New App Lets You Screen for Pancreatic Cancer Using Smartphone Selfies

August 29, 2017

Cancer is one of the most difficult illnesses to detect, especially when many forms don’t have any significant tell-tale signs. However, with the help of technology advancements, there have been many medical leaps that are changing the way we screen, test and cure cancer.

Nowadays, the newest and most accessible form may be coming right to your smartphone with an app that’s designed to screen you for pancreatic cancer.

How? Through selfies.

Introducing BiliScreen

BiliScreen uses your phone’s camera, paired with computer vision algorithms and learning tools, to scan the level of bilirubin that shows in the sclera of the eye (the white part). Bilirubin levels are said to surge within the blood levels of at-risk individuals, especially in the initial stages of pancreatic cancer. This can cause the skin and the eyes to take on a yellowish tinge. Now, there’s a chance to detect this early thanks to the beauty of technology.

The Life-Changing App

By taking and submitting a selfie, the BiliScreen app can read the levels of the bilirubin compound present before it can be naturally detected by the naked eye. As such, this app could mean that the lives of at-risk individuals may drastically change, as can the overall levels of detection and treatment for the disease.

BiliScreen was created by University of Washington researchers, with the goal of not only detecting early symptoms of the cancer, but providing an easy and accessible tool that is non-invasive and surprisingly accurate in the screening. It was made due to the prevalent levels of the disease within America, and the wonderment of whether computer visioning when paired with machine learning tools could detect colour changes that are an often-undetected symptom of pancreatic cancer.

It is changing the meaning behind taking a selfie for some people, with the app isolating parts of the eye to use in the medical prognosis. The app then calculates the information fed from the wavelengths of light being absorbed by the eye, and correlates this with the level of bilirubin present.

The Accuracy of the App

There was a lot of trialling with the app to ensure that it was presenting accurate results. In one of the first clinical studies with 70 participants, the app correctly identified 89.7% of concern cases, which was more accurate than blood tests that are currently used. The app in this case was used alongside a 3D printed box that controls the eye’s exposure to light.

This is a massive step forward, especially in relation to the current statistics of pancreatic cancer cases within Australia. Of an estimated 134,174 new cancer cases diagnosed this year, 3,271 of those are estimated to be new pancreatic cancer diagnosis.

Pancreatic cancer itself is one of the most undetected forms of cancer, often being left too late by the time it is found. The hope is that with this app, cases of this cancer form can be caught early and primary prevention can take place.

As the app depends on reading the light that is being reflected and absorbed by the eye, it accounts for the different lighting conditions through two different methods tested by the BiliScreen team. One of the methods used was paper glasses, printed with different coloured squares that help to calibrate the colour of the eyes. The other is a 3D box that blocks natural lighting. This method gained higher accuracy with results.

Other Apps by the Creators

The app builds on earlier work that was created by the team, which was named Bilicam. In this version of the app, levels of jaundice were detected in newborn babies by evaluating a picture of the baby’s skin. Bilicam was shown to provide accurate readings of bilirubin levels relating to jaundice in 530 infants, according to a recent study.

App development has brought a lot of exciting and helpful tools right into the palms of our hands. However, this could be one of the most ground-breaking advancements in medical apps thus far. We can’t wait to see the future of this app, and the possible expansions later down the track.