Objective 1: Advance development and validation of on-line automated whole-surface inspection systems for simultaneous safety and quality inspection of fresh produce in high-throughput commercial processing operations. Objective 2: Develop and validate user-friendly analytical sensing methods and technologies for targeted and non-targeted rapid screening of foods for microbial, chemical, and biological contaminants in laboratory, field, and/or industrial environments. Objective 3: Advance development of portable spectral imaging technologies to allow identification and detection of food contaminants, and develop sampling and inspection protocols for implementation of the developed technologies in industry and regulatory applications. Objective 4: Advance development, test and validate an automated system for detecting contaminants in produce fields, and investigate cost and sensitivity trade-offs of different potential system components and configurations with regard to production of a cost-effective commercial system.
Because cross-contamination may occur at many points throughout the production, processing, and distribution chains, our research targets reduction of food safety risks at multiple points, including both upstream (pre-harvest) processing and subsequent stages. The inspection point at single-layer processing is not intended to be a comprehensive inspection on its own but is key for some packaged fresh products. The ARS approach includes both pre- and post-harvest risk reduction measures that collectively can mitigate food safety concerns related to foodborne illnesses. The whole-surface sample presentation/imaging technologies developed in the previous project cycle along with the multitask imaging technology will be integrated on conveyor/processing systems to develop two automated whole-surface inspection platforms to simultaneously inspect produce for safety and quality attributes such as contaminants and defects. The two stand-alone commercial-grade prototype processing-inspection platforms will be transportable to produce processing facilities for testing and demonstration of the whole-surface inspection efficacies, with an ultimate goal of technology transfer. The proposed whole-surface fruit inspection will complement current industry sorting—based on quality attributes such as color and size—by the addition of safety inspection and will be used immediately after conventional color and size sorting. The proposed leafy green inspection will be used for inspection immediately prior to “value-added” processing, e.g., washing for packaged fresh-cut products.
Significant progress has been made for all objectives of the project, which fall under National Program 108. For Objective 1, enhanced line-scan image-processing algorithms to represent the whole surface of a rotating round fruit were developed and integrated with the prototype inspection system. The line-scan imaging prototype system generates whole-surface images of round fruits, visualizing in real-time a two-dimensional “map” of the entire surface of a spherical fruit. The prototype system for multispectral line-scan inspection of leafy greens was upgraded and used to conduct inspection experiments that simultaneously performed fecal contamination inspection and defect detection on leafy greens such as spinach, and romaine lettuce. Prototype development of the on-line whole-surface inspection systems for bulk processing and safety inspection of round fruits and of leafy greens has been completed. The systems have been demonstrated to industry, including a Cooperative Research and Development Agreement (CRADA) partner who has expressed interest in obtaining licensing for the ARS-patented multitask inspection technology (U.S. Patent No. 7,787,111, “Simultaneous acquisition of fluorescence and reflectance imaging techniques with a single imaging device for multitask inspection”). For Objective 2, ARS scientists in Beltsville, Maryland, in collaboration with National Agricultural Products Quality Management Service (NAQS), South Korea, developed a portable multimodal optical sensing system and method for automated chemical and biological assessment. The system includes a white ring light and RGB camera for color imaging, a UV-A ring light and monochromatic camera for fluorescence imaging, a dual Raman system with 785 and 1064 nm lasers for spectroscopy and imaging, and a programmable X-Y translation stage with back-illuminated sample holder. The system can conduct dual-band Raman spectroscopy and imaging measurements using the 785 and 1064 nm point-lasers for food and biological materials. Spectral and spatial classification models can be developed using multivariate analysis and/or artificial intelligence approaches (e.g., machine learning and deep learning techniques) for integration into the system software for real-time identification of chemical and biological contaminants (e.g., adulterants and bacteria) in food and agricultural products. Built on a 30 cm × 45 cm aluminum breadboard, the compact and portable system is suitable for rapid on-site analysis of food or other sample materials for chemical or biological contaminants. A patent disclosure was submitted in 2020. In addition, ARS scientists in Beltsville, Maryland, continued to collaborate with a CRADA partner to develop fish authentication methods and systems based on multimodal hyperspectral imaging techniques, to address deceptive labeling and substitution of fish fillets. Two major fraudulent practices in the seafood industry are the substitution of inexpensive fish for higher-priced species and the substitution of frozen-thawed product for never-frozen fresh product. This continuing study used three line-scan hyperspectral systems developed in-house by ARS scientists to collect four types of hyperspectral image data from fillet samples. The continuing work includes analysis of combinations of feature extraction and selection techniques and exhaustive data search, optimization, and fusion to determine the most important features needed to perform fish authentication using the different imaging modes. The results can be used to design and develop customized systems for industrial fish inspection applications. ARS scientists in Beltsville, Maryland, continued to collaborate with NASA Kennedy Space Center (KSC) to develop hyperspectral imaging systems to monitor plant growth/health and food safety in fresh food production systems to be used in spaceflight. ARS scientists designed and developed a compact hyperspectral imaging system prototype equipped with Vis/NIR and UV lights for reflectance and fluorescence measurements and with LabView-based control software developed in-house. The prototype system was integrated into a KSC growth chamber for ground testing and verification under lab conditions. Preliminary hyperspectral reflectance and fluorescence images of pick-and-eat salad crops grown by KSC scientists were acquired to evaluate system performance and develop real-time image processing algorithms. For Objective 2, ARS scientists in Beltsville, Maryland, continued experiments to develop spectroscopy- or imaging-based methods for nondestructive detection of contaminants or other food safety risks which food industries are seeking to manage or prevent with the use of better tools. A 1064-nm point-scan Raman imaging system and an infrared (IR) spectroscopic system were used to investigate spectral detection of Sudan Red and white turmeric (a toxic dye and a botanical additive, respectively) mixed into yellow turmeric, a common culinary seasoning and health supplement. Sudan Red was effectively detected using either Raman or IR spectra. White turmeric mixed into yellow turmeric was effectively identified by a distinct IR peak, but could not be detected by Raman spectroscopy due to overlapping peaks. Quantitative models developed using IR spectra for each mixture type estimated Sudan Red and white turmeric concentrations with correlation coefficients of 0.97 and 0.95, respectively. A new method based on airflow and laser ranging technique was developed to nondestructively evaluate beef freshness, enabling in situ on-line nondestructive testing. The spectral data collected with this method makes possible the identification of deformation characteristics from loss in freshness and distinguishing them from process parameters which are not measures of food safety freshness. Fipronil, a broad spectrum insecticide often used to kill lice and fleas, was analyzed using the 785-nm point-scan Raman system, and a new surface-enhanced Raman spectroscopy (SERS) method was developed in collaboration with Hefei Institute of Physical Science, China. A SERS material selective for fipronil enabled detection at concentrations as low as 0.1 ppm on chicken egg membranes. For Objective 3, ARS scientists in Beltsville, Maryland, continued work with multiple collaborators for application-specific testing of ARS portable handheld multispectral imaging technology for sanitation and contamination inspection, such as detection of bacterial biofilms for sanitation inspection of KSC plant growth chambers. ARS scientists have continued discussions with US Army Natick Soldier Center regarding improvements and advancements needed for effective use in sanitation inspection of commercial or military-contracted food processing facilities. In early 2020, an exclusive licensing request for the portable handheld imaging technology (US patent no. US 8,310,544) was granted to a CRADA partner for commercialization. Two ARS prototypes were also transferred, and ARS scientists are cooperating on ongoing development of a user-friendly App for image enhancement functions and inspection management. For Objective 4, ARS scientists in Beltsville, Maryland, expanded the research to determine the feasibility of using a drone-based sensing platform to detect in-field animal intrusion and fecal contamination. Although the SY leading the development of the ground-based laser-induced fluorescence-imaging prototype retired, the remaining SYs have continued with steps to begin moving the imaging techniques from ground-based implementation to drone-based implementation that will allow for faster imaging across larger field areas to be monitored and potentially enable joint operation with drone-based monitoring of irrigation water quality. For the new ARS research project, the group proposed to develop and validate an autonomous unmanned aerial vehicle with multimode imaging technologies for pre-harvest inspection of produce fields for animal intrusion and fecal contamination, and for irrigation water quality monitoring. The researchers also initiated design of the field-level ground truth measurement platform. In collaboration with the microbial water quality project team, a line-scan hyperspectral imaging camera system mounted on the bow of a GPS-equipped multipurpose floating platform (MFP) was developed and tested on an irrigation pond. A field study acquired about 80,000 near-infrared/red spectral line-scan images of the water for correlation to measurements of chlorophyll-a content, a water quality indicator, for water samples collected during imaging. Models developed to predict chlorophyll-a concentrations showed results highly correlated to the measured concentrations. This work shows that low-altitude hyperspectral imaging via MFP can provide valuable information about water quality through spatial mapping for data visualization. The hyperspectral imaging method for water quality can be further improved by additional research addressing variation arising from floating debris, aquatic organisms, and changes in sunlight intensity and cloud movement, as well as by considering other indicator measurements such as suspended solids, colored dissolved organic matter, bacteria, nutrient concentrations, and turbidity. This research will help researchers develop and optimize models and methods that can be used in field production by the fresh produce industry to help meet federal mandates for irrigation water quality.
1. Rapid identification of potential adulterants in commercial spice powders. ARS scientists in Beltsville, Maryland, have developed a 1064-nm dispersive Raman imaging system which successfully detected chemical contaminants in spice powders, such as Sudan Red in turmeric powder and both Sudan-I and metanil yellow in curry powder, with greatly reduced fluorescence interference. Raman spectral imaging is a suitable technique for non-destructive detection in heterogenous samples. No prior preparation is required for Raman measurement of the powder samples. With powder placed directly in a flat sample holder, Raman spectra are acquired over the entire surface area. The distribution of the contaminant particles can then be visualized in binary images. Given the widespread distribution of many powdered ingredients throughout food processing supply lines nationally and worldwide, this method will benefit food processors and food safety regulators seeking to ensure safety and quality of food powders.
2. A rapid method for detecting chemical contaminants in foods. ARS scientists in Beltsville, Maryland, have developed a Raman imaging method and apparatus for rapid, nondestructive detection of chemical contaminants in food materials. This rapid screening technology provides a direct and practical method to detect the animal feed chemical called ractopamine in meats without any need for chemical extraction. Now, this is a method to screen for animal feed chemicals or other veterinary drugs in animal meat products.
3. A method to detect fake eggs. Although some fake or imitation food materials produced for economic fraud contain lower quality or cheaper alternative ingredients, they are safe for consumption. Others containing non-edible or hazardous ingredients are not, including fake eggs made with harmful ingredients such as sodium alginate, tartrazine dye, gypsum powder, and paraffin wax. ARS scientists in Beltsville, Maryland, determined how to differentiate between fake and real chicken eggs. The results showed that using the Raman imaging technique, they could separate fake eggs from real eggs. The food industry can use this method to ensure that food products are safe and protect consumer health while combating fraud.
4. Detection of mislabeled fish fillet. A recent survey by the nonprofit organization Oceana found that 21 percent of fish sold in the United States were mislabeled. Fish fillets are easy to mislabel. Mixing inexpensive species with high-priced species and substituting frozen-thawed fillets for fresh fillets are two major fraudulent practices in the seafood industry. ARS scientists in Beltsville, Maryland, used the Raman imaging machine to differentiate fish fillets of different species and freshness conditions. The new handheld imaging machine can be used by regulatory agencies and the seafood industry to verify fish and other seafood products.
5. Method for automated imaging of whole-surface of round-shape agricultural products. The lack of a method to effectively and efficiently image the entire surfaces of round agro-products has been a longstanding obstacle to developing practical automated imaging-based food safety inspection technology to detect surface defects and contamination on round fruits such as apples and oranges. Traditional single-camera machine vision systems cannot effectively view all sides of a round object, while systems using multiple cameras or complex object manipulation increase the operational and instrumentation costs associated with practical implementation. ARS scientists in Beltsville, Maryland, developed a novel hyperspectral imaging system that uniquely incorporates an external optical assembly of three mirrors to view a round object from two sides and acquire images of the object while it is rotated on rollers, to construct a whole-surface image of the round object similar to a world-map projection. A whole-surface image processing algorithm was first developed using wooden spheres of different sizes and marked with “defect spots” at six surface locations, and then tested using 101 apples of various sizes and similarly marked with simulated defect spots. The results showed that the novel system accurately showed all six defect spots in 78% of the apple images, missed one of the six spots in only 4%, and showed seven spots in 18% due to partial duplication of an image area. This new system and image processing technique provide the basis for developing an effective whole-surface spectral imaging-based inspection system for round fruits and vegetables that will benefit fresh food processors seeking to ensure product safety and optimize quality-based sorting of their products.
Mukasa, P., Wahkoli, C., Mohammad, A., Park, E., Lee, J., Suh, H., Mo, C., Lee, H., Baek, I., Kim, M.S., Cho, B. 2020. Determination of the viability of retinispora (Hinoki cypress) seeds using shortwave infrared hyperspectral imaging spectroscopy. Journal of Near Infrared Spectroscopy. https://doi.org/10.1177/0967033519898890.
Li, Y., Wang, W., Long, Y., Peng, Y., Li, Y., Chao, K., Tang, X. 2019. A feasibility study of rapid nondestructive detection of total volatile basic nitrogen (TVB-N) content in beef based on airflow and laser ranging technique. Meat Science. 145:367-374.
Muhammad, M., Yao, G., Zhong, J., Chao, K., Aziz, M., Huang, Q. 2020. A facile and label-free SERS approach for inspection of fipronil in chicken eggs using SiO2@Au core/shell nanoparticles. Talanta. 207:120324. https://doi.org/10.1016/j.talanta.2019.120324.
Chao, K., Dhakal, S., Schmidt, W.F., Qin, J., Kim, M.S., Peng, Y., Huang, Q. 2020. Raman and IR spectroscopic modality for authentication of turmeric powder. Journal of Food Chemistry. 320:126567. https://doi.org/10.1016/j.foodchem.2020.126567.
Qin, J., Vasefi, F., Hellberg, R.S., Akhbardesh, A., Issacs, R.B., Yilmaz, A., Hwang, C., Baek, I., Schmidt, W.F., Kim, M.S. 2020. Detection of fish fillet substitution and mislabeling using multimode hyperspectral imaging techniques. Food Control. 114:107234. https://doi.org/10.1016/j.foodcont.2020.107234.
Joshi, R., Lohumi, S., Joshi, R., Kim, M.S., Qin, J., Baek, I., Cho, B. 2019. Raman spectral analysis for non-invasive detection of external and internal parameters of fake eggs. Sensors and Actuators B: Chemical. 303:127243. https://doi.org/10.1016/j.snb.2019.127243.
Pyo, J., Hong, S., Kwon, Y., Kim, M.S., Cho, K. 2020. Estimation of heavy metals using deep neural network with visible and infrared spectroscopy of soil. Journal of Hazardous Materials. https://doi.org/10.1016/j.scitotenv.2020.140162.
Yasmin, J., Ahmed, M., Lohumi, S., Wakholi, C., Kim, M.S., Cho, B. 2019. Classification method for viability screening of naturally aged watermelon seeds using FT-NIR spectroscopy. Sensors. 19(5):1190. https://doi.org/10.3390/s19051190.
Song, S., Liu, X., Huang, M., Zhu, Q., Qin, J., Kim, M.S. 2019. Detection of fish bones in fillets by Raman hyperspectral imaging technology. Journal of Food Engineering. 272:109808. https://doi.org/10.1016/j.jfoodeng.2019.109808.
Jeon, D., Stocker, M.D., Sokolova, E., Lee, H., Baek, I., Kim, M.S., Pachepsky, Y.A. 2020. Accounting for the three-dimensional distribution of E. coli concentrations in pond water in simulations assessment of microbial quality of water withdrawn for irrigation. Irrigation Science. 12(6):1708. https://doi.org/10.3390/w12061708.
Yasmin, J., Lohumi, S., Ahmed, M.R., Kandpal, L.M., Faqeerzada, M.A., Kim, M.S., Cho, B. 2020. Improvement in purity of healthy tomato seeds using an image-based one-class classification method. Sensors. 20(9):2690. https://doi.org/10.3390/s20092690.
Joshi, R., Joshi, R., Mo, C., Faqeezada, M., Amanah, H., Masithoh, R., Kim, M.S., Cho, B. 2020. Raman spectral analysis for quality determination of grignard reagent. Applied Sciences. 10(10):3545. https://doi.org/10.3390/app10103545.
Pyo, J., Duan, H., Baek, S., Kim, M.S., Jeon, T., Kwon, Y., Lee, H., Cho, K. 2020. A convolutional neural network regression for quantifying harmful cyanobacteria using hyperspectral imagery. Remote Sensing of Environment. 233:111350. https://doi.org/10.1016/j.rse.2019.111350.
Morgan, B.J., Stocker, M.D., Valdes-Avellan, J., Kim, M.S., Pachepsky, Y.A. 2019. Drone-based imaging to assess the microbial water quality in an irrigation pond: a pilot study. Science of the Total Environment. 716:135757. https://doi.org/10.1016/j.scitotenv.2019.135757.