App for early detection
A Baylor professor is hoping a newly launched iPhone app he helped create will speed up the detection of a form of eye cancer in children, helping families get critical and potentially life-saving treatment sooner.
The free White Eye Detector app, created by Baylor chemistry professor Bryan Shaw and computer science professor Greg Hamerly, was officially launched in Apple’s App Store on Oct. 22. It scans photos on an iPhone or iPad and searches for white pupils, or leukocoria, which is a symptom of retinoblastoma, a cancer that affects children up to age 5.
The app can also use the camera on the phone or tablet to directly scan a child’s eye and detect leukocoria. Leukocoria becomes apparent in photos as the camera flash bounces off tumors in the back of the eye, creating the white pupil in the image.
A white eye can also indicate cataracts in children and adults, as well as Coat’s disease, a condition the causes abnormal growth in blood vessels behind the retina and can lead to blindness.
“Think about it this way: A mom scans the retina of her kid every time she takes a picture, and she does that every day,” Shaw said. “A pediatrician will look at him once a month, once every three months . . . It’s pretty obvious that a pediatrician’s never going to be able to compete with a mom for detecting retinoblastoma.”
Shaw’s sentiments are rooted in his own experience dealing with his son Noah’s retinoblastoma diagnosis in 2008 at 4 months old. Shaw and his wife, Elizabeth, first noticed it when Noah was about 3 months old and informed his pediatrician, who confirmed the diagnosis at his next doctor’s visit.
Noah had to undergo chemotherapy and radiation for the rare cancer, which caused tumors behind both of his eyes, and ultimately had to have his right eye removed.
Though the now-6-year-old is thriving and in remission, Shaw said a quicker diagnosis likely would have allowed his son to avoid losing one eye.
He first came up with the idea of the app after studying some 9,800 baby photos of Noah, looking for white eye and discovering that the condition was first apparent when his son was about 12 days old, meaning he was born with the tumors.
“Up until that point, everyone had always thought white eye was a symptom of the later stage of retinoblastoma,” Shaw said. “By having all of these pictures and analyzing them, I showed that it started occurring early on, and that it increased in frequency, started showing up in more and more pictures as the tumors got bigger and as they multiplied.”
After he began teaching at Baylor in 2010, he reached out to the computer science department for help in designing the app, ultimately partnering with Hamerly. Fellow computer science professor Erich Baker also assisted in the early research for the project, while former graduate student Ryan Henning did most of the actual app design.
It cost about $6,000 to build the app, with Shaw giving $2,500 of his own money and the rest donated by a retinoblastoma survivor who works at Pfizer and a Marine Corps veteran who learned of the project through an NPR story. Hamerly said the team is looking for funding support to develop an Android version of the app to expand its availability.
Using a technology called machine learning, Hamerly’s team used examples of white pupils and normal eyes to train the neural network of the app to distinguish between the two.
That process was done using mainly images of Noah and public domain photos found online, but Baylor now has thousands of images of leukocoria in other children with retinoblastoma submitted by families across the country that will be used in future versions of the app.
So far, more than 200 people have downloaded the app, including residents in Latin America, Europe and parts of Africa. Hamerly noted that in developing countries, a retinoblastoma diagnosis often ends in the death of the child because the cancer is often detected in later stages.
Shaw said retinoblastoma is typically discovered when a child is between 9 and 12 months old. Shaw hypothesized that when tumors are smaller, a light must hit the eye at a certain angle to reflect the white pupils, but the range of angles increases as the tumors grow.
The duo opted to make the app free so that parents would have easy access to it and to boost awareness about the signs of retinoblastoma.
‘Information campaign’
For example, Hamerly said he was not aware that white eye was a symptom of the cancer when his first son was born, and Shaw said his wife only knew to look for it after reading an article about retinoblastoma in a parenting magazine, not from a physician.
“It’s something that any pediatrician is going to know about, and it’s something that any pediatrician is going to scan for, and it’s very easily detectable, so part of this is an information campaign,” Hamerly said.
Hamerly said a long-term goal would be partnering with social media sites like Twitter or Facebook and embedding the app’s technology into those platforms, which would allow all photos to be scanned as they are uploaded.
“We could have this software running on the server side rather than someone having to download it,” Hamerly said. “And that way, we can really bridge the gap between, if you know about it, then you can download the app and scan for it versus not knowing about it, so not knowing you should download it and scan for it.”
After enduring the emotional toil of Noah’s diagnosis and treatment, Shaw hopes the app will lead to less stress and happier results for other families dealing with retinoblastoma.
“If this can help speed up the diagnosis of one kid, it’s going to be a success,” Shaw said. “If it saves the life of one kid, my mission has been accomplished. If it is the first step in eradicating metastatic retinoblastoma, if it the first step in eradicating the deaths associated with retinoblastoma, my dreams are going to come true.”
More information
To download the White Eye Detector app to your iPhone or iPad, click
https://itunes.apple.com/us/app/white-eye-detector/id904042354?mt=8
or search in the iTunes store.
(Quelle: http://www.wacotrib.com)