The December 2022 issue of IEEE Spectrum is here!

Close bar

AI-Powered Microscope Counts Malaria Parasites in Blood Samples

Silicon Valley teams up with a Chinese microscope manufacturer to deploy deep learning to diagnose malaria

4 min read
A woman inserts a malaria slide into an AI powered microscope.
Roxanne Rees-Channer, a research biochemist, inserts a cassette into the EasyScan GO at the Hospital for Tropical Diseases in London, where the AI-powered microscope is being tested.
Photo: Andrew H. Kim/Intellectual Ventures

Today, a Chinese manufacturer and a venture backed by Bill Gates will announce plans to commercialize a microscope that uses deep learning algorithms to automatically identify and count malaria parasites in a blood smear within 20 minutes. AI-powered microscopes could speed up diagnosis and standardize detection of malaria at a time when the mosquito-borne disease kills almost half a million people per year.

An experimental version of the AI-powered microscope has already shown that it can detect malaria parasites well enough to meet the highest World Health Organization microscopy standard, known as competence level 1. That rating means that it performs on par with well-trained microscopists, although the researchers note that some expert microscopists can still outperform the automated system.

That previous research, presented at the International Conference on Computer Vision [PDF] in October, has inspired the Global Good Fund—a partnership between the company Intellectual Ventures and Bill Gates—and a Chinese microscope manufacturer called Motic to take the next big commercialization step.

Such microscopes could prove especially helpful in tracking the treatment of multidrug-resistant strains of malaria spreading in Southeast Asia. “This multidrug resistance monitoring relies on very reliable microscopy to see how quickly the malaria drugs have reduced the amount of parasites in the blood,” says David Bell, director of global health technology at the Global Good Fund. “We saw that machine learning could bring more accuracy and standardization in this area and allow countries to implement monitoring more effectively.”

The EasyScan GO microscope under development would combine bright-field microscope technology with a laptop computer running deep learning software that can automatically identify parasites that cause malaria. Human lab workers would mostly focus on preparing the slides of blood samples to view under the microscope and verifying the results. The Global Good Fund announcement was made during the MEDICA 2017 trade fair being held in Düsseldorf, Germany from 13 to 16 November 2017.

Malaria parasites present a tricky “rare object problem” for deep learning algorithms that typically require huge amounts of training data to accurately identify objects, says Ben Wilson, a principal investigator at Intellectual Ventures in Bellevue, Wash. The tiny malaria parasites may only show up a handful of times within hundreds of microscope images of blood smears.

In cases of very low infection levels, just a single malaria parasite might appear among 100,000 red blood cells. Sebastian Nunnendorf, general manager of Motic in China, compared the challenge to “finding marbles in a standard football pitch.”

Global Good researchers work on the EasyScan GO AI-powered microscope. To train the algorithm, the team collected and annotated thousands of malaria slides from all over the world.Global Good researchers work on the EasyScan GO AI-powered microscope. To train its algorithm to detect malaria, the team collected and annotated thousands of slides of blood smears from all over the world.Photo: Andrew H. Kim/Intellectual Ventures

The solution required a combination of both deep learning and traditional computer algorithms used for segmenting things of interest within images. It also required a lot of training data based on prepared microscope slides. Wilson and his colleagues even asked a few labs to intentionally prepare slides poorly so that they could train the deep learning AI to do its job under less-than-ideal circumstances. “As far as machine learning for microscopy, this is pretty unique in terms of malaria and infectious diseases in general,” Wilson says.

The speed at which the prototype microscopes scan each slide is about on par with expert human microscopists, at 20 minutes per slide. But Wilson anticipates being able to eventually cut that scanning time in half to just 10 minutes per slide. More importantly, even existing versions of the microscope can supplement the very limited number of trained microscopists available to identify malaria and track multidrug-resistant malaria. “Essentially it is a massive efficiency gain, not a robotic replacement for lab technicians,” says Nunnendorf at Motic.

If EasyScan GO microscopes become widespread, they could also benefit researchers who monitor the spread of infectious diseases and the effectiveness of drug treatments. That’s because the automated microscopes could provide standardized detection results that allow for direct comparisons between different regions and across many years.

2 open views of the EasyScan GO AI-powered microscope.Photo: Andrew H. Kim/Intellectual Ventures

Early testing with the deep learning algorithms relied upon custom-built microscope hardware. The Global Good researchers are currently working with Motic to ensure that the commercial versions of the EasyScan GO microscopes—based in part on a pre-existing “Motic EasyScan” product line—can still meet the highest standard for expert microscopists set by the World Health Organization.

The final price tag and timeline for market rollout has not yet been determined. But Global Good and Motic anticipate a fairly inexpensive microscope closer in price to a base model rather than a high-end digital slide scanner. The cost will be key for many countries struggling to combat the spread of malaria and other infectious diseases. “In order to make a real impact on this problem, the pricing is specially adapted for low income and middle-income countries,” Nunnendorf says.

Motic also plans to update the EasyScan GO system software down the line so that it can diagnose other diseases such as dengue fever, Chagas disease, microfilaria, and sickle cell anemia. That would build on the early successes of using deep learning AI to accurately recognize malaria parasites.

“Success with the most difficult-to-identify disease, malaria, paves the way for it to excel at almost any bright-field microscopy task, including other parasites and traits commonly found on a blood film, as well as other sample types, such as sputum and feces,” Nunnendorf says.

Editor’s Note: The original version of this story incorrectly stated that Global Good is funded by the Bill & Melinda Gates Foundation. Global Good is directly funded by Bill Gates as part of a shared vision with Nathan Myhrvold, CEO and founder of Intellectual Ventures.

The Conversation (0)

Will AI Steal Submarines’ Stealth?

Better detection will make the oceans transparent—and perhaps doom mutually assured destruction

11 min read
A photo of a submarine in the water under a partly cloudy sky.

The Virginia-class fast attack submarine USS Virginia cruises through the Mediterranean in 2010. Back then, it could effectively disappear just by diving.

U.S. Navy

Submarines are valued primarily for their ability to hide. The assurance that submarines would likely survive the first missile strike in a nuclear war and thus be able to respond by launching missiles in a second strike is key to the strategy of deterrence known as mutually assured destruction. Any new technology that might render the oceans effectively transparent, making it trivial to spot lurking submarines, could thus undermine the peace of the world. For nearly a century, naval engineers have striven to develop ever-faster, ever-quieter submarines. But they have worked just as hard at advancing a wide array of radar, sonar, and other technologies designed to detect, target, and eliminate enemy submarines.

The balance seemed to turn with the emergence of nuclear-powered submarines in the early 1960s. In a 2015 study for the Center for Strategic and Budgetary Assessment, Bryan Clark, a naval specialist now at the Hudson Institute, noted that the ability of these boats to remain submerged for long periods of time made them “nearly impossible to find with radar and active sonar.” But even these stealthy submarines produce subtle, very-low-frequency noises that can be picked up from far away by networks of acoustic hydrophone arrays mounted to the seafloor.

Keep Reading ↓Show less