Towards Sustainable Magnetic Resonance Imaging: Insights from long-term, high-resolution energy recordings across an entire scanner fleet
Authors: Florian Leonhard Raab, Fiona Mankertz, Nour Maalouf, Josephine Berger, Andreas Lingg, Reza Dehdab, Sebastian Werner, Judith Herrmann, +12 more
Organizations: Medical Image and Data Analysis (MIDAS.lab), Tübingen University Hospital · Department of Diagnostic and Interventional Radiology, Tübingen University Hospital · Faculty of Computer Science, Eberhard Karls University of Tübingen · Magnetic Resonance, Siemens Healthineers AG
Magnetic resonance imaging (MRI) is among the most energy-intensive diagnostic modalities in healthcare, yet its energy consumption and the factors influencing it remain insufficiently understood. This study aims to establish a comprehensive baseline of MRI energy consumption by characterizing energy demand across a scanner fleet, examining scanner utilization and operating patterns in clinical practice. Concurrently, it investigates the relationships between energy consumption and relevant operational and acquisition features. On average, a single MRI measurement consumed 0.43 kWh, while a complete examination consumed 13.50 kWh. In general, substantial differences in energy consumption were observed between MRI scanners and their corresponding operating modes (scan, idle, and eco-power mode). These variations may be related to differences in scanner operating patterns, employed examination protocols, and their resulting acquisition parameters. Idle and eco-power modes accounted for more cumulative energy consumption than active scanning. However, these energy shares should always be interpreted in relation to scanner occupancy, as utilization patterns strongly influence the distribution of energy across those operating modes. Lastly, linear regression analysis showed that energy consumption was more strongly associated with scan duration than with average power, suggesting that scan duration may be an important factor influencing MRI energy consumption. ...
Figures & tables
Figure 1: An overview of the data processing architecture. Power meters recorded the energy and power consumption of the MRI scanner and its corresponding reconstruction workstation, while MRI system log files were acquired. These complementary data sources were subsequently integrated into a unified dataset.
Figure 2: Daily scanner-specific examination and measurement volumes over the study period. MRI scanners are color-coded for comparison. Seven MRI scanners across two sites (CRONA and UFK) were consecutively onboarded. On average, 42.80 examinations and 1,329.34 measurements were performed per day. Inter-scanner differences in workload and periods of missing data due to data recording interruptions can be seen. (A) Average daily number of MRI examinations. (B) Average daily number of MRI measurements.
Figure 3: Scanner-specific MRI energy consumption aggregated across weekdays. Average energy consumption amounted to 13.02 kWh per examination and 0.50 kWh per measurement. Energy consumption varied considerably between MRI scanners, reflecting differences in scanner characteristics, applied protocols, and the number of measurements. Temporal differences reflected scanner-specific operating schedules, with reduced consumption during weekends. (A) Average energy consumption per MRI examination. (B) Average energy consumption per MRI measurement.
Figure 4: Scanner-specific occupancy. Scanner occupancy was defined as the cumulative daily scan time relative to the total time per day ( 24 hours⋅60 minutes⋅60 seconds ). (A) Mean occupancy by weekday demonstrates temporal variations in scanner utilization and reduced occupancy during weekends. (B) Mean occupancy by scanner over the study period. Averaged Pie chart slice sizes are proportional to the average occupancy of each scanner. Differences in scanner occupancy indicate variations in workload distribution and patient throughput, which influence energy consumption patterns.
Figure 5: (A) Average scanner-specific energy consumption by operating mode and weekday. Total energy consumption is shown for each MRI scanner, and fill patterns indicate the operating mode: solid bars represent scan mode, striped bars represent idle mode, and dotted bars represent eco-power mode. Temporal variations in energy consumption across weekdays and weekends were observed, primarily driven by differences in scan-mode utilization, while idle and eco-power energy consumption remained relatively stable due to continuous cooling requirements. (B) Average scanner-specific energy consumption by operating mode. Energy consumption is expressed as the proportion of the total energy consumption for each scanner. Blue slices represent scan mode, red slices represent idle mode, and green slices represent eco-power mode. Idle and eco-power modes cumulatively accounted for a larger proportion of total energy consumption than active scanning, highlighting the substantial contribution of non-scanning periods to overall MRI energy demand.
Figure 6: Average scanner-specific energy consumption per measurement grouped by examined body region (i.e. scan mode). Pie charts illustrate the distribution of energy consumption across body regions for each MRI scanner. Energy consumption varied substantially between both body regions and scanners, indicating scanner-specific differences in energy demand across examination types. While body-region-based grouping provides an overview of energy demand across clinical applications, it does not fully account for variations caused by differences in sequences and acquisition parameters, highlighting the need for future sequence-based, data-driven analyses.
Figure 7: Scanner-specific scatter plots were used to illustrate the relationship between the energy consumption of individual MRI sequences and their average scan duration and active power consumption. A: Relationship between average scan duration and average energy consumption. Linear regression demonstrated a strong positive relationship between scan duration and energy consumption ( y=0.008x+0.020 , R2=0.769 ) over all scanners. For improved visualization, the y-axis limits were determined based on the regression results, resulting in the exclusion of outliers. B: Relationship between average active power and average energy consumption. A weaker positive relationship was observed based on linear regression ( y=0.048x−0.669 , R2=0.426 ). The same y-axis limits as in A were used to ensure comparability between the two plots.
Figure 8: Screenshots of the hosted energy dashboard. (A) Energy consumption for each measurement. The buttons below the plot allow users to individually adjust the aggregation of the displayed data. Hovering over the measurements provides additional information, while clicking on entries in the legend allows specific groups to be selected and analyzed. (B) Relationship between energy consumption and scan duration for the individual sequences. By selecting a specific area within the plot, users can zoom in on that region and investigate the corresponding sequences in greater detail. (C) Live data streamed directly from the hospital, showing the power consumption of various MRI scanners over the last three hours.
Background: Existing MRI LLM benchmarks rely mainly on review-book multiple-choice questions, where top proprietary models already score highly, limiting discrimination. No systematic benchmark has evaluated vendor-specific scanner operational knowledge central to research MRI practice. Purpose: We developed MRI-Eval, a tiered benchmark for relative model comparison on MRI physics and GE scanner operations knowledge using primary multiple-choice questions (MCQ), with stem-only and primed diagnostic conditions as complementary analyses. Methods: MRI-Eval includes 1365 scored items across nine categories and three difficulty tiers from textbooks, GE scanner manuals, programming course materials, and expert-generated questions. Five model families were evaluated (GPT-5.4, Claude Opus 4.6, Claude Sonnet 4.6, Gemini 2.5 Pro, Llama 3.3 70B). MCQ was primary; stem-only removed options and used an independent LLM judge; primed stem-only tested responses to incorrect user claims. Results: Overall MCQ accuracy was 93.2% to 97.1%. GE scanner operations was the lowest category for every model (88.2% to 94.6%). In stem-only, frontier-model accuracy fell to 58.4% to 61.1%, and Llama 3.3 70B fell to 37.1%; GE scanner operations stem-only accuracy was 13.8% to 29.8%. Conclusion: High MCQ performance can mask weak free-text recall, especially for vendor-specific operational knowledge. MRI-Eval is most informative as a relative comparison benchmark rather than an absolute competency measure and supports caution in using raw LLM outputs for GE-specific protocol guidance.
Perry E. Radau
Department of Radiology, University of Calgary · Child and Adolescent Imaging Research (CAIR) Program · Alberta Children’s Hospital Research Institute +1
MRI provides excellent soft-tissue contrast without ionizing radiation, but long acquisition times increase patient discomfort while also raising exam costs and limiting scanner throughput. A common approach to reduce scan time is to acquire fewer measurements, which yields an ill-posed linear inverse problem; recovering diagnostic-quality images therefore requires incorporating prior knowledge beyond the measured data. In follow-up exams, the most recent prior scan of a patient can provide a highly informative subject-specific context, but practical use is complicated by temporal changes (including pathology progression), misalignment between scans, and protocol drift across acquisitions. In this work, we introduce L-TGVN, a Longitudinal Trust-Guided Variational Network that leverages prior scans as side information to reconstruct the current scan from heavily undersampled measurements. Crucially, L-TGVN constrains the influence of prior scans to be consistent with the acquired measurements. Unlike many existing longitudinal reconstruction methods, it does not require explicit pre-registration between prior and current scans. It further accommodates differences in acquisition protocols across visits (e.g., changes in sequence parameters). We evaluate L-TGVN against matched-capacity baselines, including prior-guided methods and methods that do not use longitudinal priors, and observe consistent improvements in standard quantitative metrics together with better preservation of fine structures at challenging accelerations. Source code is available at github.com/sodicksonlab/L-TGVN.
Arda Atalık, Sumit Chopra, Daniel K. Sodickson
NYU Center for Data Science, NY, USA · Center for Advanced Imaging Innovation and Research (CAI²R), Department of Radiology, NYU Grossman School of Medicine, NY, USA · Courant Institute of Mathematical Sciences, NY, USA +1
Reliable quality control (QC) of magnetic resonance imaging (MRI) is essential for reliable diagnostic neuroimaging, yet standard manual assessment is subjective and time-consuming. MRIQC has established standardized automated extraction of image-quality metrics (IQMs), but its reliance on local computational imaging skills and capacity including high-performance computing, limits its adoption in resource-constrained settings (RCS). We present WebMRIQC (webmriqc.mailab.io), an open-source browser-based platform that wraps the validated MRIQC engine behind a zero-installation web interface. WebMRIQC automates the DICOM-to-BIDS conversion of de-identified MRI scans, executes the unmodified containerized MRIQC pipeline on a shared compute node governed by a fair-share job queue, and returns an interactive in-browser dashboard. The dashboard grounds every IQM in published quality thresholds, benchmarks each scan against the normative distribution of high-resource open datasets, and supports cross-site multicentre implementation of optimized scan protocols in RCS.We describe the system architecture and a validation framework establishing measurement equivalence between WebMRIQC and native MRIQC across thirteen IQMs on the BraTS-Africa and BraTS 2021 datasets. Preliminary results indicate strong agreement for contrast-, signal and noise-based metrics, demonstrating that web-based implementation lowers the barrier to standardized MRI QC and provides a foundation for harmonized, regionally adapted quality benchmarks across RCS imaging sites. The code is publicly available here https://github.com/CAMERA-MRI/WebMRIqc.
Philip Nkwam, Ifeoluwa Oladeji, Sekinat Zurakat-Aderibigbe +9
Department of Radiography, Faculty of Health Professions, College of Medicine, University of Lagos, Nigeria. · Medical Artificial Intelligence Laboratory, Crestview Radiology Ltd., Lagos, Nigeria. · Department of Neurology and Neurosurgery, Montreal Neurological Institute, McGill University, Montreal, Canada. +3