications of these three methods. Typical procedures for separating and identifying the main bitter compounds in foodstuffs have also been summarized. Additionally, the activation of human bitter taste receptors (TAS2Rs) and the mechanisms of bitter taste transduction are outlined. Ultimately, a conclusion has been drawn to highlight the current problems and propose potential directions for further research.

Non-targeted metabolomics and electronic tongue analysis reveal the effect of rolling time on the sensory quality and nonvolatile metabolites of congou black tea

Rolling is a crucial step in congou black tea manufacturing. Herein, the influence of rolling time (50, 75, 100, 125 min) on the sensory quality and chemical composition of black tea was investigated by human sensory evaluation, electronic tongue, colorimetric analysis, tea pigments quantification and liquid chromatography-mass spectrometry (LC-MS) based non-targeted metabolomics. With increasing rolling time, tea liquor color attributes a*, b*, C* and pigments theaflavins (TFs), thearubigins (TRs), and theabrownins (TBs) were significantly increased. Fifty-two key differential metabolites were identified among four treatments. Catechins, dimeric/trimeric catechins, amino acids were significantly declined. The changes of flavonol/flavone glycosides depended on the linkage pattern. Flavonol O-glycosides such as myricetin 3-O-glucoside were decreased, whereas flavone C-glycosides such as vitexin were increased. These results suggested that different rolling times influenced the catechins oxidative polymerization, flavonol O-glycosides hydrolysis and amino acids oxidation. Black tea by 100-min rolling rendered optimal sensory quality, with highest umami and low bitter intensities in electronic tongue, and higher quality index (10 TFs + TRs)/TBs. Similar result was obtained in a validation batch in a different year. This study provides novel insights into the effect of rolling on black tea quality, which may serve as guidance for the manufacturing of high-quality congou black tea.