Enteric Protozoan Parasite Detection System: Linking to Neurodegenerative Diseases via the Gut-Brain Axis

Parasitic diseases are more prevalent in tropical and subtropical climates due to favorable climatic conditions that enhance species propagation, survival, and genetic diversity. These conditions, coupled with declining water quality and complex host-parasite relationships, contribute to disease persistence. Climate change further exacerbates these issues, particularly amplifying waterborne protozoan parasites like Entamoeba histolytica Giardia and Cryptosporidium. These parasites, commonly found in facial-contaminated water surface, causes a spectrum of health issues from dysentery, colitis, liver abscesses to life-threatening conditions. The parasites exist as infective cyst/oocyst stages capable of long- term survival in the environment. Therefore, early identification of cyst/oocyst in water samples is crucial for disease prevention and management in both human and veterinary health. Traditionally, cyst detection has relied on conventional microscopy, which is labor-intensive and prone to errors. However, recent advancements in Deep Learning and Computer Vision offer promising solutions. This study utilizes four deep neural network models to detect these three protozoan parasites from microscopic cyst/oocyst images of E. histolytica, Giardia and Cryptosporidium. The clarity of microscopic images directly impacts the accuracy of neural network models, highlighting the interconnectedness of image quality and diagnostic precision in protozoan parasite identification systems. The models not only detect the presence of protozoan parasites but also provide deep visions into the type and probability of infection. The DenseNet169 model, among others tested, demonstrates superior performance with a testing accuracy of 98.81%.