University of California Irvine launches AI center for healthcare
A new multispecialty initiative that will initially focus on radiology will aim to apply deep learning to the practice of healthcare.
Physicians at the University of California, Irvine, and UCI Health System have established the UCI Center for Artificial Intelligence in Diagnostic Medicine to develop and integrate artificial intelligence technology into routine clinical practice throughout the health system.
The center, led by Peter D. Chang, MD, and Daniel S. Chow, MD, neuroradiologists in UCI School of Medicine’s Department of Radiological Sciences, seeks to advance patient care, improve health outcomes and lower costs by applying machine learning neural networks to healthcare applications, such as diagnostics, disease prediction and therapy planning.
“Our goal is to empower healthcare providers, researchers and patients through the use of artificial intelligence,” says Chang, who is also a software engineer.
The center intends to offer expertise and other resources to support artificial intelligence research for any group on campus interested in incorporating deep learning into their research.
The initial focus of the center will be in radiology, applying deep learning for clinical use in the UCI Medical Center emergency department in treating intracranial hemorrhage. UCI has already conducted research in this area, creating a customized deep learning system with more than 97 percent accuracy in near real time interpretation of the images and detection of brain hemorrhage on non-contrast head exams. The software was applied to more than 10,000 UCI Health imaging exams to test its efficiency and accuracy and validated with prospectively acquired data.
Chang was recognized by the American Society of Neuroradiology with the 2018 Cornelius G. Dyke Memorial Award for this work.
“Intracranial hemorrhages are significant medical emergencies that results in 40% patient mortality, despite aggressive care. Early and accurate diagnosis is necessary for the management of life-threatening brain bleeds and to improve the odds of recovery,” says Chang.
The Center is now preparing this software for clinical applications.
“The research is an example of how we can use machine learning technology to improve the delivery of acute care in an emergency department by expediting triage of patient care and offering more detailed information to guide clinical decision making,” says Chow. “An AI-based imaging may be used either as a triage system to assist radiologists in identifying high-priority exams for interpretation or as a method to rapidly quantify ICH volume, or both.”
Another area where the center may soon apply deep learning to clinical care is in dealing with brain cancer. Chang, Chow and others have developed and trained a deep learning algorithm to virtually conduct biopsies by independently classifying gliomas—tumors that originate in the brain’s supportive tissue—in magnetic resonance imaging. The software’s accuracy rate was 83 percent to 94 percent. The software’s ability to accurately analyze genetic mutations in the tumors from the images can provide complementary insight to tissue analysis of tumor tissue obtained during an actual biopsy. This research was recently published in the American Journal of Neuroradiology.