| Título: | Discriminant analysis of sunflower seeds for fatty acid composition using NIR spectroscopy. |
| Autores: |
GRUNVALD, A. K. CARVALHO, C. G. P. de MANDARINO, J. M. G. GONÇALVES, J. L. ANDRADE, C. A. de B. SCAPIM, C. A. |
| Fecha: |
2012 2012-06-08 |
| Publicador: | Repositório Acesso Livre à Informação Científica da Embrapa (ALICE) |
| Fuente: |
 |
| Tipo: | Artigo em periódico indexado (ALICE) |
| Tema: |
Girassol Óleo vegetal Helianthus annuus Vegetable oil |
| Descripción: |
Near-infrared reflectance (NIR) spectroscopy data from sunflower seeds of genotypes differing in oleic and linoleic fatty acid contents were evaluated by discriminant analysis. A single husked seed was used for spectral analysis of each genotype. We applied two clustering strategies to discriminate the genotypes. First, we fit three calibration curves for high oleic × low oleic, high oleic × mid-oleic and mid-oleic × low oleic genotype-derived fatty acids. Next, we fit just one curve to all these three groups. All these curves were established using principal component analysis. The clustering precision was determined by d, the Mahalanobis distance between the genotype and the closest group, and D, the distance up to the next group. The spectral range used for calibration was 1,651?1,952 nm. The genotypes were well classified when d was close to one and the ratio D/d was close to or higher than two. The discriminant analysis allowed the classification of the high oleic and mid-oleic genotypes even though they had similar signals in gas chromatography (GC). Although both clustering strategies were similar, the first one allowed a better discrimination between the mid-oleic and low oleic genotypes. 2012 |
| Idioma: | en |