Plasmodium Autocount

Plasmodium Autocount automates quantification of parasitaemia from Giemsa-stained blood smear images to measure the percentage of Plasmodium-infected erythrocytes.

Key Features:

• Automated Image Analysis: Uses a camera attached to a microscope to capture Giemsa-stained blood smear images and applies foreground detection and the circular Hough transform to identify red blood cells (RBCs).
• Infection Detection: Detects stained spots within RBCs as indicators of Plasmodium infection and incorporates a calibration process using representative images to adjust parameters.
• Spurious Hit Filtering: Filters misshapen cells (debris) and overly stained cells (likely white blood cells) to reduce false positives.
• Validation and Correlation: Validated for estimating parasitaemia in mouse models infected with Plasmodium yoelii and shows high correlation with manual counting, with discrepancies comparable to inter-observer variability.
• Complementary Manual Aid: Includes a supplementary Counting Aid programme developed in Visual Basic to assist manual counting when automated analysis is compromised.

Scientific Applications:

• Rapid and Accurate Parasitaemia Measurement: Enables rapid determination of parasitaemia for monitoring infection progression in experimental settings.
• Reduction of Human Error: Reduces person-to-person variability in microscopic enumeration by automating counting.
• Enhanced Research Efficiency: Decreases time and labour required for parasitaemia determination in experimental workflows.

Methodology:

Developed in Python; implements foreground detection and the circular Hough transform for RBC detection, detects stained intraerythrocytic spots for infection calling, and uses a calibration procedure with representative images; a supplementary Counting Aid programme is implemented in Visual Basic.