Improving MRI Access and Efficiency through Changeover-Aware Scheduling
Abstract
Purpose
To evaluate dynamic MRI scheduling policies for reducing patient changeover time and shortening daily clinic operating hours across multiple imaging centers.
Methods
Historical MRI clinic data from three sites with different numbers of MRI systems were analyzed to estimate empirical distributions of daily arrivals, procedure types, exam durations, and changeover times. Changeover time was defined as the time between the last acquired image of a MR exam and the first acquired image of the following exam. Discrete-event simulation models were developed in Python to represent parallel MRI scanners serving a fixed daily patient pool. Three scheduling policies were compared: random sequencing, grouping patients with the same procedure, and a greedy policy that minimizes expected changeover time using an empirical pairwise changeover-time matrix. Performance metrics included total changeover time and total clinic operating hours. Experiments were replicated to construct confidence intervals and assess robustness across clinics.
Results
Across all clinics, the greedy policy consistently reduced total changeover time and daily operating hours relative to random scheduling, freeing substantial capacity. Average daily changeover-time reductions ranged from approximately 85 to 190 minutes, corresponding to annual savings exceeding 500 to 1,100 hours per site. Operating hours decreased by roughly 40 to 46 minutes per day, creating capacity for nearly one additional patient per scanner daily without extending hours. Grouping the same procedures also improved performance but was dominated by the greedy strategy in most settings.
Conclusion
Dynamic, data-driven MRI sequencing policies can substantially improve operational efficiency using existing resources. Greedy changeover-minimizing policies reduce scan-day length and create meaningful capacity for additional patients, with large projected annual time savings. These results support pilot clinical implementation and motivate further optimization-based policies across broader procedure sets and longer operational horizons.