2010 Annual Report
1a.Objectives (from AD-416)
(1) Develop spectroscopic imaging procedures (hyperspectral or multispectral) that can be used to assess quality of small grains (emphasizing wheat, rice, and barley) and oilseeds in bulk for grade, class, foreign material, and damage (mold, black point, heat, frost, and insect). (1a). Develop hyperspectral image analysis algorithms for determination of the level of mold damage from Fusarium Head Blight on wheat kernels. This will involve the development of image processing routines that identify the infected kernels in representative samples of intact wheat kernels and determine the regions of Fusarium damage within each infected kernel. (1b) Develop hyperspectral image analysis algorithms for identification of wheat kernels damaged by heat, frost, black point, and insects, as defined by official inspection criteria. (1c) Develop hyperspectral image analysis algorithms for the prediction of flour yield and break flour yield in soft red winter wheat. (2) Develop optical and mechanical methods and instrumentation for grain quality measurement that are applicable at points of sale, such as elevators, terminals, and mills. (2a) Develop rapid and objective optical methods for prediction of starch quality indicators in wheat, such as the ratio of amylose-to-amylopectin, and the identification of wheat into three states of waxiness: waxy, partial waxy, and wild type. (2b) Develop a near-infrared (NIR) spectroscopy procedure for wheat gluten quality determination for use in commerce.
1b.Approach (from AD-416)
Fusarium-inoculated hard red spring and hard red winter wheat samples will be imaged using an in-house near-IR hyperspectral system. Image analysis will be a multistep process. First, for each kernel a mask will be created from one of images whose wavelength creates a strong contrast between kernel and background. The mask will be applied to the images at all other wavelengths in order to remove the background. Principal component analysis (PCA) loadings from images of damaged and normal regions will be examined to identify the wavelengths at local minima and maxima, which inherently possess the greatest contrast between Fusarium damage and healthy endosperm.
Hyperspectral image analysis will also be used to examine three wheat milling properties: milling yield (% straight grade flour) defined as the percent by mass of all flour fractions recovered through a 94-mesh screen; solvent retention capacity in 50% (w/w) sucrose solution, a measure of the water affinity of the macro-polymers (starch, arabinoxylans, gluten, and gliadins); and solvent retention capacity in 5% (w/w) lactic acid, an indicator of gluten strength.
Near-IR spectroscopy will explored as a method for measuring the degree of waxiness in hexaploid wheat. Wild type, partial waxy (waxy null alleles in one or two genomes), and waxy samples (null alleles in all genomes), drawn from breeders' advanced lines of hexaploid wheat, will be used. Gel electrophoresis will be used to identify the waxy protein (granule bound starch synthase, GBSS) in each sample.
Lastly, a near-IR procedure for wheat gluten quality will be developed in conjunction with a rheological procedure. The wheat samples consist of approximately 50 lines grown in field replicated (3x) plots over three consecutive seasons. Half of these lines are transgenic, in which the gene construct modifies the length of the central repeat region within the high molecular weight (HMW) glutenin subunits. Different levels of gene expression, hence, level of glutenin protein, are represented as a function of the transgenic ancestor. Thus, this set will contain a much wider range in the ratio of glutenin-to-gliadin than naturally encountered. Flour from these samples will be evaluated for glutenin and gliadin contents by SE-HPLC another ARS laboratory. At Beltsville, the flour will be scanned in the NIR and FT-mid-IR regions. Rheological properties, such as the recovery response for a gluten specimen subjected to a controlled regiment of compressive force and hold time, will be measured at a third laboratory. Spectral calibrations for glutenin and gliadin concentrations, as well as calibrations for the rheological parameters (percent recovery and recovery time constant), will be developed using partial least squares regression. Additionally, classification algorithms (PLS discriminant analysis and SVM) algorithms will be developed that will identify the genetically modified lines based on their spectral response.
During the short period (June-September 2010) of this five-year project’s life effort continued on the use of near-infrared reflectance spectroscopy to distinguish waxy and partial waxy states in common (hexaploid) wheat. The waxy condition describes the biochemical circumstance of the absence of amylose in starch that would otherwise, in natural circumstances, consist of the two large polysaccharide molecules, amylose and amylopectin. Waxiness is controlled by three identical genes originating from the three genomes in hexaploid wheat, such that all three must be null in order for the plant to produce kernels whose starch is completely waxy. Otherwise, when one or more genes are active, granule bound starch synthase, the enzyme that regulates amylose synthesis, is produced and the resulting starch is a blend of the two macromolecules. The analysis of data from two consecutive years of breeders trials of hard winter wheat was completed after ground truth genotyping data by polymerase chain reaction (DNA profiling) and protein gel electrophoresis (profiling of the enzyme that regulates the synthesis of amylose) were supplied from cooperators. This set of data was unique for NIR analysis in that all eight genotypes representing the natural, partial waxy, and fully waxy conditions were available for analysis. The results indicate that the fully waxy condition is routinely identified at a 95% success rate, using linear discriminant analysis applied to the NIR spectra of wheat that has been prepared in one of three formats - ground meal, bulk kernel, or single kernel. Far more difficult is the ability to distinguish the partial waxy states and to distinguish these from the natural state.
Improvements were made to an inhouse-designed near-infrared (1000-1700 nm) hyperspectral imaging system. This system, along with other applications, is being used in developing methods for screening wheat kernels for presence of Fusarium Head Blight. This fungal disease reduces milling and baking quality and is a food safety concern because it produces the mycotoxin, deoxynivalenol (DON). An existing extended visible (400-1000 nm) system was similarly used. Exhaustive searches were performed on the 144 and 125 wavelength pair images that respectively comprised the NIR and visible systems to determine accuracy of classification using a linear discriminant analysis classifier. On a limited set of wheat samples, the best wavelength pairs, either with visible or NIR wavelengths, were able to discriminate Fusarium-damaged kernels from sound kernels at an average accuracy of approximately 95%. Accuracy dropped off substantially when the visual contrast between the two kernel conditions became imperceptible. The NIR region was slightly better than the visible region in its broader array of acceptable wavelength pairs. Further, the region of interest (ROI) defined as the whole kernel was slightly better than ROIs limited to either a portion of the endosperm or the germ tip. For the NIR region, the spectral absorption near 1200 nm, attributed to ergosterol (a primary constituent in fungi cell membranes), was shown to be useful in spectral recognition of Fusarium.