Distributed medical image analysis and diagnosis through crowd-sourced games: a malaria case study

Area of Science:

• Biomedical Diagnostics
• Human-Computer Interaction
• Computational Biology

Background:

• Microscopic analysis is crucial for disease diagnosis.
• Expert interpretation requires extensive training.
• Current diagnostic methods can be time-consuming and resource-intensive.

Purpose of the Study:

• To explore untrained human visual recognition for microscopic analysis.
• To develop crowd-sourced gaming platforms for medical diagnostics.
• To assess the accuracy of non-expert diagnoses compared to medical professionals.

Main Methods:

• Designed engaging digital games interfaced with AI.
• Utilized crowd-sourcing for data collection and analysis.
• Compared gamer diagnoses of malaria-infected red blood cells to expert decisions.

Main Results:

• Non-expert gamers achieved diagnostic accuracy comparable to medical experts.
• The accuracy for malaria diagnosis was within 1.25% of professional diagnoses.
• Demonstrated the potential of gamified crowd-sourcing for binary medical decisions.

Conclusions:

• Innate human visual capabilities can be harnessed for reliable microscopic diagnostics.
• Gamified crowd-sourcing offers a scalable and accessible diagnostic tool.
• This approach can augment traditional diagnostic workflows, especially in resource-limited settings.