Convolutional Neural Networks for Automated Detection and Classification of Plasmodium Species in Thin Blood Smear Images

There has been a continued transmission of malaria throughout the world due to protozoan parasites from the Plasmodium species. As for treatment and control, it is very important to make correct and more efficient diagnostic. In order to observe the efficiency of the proposed approach, This Research built a Convolutional Neural Network (CNN) model for Automated detection and classification on thin blood smear images of Plasmodium species. This model was built on a corpus of 27558 images, included five Plasmodium species. Our CNN model got an overall accuracy of 96% for the cheating detection with an $F 1$ score of 0.94. In the detection of the presence of malaria parasites the test accuracy conducted was as follows: 8%. Species-specific classification accuracies were: P. falciparum (95.7%), P. vivax (94.9%), P. ovale (93.2%), P. malaria (92.8%) and P. Knowles (91, 5%). As for the model SL was found to have sensitivity of 97.3% And the specificity in this case is $\mathbf{9 6. 1 \%}$. The proposed CNN-based approach provides a sound and fully automated solution for malarial parasite detection and species determination, which could lead to better diagnostic performances in day-to-day practices.