Sarasota teen develops cancer-detecting software

Herald Health CorrespondentFebruary 4, 2014 

In 2012, Sarasota teen Brittany Wengel won first place at the Google Science Fair, a competition for ages 13 to 18 that attracts thousands of worldwide entries. The prize was $50,000 and a trip to Galapagos. PHOTO PROVIDED

When Brittany Wenger was creating a computer software program she hoped would influence the way breast cancer is biopsied, she set her alarm clock to wake up every few hours.

She was too excited to sleep through the night -- she was building a computer "brain" and wanted results in a shorter amount of time than if she only worked on the project after school and on weekends.

Wenger was a sophomore in high school at Out-of-Door Academy in Sarasota when she started developing the artificial intelligence software program that eventually would be able to analyze data from a breast tissue biopsy -- looking for patterns in cell uniformity and more -- and tell if the biopsy was benign or malignant.

It took her two years and two failures before she got the results she was seeking.

Sometimes her mother, Cami, would sneak into her room and turn off the alarm so her daughter would sleep. But Brittany still wanted to get up.

"She would say, 'Mom, you don't know how important this could be,' " said Cami Wenger.

Wenger is now a freshman at Duke University where she juggles a heavy schedule of classes and works in the university's epigenetic laboratory.

Much has happened since she started the project she named the Global Neural Network Cloud Service for Breast Cancer in 10th grade.

In 2012, she won first place at the Google Science Fair, a competition for ages 13 to 18 that attracts thousands of worldwide entries. The prize was $50,000 and a trip to Galapagos. She was named in Time magazine's "30 under 30" who are changing the world. She has spoken at Ted Talks, attended a Google event at a renowned computer research organization in Switzerland and was invited to be in the annual science fair at the White House.

Next summer, she will speak at the Royal Society in London and also accompany a U.S. State Department team to Russia to talk about science.

The Google prize provided the platform for her software application to be noticed, she said. Two hospitals, in Philadelphia and Italy, now are testing it.

When she was in 12th grade, Wenger expanded the software to predict relapse of leukemia. The leukemia project is still in development.

"Science takes time," she said.

Wenger first became interested in artificial intelligence and how to build a neural network in the seventh grade. She is an avid soccer player and -- through much trial and error -- built a program that could play soccer. Neural network software gets smarter as it goes along; it is able to take what it has learned and apply it to new situations. The program played soccer like 4-year-olds at first, she said, then it improved.

She was a sophomore at Out-of-Door Academy when her cousin was diagnosed with breast cancer. Witnessing her cousin's experience, Wenger turned to a more serious concept for the artificial intelligence software. She believed being tested for breast cancer should be easier.

There is a very minimally invasive test for breast cancer, the fine needle biopsy, but it has fallen out of favor. A thin needle is inserted into the breast lump to collect cell samples. The biopsy feels similar to a blood test and women go home with a small adhesive bandage.

The problem is that results can be difficult to analyze with complete accuracy and many doctors don't trust it compared to other biopsy methods.

But artificial intelligence software can detect complex patterns that humans can't. Wenger set out to build a system that would be able to accurately detect cancer patterns from fine needle biopsies.

With the help of her Out-of-Door Academy science teacher, Mike Newhams, she pored over scientific journal articles. She discovered a large data set from the University of Wisconsin, easily accessible through public domain online, that showed the characteristics of tissue samples from fine needle biopsies for suspected breast cancer.

Using the data, her software program went through 7.6 million trials. It took two years of work until the program reached 99 percent accuracy in knowing whether a tissue sample was benign or malignant.

Wenger has a site named Cloud4Cancer Breast Cancer Detection at www.cloud4cancer.appspot.com where she takes information about breast cancer samples through fine needle aspiration. The more samples the computer neural network processes, the better it will get.

"I think it's hospital-ready right now, to be honest," said Wenger. The computer program could be another tool for doctors in detecting breast cancer.

Fine needle biopsies are cheaper than other biopsy methods with less discomfort, said Wenger.

"I think it's important to revive them," she said.

Susan Hemmingway, Herald health correspondent, can be reached at shemmingway@hotmail.com.

Bradenton Herald is pleased to provide this opportunity to share information, experiences and observations about what's in the news. Some of the comments may be reprinted elsewhere in the site or in the newspaper. We encourage lively, open debate on the issues of the day, and ask that you refrain from profanity, hate speech, personal comments and remarks that are off point. Thank you for taking the time to offer your thoughts.

Commenting FAQs | Terms of Service