Doctors will soon be able to get help in the fight against cancer thanks to the Computer Vision Research Center at the University of Central Florida.
Engineers at the center have learned a computer how little dots of lung cancer should be detected in CT scans, which radiologists often find difficult to identify. The artificial intelligence system is about 95 percent accurate, compared to 65 percent when done by human eyes, the team said.
"We used the brain as a model to create our system," said Rodney LaLonde, a PhD student and captain of the UCF hockey team. "You know how the connections between neurons in the brain reinforce and learn during development? We used that blueprint to help our system understand how patterns in the CT scans should be looked for and how they can find these small tumors."
The approach is similar to the algorithms used by face recognition software. It scans thousands of faces in search of a certain pattern to find the agreement.
Ulas Bagci, scientific assistant professor, leads the group of researchers in the center that focuses on AI with potential medical applications.
The group fed more than 1,000 CT scans – provided by the National Institutes of Health through a collaboration with the Mayo Clinic – in the software they developed to help the computer learn to search for the tumors.
Graduate students working on the project, the computer had to learn different things to learn well. Naji Khosravan, who obtains his doctorate, has created the backbone of the learning system. His skill in learning new machines and computer vision algorithms led to his summer as a trainee at Netflix and helped the company with various projects.
LaLonde taught the computer to ignore other tissues, nerves and other masses that it encountered in the CT scans and to analyze lung tissues. Sarfaraz Hussein, who obtained his doctorate last summer, refines the ability of AI to differentiate cancer from benign tumors, while graduate Harish Ravi Parkash learns from this project and applies them to see if another AI system can be developed to help identify or predict brain disorders.
"I believe this will have a big impact," Bagci said. "Lung cancer is the number one cancer killer in the United States and if it is discovered in the late stages, the survival rate is only 17 percent." By finding ways to help identify earlier, I think we can help increase survival. "
The team will present its findings in September at the largest leading conference for medical imaging research – the MICCAI 2018 conference in Spain. The work of the team was published prior to the conference. https: /
The next step is to move the research project to a hospital environment; Bagci is looking for partners to make that possible. Then the technology could be a year or two away from the market, Bagci said.
"I think we all came here because we wanted to use our passio for engineering to make a difference and save lives is a big impact," said LaLonde.
Ravi Prakash agrees. He studied engineering and his applications in agriculture before he heard about Bagci and his work at UCF. Bagci's research is in the field of biomedical imaging and machine learning and their applications in clinical imaging. Previously, Bagci was staff scientist and laboratory manager at the Center for Infectious Disease Imaging Lab at the NIH, in the Department of Radiology and Imaging Sciences.