An artificial intelligence (AI)-powered microscope created by the US Department of Defense and Google is said to be able to detect cancer cells automatically.
The AI microscope is called the Augmented Reality Microscope, or ARM.
The ARM outlines cancer cells using a bright green line and its artificial intelligence systems estimate a cancer’s severity and generate a monochrome heat map on the monitor to show its boundaries.
READ MORE: Cervical cancer coverage: A top priority of PhilHealth
Imagine the implications of having a microscope that detects microscopic organisms and objects automatically. Previously, you’d have to take minuscule samples and drag your microscope slide carefully to find them.
Once the ARM enters consumer markets, it could reduce hospital workloads and improve healthcare worldwide.
What do we know about the AI microscope?
The ARM is still undergoing testing and development, so we have little information about it.
Fortunately, CNBC demonstrated the ARM at a Mitre facility in August.
The AI microscope seems like a regular microscope found in a high school biology class. The machine is beige with a large eyepiece and a tray for conventional glass slides.
However, the ARM connects to a large computer tower containing AI models. The artificial intelligence system outlines where a cancer is located once you fix a glass slide under the device.
READ MORE: Cervical cancer: What to know
Pathologists can view that outline as a bright green line through their eyepiece on a separate monitor. Moreover, the AI estimates the cancer’s severity and generates a black-and-white heat map.
The latter shows the borders of a cancer on a monitor. A senior autonomous systems engineer at Mitre, Patrick Minot, said the ARM doesn’t interfere with a pathologist’s established workflow.
The easy utility is an intentional design choice meant to deal with recent pathologist problems. Pathologists have been struggling with workforce shortages, like other healthcare sectors.
Meanwhile, their workloads have also been increasing as more people grow older. Both problems could require them to juggle several tasks, reducing their focus.
As a result, they’re more likely to make mistakes and pose serious risks to patients.
Challenges of the AI microscope
CNBC says numerous organizations have been trying to digitize pathologist workflows to boost efficiency.
However, digitization has challenges, too. For example, digitizing a single slide may require over a gigabyte (GB) of storage space. Consequently, large-scale data collection costs can rack up quickly.
Minot said he and his team did not design the AI microscope to replace digital pathology systems. Instead, it can help health organizations forego the need for them.
What are other AI healthcare solutions?
More experts use artificial intelligence to create new medical devices and treatments. For example, The British Journal of Opthalmology developed an AI health test that performs cardiovascular screenings by checking a patient’s retina.
The scientific organization created an AI algorithm called QUARTZ (QUalititative Analysis of Retinal vessels Topology and siZe) to check heart conditions via eye scans. It relies on a database of retinal images from 88,052 UK Biobank participants and 7,411 people from a European Prospective Investigation into Cancer (EPIC)-Norfolk study.
The scanner compares a patient’s eye scans to these images to diagnose potential issues. For example, if your eyes are similar to a stroke patient’s, it would determine you have a high risk for it.
Meanwhile, the LabGenius company created an AI model that expedites synthetic antibody creation. In the 1980s, the latter was a slow process requiring protein designers to sift through millions of potential amino acid combinations.
Then, they must test them all and adjust the variables until they produce an effective result. Conversely, the LabGenius AI model will choose more than 700 initial options from 100,000 potential antibodies automatically.
It will also design, build, and test these candidates while humans merely oversee the process by moving samples. “The only input you give the system as a human is, here’s an example of a healthy cell, here’s an example of a diseased cell,” the CEO stated. “And then you let the system explore the different [antibody] designs that can differentiate between them.”
The ARM could become a staple for hospitals worldwide.
At the time of this writing, it can only identify prostate cancer, breast cancer, cervical cancer, and mitosis.