Your iPhone Could Help Detect Eye Disease in Kids Simply by Taking a Photo

In yet another example of how your favourite Apple devices can detect health conditions you may not even know about, a new group of researchers have found a way to analyze photos to detect early signs of eye disease, even more reliably than doctors can in many cases.

According to Engadget, researchers have developed an app that uses machine learning to analyze a photograph of a child’s eyes in order to look for early signs of conditions such as retinoblastoma and cataracts.

In a paper titled Autonomous early detection of eye disease in childhood photographs, the researchers describe how the system looks for telltale signs of “white eye” in everyday pictures taken with the iPhone camera, with a degree of accuracy that surpasses many standard paediatric examinations. As the paper notes, doctors routinely give children a “red reflex test” to screen for a condition known as leukocoria, yet this test isn’t always effective at detecting many eye disorders.

However, since leukocoria also presents in casual photographs, the researchers have developed a free smartphone application, CRADLE — which stands for ComputeR–Assisted Detector of LEukocoria — and made it available on the App Store under the name White Eye Detector.

Since leukocoria can theoretically be detected even in normal, casual photographs, the White Eye Detector app doesn’t even require that you take a specific picture of your child, but can in fact go through your existing photo library to search for past photos that might be indicative of this condition.

Red vs White

Anybody who has ever taken a picture with a direct flash is familiar with the “red eye” effect, where a person’s pupils light up in an eerie and unnatural red colour. However, this is actually a good sign, since it’s caused by the reflection of light by healthy choroidal and retinal blood vessels in the back of the eye.

On the other hand, if the light that comes back is white or yellow-orange, this can be an indication of any of several childhood eye disorders, including retinoblastoma, paediatric cataracts, Coats’ disease, and more.

The Study

In their research, a study was conducted using the app to analyze 52,982 longitudinal photographs of children that had been collected by parents from before they enrolled in the study. The study group included 20 children who had already been diagnosed with retinoblastoma and other eye conditions, as well as 20 control children. The study revealed that the app was able to detect early signs of leukocoria in photographs taken of the child up to 1.3 years before they were actually diagnosed with the condition.

Retinoblastoma can become serious, migrating to the brain, in less than six months after white eye first becomes visible, so early diagnosis is especially important and such an app could definitely save lives or at least improve the chances of dealing with the condition more effectively.

As the researchers note, however, the CRADLE application is not FDA-approved and shouldn’t be treated as a diagnosis; it’s merely intended to “allow parents to augment clinical leukocoria screening with photography.” Much like Apple’s ECG app, however, it could provide an early warning of a possible condition that would encourage a visit to a paediatric ophthalmologist.