Abstract
The reports about predictive model developed between textural and pasting properties are insufficient, and the application and feasibility of predictive model in rice blend have not been reported. The purpose of the study was to determine the extent to which the pasting parameters can be used to predict the textural properties of cooked rice and to make analysis of the application and feasibility of predictive model in rice blends. The results showed that viscosity hold, viscosity end, breakdown, and setback had a significant correlation with textural properties. The principal component factor analysis results indicated that the loadings were large for viscosity hold, viscosity end, setback on the first component, and for breakdown on the second component. The predictive models for estimating the texture of cooked rice with high accuracy were developed, which were validated with acceptance for application in rice blends by conducting 95% confidence limits validation of statistical analysis system procedure on simplex-lattice design rice blends. The optimized predictive models could be extensively used in single rice cultivars as well as rice blends, which is helpful to improve the use of low quality rice cultivars by blending high quality with low quality rice to meet the needs of consumers at a relatively lower cost.
INTRODUCTION
Conventionally, preference sensory evaluation has been used for sensory quality of rice; however, sensory methods are complicated and time consuming and the results may be biased by cultural tradition and attitudes of the panelists.Citation[1– Citation3] Oppositely, the instrumental methods to evaluate the texture of cooked rice are objective, constant, stable, and of good repetition. Some common instrumental parameters, including hardness, adhesiveness, cohesiveness, springiness, and gumminess, are well grounded in accepted rheological definitions developed by Szczesniak et al.Citation[4] Different instruments and procedures are capable of measuring these parameters. For example, Perez and JulianoCitation[5] used an Instron testing machine with an Ottawa texture measuring system (OTMS) cell (Canners Machinery Ltd., Simcoe, ON, Canada). Champagne et al.Citation[6] reported the texture profile analysis (TPA) values from a table-top TA.XT2 texture analyzer (Texture Technologies Corp., Scarsdale, NY, USA).
Besides instrumental measurement of the texture of cooked rice, amylography is one of the objective property tests that serve as indices for sensory quality. Furthermore, amylography is used for determining gelatinization and paste viscosity characteristics,Citation[7, Citation8] and traditionally it was performed on a Brabender Viscoamylograph and required approximately 1.5 h to perform. Today the test is routinely conducted with a Rapid Visco Analyser in approximately 15 min (RVA; Newport Scientific, Warriewood, Australia), and viscosity properties measured on an RVA are similar to those measured with a Brabender Viscoamylograph.Citation[9, Citation10]
Relationships have really existed among pasting properties, physicochemical properties, and instrumental measurements of texture of cooked rice. Only some reports have reported the relationship between physicochemical properties and textural parameters or between physicochemical properties and pasting properties.Citation[6, Citation11– Citation20] However, a few reports have reported about relationships between textural properties and pasting properties, which were the two main objective methods of instrumental evaluation of rice quality.
Rice blending is a practical good way of improving sensory quality of rice as well as an economic method for increasing the use of low-quality rice under the precondition of meeting consumers’ sensory needs. Choi et al.Citation[21] reported that three rice cultivars (Goami2, IIpumbyeo, and Baegjinju) with significantly different physicochemical properties were blended to promote the consumption of Goami2 on the precondition of meeting the texture and paste properties. Also, ZaidulCitation[22] had used RVA to study the pasting properties of the blends of wheat flour and potato starch and found that the pasting properties had been improved. Actually, reports about the rice blends are lacking due to the complication of rice blends study. The purpose of the study was to determine the extent to which the pasting parameters can be used to predict the textural properties of cooked rice and to make an analysis of the application and feasibility of the predictive model that relates RVA properties to textural properties in rice blends.
MATERIALS AND METHODS
Rice Samples
Samples of long-grain rice cultivars (N = 30) were planted in field experiments at Xiaogan Agricultural Institute, Hubei province, China in the summer rice growing season in 2008. Field management followed normal agronomic procedure and natural ripening of the grain occurred. The harvested paddy rice was dehulled and milled with a laboratory one-pass mill (Xin'en Precise Food Instrument Co. Ltd., Taizhou, China) according to Chinese national standards,Citation[23] and then the broken rice was removed with a 10 mesh sifter. All the varieties of head rice were stored at room temperature for subsequent analysis.
Near Infrared Reflectance (NIR) Analysis
Amylose content of rice varieties were obtained through NIR analysis. An analytical NIR spectrometer (model 6500; Foss-NIR Systems, Silver Spring, MD, USA) was used for reflectance measurements on rice samples. The whole head rice was scanned in a sample holder in the spinning cup module over a wavelength range of 400–2500 nm. Partial least squares (PLS) calculations were performed using Unscrambler (v. 7.6, Camo ASA, Oslo, Norway).
Cooking Methods
Cooked rice was prepared by mixing 100 g of whole head rice and 120 g of water (ratio of rice to water is 1:1.2) and putting it into the Media microcomputer rice cooker for precise cooking module (Media Group Co. Ltd., Foshan, China).
Instrumental Textural Analysis
TPA was conducted with a texture analyzer (TA.XT Plus, Stable Micro System Corp., Godalming, UK). Three warm rice kernels were selected from the same layer (not the bottom or top layer) and almost the same position of the Media rice cooker after the first layer of the rice cooker was removed and then were put onto the base plate. A compression probe was set at 20 mm above the base. A two-cycle compression with 70% strain, force-versus-distance program was used to allow the P36R probe to travel 20.0 mm, return, and repeat. Pretest speed, Test speed, Protest speed were 5, 0.5, and 10 mm/sec, respectively, with a trigger force of 5 g.
The test was repeated on samples (n = 8). Parameters recorded from test curves include hardness (HRD), adhesiveness (ADH), cohesiveness (COH), springiness (SPR), and resilience (RES). Gumminess (GUM) was obtained by multiplying hardness by cohesiveness; chewiness (CHW) was obtained by multiplying springiness by gumminess. Values obtained represent standard calculations of curve attributes of TPA as described by BourneCitation[24] and defined by Lyon et al.Citation[25]
RVA Analysis
RVA measurements were made with a Newport Scientific Super 3 type RVA (Foss North America, Inc., Eden Prairie, MN, USA). Rice samples were run in triplicate by adding 25 mL of distilled water directly into a metal RVA canister to which 3.00 ± 0.01 g of rice flour was added. Standard RVA experimental conditions were used. The temperature was raised from 50 to 95°C ± 1°C in 0 to 5 min. A temperature of 95°C ± 1°C was maintained for 2 min and cooled to 50°C over 7 to 12.5 min. Each experiment was initiated by a 10 sec, 960 rpm mixing period, followed by a 160 rpm paddle speed for the remainder of the data collection.
The definitions for measured properties are pasting temperature: temperature of initial viscosity increase; peak viscosity: maximum viscosity recorded during heating and holding cycles, usually occurs soon after heating cycle reaches 95°C; viscosity hold: minimum viscosity after peak viscosity; final viscosity: viscosity at test finish, corresponds to amylograph cool paste viscosity; breakdown: difference (–) between peak viscosity and viscosity hold, indication of breakdown in viscosity of paste during 95°C holding period; and setback: difference (–) between final viscosity and viscosity hold; peaktime: time required to reach peak.
Rice Blend Design
Three cultivars of highest amylose, medium amylose, and lowest amylose were selected from the 30 cultivars as the original rice varieties for rice blend design. Twenty-one formulas were generated by blending three rice cultivars according to a simple-lattice design shown in The formulas presented in were 3 pure blends at the design points of the vertices estimating main effects (F1–3), 12 binary blends at the edges estimating 2-way interaction effects (F4–15), and 6 center point blends of three components in respective proportion estimating 3-way interaction (F16–21).
Table 1 Compositions of cooked rice in formulations
Table 2 Correlation coefficients of physicochemical and textural properties
Statistical Analysis
Excel software was used for obtaining the mean value of each parameter, and the mean value of each parameter was used for statistical analysis using the SAS system. PROC CORR and PROC FACTOR were used for correlation analysis and principal analysis, respectively. Regression was conducted for each factor that loaded at least one response variable for each group. For rice blending formulas, the 95% confidence limits validation of SAS was conducted to validate if the regression can be used for rice blending and, furthermore, the validated regression could be obtained.
RESULTS AND DISCUSSION
Correlation Between Pasting and Textural Properties
showed that significant correlation coefficients among properties ranged from 0.37 for breakdown versus HRD to 0.79 for setback versus RES, which indicated that pasting properties have good correlation with textural properties. In this study, breakdown was negatively correlated with all the textural parameters. Besides that, viscosity hold, viscosity end, and setback were significantly correlated with all the textural parameters except SPR. In terms of SPR, only three parameters, including peak viscosity, breakdown, and peak time, had significant correlation. And as for peak viscosity, only SPR has significant correlation with it. In this study, the correlation between pasting temperature with all the texture properties were not significant, which confirmed the result that paste temperature seemed to have no significant correlation with the texture parameters.Citation[13] Thus, due to the correlation coefficients, the peak viscosity and peak time are not good for predicting textural properties. It can be clearly found that among all the textural properties, HRD, GUM, and RES have the strongest correlation with the pasting properties, which meant that models for predicting these properties are much more accurate. Therefore, parameters, including viscosity hold, viscosity end, breakdown, and setback, played the most important role in determining the texture among all the RVA parameters, which confirmed the indirect evaluation of texture of cooked rice with RVA parameters,Citation[2] and it also provided the basis for the development of models between these two properties.
Principal Component Factor Analysis
The complex relationship among the RVA parameters measured indicated the multifactor aspect of rice texture evaluation. Principal component factor analysis was carried out among these RVA parameters to extract simpler indicators for rice texture evaluation. showed the principal component factor analysis result. Three factors were derived explaining approximately 90% of the total variance based on the Eigenvalue, among which factor 1 explained the largest proportion of variance (51.96%), meanwhile the other two factors explained the 26.45 and 11.43% proportions, respectively.
Table 3 Loadings of factors for pasting properties of rice cultivars
Table 4 Comparison of regression analysis and variable estimate between optimized models and predicitve models for estimating cooked rice texture from pasting properties
Factor 1 explained the largest proportion of variance (51.96%) and exhibited the highest loading, which were attributes including viscosity hold (0.93366), viscosity end (0.98086), and setback (0.94837). Factor 2 explained 26.45% of data variation. Peak viscosity (0.89245) and breakdown (0.90993) loaded highest on this factor, and only pasting temperature loaded the highest on factor 3, which explained 11.43% of data variation.
Stepwise Multiple Regression Analysis
TPA attributes versus RVA properties with the strongest linear correlations, as indicated by factor loadings above, were subjected to stepwise multiple regression analysis to determine how well the pasting properties predicted cooked rice texture. Results of stepwise multiple regression analysis for the development of models for estimating all the textural attributes of cooked rice from the pasting properties were presented in . The regression analysis showed that viscosity hold, viscosity end, setback, breakdown, and peak viscosity were the major factors making significant contributions in the model development for estimating the cooked rice texture, which was in accordance with the principal component factor analysis.
Among all the texture parameters, the predictive models for estimating the cohesiveness, gumminess, and resilience of cooked rice indicated the highest values of the coefficient of determination (R 2 = 0.6129, 0.4981, and 0.4933, respectively). Meanwhile, the accuracy for hardness, adhesiveness, springiness, and chewiness were relatively lower. Variable estimate results of the predictive model for all the textural attributes of cooked rice from the pasting properties were presented in . It was noticed that all the variables for the predictive models were composed of factor 1 and factor 2 attributes, which were in accordance with the principal component factor analysis.
Among all of the estimates, nearly all of the attribute variables for cohesiveness, gumminess, chewiness, and resilience were estimated with highest accuracy (<.0001), and the intercept estimate for all the textural properties were with the highest accuracy (<.0001) except for hardness (0.0095) and gumminess (0.0001). Combined with the variable estimate, which were all accepted with variant accuracy, the predictive model for estimating all the textural attributes of cooked rice from the pasting properties were developed as follows:
Hardness = 1.08648 × setback + 1297.04724; | |||||
Adhesiveness = 0.00002563 × setback × viscosity end - 286.87574; | |||||
Springiness = −0.00022403 × peak viscosity + 1.50267; | |||||
Cohesiveness = 0.00000003492311 × setback × viscosity hold + 0.31888; | |||||
Gumminess = −1.23797 × breakdown + 0.00026584 × viscosity end × breakdown + 1935.11874; | |||||
Chewiness = 0.75428 × setback − 0.00040895 × setback × breakdown + 866.52206; | |||||
Resilience = 0.00000001600012 × viscosity end × setback + 0.09067. | |||||
Rice Blends
Due to the great contribution of amylose to textural properties, the rice cultivars of Enuo9 (amylose11.60), Guangliangyou558 (amylose24.00), and Yangliangyou419 (amylose28.00) with the largest variance of amylose content were selected as original rice cultivars for the subsequent rice blend experiment. All the pasting and textural properties of the rice blend were measured according to the rice blend design formulas.
Textural Properties of Rice Blends
The TPA textural properties of cooked rice prepared with 21 rice blends were presented in . Hardness (HRD), adhesiveness (ADH), springiness (SPR), cohesiveness (COH), gumminess (GUM), chewiness (CHW), and resilience (RES) of these cultivars ranged from 1771.2 to 3605.1, 240.95 to 48.61, 0.56 to 0.73, 0.37 to 0.47, 648.68 to 1689.78, 472.34 to 1171.46, and 0.13 to 0.23, respectively. It was obviously noticed that increasing the Enuo9 portion resulted in the decrement of hardness and gumminess, and the increment of adhesiveness, which suggested that blending Enuo9 with other cultivars could improve textural properties of hardness, gumminess, and adhesiveness of blends. The 12 binary blends at the edges and 6 center blends of three components in a certain proportion indicated that textural parameters varied depending on the proportion of each rice cultivar.
Table 5 Textural data for rice blends
Pasting Properties of Rice Blends
All the data of pasting properties for 21 formulas of simple-lattice design in this study were shown in . Pasting temperature, peak viscosity, viscosity hold, viscosity end, breakdown, setback, and peak time of rice blends ranged from 75.9 to 80.8, 3011.666667 to 3782, 1435.34 to 1924.34, 2015.34 to 3738.34, 1463 to 2094, 580 to 1901.67, and 4.38 to 5.84, respectively.
Table 6 Pasting data for rice blends
Validation for Application of Predictive Model of Single Rice Cultivars in Rice Blends
The 95% confidence limits validation of the SAS software Version 8.0 procedure was conducted to validate if the predictive regression model of single rice cultivars obtained in this study could be used for rice blends. The validation results of all the textural properties were obtained and the validation results were presented in . For each parameter, there were upper limits and lower limits provided by executing 95% confidence limits validation of the SAS procedure, and meanwhile, the measured value of each parameter for single, binary, and center rice blends would be validated if the value was between the upper limit and lower limit of 21 rice blends. The results showed that nearly all the measured value (>90%) of seven parameters for 3 single, 12 binary, and 6 center rice blends were within the range from lower limit to upper limit, which suggested that the predictive model obtained by stepwise multiple regression analysis of single rice cultivars could be applied in rice blends due to the qualified validation.
Optimization of Predictive Model
Optimization of the predictive model for estimating textural attributes of cooked rice was executed by combining the pasting data, textural data of single rice cultivars, and data of 21 rice blends designed in this study to make stepwise multiple regression analysis. Results of optimized stepwise multiple regression analysis for the development of models for estimating all the textural attributes of cooked rice from the pasting properties were presented in .
It was clearly noticed that the optimized predictive models for estimating the textural properties of cooked rice had the higher values of the coefficient of determination (R 2) than predictive models, respectively, which indicated that optimized predictive models were more accurate than predictive models in estimating the textural properties of cooked rice. Variable estimate results of the optimized predictive model for all the textural attributes of cooked rice from the pasting properties were presented in .
It was obviously noticed that among all of the variable estimate results, |t| increased and Pr > |t| decreased, which suggested that variable estimate results of the optimized predictive model for textural attributes were more accurate than predictive models. Combined with the variable estimates of the optimized model, which were all accepted with variant accuracy, the optimized predictive model for estimating all the textural attributes of cooked rice from the pasting properties were developed as follows:
Table 7 Validation of rice blends
Hardness = 1.35764 × setback + 974.1891; | |||||
Adhesiveness = 0.00002813 × setback × viscosity end − 292.68061; | |||||
Spriginess = −0.00022403 × peak viscosity + 1.50266; | |||||
Cohesiveness = 0.00000003492248 × setback × viscosity hold + 0.31888; | |||||
Gumminess = −1.23797 × breakdown + 0.00026583 × viscosity end × breakdown + 1935.15813; | |||||
Chewiness = 0.99185 × setback − 0.00043958 × setback × breakdown + 600.4648; | |||||
Resilience = 0.00000001599971 × viscosity end × setback + 0.09067. | |||||
Compared with the predictive model, these two models seemed similar except for a slight change in variable estimates; however, the optimized predictive models were with much higher accuracy and could be used in the single rice cultivars as well as rice blends, which meant that the optimized model had a much more extensive application.
CONCLUSIONS
Correlation between pasting and texture properties showed that viscosity hold, viscosity end, breakdown, and setback had significant correlation with textural properties, and the principal component factor analysis indicated that viscosity hold, viscosity end, and setback were the first factor and breakdown was the second factor among the pasting properties. As for paste temperature, in this study, it seemed to have no significant correlation with the textural parameters. The predictive models for estimating texture of cooked rice with high accuracy were developed using both first factors and second factors as variables, and these models have been validated with acceptance for rice blends by conducting a 95% confidence limits validation of a SAS procedure on simplex-lattice design rice blends. Comparing with predictive models, the finally obtained models were much more accurate and with just a slight change in variable estimate, and it also could be extensively used in single rice cultivars as well as rice blends. These results could be helpful to improve the use of low quality rice cultivars by blending high quality with low quality rice to meet the needs of consumers at a relatively lower cost.
ACKNOWLEDGMENTS
This work was supported by the following programs: National High Technology Research and Development Program of China (863 Program), contract grant numbers: 2007AA10Z310, 2007AA100408; Program for New Century Excellent Talents in University, contract grant number: NECT-2007073; Natural Science Funds for Distinguished Young Scholar of Hubei, contract grant number: 2007ABB016.
REFERENCES
- Qingyun , L. , Yeming , C. , Mikami , T. , Kawano , M. and Zaigui , L. 2007 . Adaptability of four-samples sensory tests and prediction of visual and near-infrared reflectance spectroscopy for Chinese indica rice . Journal of Food Engineering , 79 : 1445 – 1451 .
- Champagne , E.T. , Karen , L.B. , Bryan , T.V. , Mcclung , A.M. , Barton , F.E. , Karen , M. , Steve , L. and Kent , M. 1999 . Correlation between cooked rice texture and rapid visco analyser measurements . Cereal Chemistry , 76 : 764 – 771 .
- Suwannaporn , P. and Linnemann , A. 2008 . Rice-eating quality among consumers in different rice grain preference countries . Journal of Sensory Studies , 23 : 1 – 13 .
- Szczesniak , A.S. , Brandt , M.A. and Friedman , H.H. 1963 . Development of standard rating scales for mechanical parameters of texture and correlation between the objective and the sensory methods of texture evaluation . Journal of Food Science , 28 : 397 – 403 .
- Perez , C.M. and Juliano , B.O. 1979 . Indicators of eating quality for non-waxy rice . Food Chemistry , 4 : 185 – 195 .
- Champagne , E.T. , Lyon , B.G. , Min , B.K. , Vinyard , B.T. , Bett , K.L. , Barton , F.E. , Webb , B.D. , McClung , A.M. , Karen , A.M. , Linscombe , S. , Mckenzie , K.S. and Kohlwey , D.E. 1997 . Effects of postharvest processing on rice texture profile analysis . Cereal Chemistry , 75 : 181 – 186 .
- Puspitowati , S. and Driscoll , R.H. 2007 . Effect of degree of gelatinisation on the rheology and rehydration kinetics of instant rice produced by freeze drying . International Journal of Food Properties , 10 : 445 – 453 .
- Israkarn , K. and Charoenrein , S. 2006 . Influence of annealing temperature on T-g’ of cooked rice stick noodles . International Journal of Food Properties , 9 : 759 – 766 .
- Blakeney , A.B. , Welsh , L.A. and Bannon , D.R. 1991 . “ Rice quality analysis using a computer controlled RVA ” . In Cereals International , Edited by: Martin , D.J. and Wrigley , C.W. 180 – 182 . Melbourne : Royal Australian Chemistry Institute .
- Saif , S.M.H , Lan , Y. and Sweat , V.E. 2003 . Gelatinization properties of rice flour . International Journal of Food Properties , 6 : 531 – 542 .
- Sandhyarani , M.R. and Bhattacharya , K.R. 1995 . Rheology of rice flour pastes: Relationship of paste breakdown to rice quality and a simplified brabender viscograph test . Journal of Texture Studies , 26 : 587 – 598 .
- Meadows , F. and Barton , E.F. 2002 . Determination of rapid visco analyser parameters in rice by near-infrared spectroscopy . Cereal Chemistry , 79 : 563 – 566 .
- Yifang , T. and Harold , C. 2002 . Factor analysis of physicochemical properties of 63 rice varieties . Journal of the Science of Food and Agriculture , 82 : 745 – 752 .
- Yifang , T. , Sun , M. and Harold , C. 2002 . Physicochemical properties of an elite rice hybrid . Journal of the Science of Food and Agriculture , 82 : 1628 – 1636 .
- Meullenet , J.F.C. , Gross , J. , Marks , B.P. and Daniels , M. 1998 . Sensory descriptive texture analyses of cooked rice and its correlation to instrumental parameters using an extrusion cell . Cereal Chemistry , 75 : 714 – 720 .
- Juliano , B.O. , Perez , C.M. , Alyo-shin , E.P. , Romanov , V.B. , Blakeney , A.B. , Welsh , L.A. , Choudhury , N.H. , Delgado , L.L. , Iwasaki , T. , Shibuya , N. , Mossman , A.P. , Siwi , B. , Damardjati , D.S. , Suzuki , H. and Kimura , H. 1984 . International cooperative test on texture of cooked rice . Journal of Texture Studies , 15 : 357 – 376 .
- Mohapatra , D. and Bal , S. 2006 . Cooking quality and instrumental textural attributes of cooked rice for different milling fractions . Journal of Food Engineering , 73 : 253 – 259 .
- Gujral , H.S. , Mehta , S. , Samra , I.S. and Goyal , P. 2003 . Effect of wheat bran, coarse wheat flour, and rice flour on the instrumental texture of cookies . International Journal of Food Properties , 6 : 329 – 340 .
- Huaisan , K. , Uriyapongson , J. , Rayas-Duarte , P. , Alli , I. and Srijesdaruk , V. 2009 . Effect of food additives on rheological and textural properties of frozen high amylose rice starch gels . International Journal of Food Properties , 12 : 145 – 161 .
- Gupta , M. , Bawa , A.S. and Semwal , A.D. 2009 . Morphological, thermal, pasting, and rheological properties of barley starch and their blends . International Journal of Food Properties , 12 : 587 – 604 .
- Choi , I.D. , Kim , D.S. , Son , J.R. , Yang , C.I. , Choi , I.S. and Kim , K.J. 2006 . Pasting and texture properties of rice blends formulated with three rice cultivars . Korean Journal of Crop Science , 51 : 292 – 296 .
- Zaidul , I.S.M. , Yamauchi , H. , Kim , S.J. , Hashimoto , N. and Noda , T. 2007 . RVA study of mixtures of wheat flour and potato starches with different phosphorus contents . Food Chemistry , 102 : 1105 – 1111 .
- Standards of the Agricultural Department, People's Republic of China . 1989 . Methods for Rice Quality Assay , Beijing : Chinese Standards Press .
- Bourne , M.C. 1982 . Food Texture and Viscosity , 2nd , New York : Academic Press .
- Lyon , B.G. , Champagne , E.T. , Vinyard , B.T. and Windham , W.R. 1999 . Sensory and instrumental relationships of texture of cooked rice from selected cultivars and postharvest handling practices . Cereal Chemistry , 77 : 64 – 69 .