Posted in | News | Medical Robotics

Machine Learning Reads Microscopy Images for Antibiotic Resistance

Researchers from the University of Cambridge have demonstrated that drug-resistant diseases can be identified using AI, which would significantly shorten the time it takes to make an accurate diagnosis. The group showed how drug-resistant bacteria may be accurately determined from microscopy images alone using a trained algorithm. This research was published in Nature Communications.

Machine Learning Reads Microscopy Images for Antibiotic Resistance
Color-enhanced scanning electron micrograph showing Salmonella Typhimurium (red) invading cultured human cells. Image Credit: Rocky Mountain Laboratories, NIAID, NIH.

Antimicrobial resistance is a growing worldwide health concern that makes many infections harder to cure and reduces the number of accessible treatments. It even raises the possibility of some infections becoming untreatable in the near future.

Being able to quickly discriminate between organisms that are resistant to therapy and those that can be treated with first-line medications is one of the issues faced by healthcare professionals.

Traditional testing involves culture of bacteria, testing them against different antimicrobial treatments, and having a laboratory technician or machine analyze the results. This process can take several days. As a result of this delay, patients are frequently treated with the wrong medication, which might have more catastrophic consequences and perhaps increase drug resistance.

Researchers from Professor Stephen Baker's lab at the University of Cambridge led a team that created a machine-learning tool that can recognize Salmonella Typhimurium bacteria resistant to the first-line antibiotic ciprofloxacin from microscopy images, even without testing the against the drug.

In extreme cases, S. Typhimurium can produce typhoid-like illness and gastrointestinal distress, with symptoms including fever, headache, nausea, tiredness, stomach discomfort, and constipation or diarrhea. It may even be fatal in extreme circumstances. Antibiotics can cure illnesses, but the bacteria are growing more resistant to them, making treatment more difficult.

The researchers examined S. Typhimurium isolates subjected to escalating ciprofloxacin dosages using high-resolution microscopy, and they determined the five most crucial imaging characteristics for differentiating between resistant and susceptible isolates.

Then, using imaging data from 16 samples, they built and evaluated a machine-learning algorithm to detect these traits.

Without exposing the germs to the medication, the algorithm was able to accurately predict whether the bacteria were susceptible to or resistant to ciprofloxacin in every instance. This was the case for isolates cultured for just six hours, as opposed to the standard 24-hour culture period in the presence of antibiotics.

S. Typhimurium bacteria that are resistant to ciprofloxacin have several notable differences to those still susceptible to the antibiotic. While an expert human operator might be able to identify some of these, on their own they would not be enough to confidently distinguish resistant and susceptible bacteria. The beauty of the machine learning model is that it can identify resistant bacteria based on a few subtle features on microscopy images that human eyes cannot detect.

Dr. Tuan-Anh Tran, University of Cambridge

While working on this research, Tran was a Ph.D. student at the University of Oxford.

To analyze a sample, such as blood, urine, or stool, using this method, the bacteria would still need to be isolated. However, since the bacteria do not need to be tested against ciprofloxacin, the entire procedure could be shortened from several days to a few hours.

The researchers claim that this specific strategy shows how strong artificial intelligence could be in the fight against antibiotic resistance, even though there are limits to how realistic and affordable it would be.

Given that this approach uses single-cell resolution imaging, it is not yet a solution that could be readily deployed everywhere. But it shows real promise that by capturing just a few parameters about the shape and structure of the bacteria, it can give us enough information to predict drug resistance with relative ease.

Dr. Sushmita Sridhar, Postdoc, University of New Mexico

Dr. Sridhar, who is also associated with the Harvard School of Public Health, initiated this project while a Ph.D. student in the Department of Medicine at the University of Cambridge.

The team is now working on larger collections of bacteria to build a more reliable experimental set and accelerate the identification process even further. This will enable them to detect antibiotic resistance in various bacterial species, including ciprofloxacin resistance.

What would be really important, particularly for a clinical context, would be to be able to take a complex sample for example blood or urine or sputum, and identify susceptibility and resistance directly from that. That is a much more complicated problem and one that really has not been solved at all, even in clinical diagnostics in a hospital. If we could find a way of doing this, we could reduce the time taken to identify drug resistance and at a much lower cost. That could be truly transformative.

Dr. Sushmita Sridhar, Postdoc, University of New Mexico

Wellcome funded the study.

Journal Reference:

Tran, T.-A., et al. (2024) Combining machine learning with high-content imaging to infer ciprofloxacin susceptibility in isolates of Salmonella Typhimurium. Nature Communications.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this news story?

Leave your feedback
Your comment type

While we only use edited and approved content for Azthena answers, it may on occasions provide incorrect responses. Please confirm any data provided with the related suppliers or authors. We do not provide medical advice, if you search for medical information you must always consult a medical professional before acting on any information provided.

Your questions, but not your email details will be shared with OpenAI and retained for 30 days in accordance with their privacy principles.

Please do not ask questions that use sensitive or confidential information.

Read the full Terms & Conditions.