Malaria Detection Using Deep Learning Techniques

Malaria, a deadly illness caused by Plasmodium parasites is usually diagnosed by a microscopic blood test. However, this manual process consumes an extended period and is liable to mistakes made by people. To address these challenges, models are trained on a comprehensive dataset containing either malaria-infected or uninfected blood samples. This study examines the effectiveness of various models of deep learning, particularly Convolutional Neural Networks (CNN), ResNet50, VGGNet19, and Inception V3, for the detection of malaria from blood smear images. The dataset is carefully curated to ensure a balanced representation of both classes, improving the ability of the Models can generalise successfully to new, previously unknown data. The ResNet50 is recognised for its deep residual learning framework, which has an accuracy of 94%, which has the highest accuracy. VGGNet19, known for its simplicity and depth, and Inception V3, known for its efficient multi-scale feature extraction, were all compared to a standard CNN. The results showed that these advanced architectures significantly outperformed traditional methods, providing a robust and reliable approach to malaria detection. This study mainly focuses on deep learning models to revolutionize the diagnostic process, making it faster, more accurate, and easier to use, especially in resource-constrained settings.