Explore UCD

UCD Home >

Research Projects in 2017-2018

Wednesday, 28 November, 2018

Effective postharvest preservation methods for mushroom (agricus bisporus) (PhD)

Zhang KX, Pu YY and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Mushroom (Agaricus bisporus) is a popular agri-food product that is being produced worldwide. However, the short shelf life of mushrooms is an issue limiting their commercial value. To extend the shelf life, preservation techniques for postharvest mushrooms quality are essential. This review briefly summarized a number of preservation methods for Agaricus bisporus mushrooms, including packaging, cooling, drying, washing, and coating.

Combination of novel extraction technologies for bioactive compounds in the food industry (PhD)

Wen L, Tiwari BK and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Novel extraction methods are considered as clean, green and efficient alternative to conventional extraction technologies. This study focuses on the combination of novel extraction technologies for bioactive compounds in the food industry. The applications of novel extraction technologies in the food industry have been widely studied in recent decades. In the first part, the novel extraction methods, including ultrasound-assisted extraction (UAE), microwave-assisted extraction (MAE) and enzyme-assisted extraction (EAE), were discussed with reference to the principles and mechanisms of action as well as the comparison with traditional methods. In the second part, a brief review was given about the combinations of these novel extraction methods which are: ultrasound-assisted enzymatic extraction (UAEE), microwave-assisted enzymatic extraction (MAEE) and ultrasonic microwave-assisted extraction (UMAE).

Quality evaluation of mackerel fillets after high pressure treatment (PhD)

Zhao YM, de Alba Ortega M, Da-Wen Sun and Tiwari B

Sponsors: CSC-UCD Scholarship Scheme

High pressure processing (HPP) is a novel, non-thermal technology, which has been widely researched in food industry. The present study aimed to investigate microbial and physicochemical parameters of fresh mackerel fillets after high pressure treatment at 100, 300 and 500 MPa during 2 and 5 min at 10℃. The results demonstrated that total viable counts (TVC) decreased with the most intense treatments, while H2S-producing bacteria load decreased to undetectable level except for treatments of 100 MPa during 2 and 5 min. Compared with untreated samples, minor changes on hardness were observed except for the treatment of 500 MPa, when significant increase (p<0.05) was detected. No significant differences on lipid oxidation were observed. L* increased and a* decreased with the most intense treatments except for 100 MPa during 2 and 5 min, however, no significant changes (P>0.05) were detected on b*. HPP was effective in inactivating microorganisms on mackerel fillets, but it also affected some physicochemical attributes.

Hyperspectral imaging as non-destructive assessment tool for the recognition of sweet potato cultivars (PhD)

Su WH and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

The reliability and accuracy of hyperspectral imaging combined with chemometrics technique were investigated for recognition and classification of sweet potato cultivars. Hyperspectral images of tuber samples were acquired and corrected. Then the extracted data were analyzed by principal component analysis (PCA) and partial least squares discriminant analysis (PLSDA) for identification of different sweet potato categories. The spectra were pre-processed by second derivative (2nd Der). Then, eight important wavelengths (1074, 1125, 1155, 1202, 1295, 1342, 1376 and 1406) were identified as characteristic wavelengths based on 2nd Der spectra. These resulting wavelengths were used in PLSDA for classification of sweet potato samples, yielding 100 % overall classification accuracy in the cross-validation set. The results obtained in this study clearly showed that the combination of hyperspectral imaging and multivariate analysis has a great potential as a novel and rapid approach for identification and authentication of sweet potato cultivars.

Rapid classification of white stripe and red muscle pixels in salmon fillet by using near-infrared hyperspectral imaging combined with PCA and MCR (PhD)

Xu JL and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

This study was carried out to investigate the potential of a near infrared (900-1700 nm) hyperspectral imaging (HSI) system for discrimination of white stripe and red muscle pixels in conventional farm-raised salmon fillets. Hyperspectral cubes were acquired, calibrated and their corresponding spectra data were extracted. Principal component analysis (PCA) was applied to explore the variance. Satisfied unsupervised classification of 650 pixels of white stripe and 650 of red muscle was obtained by using PCA. Multivariate curve resolution (MCR) was then applied to build classification map with good performance achieved. This study confirmed the capability of the hyperspectral imaging for objective and rapid categorization of the white stripe and red muscle pixels of salmon fillets.

Ripeness classificaiton of bananito fruits using visible hyperpsectral imaging (PhD)

Pu YY, Da-Wen Sun, Buccheri M, Grassi M, Cattaneo TMP and Gowen A

Sponsors: CSC-UCD Scholarship Scheme

Non-destructive and accurate determination of fruit maturity stage is important for its shelf-life prediction and marketability. This study investigated the use of visible hyperspectral imaging (Vis-HSI) technique (400-740 nm, 5 nm interval) for rapid and non-destructive classification of bananito fruits (Musa acuminata). Bananitos peel and flesh images were collected and classification models based on partial least square discriminant analysis (PLSDA) were established. Results showed that the peel Vis-HSI data performed better than flesh Vis-HSI data for ripeness stage classification. The PLSDA model based on peel spectra achieved a total correct classification rate of 93.3%, while the PLSDA model based on flesh spectra yielded a total correct classification rate of 83.3%. This study demonstrates the potential of using visible hyperspectral imaging system to acquire peel spectra for non-destructive classification of bananitos maturity stage.

Measurement of tenderness of red meats using hyperspectral imaging: A brief review (MEngSc)

Zhu WY, Su WH and Da-Wen Sun

Sponsors: University College Dublin

Hyperspectral imaging (HSI) has gained wide regard as a fast, chemical-free, and nondestructive method for detecting meat quality in the modern meat industry. It, has ability for effectively quantifying and characterizing quality attributes. This study focuses on the recent applications of hyperspectral imaging on quality assessment for sensory attributes of red meat. The first part of this review is the basic principles, major instrumental components and data analysis methods of hyperspectral imaging. The second part is to review of hyperspectral imaging on measurements of sensory attributes of red meat.

Preliminary studies on NIR hyperspectral imaging in visualizing moisture distribution of mushroom slices during microwave-vacuum drying (PhD)

Lin XH, Xu JL and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

The moisture content (MC) of the mushroom slices being processed during microwave-vacuum drying (MVD) is crucial to evaluate the qualities of final drying mushroom slices. Near-infrared (NIR) hyperspectral imaging in combination with multivariate chemometric analysis were employed for moisture prediction. A model based on the full range wavelengths was developed using partial least squares regression (PLSR) and applying spectral pre-treatment of a SNV to the mushroom slice hyperspectral images, resulted in the best model performance. An optimized PLSR model, achieving a high prediction accuracy with Rcv2 = 0.967 and RMSECV = 5.22%. The result shows that the NIR hyperspectral imaging technique is a promising tool for non-destructively and rapidly measuring and visualizing the moisture content of mushroom slices during MVD.

The effect of plasma treatments on rheological properties on tapioca starch (PhD)

Zhang K, Tiwari B and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Starch is the most abundant reserve carbohydrate in plants and the most common used polysaccharide in human diet. What’s more, it also has many industrial applications such as thickener, colloidal stabilizer, gelling agents, bulking agent, water retention agent and adhesive. There is a high potential explore its chemical modifications during food processing. The objective of my research is to investigate the effect of different gases induced plasma treatments on rheological properties of tapioca starch. In this research, 18 samples of native tapioca starch (2 control samples, 16 treated samples) were used. The samples in treatment groups were treated in a bell-jar DBD plasma system under different gas agents (composed air, CO2, Argon and helium, respectively) by 20 or 30 min. Then the changes in starch rheological properties were estimated by Rapid Visco Analysis (RVA). During RVA test, controlled temperature (50-90 °C) and shear speed (0-960 rpm) were applied on the samples. After analysis by SPSS statistic, significant differences between plasma treated tapioca starches and control samples were found on viscosity values of pleak1, trough1, and breakdown as well as peak time.

In a conclusion, different gases mediated plasma treatments have effects on tapioca starch rheological properties, in terms of peak viscosity, trough viscosity, breakdown viscosity and peak temperature. However, the mechanisms causing these changes still need to be further investigated.

Ultrasound assisted extraction of bioactive compounds from coffee silver skin with study of extraction kinetics (PhD)

Wen L, Da-Wen Sun and Tiwari B

Sponsors: CSC-UCD Scholarship Scheme

Currently employed techniques for extraction of bioactives require long extraction times and have low extraction efficiencies. There is a need to develop novel extraction methods with improved extraction rates and yields. Ultrasound is a potential novel extraction technology which can be considered as clean, green and efficient alternative to conventional extraction technologies. This study investigated the ultrasound assisted extraction of bioactive compounds including phenolics and caffeine from coffee silver skin. In addition, it studied the effect of ultrasound pre-treatment on bioactive extraction kinetics. Ultrasound pre-treatment was carried out using 6 grams of coffee silver skin powder mixed with 300ml of solvent. Ultrasound pretreatment was applied for 10 min at the ultrasound powder of 20% and 100%. Whereas, control samples were with no ultrasound pre-treatment employed. All the samples were transferred to an orbital operating at a constant speed of 210 rpm under the temperature of 50°C. Samples were withdrawn after 0.5, 1, 2, 3, 4, 5, 6 and 24 h, and then were analyzed for total phenolics. Peleg’s kinetic model was fitted to the experimental data obtained for extraction of total phenolics.

In the study, quantification was processed for caffeine content, 3-chlorigenic acid, 4-chlorogenic acid and 5-chlorogenic acid by ultra-high pressure liquid chromatography (UPLC). Antioxidant activity was tested including DPPH and FRAP. Final samples were characterized for polyphenols by LC-MS. The result showed that the yield of total phenolics and caffeine increased with increasing ultrasound amplitude level. Total phenolics content was significantly higher (p < 0.05) for 80% methanol extraction compared to water extraction for all treatments. After 24 h of shaking on an orbital shaker, the highest concentration of extracted phenolics using water as solvent was 6.30 ± 0.02 mg GAE/gdb at 100% ultrasound amplitude level. The lowest value obtained was 5.88 ± 0.09 mg GAE/gdb in the case of control. Similarly the concentration of extracted phenolics after 24 h of shaking using 80% methanol as solvent was 8.85 ± 0.09 mg GAE/gdb at 100% ultrasound amplitude level. The lowest value obtained was 8.49 ± 0.46 mg GAE/gdb in the case of control. A similar trend was observed for caffeine. The profile of the bioactive extraction curves showed an initial high rate of extraction followed by a much reduced rate asymptotically approaching an equilibrium concentration. The extraction yield of bioactives was time-dependent and increased with extraction time, especially over first 2 h of shaking. The extraction rate of solutes increases, but the increment of the extraction rate reduces with time and remains constant. At the experimental conditions investigated in this study, the yield and the kinetics of solid–liquid extraction of coffee silver skin bioactives were shown to be strongly influenced by solvent type and ultrasound amplitude level. This study demonstrates that ultrasound can be employed to enhance the efficiency of bioactive extraction from coffee silver skin. The bioactive extraction kinetics presented provides valuable data for scaling-up and process design of extraction systems.

Quality evaluation of tubers and cereal flours by using spectral imaging (PhD)

Su WH and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

Staple foods including cereals and root/tuber crops dominate the daily diet of human by providing valuable protein, starch, oils, minerals and vitamins. Combining the techniques of spectroscopy and computer vision into one system, spectral imaging technique with proposed chemometrics (such as PLSR, MLR, and SVM) was introduced in this study for simultaneous detection of both spectral and spatial information of food products. In addition to the ability for classifying tubers into different quality grades and gaining rapid information about tuber chemical components (such as moisture, starch, and dry matter) and physical attributes (texture, water binding capacity (WBC), and specific gravity (SG), infrared (IR) imaging spectroscopy is able to determine impurities of specific flour with avoidance of extensive sample preparation. Overall, it is promising for IR imaging spectroscopy to be used for food quality determination.

Performing Dimensionality Reduction in Hyperspectral Image Classification (PhD)

Xu JL and Da-Wen Sun

Sponsors: CSC-UCD Scholarship Scheme

This work provides several useful strategies for performing the dimensionality reduction in hyperspectral imaging data, with detailed command line scripts in the Matlab computing language as the supplementary data and it is freely available to be downloaded from: https://bitbucket.org/lily-xu. In this work, there are many original codes and functions developed. More importantly, a further selection function based on variance inflation factor (VIF) is proposed to diagnose and alleviate collinearity problem because collinearity and multi-collinearity are always expected to be severe in the spectral data. In this work, step-by-step procedure is provided for easy adaptation of these strategies to individual case.

One effective solution to minimize the collinearity among contiguous wavelengths and avoid singularity is the reduction of data dimensionality. Hence, a reduced image cube can be yielded to take place of the whole hypercube with massive data, and this alteration will remarkably save the subsequent data processing time and will probably enhance model performance in terms of accuracy and robustness. This work is aimed at providing a step-bystep instruction to perform different dimensionality reduction techniques on hyperspectral imaging dataset, with corresponding command line scripts written in Matlab computing language.

In this work, transformation-based methods include principal component analysis (PCA) and linear discriminant analysis (LDA), while band selection methods are comprised of partial least squares (PLS) regression combined with the variable importance in the projection (VIP) scores, selectivity ratio (SR), and significance multivariate correlation (sMC); Monte Carlo sampling (MCS) based methods including enhanced Monte Carlo variable selection (Emcvs) and competitive adaptive reweighted sampling (CARS); model population analysis (MPA) based methods from libPLS including uninformative variable elimination (UVE), random frog (RF), and PHADIA; Matlab built-in functions for feature selection including Relieff, stepwise regression and sequential feature selection (SFS); and the selection method guided by genetic algorithm (GA).

Among all these band selection methods, PLS-based methods (VIP, SR and sMC) have yielded reasonably good classification results. Compared to other band selection methods, PLS-based methods are remarkably simple, reliable and fast. These algorithms work in different manners and they have been designed for various applications. It is thus necessary to conduct trial and error study for finding the optimal one.

Modelling the kinetics of cooking loss during potato dehydration using hyperspectral imaging (ME)

Lei T, Su WH and Da-Wen Sun

Sponsors: University College Dublin

This study assesses the optimal spectral preprocessing methods for modelling cooking loss during potato dehydration using hyperspectral imaging. 6 spectral preprocessing methods combined with 3 regression models are tested. The result shows that SNV is the best preprocessing method for modelling cooking loss. Whereas, the spectra preprocessing method is not useful for Melody potato.

Determination of lipid content of fresh and frozen-thawed salmon fillets using hyperspectral image texture and spectral features (MEngSc)

Chen TJ and Da-Wen Sun

Sponsors: University College Dublin

This experiment is carried out to research the lipid content on fresh and frozen-thawed salmon fillets by applying short-wave near infrared (400-1000nm) hyperspectral imaging system. Hyperspectral cubes will be captured, and their corresponding spectra data will be analysed. Principal component analysis (PCA) is applied to explore the variance between two conditions of salmon. Partial least squares- discriminant analysis (PLS-DA) lipid content on fresh and frozen-thawed salmon fillets will be used to build classification models for recognition and authentication of the tested samples. The study will offer an analysis model for testing lipid content of salmon fillets. The results will indicate that hyperspectral imaging could be used to determine lipid content on fresh and frozen-thawed salmon fillets.

Discrimination of organic wheat flour from other flour varieties by near-infrared (NIR) hyperspectral imaging (MEngSc)

Jia WY, Su WH and Da-Wen Sun

Sponsors: University College Dublin

Wheat is a vital food crop for human activities and daily life which supply about half of the people on earth. This paper takes organic wheat flour (cassava flour, ordinary wheat flour, organic Spelt flour) as the research object and uses the near-infrared spectroscopy technology and hyperspectral technology with the advantages of rapid and non-destructive testing to test the system which includes two types flour by measuring effective chemometrics methods. A variety of chemometric methods have been used to screen flour adulteration system and to obtain a more suitable measurement model.

Investigation of the effect of thawing on salmon fillets using time series hyperspectral imaging (TS-HSI) (MEngSc)

Wu Q, Xu JL and Da-Wen Sun

Sponsors: University College Dublin

As a rapid and non-destructive analysis technology, hyperspectral imaging (HSI) has been widely used in food industry. This study was carried out to investigate the effect of three different thawing methods (air thawing, microwave thawing, and freeze thawing) on salmon using a near infrared (900-1700 nm) time series hyperspectral imaging system (TS-HSI). Principal component analysis (PCA) was applied to explore the spectral change during the thawing process and partial least squares discriminant analysis (PLS-DA) was used to discriminate salmon fillets suffered from different thawing methods. This study confirmed the potential of TS-HSI for investigation of changing characteristics of salmon fillets over time sequences using different thawing approaches.

UCD Food Refrigeration & Computerised Food Technology

Food Refrigeration and Computerized Food Technology University College Dublin Agriculture & Food Science Centre Belfield, Dublin 4, Ireland.
T: +353 1 716 7342