Improving Malaria detection using enhanced-efficientnet deep neural network approach
Abstract
Malaria detection traditionally relies on microscopic examination of blood smears, a process that is labor-intensive and prone to human error. This study aims to introduce a robust automated detection method using deep learning, designed to enhance diagnostic accuracy and reduce human effort. The research presents an innovative Enhanced-EfficientNet (EEN) deep neural network approach comprising three distinct phases: image preprocessing, feature extraction using the Enhanced-EfficientNet model, and classification using a Deep Neural Network (DNN). The proposed methodology was validated using a dataset containing 27,558 labeled blood cell images equally divided between "infected" and "uninfected" samples. The proposed EEN approach achieved superior diagnostic performance, with a maximum classification accuracy of 97.71%, precision of 97.71%, recall of 97.72%, and an F1 score of 97.71% on the test dataset. Comparative evaluation with established models, including VGG16, Xception, ResNet152, EfficientNetB3, and InceptionV3, confirmed significant performance improvements offered by the proposed method. The Enhanced-EfficientNet model effectively addresses the accuracy and reliability challenges associated with traditional malaria diagnostics, presenting a robust deep learning alternative with improved diagnostic outcomes. The study underscores deep learning's practical value as a supportive diagnostic tool, facilitating quicker, more reliable detection of malaria infections. Clinicians can leverage this technology to enhance patient care, significantly reduce diagnostic errors, and improve survival outcomes for patients.
Authors

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.