Detection Of Malarial Infected Cells Using Convolutional Neural Networks

Overview :

Malaria is a fatal illness which is solely transmitted through the bites of infected female Anopheles mosquitoes. According to the recent studies, which shows that in the year 2020 there were 241 million cases of malaria worldwide which results in the death of nearly 6,27,000 people. It is crucial that the diagnostic process be automated in order to avoid human participation during the automated diagnosis because the majority of these deaths are caused by a delayed or inaccurate diagnosis.

In order to enhance the diagnostic reliability Convolutional neural networks (CNNs) and other deep-learning technologies, such as image processing, are employed to assess parasitemia in microscopic blood slides.

We highlight some of our recent significant innovations on highly accurate classification of malaria-infected cells using deep supervised learning in deep convolutional neural networks. The first task is to outline the methodologies for image processing that can be applied to the dataset which is going to be utilized to train the model.

Then we will discuss the procedures of for training of deep neural network, as well as data augmentation methods used to significantly increase the size of the dataset and to improve the performance of our developed model.

Lastly, using the same datasets for both training and testing, we will compare the classification accuracy outcomes from deep convolutional neural networks. With the provided blood smear samples, this trained model will be utilized to forecast the presence of malaria-infected cells.

This deep learning model gives an accuracy of 97%

Dataset:

The dataset used is Malaria-dataset which is downloaded from the National Library of medicine official website: https:ceb.nlm.nih.gov/repositories/malaria-datasets/

⚠ Note:

This project is deployed in AWS EC2, But the service may be temporarily suspended due to cost considerations