Modified CNN Model for Malaria Diagnosis

Main Article Content

Mrs. Kaveri N , Dr. Narendra.B. Mustare

Abstract

Malaria could be a syndrome triggered by a sort of tiny parasite transmitted to humans from contaminated feminine mosquito bites. It could be a devastating illness that's wide in multitudinous countries of the world. Patients will be benefitted greatly from an early and speedy identification of this disease, as traditional ways to diagnosis entail arduous labour. In recent years, several proposals for automatic strategy are made, but their accuracy is questionable. With their superior performance, deep learning algorithms have altered the world. CNNs are a type of image grouping algorithm which pulls characteristics from hidden layers of the image. The diagnosis of malaria-affected RBCs from segmented microscopic blood images exercising CNN will prop in the rapid-fire opinion, and this may be salutary for countries with a lack of clinical experts. Our work has been divided into two sections. We evaluated the efficacy of different current models for effective malaria diagnosis in the initial phase. Then, a modified CNN model is designed which surpasses well-known models. Prior to model training, it uses bilateral filtering along with augmentation approaches to emphasise RBC features. The proposed model is generalised as well as prevents over-fitting. Our proposed technique is 97.86 percent accurate in identifying malaria in a thin blood smear, according to the



 results, which are based on the reference dataset.

Article Details

Section
Articles