Malaria Training Game

Game Instructions Play Now!

On April 25th, 2014, World Malaria Day, we are releasing our BioGames image library as a training and educational module, where we give each online user ~500 images per game to assess their training on identifying malaria infected cells. At the end of each game, which can be played using the link above, the gamer is given a quantified score that is based on their false positive and false negative rates. Each deck of images per game is created randomly from a sub-set of our malaria image library, characterized through BioGames, where each selected cell image also undergoes random rotations giving the gamers many opportunities to play the training module with a new set of images every time that they start a new game. At the end of each game, not only a BioGames score is assigned to the user, but also training feedback in the form of a list of images is provided, including each one of their false positives and false negatives as well as the “questionable” cells that they miscategorized. Note that some of these questionable cells might contribute a source of images that yield false diagnoses.

We hope that this training module will get widely used and expand its database as the BioGames platform continues to create gold labels for new microscopic images of thin or thick blood smears, which you can share with us. With large databases connected to user-friendly games and web-interfaces, this platform could be used for better training and education of medical personnel toward accurate reading of microscopic slides as well as for training of machine learning algorithms to automate digital diagnosis. This approach can also be scaled to other diseases besides malaria. Considering the relatively poor training of health-care workers in developing countries, this approach could be especially valuable for improving the accuracy of malaria diagnosis and measurement of parasitemia (which is typically much less than 1%) in infected patients that are on treatment.

BioGames: Crowd-Sourced Games for Telepathology

The Idea

Analysis and related diagnosis of medical images, regardless of the source and imaging modality, are tasks that require a great deal of expertise, demanding significant training of medical practitioners prior to being able to accurately interpret and diagnose such images. This is particularly true in analysis of microscopic data, creating challenges in resource-limited settings and developing countries, where properly trained health-care professionals are difficult to find.


We have shown that by utilizing the innate visual recognition and learning capabilities of human crowds it is possible to conduct reliable microscopic analysis of biomedical samples and make diagnostics decisions based on crowd-sourcing of microscopic data through intelligently designed and entertaining games that are interfaced with artificial learning and processing back-ends. We demonstrated that in the case of binary diagnostics decisions (e.g., infected vs. uninfected), using crowd-sourced games it is possible to approach the accuracy of medical experts in making such diagnoses.

Specifically, using non-professional gamers we report diagnosis of malaria infected red-blood-cells with an accuracy that is within 1.25% of the diagnostic decisions made by a trained professional.

Number of Diagnoses per Player: Red: 0-100 || Blue: 101-1000 || Green: >1000

For more details on our implementation and how we treated human gamers as components of smart telecommunication network, please click here.

Broader Vision

Together with crowd sourcing and social networks, today’s internet can be considered as a highly sophisticated “living and learning” system, where we have more than 100 trillion email messages every year, together with >300 million unique websites, with highly dynamic content. Our vision is to use this extra-ordinary medium to create a smart biomedical micro-analysis and diagnosis platform, leading the way to micro-world’s internet. For decades optical microscopy has been the workhorse of various fields including engineering, physical sciences, medicine and biology. Over the last few years, however, there has been a massive effort to create cost-effective, compact and lightweight microscope designs such that even mobile phones could be converted into microscopic analysis tools. The figure below shows examples of such compact microscopes that we have designed and built. Similar to the development of PC, this is a very important direction since it could enable wide-spread use of optical microscopy globally, with several orders of magnitude increase in the number of microscope users over the next few years. We aim to create a self-learning integrated network of microscopes toward smart biomedical micro-analysis and diagnosis. Once successful, this smart network of microscopes would deliver a paradigm-shift for medical, environmental, and biological sciences, among others, through various innovative uses of this network and its constantly expanding database.

(a) A lensfree holographic microscope that weighs ~ 45 grams. (b) A lensless holographic microscope on a mobile phone. (c) A wide-field fluorescent microscope on a phone. (d) An imaging fluorescent flow-cytometer on a mobile phone. (e) A field-portable lensfree tomographic microscope (weighing <120 grams). (f) A handheld super-resolution computational microscope comparative to a 40x microscope. On the right are lensfree images of a thin smear for malaria infected blood samples, where the infected RBCs are marked with arrows.For more information, visit our website.

According to the WHO, each year ~0.5 billion people become infected with malaria, and around 3 million die due to malaria infection. This world-map shows the regions that face the greatest risks in orange color. We aim to develop a self-learning distributed image analysis and diagnostics framework that is based on crowd-sourced games.