Automatic classification of red cells in peripheral blood smears
No Thumbnail Available
Date
Advisors
Evaluators
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Industrial de Santander
Abstract
Description
Introducción: El diagnóstico del estado eritrocitario en frotis de sangre periférica es un proceso realizado normalmente de forma manual a partir de observación microscópica, lo cual implica una considerable inversión de tiempo y recursos, además de posibles problemas de subjetividad y diicultad en la reproducibilidad del diagnóstico. Objetivo: Desarrollar una aplicación que permita la clasiicación automática de glóbulos rojos en frotis de sangre periférica, de utilidad como herramienta de ayuda diagnóstica. Metodología: Se usaron técnicas de procesamiento de imágenes para segmentar los eritrocitos en las fotografías microscópicas y medir en ellos área, perímetro, solidez, circularidad, excentricidad, textura y dimensión box-counting. Se usó una red neuronal artiicial para clasiicar los eritrocitos según sus características en siete clases, incluyendo normalidad y seis alteraciones patológicas. La red se entrenó de acuerdo con la clasiicación de 262 eritrocitos realizada por un hematólogo experto. Los desarrollos se hicieron en matlab®, una poderosa plataforma de computación cientíica. Resultados: La red escogida alcanza el 97.3% de aciertos en los datos de validación. Las equivocaciones en la red corresponden a células de dudosa clasiicación aún para un experto, por presentar características correspondientes a varias clasiicaciones patológicas. Conclusiones: La aplicación desarrollada clasiica de manera rápida y acertada los diferentes tipos de glóbulos rojos presentes en una muestra microscópica de frotis de sangre periférica, siendo de utilidad como herramienta de apoyo diagnóstico.
Introduction: The process of erythrocyte classification in peripheral blood smear is normally done manually from microscopic observation. This implies not only a considerable investment of time and resources but also brings potential problems of subjectivity and difficulty in the reproducibility of diagnosis. Objective: To develop an application that allows the automatic classification of red blood cells in peripheral blood smears, as a diagnostic aid tool. Methodology: Image processing techniques were used in order to segment erythrocytes in the microscopic photographs and to measure characteristics as area, perimeter, solidity, circularity, eccentricity, texture and box-counting dimension. An artificial neural network was used to classify the red blood cells in the images in seven classes, including normal and six pathological changes, according to their characteristics. The network was trained according to the classification of 262 erythrocytes by an expert hematologist. The developments were made in matlab ®, a powerful scientific computing platform. Results: The chosen network reaches 97.3% correct in the validation data. Mistakes in the network correspond to cells with various pathological classifications features, which make them difficult to classify even for an expert. Conclusions: The developed application classifies quickly and accurately the different types of red blood cells in a microscopic sample of peripheral blood smear, so it could be useful as a diagnostic support tool.
Introduction: The process of erythrocyte classification in peripheral blood smear is normally done manually from microscopic observation. This implies not only a considerable investment of time and resources but also brings potential problems of subjectivity and difficulty in the reproducibility of diagnosis. Objective: To develop an application that allows the automatic classification of red blood cells in peripheral blood smears, as a diagnostic aid tool. Methodology: Image processing techniques were used in order to segment erythrocytes in the microscopic photographs and to measure characteristics as area, perimeter, solidity, circularity, eccentricity, texture and box-counting dimension. An artificial neural network was used to classify the red blood cells in the images in seven classes, including normal and six pathological changes, according to their characteristics. The network was trained according to the classification of 262 erythrocytes by an expert hematologist. The developments were made in matlab ®, a powerful scientific computing platform. Results: The chosen network reaches 97.3% correct in the validation data. Mistakes in the network correspond to cells with various pathological classifications features, which make them difficult to classify even for an expert. Conclusions: The developed application classifies quickly and accurately the different types of red blood cells in a microscopic sample of peripheral blood smear, so it could be useful as a diagnostic support tool.
Keywords
Erythrocyte, classification of red blood cells, neural network, image processing, red cells, Eritrocito, clasificación eritrocitaria, procesamiento de imágenes, red neuronal, glóbulos rojos