Parasitology Unveiled: Revolutionizing Microorganism Classification Through Deep Learning

Microorganisms play a crucial role in multiple fields, such as medical diagnostics, food safety, and ecological research, accounting for a significant 60% of Earth's biomass. This paper explores the complex connection between microorganisms and ecosystems, with a focus on accurate parasite classification methods. Conventional approaches, although important, have their limitations, leading to a growing dependence on CNN s in image analysis to enhance accuracy. Utilizing advanced DL techniques, particularly pretrained CNNs, has completely transformed the field of microorganism classification, providing unparalleled accuracy and efficiency. This study delves deeper into the intersection of swarm intelligence and DL algorithms, with a specific focus on the complex task of parasite identification. The effort is backed by a comprehensive dataset consisting of 34,298 images from ten different parasite species. The study employs meticulous preprocessing techniques, including uniform scaling and normalization, to maintain the integrity and precision of the dataset. These techniques serve as the foundation for subsequent modeling efforts. In addition, a comprehensive examination of four pretrained CNN architectures offers valuable insights that are crucial for selecting a model. Dynamic visualizations provide a thorough representation of model convergence dynamics, helping to enhance understanding of classification performance. The XceptionNet model stands out as the highest performer, achieving an impressive accuracy and F1 score of 0.9489 and 0.9492, respectively. This research demonstrates notable progress in the intersection of AI and biology, leading to new possibilities in disease diagnosis and improvements in public health practices by refining microorganism classification methods.