Classify Malaria Dataset Human Blood Images Using Convolutional Neural Networks
Purnawarman Musa, Eri Prasetyo Wibowo, Matrissya Hermita, Raihan Firas Muzhaffar · 2022 4th International Conference on Cybernetics and Intelligent System (ICORIS) · 2022
Malaria is prevalent in regions around the globe, and human activity has contributed to more than 400,000 fatalities annually. A timely and accurate diagnosis is necessary for optimal disease treatment, given that malaria is a substantial problem on a global scale. Our research presents classification studies of malaria as a solution to the previously described problem. These studies use image datasets obtained from photo data online and attempt to detect malaria infected or non-infected by someone using a blood image via Convolutional Neural Network strategies and deep learning methods. The results of tests done with Convolutional Neural Network techniques show that the processes were successful, with the best model getting more than 95%. In the confusion matrix, their accuracy is 97.28%, their precision is 99.60%, their recall is 95.15%, their specificity is 99.62%, and their F1 Score is 97.32%. In addition, the prediction accuracy for identifying malaria was 100% when utilizing photographs or various datasets from the laboratory as test data.