A new high-throughput method for measuring oil content in intact, single corn kernels is demonstrated using nuclear magnetic resonance imaging (MRI) methods. This nondestructive technique enables the evaluation of relative oil content in up to 2,592 corn kernels in less than 40 min using a 1.5 T clinical MRI scanner. Custom software was developed to process and analyze 3-D magnetic resonance (MR) image data rapidly. The precision and accuracy of the MR method for measuring oil content are discussed. The precision of the MRI results is shown to be dependent on MR scanner noise. The MRI results show very good relative accuracy compared with low-field NMR, NIR transmission, and accelerated solvent extraction measurements. Minor differences between the MRI and low-field NMR experimental protocols were shown to be inconsequential to the oil content measurement. Extending the MRI method to the analysis of other oilseeds and/or the use of other magnetic field strengths is discussed, as is a comparison of this MRI method relative to other high-throughput magnetic resonance screening techniques.Paper no. J10981 in JAOCS 82, 855-862 (December 2005).
KEY WORDS:High-throughput, intact seed, magnetic resonance imaging, MRI, nondestructive, proximate oil.Quantitative analysis of oil content in single seeds is a valuable analytical method for monitoring the success of selective breeding and transgenic engineering programs focused on increasing oil content in commercial crops. Conventional analytical procedures, such as solvent extraction (1), accelerated solvent extraction (ASE) (2,3), supercritical fluid extraction (4-6), microwaveassisted extraction (7), and Soxtherm extraction (8), provide both high accuracy and high precision for measuring oil content. Unfortunately, these methods are time consuming, laborious, and involve hazardous solvents. The minimum sample quantities required using these methods often preclude measurements of single seeds, or at best provide single-seed measurements with reduced accuracy and precision. Most importantly, these methods destroy the sample, an undesirable approach if intact seeds are required for subsequent breeding experiments or for genetic investigations of the heritability of a high-oil trait.NIR transmission (NIRT) spectroscopy is a nondestructive method for evaluating bulk oil content in samples containing large numbers of seeds (9,10). Oil content is calculated using a nonlinear chemometric approach, which depends heavily on the creation of a hyperspectral model basis set. Robust NIRT results are achieved when the seeds being tested are similar in composition to the seeds used to construct the chemometric model. Trait differences between test and model seeds, i.e., differences in oil, protein, and/or starch content, can lead to poor measurement accuracy/precision for bulk samples and create difficulties using NIRT for measurement of single seeds. Since trait modification is the precise goal behind many selective breeding and transgenic programs, new NIRT chemometric models would hav...