Enhancing Classification of Parasite Microscopy Images Through Image Edge-Accentuating Preprocessing

In medical diagnostics, accurately classifying parasite species from microscopic images is challenging, especially in resource-limited areas. Our study presents a novel deep learning-based methodology that significantly enhances parasite classification accuracy in microscopic images by employing an image preprocessing technique where pixel values greater than a certain threshold are squared to enhance edge features. Using the Microscopic Images of Parasites Species dataset for testing, our approach shows exceptional performance across various parasites, overcoming obstacles like fecal impurities and blood smear variations. Our proposed method introduces “Accentuation Edge via Pixel Value Transformation” as a key innovation in the realm of parasite microscopic image classification. This edge accentuation aids deep learning models in achieving more accurate differentiation between parasitic and non-parasitic elements. Unlike traditional methods, our approach addresses previous limitations in sensitivity and specificity, leading to a notable improvement in classification performance. Our method demonstrated a groundbreaking 99.86% accuracy in parasite classification, marking a substantial advancement over existing microscopy and computational techniques. This method not only offers a scalable and effective solution for various clinical scenarios but also sets a new standard in the field of medical imaging and diagnosis of parasitic infections.