Electronic tongue for sensing taste changes with apricots during storage

Electronic tongue for sensing taste changes with apricots during storage

Available online at www.sciencedirect.com Sensors and Actuators B 131 (2008) 43–47 Electronic tongue for sensing taste changes with apricots during ...

565KB Sizes 0 Downloads 47 Views

Available online at www.sciencedirect.com

Sensors and Actuators B 131 (2008) 43–47

Electronic tongue for sensing taste changes with apricots during storage David B. Kantor a,∗ , Geza Hitka b , Andras Fekete a , Csaba Balla b a

Corvinus University of Budapest, Department of Physics and Control, H-1118, Somloi ut 14-16., Budapest, Hungary b Department of Refrigeration & Livestock Products’ Technology, H-1118, Somloi ut 14-16., Budapest, Hungary Available online 14 December 2007

Abstract The present paper is aimed at application of electronic tongue (ET) to the recognition of different apricot varieties and detecting taste changes during storage. Effect of 1-methylcyclopropene (1-MCP) treatment and controlled atmosphere (CA) storage on ripeness and taste of fruits was also investigated. Three apricot varieties and four different CA conditions were studied. The pH and Brix measurement were used as reference methods and sensory analysis was performed with three parameters. Classification of apricot samples was done by canonical discriminant analysis (CDA). Calibration models for determination of correlation between ET data and reference methods were made by partial least squares (PLS) regression. Results of sensory analysis were carried out by variance analysis (ANOVA). The electronic tongue system proved to be suitable for apricot juice measurements. Classification of apricot varieties and determination of correlations between ET, chemical properties and sensory analysis were performed. © 2007 Elsevier B.V. All rights reserved. Keywords: Electronic tongue; Apricot; Controlled atmosphere storage; 1-MCP; Taste

1. Introduction Post-harvest ripening could limit transport time, storage period and shelf-life of fruits. Apricot (Prunus armeniaca L.) is very sensitive to storage conditions. Apricot quality (texture, taste and color) is determined by the ripening conditions. The ripening process of apricots is climacteric. The fruits are able to ripe after harvest, and the ripening could be fast under uncontrolled conditions. These over-ripe apricots are very susceptible to weight losing, bruising and subsequent decay. Controlled conditions after harvest, such as cooling, controlled atmosphere (CA) [1] can prevent production and/or action of ethylene. Controlling ethylene is a key to maintaining fruit quality in storage. The plant natural hormone ethylene has an important role in the ripening process, and 1-methylcyclopropene (1-MCP) is known that antagonizes and blocks the action of it in harvested fruits. It was successfully applied to delay ripening of apricot [2–4] and peach [5]. Appearance of artificial sensor techniques accelerated research on organoleptic properties. Until now, this statement is principally valid for electronic noses (EN) due to their non-destructive measurement method. EN system was success-

fully applied to assessment apricot quality [6]. Nevertheless, electronic nose technique cannot detect the changes in taste. Lately, electronic tongue technology was developed to determine taste characteristic of liquid media. Shortly, an instrument was developed combining electronic nose and electronic tongue technology [7]. The electronic tongue is based on the array of chemical sensors (potentiometric or voltammetric) displaying cross-sensitivity to various substances [8,9]. These instruments were applied for investigation of fruit and fruit juice to characterize taste, e.g. apple juice [10], apple varieties [11] and tomatoes [12]. The main focus of the present study was to evaluate the performance of the electronic tongue for classification of different apricot varieties, to reveal differences between control and 1MCP treated samples and to detect the effect of post-harvest techniques on fruit taste and ripeness. Further, aim of the study was to determine correlations between the results of different methods and to determine correlations between ET data and reference methods by using PLS regression. 2. Experimental 2.1. Samples



Corresponding author. Tel.: +36 1 4826023; fax: +36 1 4826361. E-mail address: [email protected] (D.B. Kantor).

0925-4005/$ – see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2007.12.003

Apricots of three Hungarian varieties were studied: G¨onczi, Cegl´edi and Pann´onia. The apricots were received from a

44

D.B. Kantor et al. / Sensors and Actuators B 131 (2008) 43–47

Table 1 Controlled atmosphere conditions for apricot storage No. of chamber

O2 (%)

CO2 (%)

N2 (%)

1. (Control) 2. 3. 4.

20.96 1.5 1.5 1.5

0.03 1.0 3.0 5.0

78.0 97.5 95.5 93.5

horticultural farm. Fruits were harvested at 80–85% maturity level, which provides the most efficient conditions for 1-MCP treatment. In case of unripe fruits the characteristic taste can develop. However, in case of over ripeness this treatment cannot reduce ripening process significantly. The 1methylcyclopropene (AgroFresh Inc., Italy) was purchased in powder form and it was solved with distilled water (ratio of 1-MCP to water 1:5) and was homogenized for 2 min. The 1MCP was dispersed in a 100 L plastic box, in which fruits were enclosed. The gas concentration was 1000 ppb and this treatment lasted for 24 h. For short-term experiment, the treated and control samples were stored at 20 ◦ C, 45% relative humidity (RH) for 4 days. The aim of this shelf-life test was to simulate the conditions of a commercial distributor and to know whether the treatment reduces the ripening within this 4 days period. For long-term experiments, control and treated G¨onczi fruits were put into chambers of 1 ◦ C and 95% RH in having normal and controlled atmosphere. The goal was to follow changes during long-term storage. Table 1 shows the applied CA conditions. In the chamber no. 1 the gas composition was not controlled. Samples were measured on 14th and 28th days after the beginning of CA storage. For the measurements fifteen apricots of each variety were processed: juice was pressed from each fruit by blender and then clarified with filter paper (ALBET DF 400, Albet Inc., Spain) and diluted with distilled water in 1:1 ratio. 2.2. Measurement methods An Alpha-MOS ␣Astree I (Alpha-MOS, Toulouse, France) potentiometric electronic tongue was applied to measure the samples connected with LS 16 autosampler unit. It comprised 7 ISFET (ion sensitive field effect transistor) based potentiometric chemical sensors for food application (ZZ3401, BA3401, BB3401, CA3401, GA3401, HA3401, JB3401). Due to the polymer membrane coating sensors display sensitivity to organic acids, salts, mono- and disaccharides. Furthermore, sensors have cross-sensitivity for taste chemicals which are typically found in food stuffs and beverages. Sensor potential was measured versus a standard Ag/AgCl 3 M KCl reference electrode (Metrohm AG). Measurement time was 180 s. Each sample was measured seven times. Sensors were cleaned with distilled water between measurements till stable potential. Measurements were made under room temperature (23–24 ◦ C). Sample volume was 100 mL. Samples were run in random order. Measurements in apricot juice were carried out using classical techniques to monitor the acid and sugar content changes

during storage. A HANNA 209 (Hanna Instruments, USA) pH meter was used for pH measurement. The degree Brix was measured to assess sugar content of fruits by ATAGO PAL-1 (Atago U.S.A., Inc.) refractometer. Sensory panel of 10 non-trained persons was applied to determine the taste differences between control and treated samples. Three parameters were scored in a 100 points scale: ‘sweet flavour’, ‘sour flavour’ and ‘overall impression’. 2.3. Data processing Data processing consisted of recognition and classification of apricot samples and determination of correlations between ET data and reference methods (pH, Bx and three parameters of sensory evaluation). ET data consisted of responses of 7 sensors in each sample. This resulted in the matrix of the size 7 × 42 in the case of short-term storage and 7 × 56 in the case of long-term storage. Classification of apricot samples was done by canonical discriminant analysis (CDA). Calibration models for determination of correlations between ET data and reference methods were made by partial least squares (PLS) regression. All data were centered and standardized prior to modeling. Software packages SPSS v. 14.0 by SPSS Inc., USA and The Unscrambler v. 9.1 by CAMO AS, Norway were used. The Profisense software was used in the data collecting and processing of sensory analysis, which was performed in the Sensory Laboratory of Corvinus University of Budapest. Results were carried out by variance analysis (ANOVA), comparing samples at P = 0.05 level according the Fisher test LSD. 3. Results and discussion The main focus of the present study was to evaluate the performance of the electronic tongue for classification of different apricot varieties, to reveal differences between control and 1-MCP treated samples and to detect the effect of postharvest techniques on fruit taste and ripeness. Further aim of the study was to determine correlations between the results of different methods and to determine latent variables by using PLS regression. The main requirement for usability of ET data is the good repeatability of sensors. Relative standard deviation (R.S.D.) was used to characterize this parameter. Lower than 5% has been found for this value in each case and each sensor. Canonical discriminant analysis with respect to three apricot cultivars was performed using ET data. Plot of discriminant functions of shelf-life experiment is shown in Fig. 1. After 4 days storage at room temperature, the six groups (control and treated samples of three cultivars) were correctly classified at 71.9% in leave one out validated case. Function 1 shows the differences between apricot varieties. It is interesting to note that relative position of the results of G¨onczi and Cegl´edi apricots are quite close to each other while Pann´onia was perfectly distinguished. Function 2 shows the difference between control and treated samples. If the control and treated samples of a cultivar were taken into account to be in the same group the classification of

D.B. Kantor et al. / Sensors and Actuators B 131 (2008) 43–47

45

Fig. 3. Results of sensory analysis for apricot samples in shelf-life test. Cultivars are as follows: G: G¨onczi, C: Cegl´edi and P: Pann´onia.

Fig. 1. Canonical functions for control and 1-MCP treated apricot cultivars.

apricot varieties was correct at 80.0% level. However, for separation of data into two groups, control and 1-MCP treated resulted in correct classification at 87.5%. According to this result the treatment caused significant ripeness difference during shelf-life test. Results of pH and ◦ Bx measurements are shown in Fig. 2. There were not found significant differences in each case, but it is interesting to mention that 1-MCP treated samples had higher sugar content, due to the delay of carbohydrates degradation. Sensory analysis data had large variance and significant differences were found only in a few cases. The average points of sensory analysis are shown in Fig. 3. There was not found clear trend in the sweet flavour. However, in case of G¨onczi the treated sample was sweeter than the control one significant at 0.05. For each cultivar the results of sour flavour parameter were higher in the case of 1-MCP treated samples although this difference was significant only in case of Pann´onia variety. The

Fig. 2. Chemical properties of apricot samples in shelf-life test. Cultivars are as follows: G: G¨onczi, C: Cegl´edi and P: Pann´onia.

same phenomenon was described in the results of overall impression parameter, however, there was no significant difference. These results show that sensory panel had better impressions from treated samples thus it was able to distinguish the control and 1-MCP treated samples. These facts confirm the results of electronic tongue, pH and ◦ Bx measurements. Aim of long-term storage was to determine effect of 1-MCP and CA treatments on taste changes and ripeness level. Discrimination analysis with respect to control and 1-MCP treated apricot samples in four different CA conditions were performed using ET data sets. Plots of canonical functions for ET data after 2 and 4 weeks are shown in Figs. 4 and 5, respectively. The eight sample groups (control and treated samples of four different storage chambers) were correctly classified at 83.9% after 2 weeks and at 92.9% after 4 weeks by CDA method in leave one out validation. If the control and treated samples from a chamber were taken into account to be in the same group the classification of four groups was correct at 82.1% level after 2 weeks and 67.9% after 4 weeks storage. It is interesting to note that the samples of

Fig. 4. Canonical functions for control and 1-MCP treated apricot cultivars under different CA conditions after 2 weeks storage.

46

D.B. Kantor et al. / Sensors and Actuators B 131 (2008) 43–47

Fig. 7. Results of sensory analysis for apricot samples after 2 weeks storage.

Fig. 5. Canonical functions for control and 1-MCP treated apricot cultivars under different CA conditions after 4 weeks storage.

control chamber (no. 1) were classified worst after 2 weeks as well as after 4 weeks. Furthermore, the relative position of control samples from the other three chambers is very close to each other on both of the measurement days. The main difference between CDA plots is the increase of distance between the position of control and 1-MCP treated samples from chambers no. 2, 3 and 4. Finally, it was observed that this distance is dependent on the CO2 content in the chambers. Namely, the more CO2 was in the chambers, the shorter was the distance. Results of pH and ◦ Bx measurements are shown in Fig. 6. There were not found significant differences in each case, but it is worthy of note that the results of control chamber (no. 1) did not vary during storage. This fact confirms the result of ET measurement. The pH values were stable during the long-term storage. However, there were some differences in ◦ Bx values, but without any definite trend. The average points of sensory analysis are shown in Fig. 7 after 2 week and in Fig. 8 after 4 week storage. Significant difference was found only in two cases in overall impression parameter between the control and CA treated samples in chamber no. 4

Fig. 8. Results of sensory analysis for apricot samples after 4 weeks storage.

after 2 weeks, and in sour flavour parameter in chamber no. 4 after 4 weeks. Final goal of this study was to determine correlations between ET data and the other measurement methods by using PLS regression. The coefficient of correlation and the slope of regression models are shown in Table 2. Good results were observed in the case of refractometer measurements. Moderate correlation was found with pH measurements and two sensory analysis variables, sweet flavour and sour flavour. However, overall impression parameter shows very poor correlation Table 2 Correlation values and rise of the curves of PLS regression Storage time

Fig. 6. Chemical properties of apricot samples in long-term storage test.

Chemical parameters

Sensory parameters

pH

Brix (%)

Sweet flavour

Sour flavour

Overall impression

4 days R Slope

0.7766 0.6032

0.8210 0.6741

0.7629 0.5819

0.6303 0.3973

0.8174 0.6681

2 weeks R Slope

0.7209 0.5197

0.8139 0.6625

0.2880 0.6209

0.7537 0.5681

0.2880 0.0829

4 weeks R Slope

0.6282 0.3946

0.9151 0.8375

0.915 0.4631

0.4892 0.2393

0.5118 0.2620

D.B. Kantor et al. / Sensors and Actuators B 131 (2008) 43–47

with ET sensor responses because it shows some subjectivity. 4. Conclusions The electronic tongue system proved to be suitable for apricot juice measurements. It was demonstrated to be a promising tool for monitoring the effects of post-harvest techniques on fruit ripening process. Controlled atmosphere storage and 1MCP treatment caused considerable mature differences which were detectable by the electronic tongue. Classification of apricot varieties and determination of correlations between ET, chemical properties and sensory analysis was successful. The instrumental measurement was more sensitive to differences than the sensory analysis. Further research is suggested for a more detailed identification of the differences in the chemical background. Acknowledgement This work was supported by OTKA Foundation Grant Number MO 45745. References [1] M.T. Pretel, et al., Ripening and ethylene biosynthesis in controlled atmosphere stored apricots, Eur. Food. Res. Technol. 209 (1999) 130–134. [2] X. Fan, et al., Inhibition of ethylene action by 1-methylcyclopropene prolongs storage life of apricots, Postharvest Biol. Technol. 20 (2000) 135– 142. [3] L. Dong, et al., Effect of 1-methylcyclopropene on ripening of ‘Canino’ apricots and ‘Royal Zee’ plums, Postharvest Biol. Technol. 24 (2002) 135–145. [4] L. Palou, C.H. Crisosto, Postharvest treatments to reduce the harmful effects of ethylene on apricots, Acta Hort. 599 (2003) 31–38. [5] H. Liu, et al., The effects of 1-methylcyclopropene on peach fruit (Prunus persica L. cv. Jiubao) ripening and disease resistance, Int. J. Food Sci. Technol. 40 (2005) 1–7. [6] C. di Natale, et al., Sorting of apricots with computer screen photoassisted spectal reflectance analysis and electronic nose, Sens. Actuators, B Chem. 119 (2006) 70–77.

47

[7] F. Winquist, et al., The combination of an electronic tongue and an electronic nose, Sens. Actuators, B Chem. 58 (1999) 512–517. [8] A. Legin, et al., Recognition of liquid and flesh food using an ‘electronic tongue’, Int. J. Food Sci. Technol. 37 (2002) 375–385. [9] P. Ivarsson, et al., A voltammetric electronic tongue, Chem. Senses 30 (2005) 258–259. [10] R.N. Bleibaum, et al., Comparison of sensory and consumer results with electronic nose and tongue sensors for apple juices, Food Qual. Prefer. 13 (2002) 409–422. [11] A. Rudnitskaya, et al., Analysis of apples varieties – comparison of electronic tongue with different analytical techniques, Sens. Actuators, B Chem. 116 (2006) 23–28. [12] K. Beullens, et al., The electronic tongue and ATR-FTIR for rapid detection of sugars and acids in tomatoes, Sens. Actuators, B Chem. 116 (2006) 107–115.

Biographies David B. Kantor received his MSc degree in food engineering at Corvinus University of Budapest in 2005. Currently he is working as a post-graduate student in the last year of his PhD. His main research interest involves sensor systems and their application for food purposes. Geza Hitka received his MSc degree in food engineering at Corvinus University of Budapest in 2005. Currently he is in the third year of his doctoral studies at Corvinus University of Budapest. His main research interest involves combination of controlled atmosphere storage and 1-MCP treatment, the application of SPME technique at post-harvest field and determination the low oxygen limit of fruits and vegetables. Andras Fekete received his MSc in agricultural engineering (Agricultural University, Budapest/Godollo) in 1961, PhD in 1972, DSc in 1989. He performed R&D at the Hungarian Institute of Agricultural Engineering. He is professor at Faculty of Food Science, Corvinus University of Budapest. His research interest is in the field of physical properties of agricultural materials and foods, nondestructive methods and instrumentation, non-contact measurement methods and instruments for quality assessment, application of electronic tongue for quality assessment of liquid foods, instrumentation and control systems. Csaba Balla has been MSc in food engineering (Corvinus University of Budapest) since 1974. He received his PhD in 1996 at Corvinus University of Budapest and is now associate professor and head of Department of Refrigeration and Livestock Products Technology. His research interests are post-harvest handling, cold store of fruits and vegetables, refrigeration technology of milk, poultry, meat and meat products and freezing technology of fruits and vegetables, meat and dairy products.