Diagnóstico automático de infección por Nosemiasis en abejas melíferas mediante procesado de imágenes
Fecha
2018
Autores
Prendas Rojas, Juan Pablo
Figueroa Mata, Geovanni
Ramírez Montero, Marianyela
Calderón, Rafael
Ramírez Bogantes, Melvin
Travieso González, Carlos Manuel
Título de la revista
ISSN de la revista
Título del volumen
Editor
Instituto Tecnológico de Costa Rica
Resumen
Las abejas polinizan una gran variedad de especies de plantas, incluyendo los cultivos agrícolas. Se estima que cerca del 30% del alimento consumido por la población mundial es derivado de cultivos polinizados por abejas. La infección por Nosemiasis es una de las principales causas de la pérdida de colmenas a nivel mundial. Los métodos de laboratorio para el diagnóstico del nivel de infección por este microsporidio son lentos, caros y demandan la presencia de un experto. Se propone un sistema automático, confiable y económico de cuantificación de infección por Nosema, a partir del procesamiento digital de imágenes.
Con el uso de técnicas de segmentación de imágenes, caracterización de objetos y conteo de formas se han reproducido las técnicas de Cantwell y Hemocitómetro de manera automática. Para el conteo de esporas se implementaron tres descriptores el tamaño, la excentricidad y la circularidad, de manera tal que son invariantes a la escala y rotación de las imágenes. Se trabajó con un total de 375 fotografías agrupadas en carpetas de 5, las cuales fueron previamente etiquetadas por un experto según el nivel de infección (muy leve, leve, moderado, semifuerte y fuerte). Con ello se alcanzó un porcentaje de diagnóstico correcto de infección del 84%.
Bees pollinate a wide variety of plant species, including agricultural crops. It is estimated that about 30% of the food consumed by the world population is derived from crops pollinated by bees.Nosemiasis infection is one of the leading causes of bee hive loss worldwide. The laboratory methods for the diagnosis of the level of infection by this microsporidium are slow, expensive and require the presence of an expert for spore count. It is proposed the creation of an automatic, reliable and economical system of quantification of Nosema infection from digital image processing. Using the techniques of image segmentation, object characterization and shape counting, the Cantwell and Hemocytometer techniques have been automatically reproduced. For the counting of spores, three descriptors were implemented: size, eccentricity and circularity, in such a way that they are invariant to the scale and rotation of the images. We worked with a total of 375 photographs grouped in folders of 5, which were previously labeled according to the level of infection (very mild, mild, moderate, semi-strong and strong). The correct diagnosis rate was 84%
Bees pollinate a wide variety of plant species, including agricultural crops. It is estimated that about 30% of the food consumed by the world population is derived from crops pollinated by bees.Nosemiasis infection is one of the leading causes of bee hive loss worldwide. The laboratory methods for the diagnosis of the level of infection by this microsporidium are slow, expensive and require the presence of an expert for spore count. It is proposed the creation of an automatic, reliable and economical system of quantification of Nosema infection from digital image processing. Using the techniques of image segmentation, object characterization and shape counting, the Cantwell and Hemocytometer techniques have been automatically reproduced. For the counting of spores, three descriptors were implemented: size, eccentricity and circularity, in such a way that they are invariant to the scale and rotation of the images. We worked with a total of 375 photographs grouped in folders of 5, which were previously labeled according to the level of infection (very mild, mild, moderate, semi-strong and strong). The correct diagnosis rate was 84%
Descripción
Palabras clave
ABEJAS, BEES, COLMENAS, HIVES, ENFERMEDADES PARASITARIAS, PARASITIC DISEASES, PARÁSITOS, PARASITES