Application of NIR spectroscopy for firmness evaluation of peaches

Application of NIR spectroscopy for firmness evaluation of peaches first-author corresp ‡Corresponding Author ‡Corresponding Author †E-mail:ybying@zju.edu.cn †E-mail:ybying@zju.edu.cn ybying@zju.edu.cn College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310029, China College of Biosystems Engineering and Food Science, Zhejiang University, Hangzhou 310029, China The use of near infrared (NIR) spectroscopy was proved to be a useful tool for quality analysis of fruits. A bifurcated fiber type NIR spectrometer, with a detection range of 800~2500 nm by InGaAs detector, was used to evaluate the firmness of peaches. Anisotropy of NIR spectra and firmness of peaches in relation to detecting positions of different parts (including three latitudes and three longitudes) were investigated. Both spectra absorbency and firmness of peach were influenced by longitudes (i, ii, iii) and latitudes (A, B, C). For modeling, two thirds of the samples were used as the calibration set and the remaining one third were used as the validation or prediction set. Partial least square regression (PLSR) models for different longitude and latitude spectra and for the whole fruit show that collecting several NIR spectra from different longitudes and latitudes of a fruit for NIR calibration modeling can improve the modeling performance. In addition, proper spectra pretreatments like scattering correction or derivative also can enhance the modeling performance. The best results obtained in this study were from the holistic model with multiplicative scattering correction (MSC) pretreatment, with correlation coefficient of cross-validation r cv =0.864, root mean square error of cross-validation RMSECV =6.71 N, correlation coefficient of calibration r =0.948, root mean square error of calibration RMSEC =4.21 N and root mean square error of prediction RMSEP =5.42 N. The results of this study are useful for further research and application that when applying NIR spectroscopy for objectives with anisotropic differences, spectra and quality indices are necessarily measured from several parts of each object to improve the modeling performance. Near infrared (NIR) Anisotropy Peach Firmness Partial least square regression (PLSR) INTRODUCTION 1 Near infrared (NIR) spectroscopy technique has gained a great deal of attention in food quality analysis due mainly to its suitability for recording the spectra of solid and liquid samples at low cost without any pretreatment and in a nondestructive way (Sáiz-Abajo et al., 2004 2004 ). The main characteristics hindering its expansion at first is that NIR spectra typically consist of broad, weak, non-specific, extensively overlapped bands (Blanco et al., 2000 2000 ). Until multivariate calibration methods became widely available and accepted, NIR spectroscopy technique has been widely studied and applied. Partial least square regression (PLSR) developed by Wold and Martens in 1983 (Xu et al., 2007 2007 ) is the most widely used method for NIR spectral modeling. The objective of PLS (partial least square) is to define a set of latent variables through the projection of the process and quality spaces onto new orthogonal subspaces by maximizing the covariance between the two spaces (Li et al., 2002 2002 ). In NIR modeling process, the components of the matrix of spectral characteristics are extracted so as to maximize the covariance with the measured absorbance in the set of calibration samples. The relationships are developed from the training set and then applied to the set of unknowns (Hopke, 2003 2003 ). PLS works nicely not only for analyzing the concentrations of specific chemicals but also for analyzing sample properties, such as color, hardness or viscosity, which produce a spectral response. NIR spectroscopy combined with PLSR method was proved to be a powerful tool for quality analysis of fruits such as mango, apple, pear, cherry, citrus, watermelon, etc. by many researchers (Clark et al., 2003 2003 ; Liu et al., 2007 2007 ; Lu R., 2001 2001 ; McGlone et al., 2005 2005 ; Schmilovitch et al., 2000 2000 ; Subedi et al., 2007 2007 ; Ying et al., 2005 2005 ; Zou et al., 2007 2007 ; Lu H.S. et al., 2006 2006 ; Tian et al., 2007 2007 ). Fruits are anisotropic objects, appearing different physical and chemical properties or characteristics when measured in different directions. Wang ( 2004 2004 ) studied the mechanical properties including failure stress, failure strain, failure energy and Young’s modulus of pear tissue related to specimen orientation and location. His study showed that the mechanical properties were significantly affected by specimen orientation, latitude and depth, but not affected by the longitude of the specimen. Khan and Vincent ( 1993 1993 ) did crack-opening tests (wedge penetration tests and notch tensile tests) on apple flesh. Their results showed that the tissue had marked anisotropy in its fracture properties. The flesh of the apple is split much more easily along the fruit’s radius than, for example, in a direction parallel to the fruit’s tangent. Also Khan and Vincent ( 2006 2006 ) investigated the parenchyma of a range of apple varieties, either sectioned in a cryotome and observed under a light microscope, or freeze-dried and viewed under a scanning electron microscope. The morphology of the material varied according to the direction from which it was viewed. This anisotropy phenomenon increased from the outside to the inside of the cortex. Fruit firmness, the ability of fruit flesh to resist compressive force, is one of the most important characteristics to estimate fruit maturity and quality. Generally, firmness is detected either by touch or by hand-held penetrometer measurement. Both these two measurements are destructive and the detected firmness can vary greatly with the skill and care taken by the operator. Recently, measurement of the optical properties of fruits has been one of the most successful nondestructive techniques for quality assessment. Optical techniques, NIR spectroscopy for instance, have received considerable attentions as a means for nondestructive evaluation of fruit firmness (van Dijk et al., 2006 2006 ; Lu and Ariana, 2002 2002 ; McGlone and Kawano, 1998 1998 ; Park et al., 2003 2003 ). However, the above studies of fruit firmness evaluation using NIR spectroscopy scarcely take fruit anisotropy into account. They mostly selected 1~4 points on the fruit for spectra acquisition and quality measurement. For example, McGlone and Kawano ( 1998 1998 ) made two measurements on opposite sides of the fruit and averaged to give a mean penetrometer firmness for the fruit; Park et al.( 2003 2003 ) measured the firmness at four locations on each apple; Lu and Ariana ( 2002 2002 ) and van Dijk et al.( 2006 2006 ) performed singular compression measurement on each fruit. The objectives of this research were: (1) to investigate the anisotropy of the NIR spectra and firmness of white peaches related to different longitudes and latitudes; (2) to compare the performance of models using NIR spectra and firmness measured from different longitudes and latitudes of each fruit by PLSR method; (3) to investigate the usefulness of spectra pretreatments and nonlinear PLSR and weighted PLSR methods for improving modeling performance. MATERIALS AND METHODS 1 Samples 2 A total of 180 white peaches were purchased from Jinhua County, Zhejiang Province and transported to the laboratory on June 30, 2006. Samples were stored at 25 °C and 60% relative humidity (RH) during the several days of experiment. Two thirds of the samples (120 peaches) were used for calibration models and the remaining one third (60 peaches) were used for prediction models. For each sample, spectra acquisition and firmness measurement were carried out in 2 h. The primary characters of the peaches were described in Table 1 1 . SD : Standard deviation Max. dia. (mm) Min. dia. (mm) Height (mm) Weight (g) Max. 78.74 76.10 81.28 245.44 Min. 62.76 64.80 67.54 170.39 Average 73.89 70.10 73.66 203.92 SD 3.22 2.57 2.91 20.10 Experimentation 2 1. NIR spectra acquisition 3 NIR diffuse reflectance spectra were acquired using a Fourier transform (FT)-NIR spectrometer (Thermo Electron Corp., USA) equipped with an interferometer, an InGaAs detector, a bifurcated fiber-optic probe, and a quartz halogen light source (50 W). Software of OMINC 6.1a (Thermo Electron Corp., USA) was used for spectrometer setup and spectra acquiring operation. The parameters of spectrometer setup were as follows: resolution of 16 cm −1 , scan number of 64 times, mirror velocity of 0.9494 cm/s, and spectral range of 800~2500 nm. For each sample, nine spectra were collected corresponding to three longitudes (i, ii, iii) and three latitudes (A, B, C) as shown in Fig. 1 1 . The three longitudes were about 120° apart from each other. One of the three latitudes was at the equator and the other two were close to the stem end and the calyx end. Samples were placed directly upon the surface of the fiber probe by hand. Before peach spectra acquisition, a reference spectrum was collected from a whit Teflon block. The diffuse reflectance spectra of samples were saved as log(1/ R ), where R is the ratio of light intensity reflected from the sample to that reflected from the Teflon block. 2. Firmness measurement 3 The firmness measurements were conducted using a standard 6-mm MT (Magness-Taylor) probe mounted in an Instron 5543 universal testing machine (Instron Corp., USA) with a loading rate of 20 mm/min, and interfaced to a computer to obtain continuous force-deformation curves. The firmness was measured at the corresponding locations where NIR diffuse reflectance spectra were acquired on each sample. Only maximum force (N) for penetration was used for the firmness test. 3. Data analysis 3 Nine diffuse reflectance spectra were collected: (iA), (iB), (iC), (iiA), (iiB), (iiC), (iiiA), (iiiB) and (iiiC). Analysis of variance (ANOVA) was applied to observe the anisotropic differences both of spectra and firmness. Spectra pretreatments and statistical modeling were carried out using software TQ Analyst (Thermo Electron Corp., USA). Spectra pretreatments include scattering correction and derivative. PLSR method was used to establish the chemometrics models. The calibration models include two validation procedures: cross-validation and external validation. Cross-validation was performed on the calibration samples based on excluding one observation from the calibration model, which is leave-one-out cross-validation. Since the validation samples in cross-validation procedure are from the calibration sample set, a comparable kind of spectral variability can be expected. External validation was performed on the prediction sample set. In this way, the ability of the calibration model to withstand unknown variability is assessed. Beside linear PLSR, other two PLSR methods (nonlinear PLSR and weighted PLSR) were also to see whether the modeling performance could be improved. The accuracy of the models was defined as correlation coefficient of calbration and cross-validation ( r and r cv ), root mean square errors of calibration, cross-validation and prediction ( RMSEC , RMSECV and RMSEP ). RESULTS AND DISCUSSION 1 Spectra anisotropic differences 2 NIR diffuse reflectance spectra were recorded at 1102 points in the region of 800~2500 nm. Fig. 2 2 shows the average spectrum of all the samples and the standard deviation (variance) of spectral data at 1102 points. The standard deviation obviously decreased above 1400 nm and revealed two vales around 1450 nm and 1940 nm corresponding to the two peaks on average spectrum, which are related to O-H first overtones and O-H stretching vibration and deformation vibration of water. In order to investigate spectra anisotropic difference of peaches, ANOVA was applied to each sample by calculating the F values of three average spectra in the full wavelength range related to longitude or latitude. To take longitude (i, ii, iii) for example, the three spectra at each longitude were averaged, and the F value of the absorbance of three average spectra (S-i, S-ii, S-iii) were calculated by ANOVA method. The result revealed an F value and an F critical value ( α =0.05). By comparing the two values, it can be clearly found out whether the absorbance of the spectra at different longitudes changed obviously or not. The same procedure was applied to three latitudes (A, B, C). Each sample has two F values corresponding to longitude and latitude, respectively. Because F critical value is only related to the freedom degrees ( df ) of the dataset for ANOVA procedure, this value was the same for all the samples. Fig. 3 3 shows the F values of longitude and latitude variance analyses of spectra in the full wavelength range for each sample. According to F -verification principle, the absorbance of the samples above the F critical line (more than two thirds) changed obviously with longitude or latitude variance, while the samples below the F critical line did not. Therefore, it is hard to conclude that longitude or latitude will influence the spectral absorbance. Another way was tried to investigate this problem. To take longitude (i) for example, the average absorbance of the averaged spectrum at longitude (i) of each sample was calculated firstly, and 180 average absorbance values were obtained. Applying the same procedure to the other two longitudes (ii, iii), a matrix (3×180) can be obtained, 3 representing three longitudes and 180 representing the number of samples. Also using ANOVA method, the F value can be obtained to determine whether the average absorbency changed obviously with longitudes or not. The same procedure was used to observe the spectral changes with latitudes. Table 2 2 shows the ANOVA results of average absorbency of different longitudes and latitudes. Both for longitude and latitude, F 0.05 < F < F 0.01 , concluding that the average absorbency changed obviously with longitude and latitude at level 0.05 but not at level 0.01. 2 2 Source 2 SS 2 df 2 MS 2 F value 2 F critical value F 0.05 F 0.01 Longitude Factor 0.027863 2 0.013931 4.599812 3.012507 4.64489 Error 1.626393 537 0.003029 Total 1.654256 539 Latitude Factor 0.024709 2 0.012354 4.528424 3.012507 4.64489 Error 1.465041 537 0.002728 Total 1.489750 539 Firmness anisotropic differences 2 Corresponding to the spectra collecting positions of each sample, nine firmness measurements were carried out. Firmness anisotropic differences were also analyzed by ANOVA method. The ANOVA results ( F values) for each longitude and latitude factor were shown in Fig. 4 4 . All F values were greater than F critical value at level 0.01, which indicated that firmness of peaches was influenced by longitude and latitude. And latitude effect was more evident than longitude effect. NIR modeling 2 1. Influence of anisotropic differences 3 From the results of the above section, it is necessary to observe the influence of anisotropic differences on NIR modeling results, for both absorbance of NIR spectra and firmness are changed in different degrees along with longitudes and latitudes. Seven calibration models were established using PLSR method. For each sample, three spectra at each longitude were averaged. For a certain longitude or latitude model, the average spectra of all 180 samples corresponding to this longitude or latitude were used. In addition, a holistic model was established using the mean spectrum of nine spectra (corresponding to three longitudes and three latitudes) of each sample. Table 3 3 shows the results of these models, including r , r cv , RMSEC , RMSECV and RMSEP . The results of the holistic model were better than those of other models for a certain longitude or latitude, with relatively high r and r cv . The small differences between RMSEC , RMSEP and RMSECV indicate that the model was robust not only for the observations in the calibration dataset but also for external samples. 2 Factors r cv RMSECV (N) r RMSEC (N) RMSEP (N) Longitude i 5 0.810 8.20 0.834 7.70 7.82 ii 2 0.867 6.23 0.875 6.06 7.18 iii 5 0.800 8.51 0.829 7.93 7.49 Latitude A 6 0.780 8.89 0.846 7.52 7.11 B 5 0.871 6.78 0.889 6.34 7.04 C 5 0.823 7.16 0.846 6.72 7.61 2 Holistic model 6 0.871 6.52 0.894 5.94 5.96 2. Improvement of the holistic model 3 The models established were based on original spectra using linear PLSR method. In order to improve the modeling performance, spectra pretreatments including scattering correction and derivative and the other two PLSR methods (nonlinear PLSR and weighted PLSR) were applied to the holistic model. Multiplicative scattering correction (MSC) or the first derivative treatments can improve the modeling performance, with r increasing to 0.909 or 0.948. When both MSC and the first derivative treatments were applied, r increasing to 0.981, but the difference between RMSEC and RMSEP also increasing ( RMSEC =4.21 N, RMSEP =7.48 N). Nonlinear PLSR method educed the same results as linear PLSR method did, while weighted PLSR method educed worse results. Fig. 5 5 shows the PLSR modeling results with MSC pretreatment of the holistic model. The RMSEP of this model was 5.42 N. CONCLUSION 1 This study indicates the anisotropic differences of NIR spectra and firmness of white peach fruit. NIR spectroscopy together with PLSR method has been proved to be a valid tool for peach firmness prediction. PLSR results of models for different longitudes and latitudes spectra and the holistic model show that collecting several NIR spectra from different longitudes and latitudes of a fruit for NIR calibration modeling can improve the modeling performance. In addition, proper spectra pretreatments like scattering correction and derivative also can enhance the modeling performance. The best results obtained in this study were from the holistic model with MSC pretreatment, with r cv =0.864, RMSECV =6.71 N, r =0.948, RMSEC =4.21 N and RMSEP =5.42 N. The results of this study are useful for further research and application that when applying NIR spectroscopy for objectives with anisotropic differences, spectra and quality indices are necessarily measured from several parts of each object to improve the modeling performance. References M Blanco , J Coello , H Iturriaga , S Maspoch , J Pages . NIR calibration in non-linear systems: different PLS approaches and artificial neural networks . Chemometrics and Intelligent Laboratory Systems , 2000 . 50 ( 1 ): 75 - 82 . DOI: 10.1016/S0169-7439(99)00048-9 http://doi.org/10.1016/S0169-7439(99)00048-9 . NIR calibration in non-linear systems: different PLS approaches and artificial neural networks CJ Clark , VA McGlone , C Requejo , A White , AB Woolf . Dry matter determination in ‘Hass’ avocado by NIR spectroscopy . Postharvest Biology and Technology , 2003 . 29 ( 3 ): 301 - 308 . DOI: 10.1016/S0925-5214(03)00046-2 http://doi.org/10.1016/S0925-5214(03)00046-2 . Dry matter determination in ‘Hass’ avocado by NIR spectroscopy PK Hopke . The evolution of chemometrics . Analytica Chimica Acta , 2003 . 500 ( 1-2 ): 365 - 377 . DOI: 10.1016/S0003-2670(03)00944-9 http://doi.org/10.1016/S0003-2670(03)00944-9 . The evolution of chemometrics AA Khan , JFV Vincent . Anisotropy in the fracture properties of apple flesh as investigated by crack-opening tests . Journal of Materials Science , 1993 . 28 ( 1 ): 45 - 51 . DOI: 10.1007/BF00349031 http://doi.org/10.1007/BF00349031 . Anisotropy in the fracture properties of apple flesh as investigated by crack-opening tests AA Khan , JFV Vincent . Anisotropy of apple parenchyma . Journal of the Science of Food and Agriculture , 2006 . 52 ( 4 ): 455 - 466 . DOI: 10.1002/jsfa.2740520404 http://doi.org/10.1002/jsfa.2740520404 . Anisotropy of apple parenchyma B Li , J Morris , EB Martin . Model selection for partial least squares regression . Chemometrics and Intelligent Laboratory Systems , 2002 . 64 ( 1 ): 79 - 89 . DOI: 10.1016/S0169-7439(02)00051-5 http://doi.org/10.1016/S0169-7439(02)00051-5 . Model selection for partial least squares regression YD Liu , YB Ying , XP Fu , HS Lu . Experiments on predicting sugar content in apples by FT-NIR technique . Journal of Food Engineering , 2007 . 80 ( 3 ): 986 - 989 . DOI: 10.1016/j.jfoodeng.2006.06.035 http://doi.org/10.1016/j.jfoodeng.2006.06.035 . Experiments on predicting sugar content in apples by FT-NIR technique HS Lu , HR Xu , YB Ying , XP Fu , HY Yu , HQ Tian . Application Fourier transform near infrared spectrometer in rapid estimation of soluble solids content of intact citrus fruits . J Zhejiang Univ Sci B , 2006 . 7 ( 10 ): 794 - 799 . 16972321 DOI: 10.1631/jzus.2006.B0794 http://doi.org/10.1631/jzus.2006.B0794 . Application Fourier transform near infrared spectrometer in rapid estimation of soluble solids content of intact citrus fruits R Lu . Predicting firmness and sugar content of sweet cherries using near-infrared diffuse reflectance spectroscopy . Transactions of the ASAE , 2001 . 44 ( 5 ): 1265 - 1271 . . Predicting firmness and sugar content of sweet cherries using near-infrared diffuse reflectance spectroscopy R Lu , D Ariana . Determination of firmness of intact apple fruit by near-infrared reflectance spectroscopy . Applied Engineering in Agriculture , 2002 . 18 ( 5 ): 585 - 590 . . Determination of firmness of intact apple fruit by near-infrared reflectance spectroscopy VA McGlone , S Kawano . Firmness, dry-matter and soluble-solids assessment of postharvest kiwifruit by NIR spectroscopy . Postharvest Biology and Technology , 1998 . 13 ( 2 ): 131 - 141 . DOI: 10.1016/S0925-5214(98)00007-6 http://doi.org/10.1016/S0925-5214(98)00007-6 . Firmness, dry-matter and soluble-solids assessment of postharvest kiwifruit by NIR spectroscopy VA McGlone , PJ Martinsen , CJ Clark , RB Jordan . On-line detection of Brownheart in Braeburn apples using near infrared transmission measurements . Postharvest Biology and Technology , 2005 . 37 ( 2 ): 142 - 151 . DOI: 10.1016/j.postharvbio.2005.04.011 http://doi.org/10.1016/j.postharvbio.2005.04.011 . On-line detection of Brownheart in Braeburn apples using near infrared transmission measurements B Park , JA Abbott , KJ Lee , CH Choi , KH Choi . Near-infrared diffuse reflectance for quantitative and qualitative measurement of soluble solids and firmness of Delicious and Gala apples . Transactions of the ASAE , 2003 . 46 ( 6 ): 1721 - 1731 . . Near-infrared diffuse reflectance for quantitative and qualitative measurement of soluble solids and firmness of Delicious and Gala apples MJ Sáiz-Abajo , JM González-Sáiz , C Pizarro . Classification of wine and alcohol vinegar samples based on near-infrared spectroscopy. Feasibility study on the detection of adulterated vinegar samples . Journal of Agricultural and Food Chemistry , 2004 . 52 ( 25 ): 7711 - 7719 . 15675824 DOI: 10.1021/jf049098h http://doi.org/10.1021/jf049098h . Classification of wine and alcohol vinegar samples based on near-infrared spectroscopy. Feasibility study on the detection of adulterated vinegar samples Z Schmilovitch , A Mizrach , A Hoffman , H Egozi , Y Fuchs . Determination of mango physiological indices by near-infrared spectroscopy . Postharvest Biology and Technology , 2000 . 19 ( 3 ): 245 - 252 . DOI: 10.1016/S0925-5214(00)00102-2 http://doi.org/10.1016/S0925-5214(00)00102-2 . Determination of mango physiological indices by near-infrared spectroscopy PP Subedi , KB Walsh , G Owens . Prediction of mongo eating quality at harvest using short-wave near infrared spectrometry . Postharvest Biology and Technology , 2007 . 43 ( 3 ): 326 - 334 . DOI: 10.1016/j.postharvbio.2006.09.012 http://doi.org/10.1016/j.postharvbio.2006.09.012 . Prediction of mongo eating quality at harvest using short-wave near infrared spectrometry HQ Tian , YB Ying , HS Lu , XP Fu , HY Yu . Measurement of soluble solids content in watermelon by Vis/NIR diffuse transmittance technique . J Zhejiang Univ Sci B , 2007 . 8 ( 2 ): 105 - 110 . 17266185 DOI: 10.1631/jzus.2007.B0105 http://doi.org/10.1631/jzus.2007.B0105 . Measurement of soluble solids content in watermelon by Vis/NIR diffuse transmittance technique C van Dijk , C Boeriu , T Stolle-Smits , LMM Tijskens . The firmness of stored tomatoes (cv. Tradiro ). 2. Kinetic and near infrared models to describe pectin degrading enzymes and firmness loss . Journal of Food Engineering , 2006 . 77 ( 3 ): 585 - 593 . DOI: 10.1016/j.jfoodeng.2005.07.017 http://doi.org/10.1016/j.jfoodeng.2005.07.017 . The firmness of stored tomatoes (cv. Tradiro). 2. Kinetic and near infrared models to describe pectin degrading enzymes and firmness loss J Wang . Mechanical properties of pear as a function of location and orientation . International Journal of Food Properties , 2004 . 7 ( 2 ): 155 - 164 . DOI: 10.1081/JFP-120025392 http://doi.org/10.1081/JFP-120025392 . Mechanical properties of pear as a function of location and orientation L Xu , JH Jiang , WQ Lin , YP Zhou , HL Wu , GL Shen , RQ Yu . Optimized sample-weighted partial least squares . Talanta , 2007 . 71 ( 2 ): 561 - 566 . DOI: 10.1016/j.talanta.2006.04.039 http://doi.org/10.1016/j.talanta.2006.04.039 . Optimized sample-weighted partial least squares YB Ying , YD Liu , JP Wang , XP Fu , YB Li . Fourier transform near-infrared determination of total soluble solids and available acid in intact peaches . Transactions of the ASAE , 2005 . 48 ( 1 ): 229 - 234 . . Fourier transform near-infrared determination of total soluble solids and available acid in intact peaches XB Zou , JW Zhao , XY Huang , YX Li . Use of FT-NIR spectrometry in non-invasive measurements of soluble solid contents (SSC) of ‘Fuji’ apple based on different PLS models . Chemometrics and Intelligent Laboratory Systems , 2007 . 87 ( 1 ): 43 - 51 . DOI: 10.1016/j.chemolab.2006.09.003 http://doi.org/10.1016/j.chemolab.2006.09.003 . Use of FT-NIR spectrometry in non-invasive measurements of soluble solid contents (SSC) of ‘Fuji’ apple based on different PLS models 10.1631/jzus.B0720018 Project supported by the National Natural Science Foundation of China (No. 30671197) and the Program for New Century Excellent Talents in University, China (No. NCET-04-0524) 2007-12-28 2008-04-06 2022-01-05 Copyright © 2008, Journal of Zhejiang University Science 2008 Journal of Zhejiang University-SCIENCE B (Biomedicine and Biotechnology) 07 09