The ability to keep radiology image acquisition on time affects all other parts of the radiology workflow, noted presenter Oleg Pianykh, PhD, of Massachusetts General Hospital (MGH) in Boston.
"Consider a simple example: Scheduling the most duration-variable exam first thing in the morning will most likely destroy the rest of the daily schedule," Pianykh told AuntMinnie.com. "So, if we still have to do this exam, can we find a better time, in the most rational way?"
However, imaging exams have highly variable durations and involve many participants, including patients, physicians, and technologists. Even devices have to be reconfigured and beds prepared, he added.
"This creates a random, unpredictable component in each exam time, and classical scheduling techniques cannot work well with randomly deviating tasks," Pianykh said.
The goal is to put exams in the most stable order, so that random deviations will have minimal effect on the daily schedule, he explained. The MGH researchers used two data-driven algorithms to tackle this problem. The first algorithm learned the real distribution of exam durations based on analysis of a full year of MRI scheduling data at their institution, while the second discovered the optimal sequencing of exams during a day.
"As a result, we can consider a single scanner, or a network of several interconnected scanners/facilities, and reorder their examinations in such as way that we achieve the most stable schedule," Pianykh said. "Note, we do not change any examinations; we simply find the optimal order. Once this optimized exam order is implemented, it reduces patient wait time, it reduces processing disruptions and stress, and sometimes it may gain additional time to process more patients."
