A new collaboration between processor technology company Intel and medical professionals at the University of Pennsylvania uses artificial intelligence models to identify brain tumors by utilizing the skills and technology of 29 international healthcare and research institutions.
The project aims to train a new AI with a process known as federated learning, first introduced by Google in 2017. Federated learning uses decentralized servers to train the proposed AI algorithm, which allows researchers to collaborate their data and train an AI algorithm with a larger data set.
The benefit of using a federated learning model is that it trains an AI algorithm by encouraging participants of this data to submit the data without actually knowing the ownership of or revealing the raw sensitive or proprietary data to one another.
And because so many healthcare institutions plan to contribute, the project allows for a far greater accumulation of data than any single facility might obtain on their own, improving the ability to ID brain tumors.
“It is widely accepted by our scientific community that machine learning training requires ample and diverse data that no single institution can hold,” said Dr. Spyridon Bakas at Penn’s Center for Biomedical Image Computing and Analytics, and principal investigator on the project.
“This year, the federation will begin developing algorithms that identify brain tumors from a greatly expanded version of the International Brain Tumor Segmentation challenge dataset. This federation will allow medical researchers access to vastly greater amounts of healthcare data while protecting the security of that data.”
In a statement, Intel principal engineer Jason Martin said, “AI shows great promise for the early detection of brain tumors, but it will require more data than any single medical center holds to reach its full potential.”
The project receives its funding from the National Institute of Health’s National Cancer Institute. The grant offers $1.2 million under a three-year term. Healthcare professionals and researchers from universities and healthcare centers in North America, Europe, and Asia plan to participate in sharing their data.
Some institutions involved in the first phase of the project include the Hospital of the University of Pennsylvania, Washington University the University of Pittsburgh Medical Center, Vanderbilt University, and Queen’s University, among several others.
Active researchers will develop local models in their facilities, then share them on decentralized servers to generate even larger data sets in collaboration with other researchers.
Data from the American Brain Tumor Association (ABTA), shows that nearly 80,000 people will be diagnosed with a brain tumor this year. More than 4,600 of these are children.
The federated learning approach by Intel and the University of Pennsylvania make the early identification of brain tumors a more streamlined research process. As Dr. Spyridon Bakas from the Perelman School of Medicine says, these algorithms “will allow medical researchers access to vastly greater amounts of health care data while protecting the security of that data.”