Proceedings of the 6th Computer Science Research Days, JRI 2023, 18-20 December 2023, Ouagadougou, Burkina Faso

Research Article

Convolutional Neural Networks Deep Learning Based for Malaria Detection and Diagnosis

Download10 downloads
  • @INPROCEEDINGS{10.4108/eai.18-12-2023.2348130,
        author={Kiswendsida Kisito  Kabore and Josue  Ouedraogo and Ferdinand T.  Guinko},
        title={Convolutional Neural Networks Deep Learning Based for Malaria Detection and Diagnosis},
        proceedings={Proceedings of the 6th Computer Science Research Days, JRI 2023, 18-20 December 2023, Ouagadougou, Burkina Faso},
        publisher={EAI},
        proceedings_a={JRI},
        year={2024},
        month={6},
        keywords={malaria deep learning object detection yolov5 plasmodium falciparum},
        doi={10.4108/eai.18-12-2023.2348130}
    }
    
  • Kiswendsida Kisito Kabore
    Josue Ouedraogo
    Ferdinand T. Guinko
    Year: 2024
    Convolutional Neural Networks Deep Learning Based for Malaria Detection and Diagnosis
    JRI
    EAI
    DOI: 10.4108/eai.18-12-2023.2348130
Kiswendsida Kisito Kabore1,*, Josue Ouedraogo1, Ferdinand T. Guinko1
  • 1: Université Joseph KI-ZERBO
*Contact email: kisito@ujkz.bf

Abstract

Malaria is a disease that occurs worldwide, especially in tropical regions where a high prevalence is observed. Difficulties are encountered especially in developing countries where resources in terms of equipment and trained personnel are limited. Until today, microscopic analysis is the standard method for diagnosing Plasmodium falciparum, which is the causative agent of malaria. In this paper, we proposed a malaria detection and diagnosis system using a deep learning technique which is a convolutional neural network called YOLOv5. Model learning was performed using a combination of two given image sources, Delgado Dataset B and Dijkstra Dataset, as a dataset containing thin smear images. We then evaluated the performance of the model by comparing it with other state-of-the-art results on deep learning. We obtained for the detection, a mean Average Precision of 96.71%.