Plasmodium Parasite Detection in Thin Blood Smear: Combining Yolo and Deep Learning Technique

Malaria is a perilous malady arising from the bite of a contaminated mosquito. This ailment is a high-risk disease and a major public health concern. Prompt identification of Plasmodium is crucial to support effective medical intervention and avoiding critical health issues. Microscopy is the preferred method for malaria diagnosis due to its accuracy and reliability. The result of microscopic analysis takes time and depends on factors such as the equipment quality, the luminosity, the expertise, and the skills of trained specialists. In addition to conventional diagnostic methods, investigators have explored analyzing cell images using deep learning models to identify Plasmodium Pathogens. This study offers a two-stage malaria detection framework to detect the Plasmodium parasite. The proposed approach uses a combination of object detection models and deep learning such as YOLO and CNN models to enhance malaria detection. First, YOLO is trained to distinguish diseased and healthy cells. The detected cell images are then used as input for a CNN model, which performs the final classification. Two open-access datasets are used for model training and evaluation. Multiple performance metrics were considered to assess effectiveness of the devised approach. The developed framework surpasses earlier methods with an accuracy of 96.30 %. From the results available, the proposed model is distinctly better rendered in response to prior works on malaria diagnosis.