This PDF file contains the front matter associated with SPIE Proceedings Volume 10217, including the Title Page, Copyright information, Table of Contents, Introduction (if any), and Conference Committee listing.

This paper investigates a phage-based biomolecular filter that enables the evaluation of large volumes of liquids for the presence of small quantities of bacterial pathogens. The filter is a planar arrangement of phage-coated, strip-shaped magnetoelastic (ME) biosensors (4 mm × 0.8 mm × 0.03 mm), magnetically coupled to a filter frame structure, through which a liquid of interest flows. This "phage filter" is designed to capture specific bacterial pathogens and allow non-specific debris to pass, eliminating the common clogging issue in conventional bead filters. ANSYS Maxwell was used to simulate the magnetic field pattern required to hold ME biosensors densely and to optimize the frame design. Based on the simulation results, a phage filter structure was constructed, and a proof-in-concept experiment was conducted where a Salmonella solution of known concentration were passed through the filter, and the number of captured Salmonella was quantified by plate counting.

This paper demonstrates a highly sensitive surface-scanning detector used for magnetoelastic (ME) biosensors for the detection of Salmonella on the surface of a polyethylene (PE) food preparation surface. The design and fabrication methods of the new planar spiral coil are introduced. Different concentrations of Salmonella were measured on the surface of a PE board. The efficacy of Salmonella capture and detection is discussed.

This paper investigates the effects of surface-scanning detector position on the resonant frequency and signal amplitude of a wireless magnetoelastic (ME) biosensor for direct pathogen detection on solid surfaces. The experiments were conducted on the surface of a flat polyethylene (PE) plate as a model study. An ME biosensor (1 mm × 0.2 mm × 30 μm) was placed on the PE surface, and a surface-scanning detector was brought close and aligned to the sensor for wireless resonant frequency measurement. The position of the detector was accurately controlled by using a motorized three-axis translation system (i.e., controlled X, Y, and Z positions). The results showed that the resonant frequency variations of the sensor were -125 to +150 Hz for X and Y detector displacements of ±600 μm and Z displacements of +100 to +500 μm. These resonant frequency variations were small compared to the sensor's initial resonant frequency (< 0.007% of 2.2 MHz initial resonant frequency) measured at the detector home position, indicating high accuracy of the measurement. In addition, the signal amplitude was, as anticipated, found to decrease exponentially with increasing detection distance (i.e., Z distance). Finally, additional experiments were conducted on the surface of cucumbers. Similar results were obtained.

Fishes are a widely used food material in the world. Recently about 4% of the fishes are infected with Kudoa thyrsites in Asian ocean. Kudoa thyrsites is a parasite that is found within the muscle fibers of fishes. The infected fishes can be a reason of food poisoning, which should be sorted out before distribution and consumption. Although Kudoa thyrsites is visible to the naked eye, it could be easily overlooked due to the micro-scale size and similar color with fish tissue. In addition, the visual inspection is labor intensive works resulting in loss of money and time. In this study, a portable microscopic camera was utilized to obtain images of raw fish slices. The optimized image processing techniques with polarized transmittance images provided reliable performance. The result shows that the portable microscopic imaging method can be used to detect parasites rapidly and non-destructively, which could be an alternative to manual inspections.

Proper chemical analyses of materials in sealed containers are important for quality control purpose. Although it is feasible to detect chemicals at top surface layer, it is relatively challenging to detect objects beneath obscuring surface. This study used spatially offset Raman spectroscopy (SORS) method to detect urea, ibuprofen and acetaminophen powders contained within one or more (up to eight) layers of gelatin capsules to demonstrate subsurface chemical detection and identification. A 785 nm point-scan Raman spectroscopy system was used to acquire spatially offset Raman spectra for an offset range of 0 to 10 mm from the surfaces of 24 encapsulated samples, using a step size of 0.1 mm to obtain 101 spectral measurements per sample. With increasing offset distance, the fraction of information from the deeper subsurface material increased compared to that from the top surface material. The series of measurements was analyzed to differentiate and identify the top surface and subsurface materials. Containing mixed contributions from the powder and capsule, the SORS of each sample was decomposed using self modeling mixture analysis (SMA) to obtain pure component spectra of each component and corresponding components were identified using spectral information divergence values. Results show that SORS technique together with SMA method has a potential for non-invasive detection of chemicals at deep subsurface layer.

Excessive use of benzoyl peroxide (BPO, a bleaching agent) in wheat flour can destroy flour nutrients and cause diseases to consumers. A macro-scale Raman chemical imaging method was developed for direct detection of BPO mixed in the wheat flour. A 785 nm line laser was used in a line-scan Hyperspectral Raman imaging system. Raman images were collected from wheat flour mixed with BPO at eight concentrations (w/w) from 50 to 6,400 ppm. A sample holder (150×100×2 mm³) was used to present a thin layer (2 mm thick) of the powdered sample for image acquisition. A baseline correction method was used to correct the fluctuating fluorescence signals from the wheat flour. To isolate BPO particles from the flour background, a simple thresholding method was applied to the single-band fluorescence-free images at a unique Raman peak wavenumber (i.e., 1001 cm⁻¹) preselected for the BPO detection. Chemical images were created to detect and map the BPO particles. Limit of detection for the BPO was estimated in the order of 50 ppm, which is on the same level with regulatory standards.

Foodborne illness represents a significant health burden worldwide. While monitoring the freshness of food before consumption could significantly improve the current predicament, there is a lack of a simple system that one can use to accurately assess the freshness of their food. Currently, the most common practice for food quality determination is by visual or odor inspection which lacks objectivity, accuracy and precision. Near infrared (NIR) spectroscopic techniques can help address this problem by providing rapid and non-destructive means to estimate the freshness state of various foods based on the changes to their characteristic spectra in the NIR region. Recent advancements in the development of portable NIR spectrometers are also enabling the realization of this technique at the point-of-need. In this study, we have evaluated the feasibility of using NIR spectroscopy at the point-of-need to estimate the freshness of various foods including: beef sirloin, beef eyeround, pork sirloin, bass, salmon, corvina, tomato and watermelon. Using a commercial portable NIR spectrometer, we periodically scanned and collected NIR spectra from the food items that were stored at 4°C inside a refrigerator for up to 30 days. For these food items, we show that the NIR spectra can be classified by the foods’ aging day as well as by the levels of chemical/microbial indicators (i.e., thiobarbituric acid, volatile basic nitrogen and bacteria levels) with high accuracy, which represents high prospects of NIR spectroscopy for point-of-need freshness assessment of meat, fish, vegetables and fruits.

The application of in-motion optical sensor measurements was investigated for inspecting livestock soundness as a means of animal well-being. An optical sensor-based platform was used to collect in-motion, weight-related information. Eight steers, weighing between 680 and 1134 kg, were evaluated twice. Six of the 8 steers were used for further evaluation and analysis. Hoof impacts caused plate flexion that was optically sensed. Observed kinetic differences between animals’ strides at a walking or running/trotting gait with significant force distributions of animals’ hoof impacts allowed for observation of real-time, biometric patterns. Overall, optical sensor-based measurements identified hoof differences between and within animals in motion that may allow for diagnosis of musculoskeletal unsoundness without visual evaluation.

Hyperspectral sensing has been proven to be useful to determine the quality of food in general. It has also been used to distinguish naturally and artificially ripened mangoes by analyzing the spectral signature. However the focus has been on improving the accuracy of classification after performing dimensionality reduction, optimum feature selection and using suitable learning algorithm on the complete visible and NIR spectrum range data, namely 350nm to 1050nm. In this paper we focus on, (i) the use of low wavelength resolution and low cost multispectral sensor to reliably identify artificially ripened mango by selectively using the spectral information so that classification accuracy is not hampered at the cost of low resolution spectral data and (ii) use of visible spectrum i.e. 390nm to 700 nm data to accurately discriminate artificially ripened mangoes. Our results show that on a low resolution spectral data, the use of logistic regression produces an accuracy of 98.83% and outperforms other methods like classification tree, random forest significantly. And this is achieved by analyzing only 36 spectral reflectance data points instead of the complete 216 data points available in visual and NIR range. Another interesting experimental observation is that we are able to achieve more than 98% classification accuracy by selecting only 15 irradiance values in the visible spectrum. Even the number of data needs to be collected using hyper-spectral or multi-spectral sensor can be reduced by a factor of 24 for classification with high degree of confidence

Spatial-frequency domain (SFD) imaging technique allows to estimate the optical properties of biological tissues in a wide field of view. The technique is, however, prone to error in measurement because the two crucial assumptions used for deriving the analytical solution to diffusion approximation cannot be met perfectly in practical applications. This research was mainly focused on modeling light transfer in turbid media under the normal incidence of structured illumination using finite element method (FEM). Finite element simulations were performed for 50 simulation samples with different combinations of optical absorption and scattering coefficients for varying spatial frequencies, and the results were then compared with analytical method and Monte Carlo simulation. Relationships between diffuse reflectance and dimensionless absorption and dimensionless scattering coefficients were investigated. The results indicated that FEM provided reasonable results for diffuse reflectance, compared with the analytical method. Both FEM and analytical method overestimated the reflectance for μ_tr/f_x values of greater than 2 and underestimated the reflectance for μ_tr/f_x values of smaller than 2. Larger values of μ′_s/μ_a yielded better estimations of diffuse reflectance than did those of smaller than 10. The reflectance increased nonlinearly with the dimensionless scattering, whereas the reflectance decreased linearly with the dimensionless absorption. It was also observed that diffuse reflectance was relatively stable and insensitive to μ_s′ when the dimensionless scattering was larger than 50. Overall results demonstrate that FEM is effective for modeling light transfer in turbid media and can be used to explore the effects of crucial parameters for the SFD imaging technique.

Raman and Brillouin spectroscopic provide with a powerful way to non-invasively assess both chemical and physical (viscoelastic) properties. In this report, Brillouin microspectroscopy was used for real time analysis of elastic properties of Populus and Geranium leaves, while Raman spectroscopy and imaging were employed for assessment of their chemical variation during drying. When used together, those techniques can improve our understanding of mechanochemical changes of plants in response to environmental stress and pathogens at microscopic (cellular) level. Our results have demonstrated for the first time the ability of multimodal assessment of elasticity modulus, hydraulic conductance and interatomic vibrational modes in plants as emerging new markers for real time quantitative assessment of agricultural crops.

This paper presents the development and application of a hyper-spectral imaging system for root phenotyping. For sustainable plant production root systems optimized for growing conditions in the field are required. Therefore, the presented system is used for the research in the field of plant drought resistance. The system is used to acquire spatially resolved near infrared (NIR) spectroscopy data of rhizoboxes. In contrast to using visible light (380 nm-780 nm) the NIR wavelength range (900 nm-1700 nm) allows to discriminate essential features for the root segmentation and water distribution mappings. The increased image contrast in the NIR range allows roots to be segmented from soil and additional information, e.g. basic root-architecture, to be extracted. In addition, the water absorption bands in the NIR wavelength range can be used to determine the water content and to estimate the age of the roots. In this paper the hardware setup of the hyper-spectral root imaging system, the data analysis, the soil water content estimations and the root segmentation using different methods to optimize separation between roots and soil, both constituting complex materials of variable properties, are presented.

A non-destructive method of estimating the freshness of fish is required for appropriate price setting and food safety. In particular, for determining the possibility of eating raw fish (sashimi), freshness estimation is critical. We studied such an estimation method by capturing images of fish eyes and performing image processing using the temporal changes of the luminance of pupil and iris. To detect subtle non-visible changes of these features, we used UV (375 nm) light illumination in addition to visible white light illumination. Polarization and two-channel LED techniques were used to remove strong specular reflection from the cornea of the eye and from clear-plastic wrap used to cover the fish to maintain humidity. Pupil and iris regions were automatically detected separately by image processing after the specular reflection removal process, and two types of eye contrast were defined as the ratio of mean and median pixel values of each region. Experiments using 16 Japanese dace (Tribolodon hakonensis) at 23℃ and 85% humidity for 24 hours were performed. The eye contrast of raw fish increase non-linearly in the initial period and then decreased; however, that of frozen-thawed fish decreased linearly throughout 24 hours, regardless of the lighting. Interestingly, the eye contrast using UV light showed a higher correlation with time than that using white light only in the case of raw fish within the early 6- hour period postmortem. These results show the possibility of estimating fish freshness in the initial stage when fish are eaten raw using white and UV lightings.

Three-dimensional (3-D) shape information is valuable for fruit quality evaluation. This study was aimed at developing phase analysis techniques for reconstruction of the 3-D surface of fruit from the pattern images acquired by a structuredillumination reflectance imaging (SIRI) system. Phase-shifted sinusoidal patterns, distorted by the fruit geometry, were acquired and processed through phase demodulation, phase unwrapping and other post-processing procedures to obtain phase difference maps relative to the phase of a reference plane. The phase maps were then transformed into height profiles and 3-D shapes in a world coordinate system based on phase-to-height and in-plane calibrations. A reference plane-based approach, coupled with the curve fitting technique using polynomials of order 3 or higher, was utilized for phase-to-height calibrations, which achieved superior accuracies with the root-mean-squared errors (RMSEs) of 0.027- 0.033 mm for a height measurement range of 0-91 mm. The 3rd-order polynomial curve fitting technique was further tested on two reference blocks with known heights, resulting in relative errors of 3.75% and 4.16%. In-plane calibrations were performed by solving a linear system formed by a number of control points in a calibration object, which yielded a RMSE of 0.311 mm. Tests of the calibrated system for reconstructing the surface of apple samples showed that surface concavities (i.e., stem/calyx regions) could be easily discriminated from bruises from the phase difference maps, reconstructed height profiles and the 3-D shape of apples. This study has laid a foundation for using SIRI for 3-D shape measurement, and thus expanded the capability of the technique for quality evaluation of horticultural products. Further research is needed to utilize the phase analysis techniques for stem/calyx detection of apples, and optimize the phase demodulation and unwrapping algorithms for faster and more reliable detection.

Total volatile basic nitrogen (TVB-N) content is one of the important factors to measure the quality of meat. However, conventional chemical analysis methods for measuring TVB-N contents are time-consuming and labor-intensive, and are destructive procedures. The objective of this study is to investigate the possibility of fluorescence hyperspectral imaging techniques for determination of total volatile basic nitrogen (TVB-N) in beef meat. High intensity LED lights at 365 nm and 405 nm were used as the excitation for acquiring fluorescence images. Prediction algorithms based on simple band-ratio, partial least square discriminant analysis (PLS-DA) have been developed. This study shows that fluorescence hyperspectral imaging system has a good potential for rapid measurement of TVB-N content in meat.

Aflatoxin contamination in peanut products has been an important and long-standing problem around the world. Produced mainly by Aspergillus flavus and Aspergillus parasiticus, aflatoxins are the most toxic and carcinogenic compounds among toxins. This study investigated the application of fluorescence visible near-infrared (VNIR) hyperspectral images to assess the spectral difference between peanut kernels inoculated with toxigenic and atoxigenic inocula of A. flavus and healthy kernels. Peanut kernels were inoculated with NRRL3357, a toxigenic strain of A. flavus, and AF36, an atoxigenic strain of A. flavus, respectively. Fluorescence hyperspectral images under ultraviolet (UV) excitation were recorded on peanut kernels with and without skin. Contaminated kernels exhibited different fluorescence features compared with healthy kernels. For the kernels without skin, the inoculated kernels had a fluorescence peaks shifted to longer wavelengths with lower intensity than healthy kernels. In addition, the fluorescence intensity of peanuts without skin was higher than that of peanuts with skin (10 times). The fluorescence spectra of kernels with skin are significantly different from that of the control group (p<0.001). Furthermore, the fluorescence intensity of the toxigenic, AF3357 peanuts with skin was lower than that of the atoxigenic AF36 group. Discriminate analysis showed that the inoculation group can be separated from the controls with 100% accuracy. However, the two inoculation groups (AF3357 vis AF36) can be separated with only ～80% accuracy. This study demonstrated the potential of fluorescence hyperspectral imaging techniques for screening of peanut kernels contaminated with A. flavus, which could potentially lead to the production of rapid and non-destructive scanning-based detection technology for the peanut industry.

Beef production is the main agricultural industry in Texas, and livestock are managed in pasture and rangeland which are usually huge in size, and are not easily accessible by vehicles. The current research method for livestock location identification and counting is visual observation which is very time consuming and costly. For animals on large tracts of land, manned aircraft may be necessary to count animals which is noisy and disturbs the animals, and may introduce a source of error in counts. Such manual approaches are expensive, slow and labor intensive. In this paper we study the combination of small unmanned aerial vehicle (sUAV) and machine vision technology as a valuable solution to manual animal surveying. A fixed-wing UAV fitted with GPS and digital RGB camera for photogrammetry was flown at the Welder Wildlife Foundation in Sinton, TX. Over 600 acres were flown with four UAS flights and individual photographs used to develop orthomosaic imagery. To detect animals in UAV imagery, a fully automatic technique was developed based on spatial and spectral characteristics of objects. This automatic technique can even detect small animals that are partially occluded by bushes. Experimental results in comparison to ground-truth show the effectiveness of our algorithm.

Meat and bone meal (MBM) has been banned as animal feed for ruminants since 2001 because it is the source of bovine spongiform encephalopathy (BSE). Moreover, many countries have banned the use of MBM as animal feed for not only ruminants but other farm animals as well, to prevent potential outbreak of BSE. Recently, the EU has introduced use of some MBM in feeds for different animal species, such as poultry MBM for swine feed and pork MBM for poultry feed, for economic reasons. In order to authenticate the MBM species origin, species-specific MBM identification methods are needed. Various spectroscopic and spectral imaging techniques have allowed rapid and non-destructive quality assessments of foods and animal feeds. The objective of this study was to develop rapid and accurate methods to differentiate pork MBM from poultry MBM using short-wave infrared (SWIR) hyperspectral imaging techniques. Results from a preliminary investigation of hyperspectral imaging for assessing pork and poultry MBM characteristics and quantitative analysis of poultry-pork MBM mixtures are presented in this paper.

This paper reports on the development, calibration and evaluation of a new multipurpose, multichannel hyperspectral imaging probe for property and quality assessment of food products. The new multichannel probe consists of a 910 μm fiber as a point light source and 30 light receiving fibers of three sizes (i.e., 50 μm, 105 μm and 200 μm) arranged in a special pattern to enhance signal acquisitions over the spatial distances of up to 36 mm. The multichannel probe allows simultaneous acquisition of 30 spatially-resolved reflectance spectra of food samples with either flat or curved surface over the spectral region of 550-1,650 nm. The measured reflectance spectra can be used for estimating the optical scattering and absorption properties of food samples, as well as for assessing the tissues of the samples at different depths. Several calibration procedures that are unique to this probe were carried out; they included linearity calibrations for each channel of the hyperspectral imaging system to ensure consistent linear responses of individual channels, and spectral response calibrations of individual channels for each fiber size group and between the three groups of different size fibers. Finally, applications of this new multichannel probe were demonstrated through the optical property measurement of liquid model samples and tomatoes of different maturity levels. The multichannel probe offers new capabilities for optical property measurement and quality detection of food and agricultural products.

Various types of optical imaging techniques measuring light reflectivity and scattering can detect microbial colonies of foodborne pathogens on agar plates. Until recently, these techniques were developed to provide solutions for hypothesis-driven studies, which focused on developing tools and batch/offline machine learning methods with well defined sets of data. These have relatively high accuracy and rapid response time because the tools and methods are often optimized for the collected data. However, they often need to be retrained or recalibrated when new untrained data and/or features are added. A big-data driven technique is more suitable for online learning of new/ambiguous samples and for mining unknown or hidden features. Although big data research in hyperspectral imaging is emerging in remote sensing and many tools and methods have been developed so far in many other applications such as bioinformatics, the tools and methods still need to be evaluated and adjusted in applications where the conventional batch machine learning algorithms were dominant. The primary objective of this study is to evaluate appropriate big data analytic tools and methods for online learning and mining of foodborne pathogens on agar plates. After the tools and methods are successfully identified, they will be applied to rapidly search big color and hyperspectral image data of microbial colonies collected over the past 5 years in house and find the most probable colony or a group of colonies in the collected big data. The meta-data, such as collection time and any unstructured data (e.g. comments), will also be analyzed and presented with output results. The expected results will be novel, big data-driven technology to correctly detect and recognize microbial colonies of various foodborne pathogens on agar plates.

The present work explores the possible utilization of hyperspectral devices, following a proximity based approach, for the diagnosis of Peronospora infection in the vineyards. It compares the performance of two hyperspectral cameras, characterized by different spectral acquisition ranges, in the identification of different levels of infection as detectable from the analysis of the leaf surface. For this purpose, healthy grapevine leaves and leaves affected by a different grade of Peronospora infection have been acquired in laboratory conditions using two different sensing devices: a Specim Imspector V10™ and a Specim Spectral Camera N17™ working in the region between 400-1000 nm and 1000-1700 nm, respectively. A Partial Least Squares Discriminant Analysis (PLS-DA) model has been built to perform the classification of healthy, infected and necrotic leaves.

This study aimed to evaluate the possibility to perform a fast, reliable and robust non-destructive monitoring of kiwifruits characteristics adopting an HyperSpectral Imaging (HSI) based approach. HSI was thus utilized for two different purposes: i) to test whether the postharvest ripeness of kiwifruits could be non-destructively determined and ii) for the diagnosis of pseudomonas infection in the Kiwi orchards. To reach the 1st goal (i.e. fruit ripening evaluation) a NIR Spectral Camera acting in the range between 900 and 1700 nm has been used. To reach the 2nd goal a hyperspectral camera working in the VIS-NIR range (400 nm – 1000 nm) was used. For both the approaches "only" significance and robustness of the collected data, in respect of the selected operative conditions, was investigated and the results have been evaluated in terms of different Principal Components (PC) images.

Pesticide residues in the fruits, vegetables and agricultural commodities are harmful to humans and are becoming a health concern nowadays. Detection of pesticide residues on various commodities in an open environment is a challenging task. Hyperspectral sensing is one of the recent technologies used to detect the pesticide residues. This paper addresses the problem of detection of pesticide residues of Cyantraniliprole on grapes in open fields using multi temporal hyperspectral remote sensing data. The re ectance data of 686 samples of grapes with no, single and double dose application of Cyantraniliprole has been collected by handheld spectroradiometer (MS- 720) with a wavelength ranging from 350 nm to 1052 nm. The data collection was carried out over a large feature set of 213 spectral bands during the period of March to May 2015. This large feature set may cause model over-fitting problem as well as increase the computational time, so in order to get the most relevant features, various feature selection techniques viz Principle Component Analysis (PCA), LASSO and Elastic Net regularization have been used. Using this selected features, we evaluate the performance of various classifiers such as Artificial Neural Networks (ANN), Support Vector Machine (SVM), Random Forest (RF) and Extreme Gradient Boosting (XGBoost) to classify the grape sample with no, single or double application of Cyantraniliprole. The key finding of this paper is; most of the features selected by the LASSO varies between 350-373nm and 940-990nm consistently for all days. Experimental results also shows that, by using the relevant features selected by LASSO, SVM performs better with average prediction accuracy of 91.98 % among all classifiers, for all days.

The improvement of living standards has urged consumers to pay more attention to the quality and nutrition of meat, so the development of nondestructive detection device for quality and nutritional parameters is commercioganic undoubtedly. In this research, a portable device equipped with visible (Vis) and near-infrared (NIR) spectrometers, tungsten halogen lamp, optical fiber, ring light guide and embedded computer was developed to realize simultaneous and fast detection of color (L*, a*, b*), pH, total volatile basic nitrogen (TVB-N), intramuscular fat (IF), protein and water content in pork. The wavelengths of dual-band spectrometers were 400~1100 nm and 940~1650 nm respectively and the tungsten halogen lamp cooperated with ring light guide to form a ring light source and provide appropriate illumination intensity for sample. Software was self-developed to control the functionality of dual-band spectrometers, set spectrometer parameters, acquire and process Vis/NIR spectroscopy and display the prediction results in real time. In order to obtain a robust and accurate prediction model, fresh longissimus dorsi meat was bought and placed in the refrigerator for 12 days to get pork samples with different freshness degrees. Besides, pork meat from three different parts including longissimus dorsi, haunch and lean meat was collected for the determination of IF, protein and water to make the reference values have a wider distribution range. After acquisition of Vis/NIR spectra, data from 400~1100 nm were pretreated with Savitzky-Golay (S-G) filter and standard normal variables transform (SNVT) and spectrum data from 940~1650 nm were preprocessed with SNVT. The anomalous were eliminated by Monte Carlo method based on model cluster analysis and then partial least square regression (PLSR) models based on single band (400~1100 nm or 940~1650 nm) and dual-band were established and compared. The results showed the optimal models for each parameter were built with correlation coefficients in prediction set of 0.9101, 0.9121, 0.8873, 0.9094, 0.9378, 0.9348, 0.9342 and 0.8882, respectively. It indicated this innovative and practical device can be a promising technology for nondestructive, fast and accurate detection of nutritional parameters in meat.

The plate count method is commonly used to detect the total viable count (TVC) of bacteria in pork, which is timeconsuming and destructive. It has also been used to study the changes of the TVC in pork under different storage conditions. In recent years, many scholars have explored the non-destructive methods on detecting TVC by using visible near infrared (VIS/NIR) technology and hyperspectral technology. The TVC in chilled pork was monitored under high oxygen condition in this study by using hyperspectral technology in order to evaluate the changes of total bacterial count during storage, and then evaluate advantages and disadvantages of the storage condition. The VIS/NIR hyperspectral images of samples stored in high oxygen condition was acquired by a hyperspectral system in range of 400~1100nm. The actual reference value of total bacteria was measured by standard plate count method, and the results were obtained in 48 hours. The reflection spectra of the samples are extracted and used for the establishment of prediction model for TVC. The spectral preprocessing methods of standard normal variate transformation (SNV), multiple scatter correction (MSC) and derivation was conducted to the original reflectance spectra of samples. Partial least squares regression (PLSR) of TVC was performed and optimized to be the prediction model. The results show that the near infrared hyperspectral technology based on 400-1100nm combined with PLSR model can describe the growth pattern of the total bacteria count of the chilled pork under the condition of high oxygen very vividly and rapidly. The results obtained in this study demonstrate that the nondestructive method of TVC based on NIR hyperspectral has great potential in monitoring of edible safety in processing and storage of meat.

Soluble solid content (SSC) is a major quality parameter to fruit, which has influence on its flavor or texture. Some researches on the on-line non-invasion detection of fruit quality were published. However, consumers desire portable devices currently. This study aimed to develop a portable device for accurate, real-time and nondestructive determination of quality factors of fruit based on diffuse reflectance Vis/NIR spectroscopy (520-950 nm). The hardware of the device consisted of four units: light source unit, spectral acquisition unit, central processing unit, display unit. Halogen lamp was chosen as light source. When working, its hand-held probe was in contact with the surface of fruit samples thus forming dark environment to shield the interferential light outside. Diffuse reflectance light was collected and measured by spectrometer (USB4000). ARM (Advanced RISC Machines), as central processing unit, controlled all parts in device and analyzed spectral data. Liquid Crystal Display (LCD) touch screen was used to interface with users. To validate its reliability and stability, 63 apples were tested in experiment, 47 of which were chosen as calibration set, while others as prediction set. Their SSC reference values were measured by refractometer. At the same time, samples' spectral data acquired by portable device were processed by standard normalized variables (SNV) and Savitzky-Golay filter (S-G) to eliminate the spectra noise. Then partial least squares regression (PLSR) was applied to build prediction models, and the best predictions results was achieved with correlation coefficient (r) of 0.855 and standard error of 0.6033° Brix. The results demonstrated that this device was feasible to quantitatively analyze soluble solid content of apple.

Pesticide residue is one of the major challenges to fruits safety, while the traditional detection methods of pesticide residue on fruits and vegetables can’t afford the demand of rapid detection in actual production because of timeconsuming. Thus rapid identification and detection methods for pesticide residue are urgently needed at present. While most Raman detection systems in the market are spot detection systems, which limits the range of application. In the study, our lab develops a Raman detection system to achieve area-scan thorough the self-developed spot detection Raman system with a control software and two devices. In the system, the scanning area is composed of many scanning spots, which means every spot needs to be detected and more time will be taken than area-scan Raman system. But lower detection limit will be achieved in this method. And some detection device is needed towards fruits and vegetables in different shape. Two detection devices are developed to detect spherical fruits and leaf vegetables. During the detection, the device will make spherical fruit rotate along its axis of symmetry, and leaf vegetables will be pressed in the test surface smoothly. The detection probe will be set to keep a proper distance to the surface of fruits and vegetables. It should make sure the laser shins on the surface of spherical fruit vertically. And two software are used to detect spherical fruits and leaf vegetables will be integrated to one, which make the operator easier to switch. Accordingly two detection devices for spherical fruits and leaf vegetables will also be portable devices to make it easier to change. In the study, a new way is developed to achieve area-scan result by spot-scan Raman detection system.

Phage based magneto-elastic (ME) biosensors have been shown to be able to rapidly detect Salmonella in various food systems to serve food pathogen monitoring purposes. In this ME biosensor platform, the free-standing strip-shaped magneto-elastic sensor is the transducer and the phage probe that recognizes Salmonella in food serves as the bio-recognition element. According to Sorokulova et al. at 2005, a developed oligonucleotide probe E2 was reported to have high specificity to Salmonella enterica Typhimurium. In the report, the specificity tests were focused in most of Enterobacterace groups outside of Salmonella family. Here, to understand the specificity of phage E2 to different Salmonella enterica serotypes within Salmonella Family, we further tested the specificity of the phage probe to thirty-two Salmonella serotypes that were present in the major foodborne outbreaks during the past ten years (according to Centers for Disease Control and Prevention). The tests were conducted through an Enzyme linked Immunosorbent Assay (ELISA) format. This assay can mimic probe immobilized conditions on the magnetoelastic biosensor platform and also enable to study the binding specificity of oligonucleotide probes toward different Salmonella while avoiding phage/ sensor lot variations. Test results confirmed that this oligonucleotide probe E2 was high specific to Salmonella Typhimurium cells but showed cross reactivity to Salmonella Tennessee and four other serotypes among the thirty-two tested Salmonella serotypes.

Bacterial biofilm formed by pathogens on fresh produce surfaces is a food safety concern because the complex extracellular matrix in the biofilm structure reduces the reduction and removal efficacies of washing and sanitizing processes such as chemical or irradiation treatments. Therefore, a rapid and nondestructive method to identify pathogenic biofilm on produce surfaces is needed to ensure safe consumption of fresh, raw produce. This research aimed to evaluate the feasibility of hyperspectral fluorescence imaging for detecting Escherichia.coli (ATCC 25922) biofilms on baby spinach leaf surfaces. Samples of baby spinach leaves were immersed and inoculated with five different levels (from 2.6x10⁴ to 2.6x10⁸ CFU/mL) of E.coli and stored at 4°C for 24 h and 48 h to induce biofilm formation. Following the two treatment days, individual leaves were gently washed to remove excess liquid inoculums from the leaf surfaces and imaged with a hyperspectral fluorescence imaging system equipped with UV-A (365 nm) and violet (405 nm) excitation sources to evaluate a spectral-image-based method for biofilm detection. The imaging results with the UV-A excitation showed that leaves even at early stages of biofilm formations could be differentiated from the control leaf surfaces. This preliminary investigation demonstrated the potential of fluorescence imaging techniques for detection of biofilms on leafy green surfaces.