Mask R-CNN for Automated Multi-Species Malaria Parasite Detection

Eugenia Mawuenya Akpo, Carine Mukamakuza

Keywords: Mask R-CNN, malaria parasite detection, Plasmodium species, deep learning, instance segmentation, microscopic image analysis

TL;DR: Mask R-CNN applied to detect and segment multiple malaria parasite species in microscopic images, achieving high accuracy and showing promise for improved clinical malaria diagnosis.

Abstract: This study investigates the automatic detection and segmentation of malaria parasites across various Plasmodium species using Mask R-CNN, an advanced deep-learning architecture. Expanding on earlier studies in digital malaria diagnosis, we apply pixel-level segmentation to overcome the drawbacks of previous approaches. 971 microscopic pictures of four Plasmodium species—P. falciparum, P. malariae, P. ovale, and P. vivax—taken from Rwanda's healthcare facilities make up our dataset. This dataset was used to train the Mask R-CNN model, which produced excellent mean average precision (mAP) scores for all species, with P. vivax and P. malariae showing the most excellent performance with mAP 0.9575 and mAP 0.9459, respectively. Compared to earlier techniques, this method shows notable advances in parasite localization and delineation, suggesting the possibility of more precise and effective malaria diagnosis in clinical settings.

Primary Area: applications to computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 11618