Development of intelligent suite for malaria pathogen detection in microscopy images

Area of Science:

• Medical diagnostics
• Parasitology
• Computer vision

Background:

• Microscopy of blood smears is the gold standard for malaria diagnosis.
• Expert microscopists are scarce in malaria-endemic remote areas.
• Automating pathogen detection in microscope images is crucial for widespread diagnosis.

Purpose of the Study:

• To develop an easy-to-use and deployable software for automated malaria pathogen detection.
• To create a user-friendly interface for interacting with an optimized deep learning model.
• To facilitate malaria diagnosis in resource-limited settings.

Main Methods:

• Developed the Intelligent Suite software with a graphical user interface (GUI) using the 'cvui' library.
• Integrated OpenVINO's inference engine for model optimization and deployment.
• Utilized a custom YOLO-mp-3l model trained on the Darknet framework for pathogen detection in thick smear images.

Main Results:

• The Intelligent Suite enables user selection of inference devices and parameter alteration.
• The software generates detection reports with model performance metrics.
• Executed on a CPU with inference on a Neural Compute Stick (NCS2), demonstrating practical deployment.

Conclusions:

• The Intelligent Suite offers a deployable solution for automated malaria detection from microscope images.
• The software addresses the scarcity of expert microscopists in endemic regions.
• This approach can enhance the accessibility and efficiency of malaria diagnosis.