It is important to assess the bias and accuracy of grain protein measurement by emerging methods such as NIR (near-infrared) based technology. This study conducted a comparative analysis of NIR based grain protein measurement methods and evaluated the inter-method bias. The study included 100 wheat grain samples with protein content ranging from 7.16 to 17.24%. Grain protein content was determined using three NIR based methods – on-combine H3 GrainGage™ (Logan, UT), FOSS Infratec benchtop NIR (FOSS Analytics, Denmark), and GrainSense handheld protein analyzer (GrainSense Ltd, Finland), and a standard laboratory method – Dumas Combustion, which was used as a reference method. An additional 500 samples with protein content ranging from 8.7 to 15.7% were also included for further comparison between the H3 GrainGage™ and the FOSS Infratec NIR. Despite strong correlation, all NIR based methods showed significant bias, with the FOSS NIR showing the lowest mean bias compared to the Dumas Combustion. The H3 GrainGage™ showed greater bias for samples with high protein content while the GrainSense grain protein analyzer showed greater bias for samples with low protein content. When the H3 GrainGage™ was compared with the FOSS Infratec NIR, the mean bias was relatively lower than when compared with the Dumas Combustion. Despite their broad applicability, NIR based instruments require improved accuracy and robustness for grain protein determination, particularly for low and high protein ranges. NIR based methods may be adequate where a certain bias is tolerable, whereas standard laboratory methods may be necessary when high precision is required.
Mahato et al. (Fri,) studied this question.