Monitoring storage shelf life of tomato using electronic nose technique

Monitoring storage shelf life of tomato using electronic nose technique

Available online at Journal of Food Engineering 85 (2008) 625–631 Monitoring storage shelf li...

459KB Sizes 3 Downloads 192 Views

Available online at

Journal of Food Engineering 85 (2008) 625–631

Monitoring storage shelf life of tomato using electronic nose technique Antihus Herna´ndez Go´mez, Jun Wang *, Guixian Hu, Annia Garcı´a Pereira Department of Agricultural Engineering, Zhejiang University, 268 Kaixuan Road, Hangzhou 310029, PR China Received 22 March 2006; received in revised form 2 April 2007; accepted 23 June 2007 Available online 7 September 2007

Abstract Electronic nose technology offers non-destructive alternative to sense aroma, can be used to assess fruit ripening stage during shelf life. The objective of this study was to monitor tomato storage shelf life during two storage treatments using PEN 2 electronic nose (E-nose). Principal component analysis (PCA) and linear discriminant analysis (LDA) were used to distinguish the different tomato storage time. The obtained results proved that tomato with different storage time can be monitored by the E-nose, but very clear separation among all groups of different storage time was not achieved. By PCA and LDA, E-nose could more clearly discrimination storage time of tomato in carton box than one in folded bag. The correlations between the measured and predicted values of fruit quality attribute (soluble solids content, pH, and puncture force) showed poor prediction performance on the base of signals of E-nose sensors. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Electronic nose; Shelf life; Storage; Tomato; Monitoring

1. Introduction In recent years there has been a considerable increase in demand for better quality fruit due to globalization of market. Consequently, it is important to evaluate fruit maturation stage and storage shelf life. Many methods of monitoring maturation and shelf life have already been proposed. The main disadvantage of these techniques is that they are not practical for cultivars or storage stations, and most of them require the destruction of the samples used for analysis. This may be reason that predictions of shelf life are mainly based on practical experience (Ka¨lvia¨inen, Roininen, & Tuorila, 2003). A strategy for determining the maturation and shelf life consists of sensing the aromatic volatiles emitted from the fruit by using electronic olfactory systems (Benady, Simon, Charles, & Miles, 1995). Metabolic changes are mostly due to the following four items: post-harvest ripening, respiration, fermentation, and phenolic oxidation (Young, Gil*

Corresponding author. Tel.: +86 571 86971881; fax: +86 571 86971139. E-mail address: [email protected] (J. Wang). 0260-8774/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jfoodeng.2007.06.039

bert, Murria, & Ball, 1996). Flavor, comprised of aroma, is an important food quality attribute. The aroma or odor of a food product is detected when its volatiles enter the nasal passages at the back of the throat and are perceived by receptors of the olfactory system (Oshita et al., 2000). Concerning the exploitation of the information contained in the headspace of fruits, it has been studied with the conventional analytical chemistry equipment, and the correlation between the state of over-ripening and the fruit aroma has also been found both in quantitative and qualitative terms. Beside, some specific compounds have been identified as the responsible of the aroma of particular fruit. In the last decade, the electronic nose technology has opened the possibility to exploit, from a practical point of view, the information contained in the headspace in many different application fields. The electronic nose offers a fast and non-destructive alternative to sense aroma, and, hence, may be advantageously used to predict the optimal harvest date. Commercially available electronic noses use an array of sensors combined with pattern recognition software. There have been several reports on electronic sensing in the food industry (Keller, Kangas, Liden, Hashem, & Kouzes, 1995; Schaller, Bosset, & Escher, 1998). Some


A.H. Go´mez et al. / Journal of Food Engineering 85 (2008) 625–631

authors reported positive applications of electronic nose technology to the discrimination of fruits of different quality or ripeness, such as: oranges (Brezmes et al., 2001); tomatoes (Brezmes, Llobet, Vilanova, Saiz, & Correig, 2000); apples (Di Natale et al., 2001; Di Natale et al., 1998; Magan & Evans, 2000; Saevels et al., 2003); grain (Young, Rossiter, Wang, & Miller, 1999), but still few literatures are referred to monitoring of fruit shelf life and quality attribute. The objectives of this study were: (1) to monitoring the tomato shelf life using PEN 2 E-nose by PCA and LDA; (2) to predicted soluble solids content, pH, and puncture force of tomato storage on the base of signals of E-nose sensors. 2. Materials and methods

2.2. Storage condition Tomatoes were stored at 20 ± 1.0 °C, 50–60% RH, and were placed in 4 plastic fold bags and 4 carton boxes (30 each tomatoes for each a bag or a box). Measurements spanned over 12 days after harvest. One bag and one box were removed on 0, 3, 6, 9 and 12 days shelf life and 60 tomatoes evaluated. 2.3. Soluble solids content (SSC) and pH measurement Soluble solid content of juice for each fruit were measured with temperature compensating refractometer (Digital refractometer WYT-J 0-32% Beijing, China). pH of juice was measured by a pH meter (Manufacturer: Sartorius AG, PB-20 (PB-s), Geottingen, Germany, pH/mv = 0 ± 0.3 mv).

2.1. Experimental material 2.4. Electronic nose data acquisition and analysis Tomato ‘Heatwave’ (Lycopersicum esculentum) was selected to experiments. The samples were hand harvested on November 9, 2004 from the experimental orchard located at Department of Horticulture, Zhejiang University. Tomatoes were sorted and selected according to uniform size, weight and color (green) approximately. The sensory panel used to classify the fruit by color was a selected and specialized profile panel of five assessors with previous experience in tomato assessment. The overripe (red skin), ripe (light-red skin) and unripe (green skin) tomatoes were obtained, respectively. The 300 ripe (lightred skin) tomatoes were conducted by E-nose experiment, 150 samples stored in plastic folded bags and 150 samples stored in carton boxes. The quality of fruit is mainly determined by appearance (color, visual aspects), firmness, flavor. Firmness is a criterion often used to assess fruit quality because it is directly related to fruit ripeness and storage state (De Ketelaere et al., 2004; Go´mez, Wang, Hu, & Pereira, 2006; Lesage & Destain, 1996; Wang, Teng, & Yu, 2004). So the firmness was applied to determine the maturity of tomatoes. Fruit firmness was carried out after the E-nose measurements with a Universal Testing Machine (Model 5543 Single Column, Instron Corp., Canton MA, USA). The method of puncture was used. After E-nose measurement and just before firmness measurement, fruits were sliced 7 mm thick using a rotary meat slicer. Both end slices were discarded and the center slice was used for puncture test measuring. Three measurements at intervals of 120° approximately were made on each slice at a junction of outer and radial pericarp, avoiding visible vascular bundles’ fissures, and locular tissue (Wu & Abbott, 2002). The puncture process was auto recorded by computer. A 6 mm diameter stainless steel cylindrical probe with a flat end was used. The puncture force was defined as the average of three maximum force required to push the probe to a depth of 3 mm at a speed of 5 mm s1.

An electronic nose device PEN 2, provided by (WMA Airsense Analysentechnik GmbH) Schwerin, Germany, was used. The portable electronic nose PEN 2 has an array of 10 different metal oxide sensors positioned into small chamber (V = 1.8 ml). Fig. 1 shows schematic diagram of the electronic-nose measurements and gas flow of PEN 2 during the experiments. Table 1 lists all used sensors and their main applications. This table contains current known or specified reaction. Each fruit was placed into an airtight glass jar with a volume of 1 L (concentration chamber). The glass jar was then closed and the headspace inside it was equilibrated for 1 h. Preliminary experiments showed that after 0.5 h of equilibration the headspace reached a steady state and experiments were conducted after 0.5 h of equilibration. One luer-lock needle (20 g) connected to a Teflon-tubing (3 mm) was used to perforate the seal (plastic) of the vial and to absorb the air accumulated inside it, during the measurements. The headspace gas was pumped over the sensors of the electronic nose with a flow of 400 ml/min; during the measurements process, three different phases can be distinguished: concentration, measurement and stand-by. The electro valves, controlled by a computer program, guide the air though different circuits depending on the measurement phase. No matter the phase, airflow is always kept constant though the measurement chamber. During the measurement phase, the bomb pushes the volatiles though

Fig. 1. Schematic diagram of the E-nose.

A.H. Go´mez et al. / Journal of Food Engineering 85 (2008) 625–631 Table 1 Sensors used and their main applications in PEN 2 Number in array


General description




Aromatic compounds



















Very sensitive, broad range sensitivity, react on nitrogen oxides, very sensitive with negative signal Ammonia, used as sensor for aromatic compounds Mainly hydrogen, selectively, (breath gases) Alkanes, aromatic compounds, less polar compounds Sensitive to methane (environment) ca. 10 ppm. Broad range, similar to No. 8 Reacts on sulfur compounds, H2S 0.1 ppm. Otherwise sensitive to many terpenes and sulfur organic compounds, which are important for smell, limonene, pyrazine Detects alcohol’s, partially aromatic compounds, broad range Aromatics compounds, sulfur organic compounds Reacts on high concentrations >100 ppm, sometime very selective (methane)

Toluene, 10 ppm NO2,1 ppm

Benzene 10 ppm H2, 100 ppb Propane 1 ppm CH3, 100 ppm H2S, 1 ppm

CO, 100 ppm H2S, 1 ppm CH3, 10 CH3, 100 ppm

a closed loop that includes the measurement and concentration chambers. No air enter, no exits the loop. The measurement phase lasts 60 s, time enough for sensors to reach a stable value. The collected data interval was 1 s. When a measurement is completed, a stand-by phase is activated (60 s). It is main purpose is to clean the circuit and return sensors to their baseline. Clean air enters the circuit, crosses the measurement chamber first, the empty concentration chamber afterwards, and pushes the remaining volatiles out of the circuit. E-nose was held at the temperature of 20 ± 1 °C and 50– 60% RH during all experiments. When the sensors are exposed to volatiles, during the measurement phase, the computer records the resistance changes that the sensors experience. When the measurement was completed, the acquired data was properly stored for later use. The set of signals of all sensors during measurement of a sample is a pattern. Pattern of multiple measurements dealing with the same problem are stored in a pattern file and act as the training set. The pattern data were recorded, checked visually and analyzed using WinMuster (version Jun 2003, copyright 1996–2002 WMA Airsense Analysentechnik GmbH 2003).


2D or 3D coordinates. This is carried out through the data reduction that extracts the most important information from the database as a result. The results of training phase can be displayed in a two dimensional view. PCA is based on a linear project of multidimensional data into different coordinates based on maximum variance and minimum correlation. Training pattern from measurements of similar samples will be located close to each other after transformation. Hence, the graphical output can be used for determining the difference between groups and comparing this difference to the distribution of pattern within one group. The linear discriminant analysis (LDA) is the first step of the discriminant function analysis (DFA). The LDA calculates the discriminant functions and similar to the PCA a 2 or 3 dimensional display of the training set data. The difference between PCA and LDA is, that PCA does not care about the relation of a data points to the specified classes, while the LDA calculation uses the class information that was given during training. The LDA takes care about the distribution within classes and the distances between them. Therefore, the LDA is able to collect information from all sensors in order to improve the resolution of classes. 2.6. Evaluation models for E-nose prediction For fruit quality indices, different calibration models were used by partial least square (PLS). The quality indices of the calibration model were quantified by standard error of calibration (SEC), standard error of prediction (SEP) and correlation coefficient (r) between the predicted and measured parameters. A good model should have a low SEC and SEP, a high correlation coefficient. A large difference indicated that too many latent variables were used in the model and noise was modeled. SEC and SEP were defined as follows: vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u Ip u 1 X 2 SEP ¼ t ð1Þ ð^y i  y i  biasÞ I p  1 i¼1 vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi u Ic u 1 X 2 ð2Þ SEC ¼ t ð^y i  y i Þ I p  1 i¼1 where yˆi yi Ic Ip

predicted value of the ith observation measured value of the ith observation number of observations in the calibration set number of observations in the validation set

Bias-systematic difference between predicted and measured values: Ip 1 X ð^y i  y i Þ I p i¼1

2.5. Principal component analysis, linear discriminant analysis

Bias ¼

Using the principal component analysis (PCA) the measurement data, previously trained will be transformed into

To evaluate the electronic nose prediction, the results by E-nose response were compared with those derived from


A.H. Go´mez et al. / Journal of Food Engineering 85 (2008) 625–631


well-established traditional techniques such as pH, soluble solids content, and compression test. The calculations were carried out using ‘The Unscrambler V8.0.5 1986–2003’ (CAMO, Process, AS, OSLO, Norway), a statistical software package for multivariate calibration. Before the calibration, the relative conductance variation of the data was analyzed by principal component analysis (PCA) and defective sensor response was eliminated, partial least square (PLS) was used to build the prediction models. The latter is a projection method, which use the independent variable and the dependent variables to regress the dependent variables on the latent variables (factors) (De Jong, 1993). 3. Results and discussion 3.1. Electronic nose response to fruit aroma Fig. 2 shows a typical response of 10 sensors for tomato. The data obtained are the ratio of conductivity between G and G0 (the conductivity of the sensors when the sample gas or zero gas blows over). Each curve represents a different sensor transient. The curves represent sensor conductivity of each sensor in array against time when the volatiles from the fruit reach the measurement chamber. In that transition, the clean airflow that reaches the measurement chamber is substituted by airflow that comes from the concentration chamber, closing a loop circuit between both chambers. It can be shown that conductivity increases sharply and then stabilizes after 30 s after an initial period of low and stable conductivity (when only clean air is crossing the measurement chamber). The each sensor

Fig. 2. Ten sensors response to tomato fruit aroma.

signal generally stabilizes and was considered to use in analysis of electronic nose. In this research, the signal of each sensor at response 42 s was used in analysis of electronic nose. 3.2. Signal analysis Fig. 3 shows the evolution of the signals generated by the sensor array. Each line represents the average signal variation of 30 tomatoes respectively for one sensor of the array (10 sensors), linking to the measurements of conductance increase or decrease experienced by the sensor as maturity state that vapors from the fruit reached the measurement chamber. The result was similar to those obtained by Brezmes et al. (2000) testing peach and pear. The G/G0 values response to volatiles of tomatoes kept in bag was less than that in box. In both treatments the sensors with minor response to fruit volatile have smaller erratic behavior during the experiment. It is inferred that sensors 2, 9 have greater response to fruit volatile, and sensors 6, 7 and 8 have higher response. 3.3. Classification of tomato using PCA and LDA 3.3.1. Tomato stored in box Fig. 4a showed PCA results for tomato stored in box, and represent the variation for different storage time. The processed data showed a shift erratic of the groups and no particular trend with storage time along the first principal component, axis-x (PC1), which explains 79.66% of the total variance with value 95.64%. The second principal component (PC2) explains 15.99%. The system is able to distinguish the different tomato storage time, and the five groups are clearly separated. After analyzed the same data set using LDA, the five groups were also distinguishable each other. In this plot about 54.86% of the total variance of the data is displayed (Fig. 4b). LDA function 1 (LD1) and function 2 (LD2) accounted for 29.63 and 25.23% of the variance respectively. This method is efficient to separate the tomatoes according the different storage time when they are stored in box.

Fig. 3. Mean value of sensors response to storage time of tomato: (a) bag; (b) box.

A.H. Go´mez et al. / Journal of Food Engineering 85 (2008) 625–631


Fig. 4. Tomato with different storage time in box: (a) PCA (b) LDA.

Fig. 5. Tomato with different storage time in bag: (a) PCA (b) LDA.

3.3.2. Tomato stored in plastic fold bag In Fig. 5, about 96.31% of the total variance of the data is displayed using PCA analysis. PCA function 1 (PC1) and function 2 (PC2) accounted for 85.84% and 10.47% of the variance respectively. The processed data shows a shift erratic of the groups at different storage time along the first principal component, and no particular trend with the tomato storage time along the axis-y. Except day 0 and day 3, day 0 and day 6 are overlapped lightly, the system was able to distinguish from the rest of groups stored in bag. After analyzed the same data set using LDA, each group were clearly distinguishable from the groups except day 3 and day 6 (Fig. 5). The processed data shows a shift erratic of the groups by different storage time along the first and second functions (LD1 and LD2). In this plot about 79.83% of the total variance of the data is displayed. LD1 and LD2 accounted for 70.79% and 9.03% of the variance, respectively. The separated result is better using LDA than PCA. 3.4. Prediction of fruit quality characteristics In order to observe the electronic nose performance of prediction for fruit quality attributes, olfactory system measurements were coupled with the values obtained from quality indices at the same measurement session.

The relative conductance values for each sensor at 42 s were related to each fruit quality characteristics such as: soluble solids content (SSC), pH, maxim puncture force. Three hundred tomatoes (150 tomatoes in bag or box) were separated randomly into two groups: a calibration set used to develop the calibration models (250 tomatoes) and the remaining samples were used to prediction set (external validation) (50 tomatoes). The calibration models were validated using full cross-validation. Measurements spanned over 12 days after harvest for each storage in bag or box, 50 samples in 60 tomatoes (25 sample in 30 tomatoes for each treatment) were removed on shelf life of 0, 3, 6, 9 and 12 days and used to develop the calibration models, 10 samples in 60 tomatoes (5 samples in 30 tomatoes for each treatment) were used to prediction set. In the validation method, some samples are kept out of the calibration and used for prediction. Validation residual variance can then be computed from the prediction residuals. In segmented cross-validation, the samples are divided into subgroups or ‘‘segments”. One segment at a time is kept out of the calibration. There are as many calibration rounds as segments, and predictions can be made on all samples. A final calibration is performed with all samples. In full cross-validation, only one sample a time is kept out of the calibration. The correlation between the measured and predicted values of the three fruit parameters show poor to reasonable

A.H. Go´mez et al. / Journal of Food Engineering 85 (2008) 625–631


Table 2 Results of calibration, cross-validation and prediction for fruit quality property on the base of electronic nose signal Parameter















7 5 6

0.840 0.803 0.756

0.213 0.087 1.249

7.42 e7 1.16 e7 7.79 e6

0.770 0.743 0.704

0.247 0.098 1.358

0.004 0.69 e3 0.129

0.720 0.658 0.660

0.506 0.302 6.07

0.008 0.004 0.070

Fig. 6. Prediction results from PLS models for soluble solids content, acidity and puncture force.

prediction performance with values between 0.756 and 0.840 during the calibration model construction and with values between 0.704 and 0.770 for the internal validation set (Table 2). PLS prediction results for soluble solids content, acidity, compression force and puncture test are presented in Fig. 6. In Fig. 6, the ordinate and abscissa axes represent the predicted and measured fitted values of the appropriate parameters, respectively. During the prediction (external validation) poor correlation coefficients were obtained with values between 0.658 and 0.720, the best correlation coefficient was obtained for SSC and the worst value in pH. In previous research work, Saevels et al. (2003) in apples, a poor correlation between fruit quality indices (firmness, acidity and soluble solids) and nose signal response was also obtained with correlation coefficient values between (0.66–0.76). However, Brezmes et al. (2001) using electronic nose signal to predict firmness and pH in pink lady apple obtained reasonable well prediction performance with correlation coefficient values of 0.94 and 0.84, respectively.

Although it may seem surprising to see that physical measurements such firmness measured by compression force can be quiet predicted with sensor responses to organic volatiles generated by fruit, such results are meaningful since the physiological characteristic of fruit are closely related to chemical processes that take place during the ripening process of fruit. In other words, the electronic nose does not measure firmness directly; it actually measures volatiles that are well correlated with the firmness of the fruits. 4. Conclusions (1) The results obtained prove that the different storage time of tomatoes could be using the electronic nose PEN 2. It is considered that the separation achieved among all groups at different storage time was clear. (2) During tomato storage in carton box and plastic folder bag, PCA and LDA analysis were able to classify the tomato in groups according the storage time, achieving slightly better results when PCA is used.

A.H. Go´mez et al. / Journal of Food Engineering 85 (2008) 625–631

(3) The correlation between the measured and predicted values of fruit parameters show poor to reasonable prediction performance on the base of electronic nose signals. Acknowledgement The authors acknowledge the financial support of Chinese National Foundation of Nature and Science (30571076 and 30771246) and the financial support of Program for New Century Excellent Talents in Chinese University (NCET-04-0544). References Benady, M., Simon, J. E., Charles, D. J., & Miles, G. E. (1995). Fruit ripeness determination by electronic sensing of aromatic volatiles. Transaction of the ASAE, 38, 251–257. Brezmes, J., Llobet, E., Vilanova, X., Orts, J., Saiz, G., & Correig, X. (2001). Correlation between electronic nose signals and fruit quality indicators on shelf-life measurements with pinklady apples. Sensors and Actuators B, 80, 41–50. Brezmes, J., Llobet, E., Vilanova, X., Saiz, G., & Correig, X. (2000). Fruit ripeness monitoring using an electronic nose. Sensors and Actuators B, 69, 223–229. De Jong, S. (1993). PLS fits closer than PCR. Journal of Chemometrics, 7, 551–557. De Ketelaere, B., Lammertyn, J., Molenberghs, G., Desmet, M., Nicolaı¨, B., & De Baerdemaeker, J. (2004). Tomato cultivar grouping based on firmness change, shelf life and variance during postharvest storage. Postharvest Biology and Technology, 34, 187–201. Di Natale, C., Macagnano, A., Martinelli, E., Paolesse, R., Proietti, E., & D’Amico, A. (2001). The evaluation of quality of post-harvest orange and apples by means of an electronic nose. Sensors and Actuators B, 70, 18–26. Di Natale, C., Macagnano, A., Paolesse, R., Mantini, A., Tarizzo, E., D’Amico, A., & Sinesio, F. (1998). Electronic nose and sensorial analysis: comparison of performances in selected cases. Sensors and Actuators B, 50, 246–252.


Go´mez, A. H., Wang, J., Hu, G., & Pereira, A. G. (2006). Electronic nose technique potential monitoring mandarin maturity. Sensors and Actuators B, 113(1), 347–353. Ka¨lvia¨inen, N., Roininen, K., & Tuorila, H. (2003). The relative importance of texture, taste and aroma on a yogurt-type snack food preference in the young and the elderly. Food Quality and Preference, 14, 177–186. Keller, P.E., Kangas, L.J., Liden, L.H., Hashem, S., & Kouzes, R.T. (1995). Electronic noses and their applications. In IEEE Technical Applications Conference (TAC’95) at Northcon’95, Portland, Oregon, 10–12 October. Lesage, P., & Destain, M. F. (1996). Measurement of tomato firmness by using a non-destructive mechanical sensor. Postharvest Biology and Technology, 8, 45–55. Magan, N., & Evans, P. (2000). Volatiles as an indicator of fungal activity and differentiation between species, and the potential use of electronic nose technology for early detection of grain spoilage. Journal of Stored Products Research, 36, 319–340. Oshita, S., Shima, K., Haruta, T., Seo, Y., Kagawoe, Y., Nakayama, S., & Kawana, S. (2000). Discrimination of odors emanating from ‘La France’ pear by semi-conducting polymer sensors. Computers and Electronics in Agriculture, 26, 209–216. Saevels, S., Lammertyn, J., Berna, A. Z., Veraverbeke, E. A., Di Natale, C., & Nicolaı¨, B. M. (2003). Electronic nose as a non-destructive tool to evaluate the optimal harvest date of apples. Postharvest Biology and Technology, 30, 3–14. Schaller, E., Bosset, J. O., & Escher, F. (1998). Electronic noses and their application to food: A review. Food Science Technology- Lebensm-Wiss Technology, 31, 305–316. Wang, J., Teng, B., & Yu, Y. (2004). Pear dynamic characteristics and firmness detection. European Food Research Technology, 218, 289– 294. Wu, T. X., & Abbott, J. A. (2002). Firmness and force relaxation characteristics of tomatoes stored intact or as slices. Postharvest Biology and Technology, 24, 59–68. Young, H., Gilbert, J. M., Murria, S. H., & Ball, R. D. (1996). Causal effects of aroma compounds on royal gala apple flavors. Journal of the Science of Food and Agriculture, 71, 329–336. Young, H., Rossiter, K., Wang, M., & Miller, M. (1999). Characterization of Royal Gala apple aroma using electronic nose technology potential maturity indicator. Journal of Agriculture and Food chemistry, 47, 5173–5177.