Source: WASHINGTON STATE UNIVERSITY submitted to NRP
FACT: PREDICTING WHEAT HAGBERG FALLING NUMBER FROM NEAR INFRARED SPECTROMETERS
Sponsoring Institution
National Institute of Food and Agriculture
Project Status
ACTIVE
Funding Source
AFRI COMPETITIVE GRANT
Reporting Frequency
Annual
Accession No.
1023750
Grant No.
2020-67021-32460
Cumulative Award Amt.
$499,660.00
Proposal No.
2019-07485
Multistate No.
(N/A)
Project Start Date
Aug 1, 2020
Project End Date
Jul 31, 2025
Grant Year
2020
Program Code
[A1541]- Food and Agriculture Cyberinformatics and Tools
Recipient Organization
WASHINGTON STATE UNIVERSITY
240 FRENCH ADMINISTRATION BLDG
PULLMAN,WA 99164-0001
Performing Department
Crop and Soil Sciences
Non Technical Summary
The Hagberg-Perten falling number (FN) test is the international standard for measuring sprouting damage in wheat. Sprouted grain produces starch-degrading enzymes, resulting in poor baking quality and heavy price discounts. Because enzymes are catalysts, a large amount of sound wheat can be ruined by a small amount of sprouted wheat. Worldwide losses are ~$1B/yr. The current FN test is slow, destroys the sample, and is conducted retrospectively. These features make it unable to prevent contamination during harvest, transport, and storage. A tool that combines Artificial Neural Networks (ANN) with Near-InfraRed (NIR) and HyperSpectral Imaging (HSI) promises to solve this problem. Our preliminary data indicate that ANN continuously achieve greater predictive accuracy with increasing training sample size. Using this approach, we propose to develop an online computing tool to transform three existing datasets into knowledge for: i) prediction of FN based on NIR of ground kernels; ii) non-destructive prediction of FN from HSI of intact kernels; and iii) Extension applications in spring and winter wheat breeding. The dataset resources include a USDA Preharvest Sprouting Project with 2940 samples covering 320 varieties and 7 environments spanning 3 years; Washington Variety Testing Program with 28,000 samples covering 495 varieties grown in multiple locations for 7 years; and Wheat Market Center with 1500 samples outside of Pacific Northwest. If successful, this new data-application will provide wheat growers, millers, bakers, and breeders with a rapid tool to: 1) partition sprouted/sound wheat in real-time, and 2) identify lines with resistance to low FN for sustainable development.
Animal Health Component
10%
Research Effort Categories
Basic
80%
Applied
10%
Developmental
10%
Classification

Knowledge Area (KA)Subject of Investigation (SOI)Field of Science (FOS)Percent
5031549108170%
2011549108030%
Knowledge Area
201 - Plant Genome, Genetics, and Genetic Mechanisms; 503 - Quality Maintenance in Storing and Marketing Food Products;

Subject Of Investigation
1549 - Wheat, general/other;

Field Of Science
1081 - Breeding; 1080 - Genetics;
Goals / Objectives
Our overallgoalis to improve overall end-use quality of wheat by isolating low falling number wheat and developing varieties with resistance to low falling number.Our five objectives are:Objective 1: Predict wheat falling number using near infrared spectroscopy on wheat meal.NIR spectrometers have been widely and successfully used to evaluate protein, oil, and moisture content. Although research has been conducted to predict FN from NIR on either wheat meal or kernels, the resulting prediction accuracy is still far from satisfactory. Unlike protein, oil, and moisture content, the causes of low FN are catalytic. A single enzyme can make thousands of cuts into starch and protein within seconds during dough making, so even a small amount of enzyme can cause significant damage. Detecting this enzyme in low amounts requires precise identification of the wavelength patterns related to the enzyme's molecular bonds, including N-H, C-H, and O-H. Our preliminary results showed the potential for improving prediction accuracy by increasing the training sample sizes using ANN. Our target is to limit prediction error within 40 seconds, compared to 30 seconds of FN standard deviation of technical replication.Objective 2: Predict wheat falling number using hyperspectral imaging of individual kernels.FN prediction using HSI of individual kernels not only provides a non-destructive assessment tool, but also a mechanism to improve prediction accuracy. Measuring the kernel surface provides direct access to the relevant area for enzyme activity, which is concentrated in the aleurone layer of the wheat kernel. Mixing with the inside starchy endosperm, which makes up over 80% of the wheat kernel, substantially dilutes the enzyme concentration, reducing FN prediction accuracy. Detecting and measuring only the surface components of the wheat kernel is achievable using HSI. Our target is to limit prediction error within 35 seconds, compared to 30 seconds of FN standard deviation of technical replication.Objective 3: Validate prediction accuracy on independent samples outside the Pacific Northwest.Low FN is a nationwide problem that degrades wheat quality and causes heavy financial losses to farmers, millers and bakers. As the prediction methods and tool will be developed by using samples from Pacific Northwest region, it is critical to validate these methods and tool to predict FN on independent samples outside the region.Objective 4: Develop and maintain an online computing tool to predict wheat falling number.The input of the computing tool is NIR and HSI data and the output is the prediction of FN. The calibration will be integrated to the tools and documented online. The tool can be directly used as a plugin for firmware update on devices such the detectors on combine harvesters and elevators.Objective 5: Stakeholder outreach for improving grower economics.Active engagement of wheat growers, breeders, millers, and bakers will allow them to incorporate our results as soon as they are available. For example, at the current 20% (R square) prediction accuracy using fast and nondestructive HSI screening, wheat breeders can efficiently eliminate the bottom 30% of early lines that have very little chance to get into the top 10% for field trials. Now breeders can eliminate them earlier and work on making more crosses.
Project Methods
This integrated project (research and extension) proposes to develop and apply rapid and non-destructive tools to predict Hagberg-Perten FNs for wheat. We will use three data/material resources and seven methods to achieve the five objectives of this project.NIFA PHS GWAS project (1008463):A total of 320 soft white winter wheat lines were grown in 7 environments spanning 3 years (2016 to 2018). Of these environments, two did not experience low FN, four had PHS (Pullman 2013, 2017, and 2018, Central Ferry 2016), and one had LMA (Mayview 2017). This dataset provides us with 2,240 (320x7) grain samples. The FNs, ranging from 62 to 512 seconds, were determined with two technical replicates per sample. The knowledge of PHS/LMA events in each environment will be set as the class variables to train ANN to differentiate these events.WSU variety trial on FN testing program:Since 2013, the program has evaluated 495 elite varieties over multiple years in multiple locations (Steber lab). About 4,000 samples were collected each year. Currently, there 28,000 (4000x7) samples. Each sample was measured for FN with one technical replicate. Of the soft white winter and hard red winter wheat samples from 2016 and 2018, 11 locations experienced low FN in 2018 solely due to LMA. FN ranged from 149 to 389 seconds. In contrast, samples in 2016 experienced LMA, PHS, or a combination of both. The knowledge of PHS/LMA events in each environment will be set as the class variable to train ANN to differentiate these events.Samples Outside of Pacific Northwest.The low FN problem is more severe in soft white than other class of wheat. Therefore, the cultivated areas are limited to the regions that are more dry during harvest and have less fluctuation in temperature during the filling season, such as Pacific Northwest. Solving the FN problem will make more regions available to grow soft white to improve management and increase market values. Both similarity and differences are expected between samples collected in the Pacific Northwest and samples outside the area. We will examine our prediction by using independent samples outside the area. The samples will be provided by Wheat Market Center through the Wheat Quality Council (see Support Letter from the technical Director of Wheat Market Center) at a rate of 500 samples per year. These samples cover the major wheat states, including Kansas, Colorado, Oklahoma, Taxa, North Dakota, Montana, and Ohio.Method 1: Hagberg-Perten Falling Number testThe FN test is routinely performed in CoPD Steber's laboratory. The FN test will be performed according to the AACC International Method 56-81.03. FN will be corrected for barometric pressure.Method 2: α-amylase enzyme assaysThe colorimetric α-amylase enzyme assay is also routinely performed in CoPD Steber's laboratory. The Phadebas® colorimetric enzyme assay will be used to specifically measure α-amylase activity based on the appearance of blue color. The test has been adapted to a 96-well format.Method 3: Near-Infrared Data AcquisitionWe will use a NIR spectrometer provided by our collaborator, Jinguo Hu, who is the research leader of USDA Plant Germplasm Introduction and Testing Research Unit (see support letterfrom Dr. Hu) to collect NIR spectral data using the standard protocol35. The model is Matrix-I Industrial-Type Fourier Transform from Bruker (Figure 5). Diffuse reflection scans (800-2,500 nm, every 2 nm) will be collected using the default settings of 32 repetitions per scan. Sample material, in either meal or kernel format, will be placed in a standard ring cell (97 mm in diameter × 45 mm in height) that possesses a quartz window on the face oriented toward the spectrometer. Before each sample scan, a similar scan will be made of the internal reference ceramic material. Measurements will be performed indoors at room temperature (20-22 °C) and 50-60% humidity.Method 4: Hyperspectral Imaging Data AcquisitionHyperspectral images will be acquired using Specim IQ studio. The camera provides 512 × 512-pixel resolution and 204 spectral bands over a wavelength range of ~440-1000 nm, with a spectral resolution of about 2 nm. Both black references (approximately 0% reflectance) and white references (approximately 100% reflectance) of polytetrafluoroethylene (PTFE) will be recorded to establish the baseline signals36,37. Shutter speeds for each band will be 150 milliseconds (ms), and 90 seconds will be the overall period for taking an entire image. The image will be stored as an ENVI (i.e., Environment for Visualizing Images) encoded file that can be decoded into a 3-dimension numeric matrix in Python environments. Samples will be placed on a movable plate and illuminated with two Halogen lamps, mounted on two sides of the lens, at an angle of incidence of approximately 45°. The kernels will be placed on a black high-density polyethylene plate, randomly aligned to the wells on the plate (Figure 6). Reflectance values will be calculated for each pixel by first subtracting the black reference value for the corresponding detector pixel. Second, reflectance values will be normalized according to the corresponding white reference values and the ratio of exposure times, assuming a linear response. Individual kernels will be extracted by a software developed by PD Zhang.Method 5: Artificial Neural NetworksCurrently, ANN systems are the most active area in artificial intelligence research. These systems, inspired by our brains, function through a network of neurons and connections. In ANN, the neurons are described by bias (relative to base) and the connections are described by weights. With advances in computational capability, more layers and more nodes within layers are placed between the input and output layers. This machine learning method using ANN has gained a popular name, "deep learning". Deep learning has been successfully used in a wide range of applications, including automatic driving and facial recognition. In this project, we will mainly use two techniques in the framework of ANN: autoencoders and convolutional ANN.?Method 6: Falling number predictiontoolThe input of the tool will be data from NIR and HSI and the output is FN. The tool will be developed using Python and can be used directly as plugin by firmware. Calibration will be provided to link the input data with our training data. We will provide the computation tool in two formats. One is the stand-alone version that users can access at their own computational facility. The other is the online version that users can upload their images and use the computation server provided by PD Zhang. The tool and related documentation will be hosted on PD Zhang's lab website (http://zzlab.net).Method 7: Extension to engage stakeholdersThe project will have immediate application in wheat breeding. CoPD Carter and Pumphrey are the leaders of WSU winter and spring wheat breeding Extension programs, respectively. CoPD Carter has developed 600 Double Haploid (DH) lines derived from parents in the NIFA GWAS panel40,41. Doubled haploids are homozygous 2n progeny derived from F1 female gametes (1n) by artificially doubling the chromosome number in tissue culture. CoPD Pumphrey has ~1000 advanced breeding lines. They will use FN predictions to select their breeding lines. We anticipate widely disseminating our results and notifying stakeholders about the value of integrating our research via several communication outlets.

Progress 08/01/23 to 07/31/24

Outputs
Target Audience:The project reached stakeholders in the full range from scientists to farmers. One presentation was delivered to farmers, four peer-reviewed articles were published in scientific journals, and one extension article was published in an extension magazine in the period of 8/1/2023-7/31/2024. One software package was updated to process image data, AI4EVER (https://zzlab.net/AI4EVER). Changes/Problems:There was no major change in approach. What opportunities for training and professional development has the project provided? Meijing Liang, Preharvest Sprouting in Quinoa: A New Screening Method Adapted to Panicles and GWAS Components, The International Conference of Plant and Animal Genome, January 14, 2024 Zhiwu Zhang, Modeling Complex Data Structure in GWAS Using Blink, The International Conference of Plant and Animal Genome, January 14, 2024 How have the results been disseminated to communities of interest?The research results have been disseminated through multiple platforms, including publication, extension articles, website, and conferences. Four articles on the method and software development have been published during period of 2023-2024. Lanzhi Li, Xingfei Zheng, JiaboWang, Xueli Zhang, Xiaogang He,Liwen Xiong, Shufeng Song, Jing Su, Ying Diao, Zheming Yuan,Zhiwu Zhang*& Zhongli Hu*,Joint analysis of phenotype-effect-generation identifies loci associated with grain quality traits in rice hybrids.Nature Communication, 2023,https://doi.org/10.1038/s41467-023-39534-x Chun-Peng James Chen*, Yang Hu, Xianran Li, Craig F. Morris, Stephen Delwiche, Arron H. Carter, Camille Steber,Zhiwu Zhang,An independent validation reveals the potential to predict Hagberg-Perten falling number using spectrometers.The Plant Phenome Journal, 2023,https://doi.org/10.1002/ppj2.20070 Zhou Tang, Meinan Wang, Michael Schirrmann, Karl-Heinz Dammer, Xianran Li, Robert Brueggeman, Sindhuja Sankaran, Arron H Carter, Michael O Pumphrey, Yang Hu*, Xianming Chen*, andZhiwu Zhang*,Affordable High Throughput Field Detection of Wheat Stripe Rust Using Deep Learning with Semi-Automated Image Labeling.Computers and Electronics in Agriculture, 2023,https://doi.org/10.1016/j.compag.2023.107709 Zhou Tang, Yang Hu*, andZhiwu Zhang*,ROOSTER: An image labeler and classifier through interactive recurrent annotation.F1000Research, 2023,https://doi.org/10.12688/f1000research.127953.1 What do you plan to do during the next reporting period to accomplish the goals? Analyze hyperspectral image of 200 individual grains under real time germination progression. Establish neural network model to connect alpha amylase enzyme activities for individual grains at different gemination time (0, 2, 4, 6, 8, 10, and 12 hours) and hyperspectral images. Features of the hyperspectral reflection curved will be extracted to serve as the middle layer of the neural networks. Software development Algorithm development Publications Training for graduate and postdoc Outreach stakeholders

Impacts
What was accomplished under these goals? In the period of 8/1/2023-7/31/2024, our focus is to analyze data accumulated in the previous three years. We published analytical results using traditional statistical methods, including partial least square regression, ridge regression, and Bayesian-information and Linkage-disequilibrium Iteratively Nested Keyway (BLINK). The prediction accuracy is moderate to low. We also tried a neural network with multi layer perceptron with the raw bands of hyperspectral reflection as the input layer. The prediction accuracy did not get better. Then we were inspired by the success of AlphaFolder in prediction of protein 3D structure using middle layers from extracted features from amino acids. Our middle layer features were the spline functions of the hyperspectral reflection curves.The challenge is the convolve process between the middle layer and the input layer and between the middle layer and the output layer, similar to the construction of theAlphaFolder. Solving the challenges is the focus of next year.

Publications

  • Type: Journal Articles Status: Published Year Published: 2023 Citation: 1. Lanzhi Li, Xingfei Zheng, JiaboWang, Xueli Zhang, Xiaogang He,Liwen Xiong, Shufeng Song, Jing Su, Ying Diao, Zheming Yuan, Zhiwu Zhang* & Zhongli Hu*, Joint analysis of phenotype-effect-generation identifies loci associated with grain quality traits in rice hybrids. Nature Communication, 2023, https://doi.org/10.1038/s41467-023-39534-x
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: 2. Chun-Peng James Chen*, Yang Hu, Xianran Li, Craig F. Morris, Stephen Delwiche, Arron H. Carter, Camille Steber, Zhiwu Zhang, An independent validation reveals the potential to predict HagbergPerten falling number using spectrometers. The Plant Phenome Journal, 2023, https://doi.org/10.1002/ppj2.20070
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: 3. Zhou Tang, Meinan Wang, Michael Schirrmann, Karl-Heinz Dammer, Xianran Li, Robert Brueggeman, Sindhuja Sankaran, Arron H Carter, Michael O Pumphrey, Yang Hu*, Xianming Chen*, and Zhiwu Zhang*, Affordable High Throughput Field Detection of Wheat Stripe Rust Using Deep Learning with Semi-Automated Image Labeling. Computers and Electronics in Agriculture, 2023, https://doi.org/10.1016/j.compag.2023.107709
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: 4. Zhou Tang, Yang Hu*, and Zhiwu Zhang*, ROOSTER: An image labeler and classifier through interactive recurrent annotation. F1000Research, 2023, https://doi.org/10.12688/f1000research.127953.1


Progress 08/01/22 to 07/31/23

Outputs
Target Audience:The project reached stakeholders in the full range from scientists to farmers. One presentation was delivered to farmers, five peer-reviewed articles were published in scientific journals, and one extension was published in an extension magazine. One software package has been developed to process image data, AI4EVER (https://zzlab.net/AI4EVER). Changes/Problems:There was no major change in approach. What opportunities for training and professional development has the project provided?The project provided support for the training and professional development at both graduate and postdoc levels.A graduate student and a postdoc were trained for research and career development.Two people were supported for workshops and conferencesto present their discoveries to reach stakeholders. Camille Steber, Alternatives to assess wheat falling numbers, October 8, 2021 Yang Hu, WSU Entrepreneurship Skills and Knowledge Accelerator (WESKA), June 13-17, 2022. Rao Doppa, Digital Agriculture Summit, October 6-7, 2020 James Chen, Decoding Images in minutes with GRID. WSU Plant Science Retreat, March 6, 2021. Yang Hu, Becoming an Entrepreneurial Woman at WSU, April 22, 2021. Zhiwu Zhang, Estimating falling number using hyperspectral imaging, the progress in developing a non destructive test, Pacific NorthWest Wheat Quality Council and Falling Number Workshop, January 25, 2023 How have the results been disseminated to communities of interest?The research results have been disseminated through multiple platforms, including publication, extension articles, website, and conferences. Five articles on the method and software development have been published. Chun-Peng James Chen*, Yang Hu, Xianran Li, Craig F. Morris, Stephen Delwiche, Arron H. Carter, Camille Steber,Zhiwu Zhang,An independent validation reveals the potential to predict Hagberg-Perten falling number using spectrometers.The Plant Phenome Journal, 2023,https://doi.org/10.1002/ppj2.20070 Zhou Tang, Meinan Wang, Michael Schirrmann, Karl-Heinz Dammer, Xianran Li, Robert Brueggeman, Sindhuja Sankaran, Arron H Carter, Michael O Pumphrey, Yang Hu*, Xianming Chen*, andZhiwu Zhang*,Affordable High Throughput Field Detection of Wheat Stripe Rust Using Deep Learning with Semi-Automated Image Labeling.Computers and Electronics in Agriculture, 2023,https://doi.org/10.1016/j.compag.2023.107709 Zhou Tang, Yang Hu*, andZhiwu Zhang*,ROOSTER: An image labeler and classifier through interactive recurrent annotation.F1000Research, 2023,https://doi.org/10.12688/f1000research.127953.1 Yang Hu, Stephanie M Sjoberg, Chunpen James Chen, Amber L Hauvermale, Craig F Morris, Stephen R Delwiche, Ashley E Cannon, Camille M Steber*,Zhiwu Zhang*,As the number falls, alternatives to the Hagberg-Perten falling number method: A review.Comprehensive Reviews in Food Science and Food Safety, 2022,https://doi.org/10.1111/1541-4337.12959. Chun-Peng J Chen, Gota Morota, Kiho Lee,Zhiwu Zhang, Hao Cheng,VTag: a Semi-Supervised Pipeline for Tracking Pig Activity with a Single Top-View Camera.Journal of Animal Science, 2022,https://doi.org/10.1093/jas/skac147. What do you plan to do during the next reporting period to accomplish the goals? Analyze hyperspectral image of 200 individual grains under real time germination progression. Establish neural network model to connect alpha amylase enzyme activities for individual grains at different gemination time (0, 2, 4, 6, 8, 10, and 12 hours) and hyperspectral images. Software development Algorithm development Publications Training for graduate and postdoc Outreach stakeholders

Impacts
What was accomplished under these goals? Spectrometers have been studied as a potential tool for fast falling number assessment, but these studies have not arrived at a calibration that is applicable across variable environments. In this project, we conducted a calibration experiment and a validation experiment. The calibration experiment had 462 grain samples consisting 92 varieties grown at 24 locations in 2019 and being measured by a near-infrared, hyper-spectrometer camera. The validation experiment had 39 samples collected from 10 locations in two years (2018 and 2019) that experienced PHS (7), LMA (13), or no event (19), and scanned by a hyper-spectrometer imager. A Pearson correlation of 0.72 was achieved between the observed and predicted falling numbers. Among the validation samples, samples that experienced LMA had prediction accuracies of 0.81, which was significantly higher than PHS samples (0.39). Furthermore, kernel pixel-wise predictions indicated that the low falling number pixels were concentrated in areas near the embryo for all samples. When the low falling number pixels spread throughout the kernels, the falling number became very low. These results show the potential of using spectrometers to predict FN, which could provide a faster assessment method. Using this faster method, breeders would have an efficient tool to use in the development of varieties with resistance to PHS and LMA. This study shows the potential that growers could separate damaged grain from sound grain during harvesting and transportation to preserve the value of sound grain. Low falling numbers (FN), resulted mainly from preharvest sprouting (PHS) or late maturity Alpha-Amylase (LMA), cause over one billion dollars in losses to the wheat industry annually worldwide.Currently, the FN assessment relies on aminternational standard method, the Hagberg-Perten method, which is slow, labor-intensive, and vulnerable to variability. A fast and cost-effective method will be beneficial not only to reduce the workload but open the opportunity to access a large number of breeding lines so that superior varieties can be developed with resistance to low FN. In the past year, we measured the FN with the Hagberg-Perten method on over 500 samples collected in three years (2019-2021) in Washington State, US. Each sample was imaged with ~100 grains by using hyperspectral cameras. Image analyses demonstrated that the pixels with low FN appeared on the embryo distal for all kernels. These pixels with low FN also appeared in the brush distal, crease, and everywhere in samples with low FN. The kernel average predicted FN had a prediction error of 25 seconds, which is close to the range of technical replicate error of the Hagberg-Perten method (20-30 seconds). The predicted FN demonstrated intensive variation among the kernels within the same samples. Since enzymes from PHS and LMA are catalysts, a large amount of sound (high FN) grain can be ruined by mixing with a single kernel with low FN grain. This may partially explain the variability of the Hagberg-Perten method, which needs to blend 200~300 kernels together to get a single measurement. Consequently, this study opens the opportunities to quickly evaluate grain damage to avoid inappropriate mixing, removing damaged kernels, and developing resilient wheat varieties to solve the low FN problem at the beginning of the wheat production chain.

Publications

  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Chun-Peng James Chen*, Yang Hu, Xianran Li, Craig F. Morris, Stephen Delwiche, Arron H. Carter, Camille Steber, Zhiwu Zhang, An independent validation reveals the potential to predict HagbergPerten falling number using spectrometers. The Plant Phenome Journal, 2023, https://doi.org/10.1002/ppj2.20070
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Zhou Tang, Meinan Wang, Michael Schirrmann, Karl-Heinz Dammer, Xianran Li, Robert Brueggeman, Sindhuja Sankaran, Arron H Carter, Michael O Pumphrey, Yang Hu*, Xianming Chen*, and Zhiwu Zhang*, Affordable High Throughput Field Detection of Wheat Stripe Rust Using Deep Learning with Semi-Automated Image Labeling. Computers and Electronics in Agriculture, 2023, https://doi.org/10.1016/j.compag.2023.107709
  • Type: Journal Articles Status: Published Year Published: 2023 Citation: Zhou Tang, Yang Hu*, and Zhiwu Zhang*, ROOSTER: An image labeler and classifier through interactive recurrent annotation. F1000Research, 2023, https://doi.org/10.12688/f1000research.127953.1
  • Type: Journal Articles Status: Published Year Published: 2022 Citation: Yang Hu, Stephanie M Sjoberg, Chunpen James Chen, Amber L Hauvermale, Craig F Morris, Stephen R Delwiche, Ashley E Cannon, Camille M Steber*, Zhiwu Zhang*, As the number falls, alternatives to the HagbergPerten falling number method: A review. Comprehensive Reviews in Food Science and Food Safety, 2022, https://doi.org/10.1111/1541-4337.12959.
  • Type: Journal Articles Status: Published Year Published: 2022 Citation: Chun-Peng J Chen, Gota Morota, Kiho Lee, Zhiwu Zhang, Hao Cheng, VTag: a Semi-Supervised Pipeline for Tracking Pig Activity with a Single Top-View Camera. Journal of Animal Science, 2022, https://doi.org/10.1093/jas/skac147.


Progress 08/01/21 to 07/31/22

Outputs
Target Audience:The project reached stakeholders in the full range from scientists to farmers. One presentation was delivered to farmers, five peer-reviewed articles were published in scientific journals, and one extension was published in extension magazine. One software package has been developed to process image data, AI4EVER (https://zzlab.net/AI4EVER). Changes/Problems:There was no major change in approach. What opportunities for training and professional development has the project provided?The project provided support for the training and professional development at both graduate and postdoc levels.A graduate student and a postdoc were trained for research and career development.Two people were supported for workshops and conferencesto present their discoveries to reach stakeholders. Camille Steber, Alternatives to assess wheat falling numbers, October 8, 2021 Yang Hu, WSU Entrepreneurship Skills and Knowledge Accelerator (WESKA), June 13-17, 2022 How have the results been disseminated to communities of interest?The research results have been disseminated through multiple platforms, including publication, extension articles, website, and conferences. Five articles on the method and software development has been published. Yang Hu, Stephanie M Sjoberg, Chunpen James Chen, Amber L Hauvermale, Craig F Morris, Stephen R Delwiche, Ashley E Cannon, Camille M Steber*,Zhiwu Zhang*,As the number falls, alternatives to the Hagberg-Perten falling number method: A review.Comprehensive Reviews in Food Science and Food Safety, 2022,https://doi.org/10.1111/1541-4337.12959. Chun-Peng J Chen, Gota Morota, Kiho Lee,Zhiwu Zhang, Hao Cheng,VTag: a Semi-Supervised Pipeline for Tracking Pig Activity with a Single Top-View Camera.Journal of Animal Science, 2022,https://doi.org/10.1093/jas/skac147. Jiabo Wang* andZhiwu Zhang*,GAPIT Version 3: Boosting Power and Accuracy for Genomic Association and Prediction.Genomics, Proteomics and Bioinformatics, 2021,https://doi.org/10.1016/j.gpb.2021.08.005. Chandler R. Keller, Yang Hu, Kelsey F. Ruud, Anika E. VanDeen, Steve R. Martinez, Barry T. Kahn, Zhiwu Zhang, Roland K. Chen and Weimin Li.Human Breast Extracellular Matrix Microstructures and Protein Hydrogel 3D Cultures of Mammary Epithelial Cells.Cancers, 2021,https://www.mdpi.com/2072-6694/13/22/5857. Matthew T. McGowan,Zhiwu Zhang& Stephen P. Ficklin,Chromosomal characteristics of salt stress heritable gene expression in the rice genome. BMCGenomic Data, 2021,https://doi.org/10.1186/s12863-021-00970-7. What do you plan to do during the next reporting period to accomplish the goals? Image 200 individual grains with hyperspectral cameras under real time germination progression. Measure alpha amylase enzyme activities for individual grains at different gemination time (0, 2, 4, 6, 8, 10, and 12 hours). The individual grains will be imaged with hyperspectral cameras on both crease and non-crease sides. Software development Algorithm development Publications Training for graduate and postdoc Outreach stakeholders

Impacts
What was accomplished under these goals? Low falling numbers (FN), resulted mainly from preharvest sprouting (PHS) or late maturity Alpha-Amylase (LMA), cause over one billion dollars in losses to the wheat industry annually worldwide.Currently, the FN assessment relies on aminternational standard method, the Hagberg-Perten method, which is slow, labor-intensive, and vulnerable to variability. A fast and cost-effective method will be beneficial not only to reduce the workload but open the opportunity to access a large number of breeding lines so that superior varieties can be developed with resistance to low FN. In the past year, we measured the FN with the Hagberg-Perten method on over 500 samples collected in three years (2019-2021) in Washington State, US. Each sample was imaged with ~100 grains by using hyperspectral cameras. Image analyses demonstrated that the pixels with low FN appeared on the embryo distal for all kernels. These pixels with low FN also appeared in the brush distal, crease, and everywhere in samples with low FN. The kernel average predicted FN had a prediction error of 25 seconds, which is close to the range of technical replicate error of the Hagberg-Perten method (20-30 seconds). The predicted FN demonstrated intensive variation among the kernels within the same samples. Since enzymes from PHS and LMA are catalysts, a large amount of sound (high FN) grain can be ruined by mixing with a single kernel with low FN grain. This may partially explain the variability of the Hagberg-Perten method, which needs to blend 200~300 kernels together to get a single measurement. Consequently, this study opens the opportunities to quickly evaluate grain damage to avoid inappropriate mixing, removing damaged kernels, and developing resilient wheat varieties to solve the low FN problem at the beginning of the wheat production chain.

Publications

  • Type: Journal Articles Status: Published Year Published: 2022 Citation: 1. Yang Hu, Stephanie M Sjoberg, Chunpen James Chen, Amber L Hauvermale, Craig F Morris, Stephen R Delwiche, Ashley E Cannon, Camille M Steber*, Zhiwu Zhang*, As the number falls, alternatives to the HagbergPerten falling number method: A review. Comprehensive Reviews in Food Science and Food Safety, 2022, https://doi.org/10.1111/1541-4337.12959.
  • Type: Journal Articles Status: Published Year Published: 2022 Citation: 2. Chun-Peng J Chen, Gota Morota, Kiho Lee, Zhiwu Zhang, Hao Cheng, VTag: a Semi-Supervised Pipeline for Tracking Pig Activity with a Single Top-View Camera. Journal of Animal Science, 2022, https://doi.org/10.1093/jas/skac147.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: 3. Jiabo Wang* and Zhiwu Zhang*, GAPIT Version 3: Boosting Power and Accuracy for Genomic Association and Prediction. Genomics, Proteomics and Bioinformatics, 2021, https://doi.org/10.1016/j.gpb.2021.08.005.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: 4. Chandler R. Keller, Yang Hu, Kelsey F. Ruud, Anika E. VanDeen, Steve R. Martinez, Barry T. Kahn, Zhiwu Zhang, Roland K. Chen and Weimin Li. Human Breast Extracellular Matrix Microstructures and Protein Hydrogel 3D Cultures of Mammary Epithelial Cells. Cancers, 2021, https://www.mdpi.com/2072-6694/13/22/5857.
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: 5. Matthew T. McGowan, Zhiwu Zhang & Stephen P. Ficklin, Chromosomal characteristics of salt stress heritable gene expression in the rice genome. BMC Genomic Data, 2021, https://doi.org/10.1186/s12863-021-00970-7.


Progress 08/01/20 to 07/31/21

Outputs
Target Audience:We reachedCraig Morris, the director of Western Wheat Quality Laboratory, for comments on sample selection and customer demands. Dr. Morris found that late maturity alpha-amylase (LMA) influences wheat quality differently from pre-harvesting sprouting (PHS). It is desirable to develop a non-destructive method to differentiate LMA and PHS. Changes/Problems:There was no major change in approach. What opportunities for training and professional development has the project provided?The project provided support for the training and professional development at both graduate and postdoc levels.A graduate student and a postdoc were trained for research and career development.Three people were supported for workshops and conferencesto present their discoveries to reach stakeholders. Rao Doppa, Digital Agriculture Summit, October 6-7, 2020 James Chen, Decoding Images in minutes with GRID. WSU Plant Science Retreat, March 6, 2021. Yang Hu, Becoming an Entrepreneurial Woman at WSU, April 22, 2021. How have the results been disseminated to communities of interest?The research results have been disseminated through multiple platforms, including publication, extension articles, website, and conferences. Three articles on the method and software development has been published. Yang Hu, andZhiwu Zhang*,GridFree: a python package of imageanalysis for interactive grain counting and measuring.Plant, Physiology, 2021,https://doi.org/10.1093/plphys/kiab226 Zhou Tang, Atit Parajuli, Chunpeng James Chen, Yang Hu, Samuel Revolinski, Cesar Augusto Medina, Sen Lin,Zhiwu Zhang*.Validation of UAV-based alfalfa biomass predictability using photogrammetry with fully automatic plot segmentation.Scientific Reports,doi: 10.1038/s41598-021-82797-x. Wei Huang, Ping Zheng, Zhenhai Cui, Zhuo Li, Yifeng Gao, Helong Yu, You Tang, Xiaohui Yuan, andZhiwu Zhang*.MMAP: A Cloud Computing Platform for Mining the Maximum Accuracy of Predicting Phenotypes from Genotypes.Bioinformatics, 2020.https://doi.org/10.1093/bioinformatics/btaa824 What do you plan to do during the next reporting period to accomplish the goals? Measure 200 samples with hyperspectral cameras Software development Algorithm development Publications Training for graduate and postdoc Outreach stakeholders

Impacts
What was accomplished under these goals? Spectrometers have been studied as a potential tool for fast falling number assessment, but these studies have not arrived at a calibration that is applicable across variable environments. In this project, we conducted a calibration experiment and a validation experiment. The calibration experiment had 462 grain samples consisting 92 varieties grown at 24 locations in 2019 and being measured by a near-infrared, hyper-spectrometer camera. The validation experiment had 39 samples collected from 10 locations in two years (2018 and 2019) that experienced PHS (7), LMA (13), or no event (19), and scanned by a hyper-spectrometer imager. A Pearson correlation of 0.72 was achieved between the observed and predicted falling numbers. Among the validation samples, samples that experienced LMA had prediction accuracies of 0.81, which was significantly higher than PHS samples (0.39). Furthermore, kernel pixel-wise predictions indicated that the low falling number pixels were concentrated in areas near the embryo for all samples. When the low falling number pixels spread throughout the kernels, the falling number became very low. These results show the potential of using spectrometers to predict FN, which could provide a faster assessment method. Using this faster method, breeders would have an efficient tool to use in the development of varieties with resistance to PHS and LMA. This study shows the potential that growers could separate damaged grain from sound grain during harvesting and transportation to preserve the value of sound grain.

Publications

  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Yang Hu, and Zhiwu Zhang*, GridFree: a python package of imageanalysis for interactive grain counting and measuring. Plant, Physiology, 2021, https://doi.org/10.1093/plphys/kiab226
  • Type: Journal Articles Status: Published Year Published: 2021 Citation: Zhou Tang, Atit Parajuli, Chunpeng James Chen, Yang Hu, Samuel Revolinski, Cesar Augusto Medina, Sen Lin, Zhiwu Zhang*. Validation of UAV-based alfalfa biomass predictability using photogrammetry with fully automatic plot segmentation. Scientific Reports, 2021, doi: 10.1038/s41598-021-82797-x.
  • Type: Journal Articles Status: Published Year Published: 2020 Citation: Wei Huang, Ping Zheng, Zhenhai Cui, Zhuo Li, Yifeng Gao, Helong Yu, You Tang, Xiaohui Yuan, and Zhiwu Zhang*. MMAP: A Cloud Computing Platform for Mining the Maximum Accuracy of Predicting Phenotypes from Genotypes. Bioinformatics, 2020, https://doi.org/10.1093/bioinformatics/btaa824 .