Pre-MRI Safety Screening for Implantable Medical Devices

Year Submitted: 2025
Published Date: 02 January 2026

Tags: Quality Improvement, Workflow Redesign, Data Analytics, Care Process ＆ Redesign, Technology, Digitalisation, Data Management, Data Visualization

About this Content

Aims

This quality improvement project aims to develop and implement a novel, data-driven approach to reduce the time and effort needed for pre-MRI safety screening, and to detect undisclosed AIMDs/PMDs.

Background

Magnetic Resonance Imaging procedures pose safety risks for patients with Active Implantable Medical Devices (AIMDs), such as Cardiac Implantable Electronic Devices (CIEDs), and Passive Medical Devices (PMDs), such as programmable ventriculoperitoneal shunts. A Pareto analysis of MRI incident reports at our institution revealed that 56% were implant related. Despite the known risks, a significant proportion of AIMDs/PMDs (41.09%) at our institution remained undisclosed. Given the volume and complexity of implant screening, especially for high-risk devices such as pacemakers, there is a pressing need for a systematic, data-driven approach to support accurate and timely identification of at-risk patients during pre-MRI safety screening. Our quality improvement project aims to reduce the time and effort needed for pre-MRI safety screening, and to detect undisclosed AIMDs/PMDs.

Methods

A data-driven implant detection dashboard was developed using Endeavour AI + TIBCO Spotfire. The dashboard serves to centralize EPIC data, including appointment details, implant history, MRI request forms, and radiology reports into a single location. Leveraging on TIBCO Spotfire's streaming capabilities, the platform search these data using keyword matches, rule-based logic, and regular expressions to identify patients with devices based on radiology reports, device histories and request forms. The device is then classified based on keywords and identified on the dashboard for safety clearance by radiographers.

Results

We prospectively evaluated 32,214 unique patient visits to MRI over 1 year (Nov 2023-Oct 2024). In the past, blanket screening of all scheduled patients took 400 mins a day. In comparison, the dashboard efficiently narrows the focus to just 1.5% of the daily workload (23 patients, 5 minutes) for review, closely aligning with the 1.4% of unique patient visits to MRI that truly involve devices. Therefore, blanket screening a full schedule of patients to identify this small subset is highly inefficient and a disproportionate use of radiographers time. The preview of radiology reports on the dashboard also enables a more directed implant search in EMR, reducing the average screening time from 2.5 minutes per patient to 1.5 minutes. The resulting time saving of 395 minutes is equivalent to 0.8 full-time equivalents (FTE), freeing up radiographers to concentrate on clinical and patient tasks. Early identification of undisclosed devices prevented an estimated 1280 minutes of scanner idle time and related revenue loss from on-arrival cancellations due to non-disclosure. Notably, it also helped identify undisclosed AIMDs/PMDs, as well as those inserted after MRI orders were placed.
The dashboard demonstrated exceptional reliability with consistently high accuracy (Negative Predictive Value (99.91%) and Specificity (99.87%), indicating excellent ability in filtering out patients without devices. False negatives were due to data limits or external records. A final safety interview on patient arrival remains essential.

Conclusion

The dashboard flagged at-risk patients early up to 2 weeks in advance, as first layer of MRI safety screening, reducing screening workloads by 98%. This implementation demonstrates a successful paradigm shift from blanket screening to selective screening of at-risk patients using data analytics, achieving exceptional detection accuracy.

Lessons Learnt

Our work demonstrates that data analytics support a smarter, faster way to pre-MRI safety screening:
1. This paradigm shift from 100% blanket screening to targeted, data-driven triaging that flags only 1.5% of workload, close to the 1.4% actual prevalence of devices.
2. Importantly, the dashboard helps address documentation gaps and changes in patients actual device status, such as devices inserted after MRI was ordered (n=6/32,214), demonstrating its capability to detect changes in implant status and reduce the risk of overlooked implants.
3. Initially hindered by skepticism and a deep-seated reluctance to relinquish manual checks, our management stepped in with no-blame assurance, ensuring that the radiographers will not be held accountable for any miss resulting from the dashboards limitations. Peer advocacy & on-the-ground support from the pilot group during deployment was also important in building confidence.
4. Most importantly the dashboard acts as the first layer of screening while a final safety interview by radiographers upon arrival remains a vital practice.
5. Beyond MRI safety, this dashboard has been replicated for other needs, including BMD Device screening, Duplicate imaging checks for CT/US/MRI, and Right-siting of Special MRIs.

Additional Information

NUH IC-QIX Merit Award CY25

Keywords

MRI Safety, MRI Screening, Implantable Medical Devices, Implants, Data Analytics

Innovators' Details

Connect with this contributor!

Regine Teo Ee Chin - ee_chin_teo@nuhs.edu.sg

Project Attachment

NUH QIX Merit Award - PreMRI Safety Screening.pdf