Source: SK INFRARED, LLC submitted to NRP
HANDHELD OPTICAL SPECTROMETER FOR REAL-TIME CHARACTERIZATION OF DIGESTIBLE PROTEIN IN SOYBEANS
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
COMPLETE
Funding Source
SMALL BUSINESS GRANT
Reporting Frequency
Annual
Accession No.
1019450
Grant No.
2019-33610-29762
Cumulative Award Amt.
$100,000.00
Proposal No.
2019-00409
Multistate No.
(N/A)
Project Start Date
Jul 1, 2019
Project End Date
Feb 29, 2020
Grant Year
2019
Program Code
[8.5]- Food Science & Nutrition
Recipient Organization
SK INFRARED, LLC
1275 KINNEAR ROAD STE 233
COLUMBUS,OH 432121180
Performing Department
(N/A)
Non Technical Summary
Amino acid content in soybeans is important for high quality feed, aquaculture, and human food products. While protein is ostensibly a key marker for soybean quality, the presence of certain amino acids indicates how "digestible" the protein is for the purpose of its target application. These "essential amino acids" (EAAs), including cysteine, lysine, methionine, threonine, and tryptophan, are tracked by the US Soybean Export Council in the annual report on soybean quality. In addition, selecting soybeans with an abundance of these EAAs is critical for identifying plant-based fish meal substitutes for aquaculture.On a global basis, plant-based protein is of immense importance and there is significant interest in its ability to meet growing demand for protein from non-meat sources. Protein quality data could provide a useful way to define optimal foods to meet protein requirements in low income countries, where food availability can be very limited, and the choice of adequate protein sources can be vital. Plant-based protein is preferred to animal-based protein from an environmental perspective as it is associated with a lower land use requirement, and it is generally accepted that plant-based foods produce lower levels of greenhouse gases, which are associated with climate change, than animal-based foods. Soy exhibits wide variation in protein content mainly due to genetic, environmental and agronomic factors.The need exists for a handheld, real-time, affordable system that can measure EAAs in soybeans. This approach can be particularly beneficial at the point of sale, where farmers can demand higher prices for their higher quality soybeans, which are characterized by the presence of these EAAs. Successful development of the proposed soybean sensor technology will provide several benefits that address known issues in agriculture, food, and the environment. A real time, field deployable technology will enable a transition to value-pricing of soybean commodities. Farmers (sellers) and food producers (buyers) will be able to assess the quality of the soybean product at the point of sale, without having to resort to costly and slow laboratory analyses. This shift in the economic paradigm of soybean sales will benefit farmers and society by rewarding growers who focus on high quality products for human consumption and other specialty feed products.The sensor technology may be generalizable for other agriculture and food applications. For example, the same instrumentation can be used with minor changes to the backend algorithms and training data to monitor lycopene in tomatoes as well as other types of nutrients in a wide variety of food products. An easy to use, flexible, and effective sensor can have great value within the nutrition and food production markets.
Animal Health Component
(N/A)
Research Effort Categories
Basic
(N/A)
Applied
(N/A)
Developmental
100%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
70118201010100%
Knowledge Area
701 - Nutrient Composition of Food;

Subject Of Investigation
1820 - Soybean;

Field Of Science
1010 - Nutrition and metabolism;
Goals / Objectives
We intend to demonstrate technical feasibility of an optical approach for soybean quality characterization through laboratory research and prototype system development activities. Our end product is envisioned to be a handheld, real time sensor technology that supports agriculturally-related manufacturing of soybean products for human consumption and as feed for livestock and aquaculture.The goal of this project is to develop a commercial handheld infrared spectrometer-based rapid test device that is able to grade soybeans on the basis of their essential amino acid (cysteine, methionine, threonine, lysine, and tryptophan) profile.Due to genetic, environmental, and agronomic factors soybeans exhibit a wide variation in protein content as well as in their essential amino acid profile. Traditionally the soybean market driver has been economic, i.e., higher yields per acre to maximize return for the farmer. However, recent evidences show that one of the soybean market drivers is the protein content, suggesting that food-grade quality determines the export market potential of soybeans. A corollary of this new trend also implies that in the near future the quality of the protein in terms of its essential amino acid profile, in addition to protein content, will become a driver of soybean market. The Protein Digestibility Corrected for Amino Acid Score (PDCAAS) values of soy range from 0.91 to 1.00, indicating wide variation in essential amino acid content. While handheld infrared sensor devices are currently being used to measure protein levels in situ in soybeans, these devices cannot provide information on the essential amino acid profiles.Objectives:SK Infrared will work with The Ohio State University (OSU) to accomplish the Phase I objectives outlined below:Optimize soybean sample handling processOptimize chemometric algorithm performancePerform engineering trade studies for key subsystemsDesign handheld spectrometer prototypeDevelop commercialization planSK Infrared will develop a prototype design that focuses specifically on the optical spectrometer implementation validated by previous laboratory work at OSU.SK Infrared will develop a commercialization plan that will focus on how we plan to transition the proposed sensor technology from prototype to product after completion of the SBIR Phase I and II projects.Ultimately, our long term goal is to develop a handheld sensor system that can be transitioned for commercial development. The proposed project is one step toward that long term goal and should position the team well for subsequent investment by transition partners.
Project Methods
Sample Handling: OSU will develop a sample handling process for the proposed breadboard system. This process may include sample crushing or grinding (or not), sample rotation (for multiple view angles and spatially averaged results), and/or sample compaction against a collection optic (for mid-infrared attenuated total reflectance, if used). The end result of this task will be a design for the sample handling subsystem of the breadboard system, which leverages COTS components as much as possible for rapid development and test.Optimized Chemometric Algorithm Performance: OSU will continue to develop and optimize the underlying chemometric algorithms used to convert optical spectroscopy data into percent concentrations for each essential amino acids. We will perform spectroscopy measurements of a broader set of soybean varieties (ground and whole seed) in order to expand the algorithmic training sets and increase the accuracy of the quantification results. By necessity, "truth" measurements using GC will also be performed for each new sample, a necessary step for implementing the expanding training sets. A critical element of this task is to identify the processing requirements and user interface required for the prototype system design. The end result of this task will be a design for the data processing subsystem of the prototype sensor, which will include the processor, user interface, and software components.Laboratory Methods:Near Infrared Spectroscopy (NIR): NIR spectra of the soybean samples will be collected with a handheld NIR system constructed with an optical structure for sample illumination and collection of diffuse reflected light, monolithic micro-electromechanical system (MEMS) Michelson interferometer chip and a single indium-gallium-arsenide (InGaAs) photodetector. Soybean meal (~3 g) is placed in a glass dish, rotated and the spectra collected through the glass in the range from 7718 to 3829 cm-1. The background spectrum data is collected before each sample with a highly reflective gold-ceramic standard material. For each soybean sample, two spectra is collected.Profiling and Quantification of Amino Acids: The amino acid content of soybeans is analyzed with a gas chromatography (GC/MS) after acid hydrolysis. Samples are deaerated, closed under nitrogen, placed in heater and hydrolyzed. All acquired data is analyzed with HP Chemstation version A.06.03. The extraction and derivatization of AAs in plasma is carried out. Amino acid separation is achieved using a Phenomenex Zebron ZB-A AA analysis dedicated column. A selective ion monitoring (SIM) GC/MS method is applied for the detection of 26 AAs, based on the retention time (Rt) and a qualifier ion. Quantification was carried out employing norvaline as internal standard and constructing reference curves for every AA by means of standard solutions.Data Analysis: Soybean samples (n=150), including different varieties and geographical origins, will be provided by soybean growers affiliated to the Ohio Soybean Council (OSC). We want to include sample variability and thus we propose to collect a large number of samples encompassing different genotypes, growth conditions, locations, soil among others. Due to the large number of variables (high-dimensional predictor space) in vibrational spectroscopy data from a limited number of subjects, estimation methods provide a rational simulation following Dobbin and Simon's recommendations. We will collect and analyze at least 100 samples for the model-building (training set) stage. For the model validation studies, 50 independent samples will be considered to be an upper bound. Partial Least Squares Regression (PLSR) provides optimal EAA quantification results from optical spectroscopy measurements. Thus, using the spectra obtained and reference concentrations from GC-MS; quantitative models will be generated with PLSR for each EAA. Using the NIR spectra and reference amino acid concentrations, quantitative models will be generated with PLSR (Partial Least Square Regression). Independent validation study will be conducted using approximately 75 % of sample set to generate calibration models and about 25 % of the all set to serve as independent validation set.Predictive Accuracy of the Models: Regression models will be used to generate prediction models and the accuracy and ability of these models will be examined with an independent test set representative of the classes modeled with the training set. Blind samples (the researcher will not have access to its identity before prediction) will be included to test the ability of the models to predict the levels of essential amino acids. Results from the validation testing set will be used to determine the sensitivity, specificity and positive predictive value of the patterns

Progress 07/01/19 to 02/29/20

Outputs
Target Audience:SK Infrared LLC has conducted Phase I research and technology development toward a novel, handheld optical sensor that provides real time characterization of the quality of soybeans. Specifically, the sensor quantifies total protein content and five essential amino acids (cysteine, lysine, methionine, threonine, tryptophan) that indicate digestibility of the protein within the soybean product. This technology directly responds to the stated needs of soybean farmers for tools that can verify soybean quality at point of sale, evolving the economics of soybean transactions toward value-pricing. SK Infrared has collaborated with the Ohio State University and has leveraged laboratory investigations recently funded by the Ohio Soybean Council to demonstrate the feasibility of our approach. The sensing concept involves collecting optical spectroscopy data of soybean samples and analyzing the data using advanced chemometric algorithms. The end result is an amino acid profile that can be obtained with only seconds of data collection and analysis. In addition, we have developed an engineering concept whereby low-cost, commercially available components can be integrated to produce a self-contained sensor system. This design concept was verified by development of a Preliminary Prototype, which demonstrated the sensor functionality in a laboratory environment. A subsequent Phase II effort will focus on an enhanced, field-deployable prototype that can be demonstrated at outdoor locations such as farms or grain elevators. SK Infrared intends to commercialize this research thereafter with significant economic impact and benefit to U.S. food-grade soybean producers. Changes/Problems: Nothing Reported What opportunities for training and professional development has the project provided? Nothing Reported How have the results been disseminated to communities of interest?The OSU team has had several meetings with various stakeholders in the soybean agriculture community (e.g., farmers, trade councils, researchers). Based on these interaction, we have developed the following list of "market needs" to address the demands of the potential user community for a handheld soybean quality sensor. These market needs include the following: Sensor shall measure quality of soybean samples Sensor measurements shall be accurate Sensor measurements shall be rapid Sensor shall be of small, handheld size Sensor shall be deployable at field locations (farm, grain elevator, etc.) Sensor shall be easy to operate with minimal training Sensor shall require minimal maintenance. ased on these user needs, a set of engineering requirements have been derived for a future soybean quality sensor and listed in Table 5. A sensor design follows from these requirements (Phase I) followed by development and testing of a field deployable prototype (Phase II). Table 5. Correlation of market needs with derived sensor requirements Market needs Sensor requirements Sensor shall measure quality of soybean samples Measurement of % concentrations of 5 EAAs: cysteine, lysine, methionine, threonine, and tryptophan Measurement of % concentration of protein, moisture, fat, oleic and linoleic acids Measurement using ground or whole bean samples Sensor measurements shall be accurate Measurement absolute accuracy of 0.1% concentration (depending on the attribute concentration and accuracy of the reference method) Sensor measurements shall be rapid Complete measurement in 20 seconds or less Sensor and user interface application startup sequence less than 10 seconds Sensor shall be of small, handheld size Sensor head volume less than 110 cubic inches (<1800 cm3) Sensor head weight less than 5 lbs (<2.3 kg) Sensor shall be deployable at field locations (farm, grain elevator, etc.) Sensor operation over temperature range from 23°F-104°F (?5°C-40°C) Sensor operation in presence of light precipitation Sensor withstand vibrations typically associated with tractor driving through a field Sensor shall be easy to operate with minimal training Sensor operation performed by user with high school diploma or equivalent User interface via display screen User interface can initiate measurement, halt measurement, adjust measurement parameters Measurement results provided to user interface display screen Measurement results (% concentration, measured spectrum) archived for subsequent processing User interface and data processing leverage commercial smartphone or tablet Smartphone or tablet connects to sensor head via Bluetooth and USB cable with equivalent functionality Sensor shall require minimal maintenance Sensor head operation of 8 hours on single battery charge Sensor lifetime >10,000 hours What do you plan to do during the next reporting period to accomplish the goals? Nothing Reported

Impacts
What was accomplished under these goals? The details are included in a 26 page final report being submitted to the Program Point of Contact This Phase I effort has met all of the proposed objectives toward developing a design concept for a handheld sensor that offers real-time, in situ characterization of important soybean traits, including protein and Essential Amino Acids. SK Infrared LLC, and our partners at OSU, have completed laboratory investigations focused on enhancing the measurement performance of the sensor as well as engineering design activities toward a future, field-deployable prototype to be developed under a Phase II effort. The engineering effort included assembly and testing of a Preliminary Prototype device based primarily on COTS components, which sought to verify the sensor design concept and illuminate key engineering trades to be considered further in Phase II. Our Phase I design efforts have increased confidence that a future commercial product is viable and have reduced various technical and cost risks associated with this technology. Ultimately, a sensor technology that supports increased production of high quality soybeans is within reach, and additional Phase II funding will help further this endeavor. Regression models for powdered samples (including ground soybeans, isolates, concentrates and soy products) showed excellent fit. All calibration models showed good performance statistics with low standard error and high correlation of determinations (Table below). Models using NIR spectra collected from intact soybeans showed lower modeling performance as compared to the ground samples, similar results have been reported for models on soybean amino acid (Pazdernik et al. 1997). Our promising results highlight the ability to parallel and/or outperform the quantitative techniques using benchtop systems from other research groups Our technique can be effectively applied in the industry as an analytical tool for phenotyping soybeans and simultaneously measure several quality parameters. Calibration Model Validation Model Range n Factor SECV Rcal Range n SEP Rval Powder Threonine (%) 1.34 - 3.23 31 4 0.051 0.99 1.40 - 3.16 10 0.062 0.99 Cysteine (%) 0.45 - 1.06 31 4 0.041 0.98 0.54 - 1.00 10 0.043 0.97 Methionine (%) 0.47 - 1.14 30 4 0.044 0.98 0.47 - 1.13 11 0.090 0.90 Lysine (%) 2.34 - 5.54 31 5 0.246 0.98 2.43 - 5.37 10 0.238 0.98 Tryptophan (%) 0.30 - 1.32 31 3 0.080 0.97 0.30 - 1.23 10 0.097 0.96 Crude protein (%) 33.00 - 84.30 31 6 1.630 0.99 33.35 - 79.25 10 2.801 0.98 Intact Seed Threonine (%) 1.36 - 1.53 11 5 0.012 0.97 1.40 - 1.45 3 0.011 0.90 Cysteine (%) 0.51 - 0.62 11 6 0.014 0.94 0.56 - 0.60 3 0.068 0.79 Methionine (%) 0.52 - 0.61 11 6 0.013 0.94 0.52 - 0.57 3 0.055 0.93 Lysine (%) 2.37 - 2.57 10 5 0.014 0.98 2.38 - 2.54 4 0.035 0.87 Tryptophan (%) 0.30 - 0.47 11 2 0.015 0.96 0.36 - 0.43 3 0.022 0.94 Crude Protein (%) 32.83 - 36.66 11 5 0.221 0.980 33.80 - 35.86 3 0.634 0.88 Based on user needs, a set of engineering requirements have been derived for a future soybean quality sensor and listed in Table 5. A sensor design follows from these requirements (Phase I) followed by development and testing of a field deployable prototype (Phase II). To meet the requirements we have produced a high level, block diagram design, that illustrates major subsystems and components. This design comprises two major subsystems: the sensor head, which contains the NIR spectrometer and supporting electronics, and a smartphone or tablet, which contains the processor for analyzing spectral data as well as the UI. The decision to use a smartphone or tablet was predicated on discussions with the OSC and member farmers, who indicated that smartphone/tablet interface is preferable to an integrated processor and interface. Many other, high tech devices are currently being added to the farmers' "toolbox" that run off of smartphone/tablet devices, and generally the farmers were very comfortable with the simplicity that can be designed into an appropriate UI application (app). A Commercialization plan was developed to include: The path to commercialization includes the following steps in our plan: Demonstrating feasibility - accomplished by this Phase I project Advancing the prototype - a Phase II objective to demonstrate functionality more fully and attract additional early adopters for market validation Licensing Intellectual Property - obtaining patent protection to provide a strategic advantage to the company as it launches the product Financing of the product launch including equity investment necessary to fund growth capital requirements of the enterprise Pilot Scale Production to create the competitive product and first revenue Commercial scale production for the US market initially and growth to the rest of the world via distribution channels. The management of the company has significant prior experience raising capital (more than $17M), operating world class manufacturing operations, and engaging with customers worldwide making SK Infrared a credible partner for the eventual commercial success of this project.

Publications