Recent developments in AI-based image production and enhancement techniques have allowed for aggressive undersampling of k-space data resulting in shorter acquisition times, often with preserved or even improved image quality compared to non-AI reconstructed pulse sequences. In order to take advantage of shorter MR examination times, conventional workflow strategies and AI applications can be employed. Early AI-based MR image production products have predominantly applied to spin-echo based 2D pulse sequences. Musculoskeletal MR, largely comprised of these sequences, has benefited disproportionately. These time savings, in addition to the fact most musculoskeletal MR exams are performed without intravenous contrast, mean that common exams such as routine joint MRs can have much shorter exam times than the average MR exams. Although block scheduling, where exams at an imaging site for a fixed period can be limited to certain exam types, may provide a strategy for aggressively shortening exam timeslots and increasing throughput, this may result in decreased filling of slots and increase in MR wait times for patients. Instead, conventional strategies such as optimization of MR suite layout and updating scheduling to align with patient demand should be considered. AI applications can be deployed to help address scheduling challenges. Machine learning based models to predict patient no-shows based on exam type, time or day of the week, and patient demographics. Targeted phone calls prior to the appointments can reduce no-show rates. The impact of cancellations could also be mitigated by strategically overbooking in order to ensure that MR resources are maximized.

Acknowledgements

No acknowledgement found.

References

No reference found.