- A deep learning platform designed to identify acute neurological conditions, including stroke, hemorrhage, and hydrocephalus, detected disease in CT scans in 1.2 seconds, or 150 times faster than human radiologists.
Researchers at Mount Sinai Icahn School of Medicine developed the platform by using 37,236 head CT scans and training a deep neural network to identify whether an image contained critical or non-critical findings.
The team then tested the platform in a randomized controlled trial in a simulated clinical setting, where it sorted the CT scans based on their severity.
The algorithm could preprocess an image, run its inference method, and, if needed, raise an alarm at a speed that was 150 times faster than the time it took for physicians to simply read the image.
This study is the first to use artificial intelligence to detect a wide range of acute neurological events and to demonstrate a direct clinical application.
“The application of deep learning and computer vision techniques to radiological imaging is a clear imperative for 21st century medical care,” said Burton Drayer, MD, the Charles M. and Marilyn Newman Professor and System Chair of the Department of Radiology for the Mount Sinai Health System, CEO of the Mount Sinai Doctors Faculty Practice, and Dean for Clinical Affairs of the Icahn School of Medicine.
The research team used weakly supervised learning approaches in this study, a method that leveraged the research team’s expertise in natural language processing and Mount Sinai Health System’s large clinical datasets.
The next phase of research will involve improved computer labeling of CT scans and a shift to strongly supervised learning approaches, as well as new techniques to increase data efficiency.
Researchers anticipate that they will complete the process of re-engineering the system with these changes within the next two years.
“The expression ‘time is brain’ signifies that rapid response is critical in the treatment of acute neurological illnesses, so any tools that decrease time to diagnosis may lead to improved patient outcomes,” said Joshua Bederson, MD, Professor and System Chair for the Department of Neurosurgery at Mount Sinai Health System and Clinical Director of the Neurosurgery Simulation Core.
The study was conducted by the Mount Sinai AI Consortium, or AISINAI, a team of scientists, physicians, and researchers committed to advancing the role of artificial intelligence in medicine and improving care delivery.
This current research venture adds to the work AISINAI completed earlier this year, in which natural language processing algorithms were used to identify clinical concepts in reports for CT scans.
The team expects that this platform will contribute to improvements in radiology and patient outcomes.
“With a total processing and interpretation time of 1.2 seconds, such a triage system can alert physicians to a critical finding that may otherwise remain in a queue for minutes to hours,” said senior author Eric Oermann, MD, Instructor in the Department of Neurosurgery at the Icahn School of Medicine at Mount Sinai.
“We’re executing on the vision to develop artificial intelligence in medicine that will solve clinical problems and improve patient care.”