Trending: Inside T-Mobile’s new device testing lab, where the ‘Un-carrier’ is gearing up for national 5G rollout
BiliScreen app at work
A 3-D-printed viewing box holds a smartphone in place to take a picture of the user’s eyes. The BiliScreen app analyzes the eye image to look for signs of jaundice, which could point to pancreatic cancer. (University of Washington Photo / Dennis Wise)

University of Washington researchers have created a smartphone app that can let users screen themselves for pancreatic cancer and other diseases by taking a selfie.

But not just any selfie.

The BiliScreen app is designed to focus in on the whites of your eyes. If your whites have an overly yellowish tinge, that could suggest you have increased levels of a compound known as bilirubin. That’s a sign of jaundice, and also one of the earliest indicators of pancreatic cancer.

The first effects on the whites of a person’s eyes, also known as the sclera, are too subtle to be noticeable to the naked eye. Heightened levels typically show up in blood tests, but the UW team says BiliScreen can serve as an effective, low-cost, low-impact screening tool.

An initial clinical study of 70 people found that when used with a specially designed smartphone box to control the light exposure conditions, the BiliScreen app’s readings correlated with actual bilirubin levels in the blood at an 89.7 percent level.

The findings were published in the Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies, and are the subject of a presentation set for Sept. 13 at Ubicomp 2017 in Hawaii.

“The hope is that if people can do this simple test once a month — in the privacy of their own homes — some might catch the disease early enough to undergo treatment that could save their lives,” lead study author Alex Mariakakis, a doctoral student at UW’s Paul G. Allen School of Computer Science and Engineering, said today in a news release.

UW’s Ubiquitous Computing Lab, or UbiComp, focuses on how common consumer devices, such as smartphones equipped with cameras and microphones, can be used to screen for disease. BiliScreen builds on an earlier project from UW’s Ubiquitous Computing Lab known as BiliCam.

BiliCam is a smartphone app that screens for jaundice in newborns by analyzing pictures of the baby’s skin. A study published in the journal Pediatrics reported that the BiliCam test results for 530 infants hit an overall correlation of 90 percent with bilirubin levels measured by blood tests.

After seeing the success of BiliCam, the UbiComp researchers turned their attention to pancreatic cancer, a fast-developing disease that kills more than 40,000 Americans every year. Apple co-founder Steve Jobs died of pancreatic cancer in 2011. And just this summer, Mighty AI co-founder Matt Bencke stepped down from his CEO role at the Seattle startup after being diagnosed with the disease.

The prospects for surviving pancreatic cancer are much higher if a diagnosis is made early, but the early signs of the disease often go undetected. The UbiComp team was intrigued by the idea of focusing on color changes in the eyes.

“The eyes are a really interesting gateway into the body — tears can tell you how much glucose you have, sclera can tell you how much bilirubin is in your blood,” said the study’s senior author, UW Professor Shwetak Patel.  “Our question was: Could we capture some of these changes that might lead to earlier detection with a selfie?”

BiliScreen takes advantage of computer vision and machine learning tools to estimate bilirubin levels based on the wavelengths of light that are reflected or absorbed by the sclera.

Patel and his colleagues tried a couple of approaches for imaging the sclera. One idea was to take pictures of faces while the subjects were wearing color-printed cardboard glasses. Light levels were compared to calibrate the images.

The other idea was to program a 3-D printer to make a viewing box, analogous to a Google Cardboard VR headset. Users could slip their smartphones into the box, with the camera and flash pointing at their eyes through the headset.

Viewing box and glasses
The UW team tested two different accessories for BiliScreen: a 3-D printed box to control lighting conditions and glasses that help the app calibrate colors. (UW Photo / Dennis Wise)

The pictures taken through the box turned out to produce the most reliable images for analysis.

“This relatively small initial study shows the technology has promise,” said co-author Jim Taylor, a pediatrics professor at UW Medicine whose father died of pancreatic cancer at age 70.

Now the team plans to test the app on a wider range of people at risk for jaundice and underlying conditions, which can range from cancer to hepatitis or alcoholism to a relatively harmless liver condition known as Gilbert’s Syndrome.

The bigger the database, the better BiliScreen becomes at recognizing the first signs of trouble. The researchers hope the app will become so adept at analyzing the sclera that accessories become unnecessary. In other words, they’ll expect BiliScreen to think outside the box.

In addition to Mariakakis, Patel and Taylor, the authors of “BiliScreen: Smartphone-Based Scleral Jaundice Monitoring for Liver and Pancreatic Disorders” include Megan Banks, Lauren Phillipi and Lei Yu.

Subscribe to GeekWire's Space & Science weekly newsletter


Job Listings on GeekWork

Computer Science InstructorRenton Technical College
Brand JournalistRad Power Bikes
Find more jobs on GeekWork. Employers, post a job here.