Location: Grain Quality and Structure ResearchTitle: Extended multiplicative signal correction to improve prediction accuracy of protein content in weathered sorghum grain samples
|PEIRIS, KAMARANGA H. S. - Kansas State University|
|JAGADISH, S.V. KRISHNA - Kansas State University|
Submitted to: Cereal Chemistry
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 7/29/2020
Publication Date: 9/3/2020
Citation: Peiris, K., Bean, S.R., Jagadish, S. 2020. Extended multiplicative signal correction to improve prediction accuracy of protein content in weathered sorghum grain samples. Cereal Chemistry. 97; 1066-1074. https://doi.org/10.1002/cche.10329.
Interpretive Summary: Sorghum grains may be subjected to weathering damage in the field when naked grains on the head are exposed to adverse environmental conditions during grain development or dry-down before harvest. Weathering may alter the chemical and physical properties of grains and negatively impact grain quality. Because of changes to weathered grain, near infrared reflectance spectroscopy (NIRS) spectra of weathered grains showed a flat spectra with higher absorbances in some spectral regions compared with the spectra of sound (unweathered) grains. The change in NIRS spectra creates problems when trying to accurately analyze weathered sorghum grains by NIR. Methods for correcting the spectra of weathered grains were evaluated and it was found that using extended multiplicative scatter correction (EMSC) enhanced the prediction of sorghum grain composition when sample sets included weathered grain. Including weathered grain into calibration sample sets and using EMSC techniques improved the robustness of NIR sorghum calibration curves.
Technical Abstract: Sorghum grains may be subjected to weathering damage in the field when naked grains on the head are exposed to adverse environmental conditions during grain development or dry down before harvest. Weathering may alter the chemical and physical properties of grains. Near infrared reflectance spectroscopy (NIRS) spectra of weathered grains showed a rather flat spectra with higher absorbances in 950-1050 nm spectral region compared with the spectra of sound grains. A partial least squares (PLS) calibration developed with sound grain samples using spectra preprocessed with multiplicative signal correction (MSC) predicted protein content of an external validation sample set of sound grains with a root mean squire error of prediction (RMSEP) = 0.67% and coefficient of determination (R2) = 0.83. When weathered grain samples were included in the validation set prediction performance dropped with a RMSEP = 8.35% and R2 = 0.06. When weathered grain spectra were introduced to the calibration, predictive performance improved with a RMSEP = 0.83% and R2 = 0.72. Application of extended multiplicative scatter correction (EMSC) to preprocess the spectra further enhanced the prediction performance with a RMSEP = 0.61% and R2 = 0.85. These results showed that inclusion of weathered grain spectra to the calibration and preprocessing spectra with EMSC markedly improved the robustness of the sorghum protein calibration.