Presenter Dr. Michael Muelly of Stanford University said that their research was motivated by the observation that their MR scanners were sitting idle a significant portion of the time -- despite the schedule being nearly completely filled. While a lot of work has gone into speeding up sequence acquisition, very little has gone into optimizing other portions of the scanning process, Muelly said.
"Our goal is to increase utilization of existing resources and improve the patient experience," he said. "Increased resource utilization lowers the cost of a scan, which then allows us to expand the range of applications of MRI."
After collecting sequence acquisition data for about 1.5 years, the researchers now have detailed parameters for almost 1.3 million acquisitions, Muelly said. They used this information to build a simulation platform to test the effects of changes on the utilization rate. By testing the effects of different schedule block lengths, the researchers found that they could increase the scanner utilization rate by setting the block length at 28 minutes. They also employed a neural network that could accurately predict study length based on patient characteristics.
"By adjusting the block times, we can significantly increase the utilization rate of our scanners," he said. "Further, with the help of machine learning, we can predict how long an exam can be expected to take for a specific patient. In practice, these optimizations will allow us to better use our existing resources while decreasing patient wait times and improving patient satisfaction."
Learn more about how they accomplished these goals by sitting in on this Tuesday afternoon session.
