Plasmodium Vivax Parasite Detection using Transfer Learning Techniques
S Shashikiran, Srinivas Babu N, Sunitha HD, Sandip Mondal, Syed Sadaqathulla, K S Pramod · 2024
Malaria, caused by Plasmodium parasites, is a blood disease transmitted through the bite of a female Anopheles mosquito. Instances of this occur nearly 240 million times in the US. The disease affects about 40% of people annually. There is a threat to one-third of the world's population. In general, macroscopic examinations are time-consuming. Examine both thin and thick blood smears to determine what causes a disease or condition and to identify risk factors in individuals. However, a smear's accuracy is dependent on both its quality and the situation's information. Cells with and without parasites are categorized and tallied. Manual assessment is the gold standard for diagnosis, yet it only yields 50% accuracy. It requires multiple steps to be finished. Pre-trained CNNs are being employed as a viable method for attribute extraction output, which might be used to ascertain this. Statistics back it up. Due to these modifications, automation in the search for treatment may greatly benefit from microscopy, which offers a simple, affordable, and accurate method of diagnosing malaria. Our model can identify the picture model and determine whether it is afflicted by malaria with the aid of CNN. The VGG-16 model offers 95 percent accuracy and 96 percent validation accuracy. With a 10% train loss and a 9% validation loss, VGG-19 has a 96% training accuracy and a 97% validation accuracy. Further Investigation of the Latest updated transfer learning techniques can provide better results.