In 2018, over 6,000 volunteers examined pictures of tuberculosis bacteria exposed to drugs; their answers helped researchers test resistance to 13 medicines

Medical discoveries are usually seen as something that happens in high-security laboratories, where trained scientists wearing white lab coats make all critical decisions. Instead of diagnosing patients or working with harmful bacteria, they analyzed pictures of laboratory experiments and answered a simple question: Had the bacteria grown, or had the medicine stopped them? Their combined responses helped researchers to analyze one of the most serious infectious diseases in the world. A 2022 study published in eLife compared volunteer classifications with expert assessments to examine whether crowdsourcing could support tuberculosis drug-resistance research.

Turning a complex laboratory task into a simple online challenge

BashTheBug, hosted on the Zooniverse citizen-science platform, allowed thousands of volunteers to examine laboratory images of Mycobacterium tuberculosis samples. The volunteers studied images of Mycobacterium tuberculosis bacterial samples in 96-well test plates that were exposed to different antibiotics; the volunteers’ task was to study the images of the bacteria’s response to the medicine rather than to determine whether the subject suffered from tuberculosis.

The usual way to test whether Mycobacterium tuberculosis strains are antibiotic-resistant entails a procedure carried out by trained laboratory specialists. Scientists put bacteria into small wells containing different concentrations of the medication and observe whether the bacteria are still reproducing despite the medicine. If the bacteria grow, the particular strain is resistant to antibiotics. If the bacteria stop growing, it means that the medicine is effective.

Their role was limited to identifying visible patterns of bacterial growth in laboratory images. Rather, each volunteer examined an image of one laboratory plate and determined at what point bacterial growth seemed to end. Multiple volunteer classifications allowed researchers to compare crowd assessments with expert evaluations. This approach reduced the influence of individual mistakes while making the process suitable for people without scientific training.

Why drug resistance matters

Tuberculosis is still one of the most lethal infectious diseases of humanity, and the presence of drug-resistant strains makes treatment far more complicated. Some tuberculosis strains become resistant to medicines that have been used in treatment for decades. Consequently, doctors are forced to use longer and more expensive treatment options, which often produce more severe side effects.

In order to determine the best method of treatment, laboratories require accurate information on which drugs will be able to stop the growth of bacteria. The photographs examined by BashTheBug volunteers were part of that process. By assisting scientists in the interpretation of laboratory results, citizen scientists contributed to work aimed at improving drug resistance measurement rather than diagnosing individual patients.

What the volunteer classifications revealed

The 2022 eLife study compared volunteer classifications with expert assessments and found that aggregated crowd responses could provide useful estimates of bacterial growth patterns. The researchers found that combining multiple volunteer assessments improved agreement with expert classifications. It was concluded that ordinary people could be of great help in reading laboratory images if enough people were to analyze every picture.

The study also suggested that volunteer classifications could help researchers process large numbers of laboratory images by highlighting patterns that require further analysis. Since every picture was subjected to several independent assessments, the resulting responses provided an excellent mechanism for checking the consistency of the data and identifying plates that might need additional expert review. This could help laboratories improve quality control while reducing the burden on specialists who already face large workloads.

How citizen science can support biomedical research

Most individuals will think of citizen science as the counting of birds, the classification of galaxies, and even the identification of creatures in photographs taken by camera traps. BashTheBug demonstrated how citizen science platforms can involve volunteers in analyzing biomedical research data without requiring them to work directly with laboratory samples. Volunteers participated remotely by analyzing digital images provided through the online platform.

The researchers suggested that similar approaches could support future projects involving large volumes of scientific image analysis. With the continued development of digital imaging leading to an increase in the amount of laboratory data generated, well-structured citizen science projects might serve as an alternative avenue for the processing of information. The project showed how digital platforms can contribute to scientific workflows by helping researchers analyze large datasets.