Will This AI Robot Replace Your Annual Pap Smear?
Pap smears probably aren’t on your list of favorite activities, but new technology involving robotics might make the procedure a little easier and more widely available.
Using a technique involving artificial intelligence, Dr. Xiaolei Huang, associate professor of computer science and engineering at Lehigh University, has created an image-based cervical cancer-screening method that has the potential to replace Pap smears and HPV tests. Additionally, this relatively low-cost technique could be used in less developed countries, where 80 percent of cervical cancer deaths occur, according to Huang.
There are two systems for completing the screenings, Huang explained. The first takes a single image of the cervix once it has been prepped with an application of diluted vinegar. Then, the computer takes a picture, which doctors will then examine and diagnose.
This works by comparing the image taken of the patient’s cervix with a large number of pictures of cervixes which, when read, are designed to detect the possibly precancerous change and abnormal growth of squamous cells on the surface of the cervix. While this has the potential to increase screenings in low-resource countries, Huang did note that there are some concerns over the overall effectiveness of the system due to reports of poor correlation between the AI recognition and high-grade disease as well as disagreement among experts when assessing visual findings.
The second multimodal system is more applicable in advanced regions like the United States, where there are already well-established screening programs.
“The second use for our system is that it’s going to take not just images, but other screening test results,” including other background information on the patient, their age and their pH level, Huang told SheKnows. Because this version takes more factors into account, it produces more accurate results after the first use.
Using a large data set, Huang reported “very high accuracy” with the multimodal system, reaching almost 87 percent sensitivity and 90 percent specificity. The first image-only system achieved 80 percent sensitivity and 90 percent specificity. So far, Huang and her team have not tested the screening methods on clinical data, but everything in the lab is ready for that next step.
“I am very excited about this technology, and we’re moving forward with the proposal for the clinical trial. I really hope the system can be used in practice and in the field,” Huang said.
Unlike traditional Pap smears where doctors remove some cervical tissue from the patient, Huang’s method is based on images only. The most invasive part is applying diluted vinegar to the surface of the cervix and setting up the camera.
“There’s no physical harm to the patient in any way,” Huang said.
The idea is getting to the point that this robotic system can use AI to read the biomedical images more accurately than doctors alone.
“I think it’s going to be very useful and it is a very important goal for my career,” Huang said. “I want to develop technologies — especially image-based diagnosis and screening systems — to improve the physical and mental well-being of women and children.”