An Efficient Learning Assisted Malaria Parasites Identification Methodology based on Blood Smear Images
S. D. Lalitha, P. Remya, R. Priyanka, Niveathika Rajendran, N. Sivakamasundari · 2023
Malaria remains a global health challenge, necessitating accurate and efficient methods for the identification of malaria parasites from blood smear images. In this study, we propose an innovative approach for malaria parasites identification that leverages the power of deep learning and gradient boosting. Our methodology combines the feature extraction capabilities of the MobileNet convolutional neural network with the robust classification capabilities of XGBoost. The MobileNet model is employed to extract discriminative features from blood smear images, providing a rich representation of the parasites’ morphological characteristics. Subsequently, XGBoost is utilized as the classification model to distinguish between infected and uninfected blood samples based on these extracted features. This hybrid approach showcases remarkable accuracy and efficiency, offering a reliable solution for malaria diagnosis. Experimental results demonstrate the superiority of our methodology, achieving precise identification of malaria parasites while minimizing false positives. The proposed model takes the lead with an impressive accuracy score of 0.98, signifying its exceptional capability in distinguishing between different classes within the dataset. This model, tailored for the specific task at hand, utilizes a combination of innovative techniques to achieve superior accuracy.This innovative learning-assisted approach holds great promise for enhancing malaria diagnosis, particularly in resource-constrained settings, and represents a significant step towards combatting this life-threatening disease more effectively.