Integrated recognition and quantitative detection of starch in surimi by infrared spectroscopy and spectroscopic imaging

Accepted Manuscript Integrated recognition and quantitative detection of starch in surimi by infrared spectroscopy and spectroscopic imaging

Shi-Wei Hou, Wei Wei, Yang Wang, Jian-Hong Gan, Ying Lu, Ning-Ping Tao, Xi-Chang Wang, Yuan Liu, Chang-Hua Xu PII: DOI: Reference:

S1386-1425(19)30203-3 https://doi.org/10.1016/j.saa.2019.02.080 SAA 16856

To appear in:

Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy

Received date: Revised date: Accepted date:

6 October 2018 11 January 2019 17 February 2019

Please cite this article as: S.-W. Hou, W. Wei, Y. Wang, et al., Integrated recognition and quantitative detection of starch in surimi by infrared spectroscopy and spectroscopic imaging, Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy, https://doi.org/10.1016/j.saa.2019.02.080

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Integrated recognition and quantitative detection of starch in surimi by infrared spectroscopy and spectroscopic imaging Shi-Wei Houa, Wei Weia, Yang Wangb, Jian-Hong Gana, Ying Lua, Ning-Ping Taoa, Xi-Chang Wanga, Yuan Liuc,*, Chang-Hua Xua, d, e, f,* College of Food Science & Technology, Shanghai Ocean University, Shanghai 201306, PR China

b

First Teaching Hospital of Tianjin University of Traditional Chinese Medicine, Tianjin 300112 PR China

c

Department of Food Science and Technology, School of Agriculture and Biology, Shanghai Jiao Tong

PT

a

University, Shanghai 200240, China

Shanghai Engineering Research Center of Aquatic-Product Processing & Preservation, Shanghai

RI

d

201306, China

Laboratory of Quality and Safety Risk Assessment for Aquatic Products on Storage and Preservation

SC

e

(Shanghai), Ministry of Agriculture, Shanghai 201306, China f

National R&D Branch Center for Freshwater Aquatic Products Processing Technology (Shanghai),

AC

CE

PT E

D

MA

NU

Shanghai 201306, China

*Corresponding author

(Y. Liu) E-mail: [email protected] ; Tel.: +86 18321615886 (C. Xu) E-mail: [email protected] ; Tel.: +86 21 61900380 1

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

Abstract: Surimi products have become increasingly-consumed food with prominent characteristics of high nutrition and convenience and its supply falls short of demand. However, due to exhausted fishery resource in recent years, surimi adulteration, such as addition of plant proteins, starch and other animal origin meat, is becoming serious, so recognition of these exogenous substances has become an urgent issue. In this study, Fourier transform infrared spectroscopy (FT-IR) combined with infrared spectroscopic imaging could distinguish heterogeneity in surimi qualitatively and quantitatively and obtain integral chemical images so that spatial distribution of each component in surimi could be visually displayed, thus a rapid recognition method and a prediction model were developed. The different starch contents in surimi had been primarily identified through intensity change of infrared absorption peaks at 1045 cm-1 and 988 cm-1, specifically with peak shifts to 1041 cm-1 and to 992 cm-1, respectively. In infrared imaging analysis, principal components (PCs) were separated and one key PC was confirmed as starch by characteristic peaks comparison at 1147 cm-1, 1075 cm-1, 997 cm-1 and 930 cm-1. Meanwhile, an established statistic model could predict starch content in surimi correctly with a reliable correlation coefficient (R=0.9856) and root mean square error of prediction (RMSEP=5.64). Therefore, FT-IR combined with infrared spectroscopic imaging could be applicable to integrally recognize and quantitatively detect starch in surimi. Keywords: surimi, starch, FT-IR, infrared spectroscopic imaging, recognize

2

ACCEPTED MANUSCRIPT 1. Introduction

AC

CE

PT E

D

MA

NU

SC

RI

PT

Surimi is a semi-product made from fish flesh after a process of deboning, mincing, washing, dewatering, filtering and then adding sugar and phosphate to protect the protein from frozen denaturation for storage. High-quality surimi could be used to produce different products with the characteristics of high protein, low lipid, good taste, and appropriate elasticity [1, 2]. However, due to diminishing fishery resource and a shortage of high-quality fishes, some surimi manufacturers illegally add low-quality fishes, starch or plant protein instead of actual raw material to improve yield for maximizing economic profits. More seriously, some surimi products, such as fish tofu, fish cake, fish ball, and crab stick, are made of poor-quality fish flesh with an excessive additive like flour. Surimi adulteration drives the development of detection methods, such as immunological detection, protein detection and fluorogenic quantitative PCR. Fluorescence quantitative PCR [3] was employed to detect the content of fish meat in surimi products. Special primer of muraenesox cinereus [4] was designed using its 12S rRNA to identify muraenesox cinereus in surimi products with a high sensitivity (0.1%). The s-ELISA [5] was constructed based on STI polyclonal antibody prepared in immune-feed New Zealand male rabbit to build a quantitative model predicting soybean protein in surimi products with a recovery ranging from 100.1%-122.2%. The sandwich-ELISA of microbial transglutaminase (MTG) [6] was established and the method could detect MTG in frozen surimi quantitatively with a recovery of 94%. Adulterant in grass crap surimi was discriminated through SIMCA model and the content was analyzed by NIR Spectra [7]. These methods have good precision and accuracy but require rigorous and tedious operations. Infrared spectroscopy (IR) has a wide range of applications in food detection as a rapid and nondestructive method. For beef adulteration [8], detection availability of poultry adulterated in beef products was made through mid-infrared (MIR) spectroscopy and near-infrared spectroscopy (NIR), which are considered more concise and convenient to identify meat origins compared to ultraviolet-visible and could be combined with principal component analysis (PCA) and the partial least square (PLS). The MIR integrated with chemometrics was also employed to discriminate frozen fish from fresh fish [9]. Meanwhile, a quantitative model was built with an accuracy up to 100% (calibration set) and 87.5% (validation set) between 3000 cm-1 and 2800cm-1, and 100% (calibration set) and 75% (validation set) between 1500cm-1 and 900cm-1. The VIS/NIR technology [10] was suggested to be a successful method to detect adulteration in crab meat samples adulterated with surimi-based imitation crab meat. For combined detection, the key indicator of soy sauces, amino nitrogen, was quantitatively evaluated by Tri-step infrared spectroscopy with a determination coefficient of 0.9956 and a RMSEP of 0.226 in validation set [11]. Different Formula Granules of Ginseng, Red Ginseng and American Ginseng were identified by Tri-step infrared spectroscopy rapidly and accurately [12]. Hu Wei employed a Tri-step infrared spectroscopy (Fourier transform infrared spectroscopy, second derivative infrared spectroscopy and two-dimensional 3

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

correlation infrared spectroscopy) and cluster analysis to differentiate white croaker surimi, hairtail surimi and gurnard surimi [13]. Results showed that these three kinds of surimi had different protein structures and could be classified by cluster analysis. Zhang Xianyi also employed Tri-step infrared spectroscopy and cluster analysis to discriminate hairtail surimi, red sea bream surimi, gurnard surimi and coarse fish surimi, revealing protein and lipid were the key components to classify different surimi [14]. For white croaker surimi, Tri-step infrared spectroscopy characterized and discriminated four grades of white croaker surimi (A, AA, FA, and SA) combined with soft independent modeling of class analogy (SIMCA) [15]. Infrared spectroscopic imaging is a generalized vibrational spectral image obtained by collecting each pixel spectrum simultaneously, which contains chemical information of local chemical components and displays spatial distribution in microscale. Compared to IR, overlapped spectral information in complex systems can be separated in infrared spectroscopic images so that compositions in different regions can be profiled straightforward. Therefore, chemical compounds are identified by infrared spectra in different regions. Infrared spectroscopic imaging has been widely applied in complex systems such as traditional Chinese medicine and food with advantages of non-destruction and accuracy [16-19]. Other fields like environment, materials, agriculture and physical evidence have also been investigated with the technique [20-25]. Infrared spectroscopy imaging could obtain images and chemical information from localized areas in human hairs to distinguish cuticle, cortex and medulla, and compare spatial resolution of two imaging models [26]. Cross-section of paper [27] was analyzed using infrared spectroscopy imaging and revealed that fourth principal component and fifth principal component have different content of amides in principal component loading diagram. Cotton leaf [28] was analyzed by infrared spectroscopy to obtain the distribution of pesticide residue. Compositions in Liuwei Dihuang teapills, Poria cocos and Niuhuang Jiedu pills were identified [29, 30] through near/mid-infrared spectroscopic imaging without separation-and-extraction process, proving the availability of IR imaging in Chinese herbs. In this study, infrared spectroscopy combined with micro-spectroscopic imaging were applied to integrally and directly recognize starch in surimi and establish a quantitative model to determine starch content, thus to develop an effective and practical technical method for surimi adulteration and other fake surimi products.

2. Materials and methods 2.1 Materials Homemade silver crap surimi: the surimi was made in laboratory to avoid any impurities and illegal additives such as plant protein and starch. Fresh silver craps were purchased in local aquatic product market, then processed through deboning, mincing, washing and dewatering to remove fish bones, entrails, bloods and sarcoplasmic proteins, and finally stabilized with cryoprotectants (0.1% sodium pyrophosphate and 5% white sugar). Corn starch was purchased in supermarket. 4

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

2.2 Apparatus Fourier transform infrared spectrometer and imaging system (Spotlight 400, PerkinElmer Instrument, Inc.) equipped with an ATR (Attenuated Total Reflection) accessory. Vacuum Freeze Drier (BTP-3XLOVX, Virtis, Inc.); Refrigerator (DW-25L262, Haier Inc.); Knifetec Mill (AUX J22). 2.3 Sample preparation The prepared surimi was mixed with corn starch according to proportion of 1%, 2%, 3%, 5%, 8%, 10% ,15%, 20% and 50% (starch/surimi) with 8 duplicates in each group. Surimi in control group contained no starch. All prepared samples were dispensed in polyethylene plastic bags without extra air and reserved in refrigerator at -24°C for further analysis. 2.4 Infrared spectra acquisition All samples were freeze-dried for 24 hours and then pulverized into fine powders. 1-2 mg of each sample and corn starch were blended with KBr powder to press into tablets, respectively. Then spectra were collected by 16 scans in transmission mode, in the range of 4000-600 cm-1 with a resolution of 4 cm-1. Six duplicates were made in each group. 2.5 Infrared spectroscopic image acquisition Samples containing 0, 5%, 10% and 15% starch were chosen to acquire infrared spectroscopic images. Surimi powders were pressed smoothly into tablets with a thickness of 1 mm, then the tablets were immobilized in ATR imaging accessory. IR absorption images were obtained in size of 100 ×100 μm with a spectral resolution of 4 cm–1 in range of 4000–750 cm–1. The pixel size was 1.56 μm and 2 pixels were collected per scanning. 2.6 Model establishment Forty-eight samples (containing 0, 1%, 2%, 5%, 10%, 20%, 50% and 100% starch) were analyzed to build a prediction model. All samples were divided into calibration set and validation set in proportion of 3:1. A quantitative prediction model was established based on partial least squares (PLS) analysis in Spectrum Quant. software (PerkinElmer Instrument, Inc.). 12 samples (containing 3%, 8% and 15% starch) were used to test the applicability of established models. Furthermore, t-test was used to verifying the reliability of the prediction values.

3. Results and discussion 3.1 Infrared spectra analysis 3.1.1 Corn starch Corn starch is a kind of homopolysaccharide composed of D-glucose and can be divided into amylose and amylopectin according to different bonging modes. Amylose is bonded by α-D-1-4-glucosidic bonds while amylopectin by α-D-1-4-glucosidic bonds and α-D-1-6-glucosidic bonds. Amylose and amylopectin in corn starch account for 26% and 74% [31], respectively. IR spectrum (Figure 1) of corn starch in range of 4000-600 cm-1 shows characteristic absorption peaks which reflect molecular structures. Assignment of absorption peaks in starch is summarized in Table 1. The strongest peak appears in 996 cm-1 mainly attributed to the coupled 5

ACCEPTED MANUSCRIPT

D

MA

NU

SC

RI

PT

vibration of C-C and C-O. Besides, the middle strong peaks at 1149 cm-1 and 1076 cm-1 are mainly attributed to stretching vibration of C-O and bending vibration of C-O-H. Peaks at 1125 cm-1 and 1103 cm-1 belong to stretching vibration in ring of C-O and C-C, and peak at 930 cm-1 is assigned to skeleton vibration of α-1,4 glucosidic bond. Weak absorption peaks at 861 cm-1, 763 cm-1 and 708 cm-1 are contributed by D-pyran glucosidic bond and vibration of pyranose ring. These characteristic absorption peaks and their assigned groups indicate the molecular structures of corn starch and evidences determining existence of starch in unknown samples.

PT E

Figure 1. Infrared spectra of corn starch

Table 1 Assignments of absorption peaks in infrared spectra of corn starch [32, 33]

υ(O-H)

Absorption intensity Strong

υ(C-H)

Mid strong

υδ(H-O-H)

Weak

υδ(C-H)

Weak

1336

υδ(C-H)

Mid strong

1149-1147

υ(C-O)

Strong

1125, 1103

υ(C-O)，υ(C-C)

Mid strong

1076, 1075

γ(C-O-H)

Strong

1639 1414

AC

2929

position

Base group and vibration mode

CE

Peak (cm-1) 3285

6

ACCEPTED MANUSCRIPT 999- 996

υ(C-O, C-C) coupled vibration

Strongest

930, 929

α-1,4 glycosidic bonds skeleton vibration Characteristic absorption of D-pyran glycoside bonds Stretching vibration of pyranose ring

Mid strong

861 763,708

Weak Weak

υs: symmetric stretching vibration

PT

υas: asymmetric stretching vibration δ: plane vibration,

RI

δs: scissor vibration γ: bending vibration

AC

CE

PT E

D

MA

NU

SC

3.1.2 Surimi Surimi containing 5%, 10% and 15% corn starch were chosen to collect infrared spectra as practically regular surimi products generally contained a starch proportion from 5% to 20%. Figure 2 shows spectra of surimi containing 5%, 10% and 15% corn starch in range of 1800-700 cm-1. According to the spectra, samples with higher content starch exhibit stronger peak intensity of 988 cm-1 that is originated from cryoprotectants added during surimi processing. In addition, two obvious peak shifts (988 cm-1 to 992 cm-1 and 1045 cm-1 to 1041 cm-1) can be observed with the increase of starch content. These changes could be attributed to starch because its strongest absorption peak is just around 996 cm-1. However, other parts of all samples’ spectra are closely similar, so single peak variation is still not well grounded to determine the existence of starch or other components in surimi.

Figure 2. Infrared spectra of surimi containing different content corn starch in range of 1800-650 cm-1

3.2 Infrared imaging analysis Additives in experimental groups could be primarily recognized by absorption peak comparison in original infrared spectra when compared to control group, but 7

ACCEPTED MANUSCRIPT

NU

SC

RI

PT

exact substances remain to be further analyzed through enhanced spectroscopic method. Infrared imaging is a technology that can be applied in complicated mixture system with concise sample preparations. Information from images such as infrared spectra of all pixels, distribution and size of heterogeneous components are acquired objectively. In Figure 3, optical images of each sample can be hardly distinguished, but different regions in surimi infrared images have diverse infrared absorptions, which is the base to analyze the overlapped information.

AC

CE

PT E

D

MA

Figure 3. Optical images and ATR-IR images of surimi containing starch

Figure 4. Principal components analysis of ATR-IR images of surimi containing starch

Principal components analysis (PCA) was employed to analyze infrared images containing multiple components spectral information and three principal components (PC) were analyzed. Comparison between PC-1 and PC-2 is presented while PC-3 is not be considered for its similarity with PC-2. Obviously, the two PCs are 8

ACCEPTED MANUSCRIPT

NU

SC

RI

PT

well separated and their distribution is displayed clearly (Figure 4). To figure out the chemical essence, three local pixel spectra from each image were viewed and calculated to produce eight average spectra for identification of the two PCs.

D

MA

Figure 5. The pixel spectra of principal components (P1-1: the average spectrum of pixel P11, P12 and P13; P1-2: the average spectrum of pixel P14, P15 and P16; P2-1: the average spectrum of pixel P21, P22 and P23; P2-2: the average spectrum of pixel P24, P25 and P26; P3-1: the average spectrum of pixel P31, P32 and P33; P3-2: the average spectrum of pixel P34, P35 and P36; P4-1: the average spectrum of pixel P41, P42 and P43; P4-2: the average spectrum of pixel P44, P45 and P46)

AC

CE

PT E

In pixel spectra (Figure 5) of pure surimi, P1-1 (originating from PC-1) and P1-2 (originating from PC-2) have characteristic absorption peaks at amide I band (around 1644 cm-1), amide II band (around 1544 cm-1), 1049 cm-1(C-O stretching vibration) and 990 cm-1 (-OCH), which resembles the peak profiles in acquired surimi infrared spectra in Figure 2. Therein, peaks in 1049 cm-1 and 990 cm-1 belongs to saccharides as surimi is mainly a mixture of fish protein and cryoprotectants (usually sodium pyrophosphate and sucrose), so PC-1 and PC-2 could be confirmed as fish protein mixed with a small quantity of saccharides. In surimi containing 5% and 10% corn starch, PC-1 (presented by P2-1 and P3-1) are still surimi mixture (the evidence is the same as above). However, PC-2 (presented by P2-2 and P3-2) has the characteristic absorption peaks at 1147 cm-1 (C-O stretching vibration), 1075 cm-1 (C-O-H bending vibration), 997 cm-1 (C-O, C-C coupling vibration) and 930 cm-1 (glucopyranose ring vibration), which is coincident with starch infrared spectra in Figure 1 where strongest absorption peak also locates at the band of 997 cm-1 with mid-strong peaks at 1147 cm-1, 1075 cm-1 and 930 cm-1. Therefore, PC-2 is recognized as starch by peaks profile comparison (position, shape and absorption intensity). When the starch content is increased to 15% in surimi, P4-1 (originating from PC-1) presents the characteristic peaks of protein (1646 cm-1, 1531 cm-1) and starch (994 cm-1) simultaneously, so PC-1 mainly typifies the mixture of surimi protein and starch. PC-2 is recognized as starch according to peaks profile comparison as above. Hence, it is 9

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

RI

PT

adequate to identify PC-1 and PC-2 to determine the existence of starch adulterated in surimi. In addition, starch distribution in Figure 5 shows that adulterant particle size is about 5-20 μm which is agreement with size of corn starch (5-26 μm) studied by former researchers[31]. Above all, PCA in infrared imaging can spatially recognize and detect starch added in surimi by a holistic and visual way. 3.3 Quantitative analysis 3.3.1 Original spectra In order to establish a convenient and feasible method to quantitatively detect the content of starch in surimi, infrared spectra were analyzed to build a quantitative model. Figure 6 shows the starch-added surimi infrared spectra in which main peaks are 3280 cm-1, 2927 cm-1, 1637 cm-1,1514 cm-1, 1452 cm-1, 1389 cm-1, 1149 cm-1, 1076 cm-1 and 993 cm-1, representing different infrared absorption groups (Table 2). Specifically, absorption peaks at 3280 cm-1, 2927 cm-1 are predominantly contributed by C-H asymmetric stretching vibration in CH3 or CH2; 1514 cm-1 is contributed by rocking vibration of N-H and symmetric stretching vibration of C-N, and 1389 cm-1 belongs to C-H rocking vibration of methyl groups in proteins. In addition, absorption peaks at 1149 cm-1, 1076 cm-1 and 993 cm-1, which are assigned to saccharides, can be found in spectra of samples containing more corn starch than the 5% corn starch. The high content starch strengthened the peaks at 1149 cm-1, 1076 cm-1 and 993 cm-1, suggesting a positive correlation between the peak intensity and corn starch content.

Figure 6. Infrared spectra of surimi containing different content of corn starch in range of 4000cm-1-800cm-1 Table 2 Assignments of absorption peaks in infrared spectra of surimi and corn starch [34, 35]

Peak position / cm-1

Base group and vibration mode

Main attribution

3280，2927

vas(C-H)

CH3，CH2

10

ACCEPTED MANUSCRIPT 1637

vs(C=O)

1514

ρ(N-H)，vs（C-N）

1452

ρ(N-H)，vs(C-N)

1389

ρ(C-H)

Protein methyl

1149

vs(C-O-C)

Polysaccharide

1076

ρ(C-O)

Saccharides

993

vs(C-O)

PT

Protein

Saccharides

υs:symmetric stretching vibration

RI

υas: asymmetric stretching vibration

SC

ρ: rocking vibration

AC

CE

PT E

D

MA

NU

3.3.2 Prediction model To establish a quantitative prediction model for starch in surimi, forty-eight samples (starch content ranging from 0-100%) were analyzed by partial least squares (PLS). Each spectrum was processed through baseline correction and normalization (MSC, multiplicative scatter correction) before modeling in the range of 1200-950 cm-1. Figure 7 shows the prediction results of starch content in surimi calculated by constructed quantitative model based on infrared spectra. The calibration set has a correlation coefficient of 0.9840 with a RMSEC (Root-Mean-Square Error of Calibration) of 5.57 and validation set presents a correlation coefficient of 0.9856 with a RMSEP (Root-Mean-Square Error of Prediction) of 5.64. The prediction values are close to actual starch contents in set samples.

Figure 7. Quantitative plot based on infrared spectra for predicting starch content in surimi

3.3.3 External validation Table 3 The prediction values of starch content in external validation of quantitative model 11

ACCEPTED MANUSCRIPT Prediction values /%

Sample

Starch content /%

1

2

3

4

A

3.00

2.75

4.84

1.22

4.05

3.22

B

8.00

7.88

5.30

7.75

7.69

7.16

C

15.00

14.35

16.19

14.33

15.00

14.97

AVG /%

MA

NU

SC

RI

PT

To verify the applicability and accuracy of the model, three surimi groups marked A, B and C with starch content of 3%, 8% and 15% were prepared for testing, and each group had four duplicates. Table 3 displays the prediction results. Individual predicted value has slight deviations with actual value while three average values, 3.22%, 7.16% and 14.97%, are close to their true starch contents in each group (3%, 8% and 15%, respectively). The subtle differences between prediction results and actual results could be attributed to heterogeneous mixture of surimi and starch, leading to a fluctuation among produced results. With the aim of investigating the significance of these values, the data was statistically analyzed through one sample t-test method in a basic of t-distribution theory. The results were listed in Table 4. According to t value, standard deviation and significance, prediction values in group A, B and C have no significant differences (P>0.05) between prediction values and actual values, thus verifying accuracy of prediction model.

t

Standard deviation

Significance （P）

PT E

Sample

D

Table 4 Results of t-test for external validation of quantitative model

Confidence interval (95%) Lower Upper limit limit

0.271

1.5847

0.804

-2.3067

2.7367

B

-1.364

1.2392

0.266

-2.8169

1.1269

-0.075

0.8724

0.945

-1.4207

1.3557

C

CE

A

AC

In general, detecting starch content in surimi by the constructed model is reliable after verification of external validation and t-test. Comparing to traditional methods, it requires fewer preparations, usually no chemical treatment, and only a small amount of samples to acquire infrared spectra. This is convenient, environment-friendly and cause less damage to samples, presenting promising extensive applicability. 4. Conclusion Corn starch added in surimi was recognized and determined quantitatively by infrared spectroscopy combined with spectroscopic imaging. Main absorption peaks in spectra of corn starch were at 996 cm-1, 1076 cm-1 and 1149 cm-1 while surimi at 998 cm-1 and 1045 cm-1. For samples with varied contents of starch, original infrared spectra can hardly distinguish them effectively because of the overlapping 12

ACCEPTED MANUSCRIPT

RI

PT

information in main absorption peaks. Whereas, infrared imaging combined with principal components analysis could show spatial distribution of surimi and corn starch. Comparison of peaks profiles suggested that PC-1 was surimi and PC-2 was corn starch so that they could be differentiated in a mixed system. In addition, a quantitative model was built in the range of 1200-950 cm-1 to analyze and predict starch content in surimi. Results showed a high prediction correlation coefficient (R=0.9856>0.9800) and acceptable root-mean-square error of prediction (RMSEP=5.64). Furthermore, external validation and t-test verified the veracity of prediction model. Therefore, combined infrared spectral method (spectroscopic imaging and spectral modeling) is a powerful and comprehensive tool to directly and visually identify additives and adulterants in surimi, and separate overlapped information to show their spatial distributions.

MA

NU

SC

Acknowledgement This work was supported by the National Natural Science Foundation of China (Grant No. 31401571), the National Key Research and Development Program of China (2016YFD0401501), Key Projects in the National Science & Technology Pillar Program during the Twelfth Five-year Plan Period (Grant No. 2015BAD17B01 and 2015BAD17B02), Shanghai Engineer Research Center of Aquatic-Product Processing & Preservation (Project No. 16DZ2280300), The Project of Science and Technology Commission of Shanghai Municipality (15320502100), The Key Project of Shanghai Agriculture Prosperity through Science and Technology (2016 (4-4)).

D

Reference

[1] I. Sánchez-González, P. Carmona, P. Moreno, J. Borderías, I. Sánchez-Alonso, A. Rodríguez-Casado,

PT E

M. Careche, Protein and water structural changes in fish surimi during gelation as revealed by isotopic H/D exchange and Raman spectroscopy, Food Chemistry, 106 (2008) 56-64. [2] M. Uddin, E. Okazaki, H. Fukushima, S. Turza, Y. Yumiko, Y. Fukuda, Nondestructive determination of water and protein in surimi by near-infrared spectroscopy, Food Chemistry, 96 (2006) 491-495.

CE

[3] Q. Qinfeng, Detection to main ingredient of surimi products using

adulteration detection

technology (Real-time PCR), in, Fudan University, 2011. [4] P. Hai-yun, Z. Wen-xiu, W. Zhi-he, S. Wen-zheng, Z. Yong, Study on the PCR method specific for

AC

muraenesox cinereus surimi adulteration detection technology, Science and Technology of Food Industry, 33 (2012) 325-326+387. [5] T.L. Jiang, Q.F. Cai, J.D. Shen, M.J. Huang, L.J. Zhang, G.M. Liu, M.J. Cao, Establishment of immunological methods for the detection of soybean proteins in surimi products, Lwt-Food Sci Technol, 64 (2015) 344-349. [6] L. Yi, F. Daming, G. zhennan, H. Jianlian, L. Huizhang, Z. Jianxin, C. Wei, Z. Hao, ELISA-based method to detect microbial transglutaminase in frozen surimi, Modern food science and technology, 30 (2014) 275-279. [7] Y. Qi, Rapid determination of the quality of surimi-based products using near infrared spectroscopy in, Zhejianggongshang University, 2011. [8] C. Alamprese, M. Casale, N. Sinelli, S. Lanteri, E. Casiraghi, Detection of minced beef adulteration with turkey meat by UV-vis, NIR and MIR spectroscopy, Lwt-Food Sci Technol, 53 (2013) 225-232. 13

ACCEPTED MANUSCRIPT [9] R. Karoui, B. Lefur, C. Grondin, E. Thomas, C. Demeulemester, J. De Baerdemaeker, A.S. Guillard, Mid-infrared spectroscopy as a new tool for the evaluation of fish freshness, Int J Food Sci Tech, 42 (2007) 57-64. [10] J. Gayo, S.A. Hale, S.M. Blanchard, Quantitative analysis and detection of adulteration in crab meat using visible and near-infrared spectroscopy, Agriculture and food chemistry, (2006) 1130-1136. [11] L. Zhu, Y. Yan, D.-C. Gu, Y. Lu, J.-H. Gan, N.-P. Tao, X.-C. Wang, C.-H. Xu, Rapid Quality Discrimination and Amino Nitrogen Quantitative Evaluation of Soy Sauces by Tri-Step IR and E-nose, Food Analytical Methods, 11 (2018) 3201-3210. [12] F. Shuaishuai, G. Han, T. Wei, P. Xinhua, N. Liying, Rapid Identification for 3 formula granules of

PT

Ginseng, Red Ginseng and American Ginseng by FT-IR, Drug Evaluation Research 41 (2018) 2242-2247. [13] Y. Liu, W. Hu, X.X. Guo, X.C. Wang, S.Q. Sun, C.H. Xu, Rapid discrimination of three marine fish

RI

surimi by Tri-step infrared spectroscopy combined with Principle Component Regression, Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy, 149 (2015) 516-522.

SC

[14] X.Y. Zhang, W. Wei, W. Hu, X.C. Wang, P. Yu, J.H. Gan, Y. Liu, C.H. Xu, Accelerated chemotaxonomic discrimination of marine fish surimi based on Tri-step FT-IR spectroscopy and electronic sensory, Food Control, 73 (2017) 1124-1133.

NU

[15] W. Hu, X.-X. Guo, X.-C. Wang, Y. Zhao, S.-Q. Sun, C.-H. Xu, Y. Liu, Rapid Discrimination of Different Grades of White Croaker Surimi by Tri-Step Infrared Spectroscopy Combined with Soft Independent Modeling of Class Analogy (SIMCA), Food Analytical Methods, 9 (2015) 831-839.

MA

[16] L.F. Xiao, Z.D. Schultz, Spectroscopic Imaging at the Nanoscale: Technologies and Recent Applications, Anal. Chem., 90 (2018) 440-458.

[17] S.-Q. Sun, C.-W. Li, J.-P. Wei, Q. Zhou, I. Noda, Discrimination of Chinese Sauce liquor using FT-IR and two-dimensional correlation IR spectroscopy, Journal of Molecular Structure, 799 (2006) 72-76.

D

[18] Q. Zhou, S.-Q. Sun, L. Yu, C.-H. Xu, I. Noda, X.-R. Zhang, Sequential changes of main components in different kinds of milk powders using two-dimensional infrared correlation analysis, Journal of

PT E

Molecular Structure, 799 (2006) 77-84.

[19] Q. Zhu, X. Changhua, Z. Qun, W. Jing, F. Xiang, S. Sqin, Quality Analysis of Powdered Milk by Two Dimensional Correlation Infrared Spectroscopy, Chinese Journal of Analytical Chemistry, 32 (2004) 1156-1160.

CE

[20] J. Ryu, C.U. Ro, Attenuated Total Reflectance FT-IR Imaging and Quantitative Energy Dispersive-Electron Probe X-ray Microanalysis Techniques for Single Particle Analysis of Atmospheric Aerosol Particles, Anal. Chem., 81 (2009) 6695-6707.

AC

[21] J.C. Arnold, T. Watson, S. Alston, M. Carnie, C. Glover, The use of FTIR mapping to assess phase distribution in mixed and recycled WEEE plastics, Polymer Testing, 29 (2010) 459-470. [22] M. Mauricio-Iglesias, V. Guillard, N. Gontard, S. Peyron, Application of FTIR and Raman microspectroscopy to the study of food/packaging interactions, Food Addit. Contam. Part A-Chem., 26 (2009) 1515-1523. [23] L.R. Brewer, D.L. Wetzel, Phenotypic expression in wheat revealed using FT-IR microspectroscopy, Vibrational Spectroscopy, 52 (2010) 93-96. [24] J. Chen, S. Sun, Q. Zhou, Direct observation of bulk and surface chemical morphologies of Ginkgo biloba leaves by Fourier transform mid- and near-infrared microspectroscopic imaging, Anal Bioanal Chem, 405 (2013) 9385-9400. [25] R. Bhargava, R.S. Perlman, D.C. Fernandez, I.W. Levin, E.G. Bartick, Non-invasive detection of superimposed latent fingerprints and inter-ridge trace evidence by infrared spectroscopic imaging, 14

ACCEPTED MANUSCRIPT Anal Bioanal Chem, 394 (2009) 2069-2075. [26] K.L.A. Chan, S.G. Kazarian, A. Mavraki, D.R. Williams, Fourier transform infrared imaging of human hair with a high spatial resolution without the use of a synchrotron, Appl. Spectrosc., 59 (2005) 149-155. [27] S. Genest, R. Salzer, G. Steiner, Molecular imaging of paper cross sections by FT-IR spectroscopy and principal component analysis, Anal Bioanal Chem, 405 (2013) 5421-5430. [28] B.O. Budevska, Application of Target Partial Least Squares for Analysis of Fourier Transform Infrared-Attenuated Total Reflection Hyperspectral Images for Characterization of the Distribution of Crop Protection Products on the Leaf Surface, Appl. Spectrosc., 63 (2009) 992-999.

PT

[29] J.B. Chen, S.Q. Sun, Q. Zhou, Infrared microspectroscopic identification of marker ingredients in the finished herbal products based on the inherent heterogeneity of natural medicines, Anal. Bioanal.

RI

Chem., 406 (2014) 4513-4525.

[30] J.B. Chen, S.Q. Sun, F. Ma, Q. Zhou, Vibrational microspectroscopic identification of powdered

SC

traditional medicines: chemical micromorphology of Poria observed by infrared and Raman microspectroscopy, Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy, 128 (2014) 629-637. South China University of Technology, 2010.

NU

[31] W. Qiang, Application of Two-Dimensional Infrared Correlation spectroscopy in starch analysis, in, [32] Z. Movasaghi, S. Rehman, I.U. Rehman, Fourier transform infrared (FTIR) spectroscopy of

MA

biological tissues, Appl. Spectrosc. Rev., 43 (2008) 134-179.

[33] W. Wei, Y. Yan, X.P. Zhang, Y. Liu, Y. Lu, W.Z. Shi, C.H. Xu, Enhanced chemical and spatial recognition of fish bones in surimi by Tri-step infrared spectroscopy and infrared microspectroscopic imaging, Spectroc. Acta Pt. A-Molec. Biomolec. Spectr., 205 (2018) 186-192.

D

[34] X.X. Guo, W. Hu, Y. Liu, S.Q. Sun, D.C. Gu, H.L. He, C.H. Xu, X.C. Wang, Rapid determination and chemical change tracking of benzoyl peroxide in wheat flour by multi-step IR macro-fingerprinting,

PT E

Spectroc. Acta Pt. A-Molec. Biomolec. Spectr., 154 (2016) 123-129. [35] N. Perisic, N.K. Afseth, R. Ofstad, A. Kohler, Monitoring protein structural changes and hydration in bovine meat tissue due to salt substitutes by Fourier transform infrared (FTIR) microspectroscopy,

AC

CE

Journal of agricultural and food chemistry, 59 (2011) 10052-10061.

15

RI

PT

ACCEPTED MANUSCRIPT

AC

CE

PT E

D

MA

NU

SC

Graphical abstract

16

ACCEPTED MANUSCRIPT Highlights 1. Infrared spectra showed characteristic absorption peaks of starch and surimi, reflecting molecular structures. 2. Infrared imaging revealed spatial distribution of starch and surimi to distinguish overlapped information. 3.

A holistic and comprehensive method of detecting starch adulteration in surimi products

AC

CE

PT E

D

MA

NU

SC

RI

PT

has been constructed.

17