Detection and identification of <i>Ascaris lumbricoides</i> and <i>Necator americanus</i> eggs in microscopic images of faecal samples of pigs

Abstract

Parasite egg detection and identification is an important task for diagnosis of many diseases. Manual identification of various types of parasite eggs is time consuming and prone to human error. Automation of this process can help in reducing time, human effort and error rate significantly. In this paper, we proposed a system for detection and identification of parasite eggs from microscopic images of faecal samples of pigs using image processing and machine learning techniques. We considered images of two types of parasite eggs: Ascaris lumbricoides and Necator americanus that are commonly found in pigs. The system segments different objects from the microscopic images and identify each of them as either a parasite egg of specified type or non-egg object. Identification of the segmented objects is done by convolutional neural network (CNN) models that automatically extract relevant features and classify them into three classes: Ascaris egg, Necator eggs and non-egg object.