Use of Near-Infrared Transmittance for Classifying
Mutant Grain Samples in Corn

---

Jeremy Sykes.
Dr. Mark Campbell, Faculty Mentor, 1999.

---

Most of the corn in the U.S. is fed to livestock and possesses a normal (non-mutant) endosperm type. In corn, there are a number of mutations that can alter the structure of the starch in the endosperm. These changes can result in novel grain types that have improved functional properties for food and nonfood applications. If the demand for corn hybrids possessing these mutations increases, there will be a need for commercial handlers to maintain genotypic purity. The objective of this study was to investigate the use of Near-Infrared Transmittance Spectroscopy (NITS) for classifying unknown grain samples representing ten genotypic classes. Classification models were constructed using Principle Component Analysis (PCA) which assigns samples into discrete categories based on similarities of near-infrared spectra. The performance of the models was assessed on the percentages of unknown samples that were misclassified.