Identificación de adulterantes en comino (Cuminum cyminum l.) mediante el uso de imágenes hiperespectrales
No Thumbnail Available
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Nacional de Trujillo
Abstract
El comino es una especia mayormente consumida en polvo, que tiene un alto valor económico, por lo tanto, susceptible a la adulteración con residuos de otros alimentos o con comino gastado. Puede causar efectos negativos en la salud, como reacciones alérgicas. El objetivo de esta investigación fue utilizar el sistema de HSI-NIR en combinación con análisis multivariado para discriminar, clasificar y cuantificar concentraciones de contaminación (<10%) y adulteración (10 – 40%) en muestras de comino diferentes, adulterado con cáscara de maní, nuez, pecana y comino gastado. Para cada adulterante se hizo mezclas de baja concentración (0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 7.5 y 10% p/p) y alta concentración (10 12.5, 15, 17.5, 20, 22.5, 25, 27.5, 30, 32.5, 35, 37.5, 40%). Se uso 20 repeticiones por concentración y 30 de muestras puras por cada adulterante y cada proveedor. Se tomo una imagen por cada 2 repeticiones en un rango espectral de infrarrojo cercano de 900 – 1710 nm. Para evitar fenómenos no deseados como la dispersión de la luz y ruido aleatorio se aplicó preprocesamiento espectral usando las técnicas de corrección: centrado en la media (MC) y variable normal estándar (SNV) y la primera derivada de Savitzky-Golay. Se realizó un análisis exploratorio de componentes principales (PCA) para identificar diferencias las muestras puras de las muestras adulteradas, además se seleccionó 80 % de muestras para calibración y 20 % para validación, analizadas con el método de discriminación del cuadrado de mínimos parciales (PLS-DA) y el método cuantificación de regresión de mínimos cuadrados parciales (PLS-R). Para la observación directa de la concentración de los adulterantes se elaboró mapas químicos de distribución. PLS-DA permitió discriminar la adulteración con sensibilidad, especificidad y precisión superior al 90%. Los modelos de regresión PLS-R presentaron una alta capacidad predictiva de la concentración de adulterantes en polvo de comino (R_P^2 > 0.99). Los mejores modelos identificados fueron con adulteración de cáscara de nuez y pecana. En general el uso NIR-HSI es efectivo para detectar las principales contaminaciones o adulterantes de comino en polvo de diferentes orígenes geográficos.
ABSTRACT Cumin is a spice mostly consumed in powder form, which has a high economic value, therefore, susceptible to adulteration with residues from other foods or with spent cumin. Can cause negative health effects, such as allergic reactions. The objective of this research was to use the HSI-NIR system in combination with multivariate analysis to discriminate, classify and quantify concentrations of contamination (<10%) and adulteration (10 - 40%) in samples of cumin different, adulterated with peanut shell, nutshell, pecan shell and spent cumin. For each adulterant, mixtures of low concentration (0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 7.5 and 10% w / w) and high concentration (10 12.5, 15, 17.5) were made, 20, 22.5, 25, 27.5, 30, 32.5, 35, 37.5, 40%). 20 replicates per concentration and 30 of pure samples were used for each adulterant and each supplier. An image was taken for every 2 repetitions in a near infrared spectral range of 900-1710 nm. To avoid unwanted phenomena such as light scattering and random noise, spectral preprocessing was applied using the correction techniques: mean center (MC) and standard normal variable (SNV) and the first derivative of Savitzky-Golay. An exploratory analysis of principal components (PCA) was carried out to identify differences between the pure samples and the adulterated samples, in addition, 80% of the samples were selected for calibration and 20% for validation, analyzed with the partial least square discrimination method (PLS -DA) and the partial least squares regression quantification method (PLS-R). For the direct observation of the concentration of the adulterants, chemical distribution maps were elaborated. PLS-DA allowed to discriminate adulteration with sensitivity, specificity, and precision greater than 90%. The PLS-R regression models showed a high predictive capacity of the concentration of adulterants in cumin powder (R_P^2> 0.99). The best models identified were with walnut and pecan shell adulteration. In general, the use of NIR-HSI is effective to detect the main contaminations or adulterants of cumin powder from different geographical origins.
ABSTRACT Cumin is a spice mostly consumed in powder form, which has a high economic value, therefore, susceptible to adulteration with residues from other foods or with spent cumin. Can cause negative health effects, such as allergic reactions. The objective of this research was to use the HSI-NIR system in combination with multivariate analysis to discriminate, classify and quantify concentrations of contamination (<10%) and adulteration (10 - 40%) in samples of cumin different, adulterated with peanut shell, nutshell, pecan shell and spent cumin. For each adulterant, mixtures of low concentration (0, 0.5, 1.0, 1.5, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, 7.5 and 10% w / w) and high concentration (10 12.5, 15, 17.5) were made, 20, 22.5, 25, 27.5, 30, 32.5, 35, 37.5, 40%). 20 replicates per concentration and 30 of pure samples were used for each adulterant and each supplier. An image was taken for every 2 repetitions in a near infrared spectral range of 900-1710 nm. To avoid unwanted phenomena such as light scattering and random noise, spectral preprocessing was applied using the correction techniques: mean center (MC) and standard normal variable (SNV) and the first derivative of Savitzky-Golay. An exploratory analysis of principal components (PCA) was carried out to identify differences between the pure samples and the adulterated samples, in addition, 80% of the samples were selected for calibration and 20% for validation, analyzed with the partial least square discrimination method (PLS -DA) and the partial least squares regression quantification method (PLS-R). For the direct observation of the concentration of the adulterants, chemical distribution maps were elaborated. PLS-DA allowed to discriminate adulteration with sensitivity, specificity, and precision greater than 90%. The PLS-R regression models showed a high predictive capacity of the concentration of adulterants in cumin powder (R_P^2> 0.99). The best models identified were with walnut and pecan shell adulteration. In general, the use of NIR-HSI is effective to detect the main contaminations or adulterants of cumin powder from different geographical origins.
Description
Keywords
análisis espectral, análisis de imágenes, análisis de datos, comino en polvo