Supporting Malaria Diagnosis Using Deep Learning and Data Augmentation.

Area of Science:

• Medical diagnostics
• Artificial intelligence in healthcare
• Parasitology

Background:

• Malaria diagnosis relies on time-consuming microscopic blood smear analysis.
• Delayed diagnosis impacts malaria prevention, treatment, and patient outcomes.
• Accurate and rapid parasite quantification is crucial for effective disease management.

Purpose of the Study:

• To develop a deep learning model for automated malaria parasite and leukocyte detection.
• To enable rapid parasite/μL blood count for efficient malaria diagnosis.
• To reduce diagnostic time compared to traditional microscopic methods.

Main Methods:

• Utilized the YOLOv8 algorithm for training on augmented microscopic blood smear images.
• Implemented data augmentation to enhance the dataset's diversity and size.
• Employed a counting formula for parasite quantification based on model detection.

Main Results:

• Achieved 95% accuracy in detecting malaria parasites.
• Achieved 98% accuracy in detecting leukocytes.
• Demonstrated significantly faster parasitemia reporting times compared to human experts.

Conclusions:

• Deep learning offers a highly accurate and efficient method for malaria diagnosis.
• Automated parasite counting can significantly expedite malaria detection and management.
• This AI-driven approach shows promise for improving healthcare access in resource-limited settings.