Towards Sustainable Magnetic Resonance Neuro Imaging: Pathways for Energy Optimization and Cost Reduction Strategies

Featured Application

Optimizing Magnetic Resonance Imaging (MRI) eco-efficiently using fast protocols and daily management increases the number of examinations while reducing costs. Among the idle modes, the “Restart” mode cuts consumption by 9–10.6%, saving 3581–4260 kWh annually per MRI, totaling 4759–5661 MWh for the French MRI fleet. AI protocols boost throughput by 36%, cut energy use by 32%, and enable 41 protocols in 12 h versus the standard 30. Optimized protocols on French outpatient MRIs save 7900–8800 kWh per unit annually, totaling 10,500–11,600 MWh and over 500 CO₂ tons, possibly more with energy mix variations.

Abstract

We evaluated the energy consumption of a 3T MRI using a central monitoring system, focusing on hospital energy costs during peak winter months from 2021 to 2023. We analyzed consumption during non-productive phases like end-of-day standby and assessed their impact. For active use, we compared standard and AI-enhanced protocols on phantoms, scheduling high-demand protocols during off-peak hours to benefit from lower energy prices. Standard protocols consumed 3.4 to 15 kWh, while optimized protocols used 2.3 to 10.6 kWh, reducing consumption by 32% on average. Savings per scan ranged from EUR 0.03 to EUR 3.7. The electrical consumption of a brain MRI protocol is equivalent to that of 3–4 knee protocols or 2–3 lumbar spine protocols. Using AI-optimized protocols and management, 41 protocols can be completed in 12 h, up from 30, reducing daily costs by EUR 2.38 to EUR 29.18. Annually, AI-optimized protocols could save 7900 to 8800 kWh per MRI unit, totaling 10,500 to 11,600 MWh across France’s MRI fleet, equivalent to the yearly consumption of about 4700 to 5300 people. Optimizing MRI resource use can expand patient access while significantly reducing the associated energy footprint. These findings support the implementation of more sustainable practices in medical imaging without compromising care quality.

1. Introduction

Healthcare sector’s carbon footprint is a significant environmental concern, with regional variations: 4.6% in France, 7.6% in the United States, and 5.1% in Australia [1]. Diagnostic imaging, particularly Computed Tomography (CT) and Magnetic Resonance Imaging (MRI), is a major contributor due to high energy consumption and waste generation [2,3,4]. This has driven the rise of green radiology, highlighting growing environmental awareness [5] and decarbonization policies [6].

Despite these challenges, MRI remains essential for non-invasive diagnostics. In France, there are 15.8 MRI units per million population, compared to 40.1 in the United States [7]. However, rising electricity costs pose sustainability challenges, emphasizing the need for energy-efficient strategies.

A significant portion of an MRI machine’s energy consumption occurs during idle periods, underscoring the importance of initially selecting energy-efficient models, as retrofitting is often not feasible [8]. Simple measures, such as turning off workstations and monitors, can reduce unnecessary energy consumption [9,10]. For instance, turning off scanners overnight and on Sundays can save approximately 14,000 kWh/year [11].

However, MRI scanners equipped with superconducting magnets cannot be completely shut down. In response, manufacturers offer various standby modes [12]. Carbon footprint reduction also extends to data preprocessing, with strategies such as optimizing workflows like functional MRI preprocessing tools [13].

Advances in artificial intelligence and deep learning, including technologies such as Siemens’ Deep Resolve [14], General Electric’s Air Recon DL (deep learning) [15], Canon’s AICE (Advanced intelligent Clear-IQ Engine) [16], and Philips’ SmartSpeed [17], have notably advanced MRI capabilities and significantly reduced scan times.

Previous research has mainly focused on image quality and clinical relevance, particularly in musculoskeletal imaging [18]. However, the significant reduction in scan times not only benefits non-cooperative patients (e.g., pediatric patients, those in pain) [19] by minimizing sedation needs and enhancing comfort but also improves the overall accessibility and efficiency of imaging services. Moreover, the effect of these advancements on power consumption has yet to be fully explored.

This study aims to optimize MRI energy usage by analyzing non-productive energy modes, refining operational workflows, and assessing the financial implications of increased productivity through a theoretical scheduling model that accounts for electricity pricing.

2. Materials and Methods

This single-center study uses a descriptive methodology to analyze changes in power consumption during idle standby and to assess the impact of protocol optimization.

By employing phantom models, our research bypassed the need for ethical approval and consent, as it did not involve human subjects.

2.1. MRI Machine

Our MRI system was a Philips MR7700 3T (Best, The Netherlands) installed in 2022. It has a maximum gradient strength (Gmax) of 65 mT/m and a slew rate (Rs) of 220 T/m/s and runs on software version 11.1. This MRI system is primarily used for neuroradiological studies, which accounts for approximately 70% of the scans, with the majority of which are focused on brain imaging. The possibilities for extinction and different modes of use provided by the manufacturer are listed in

Table 1. Definitions and procedures for MRI scanner modes.

System State	Definition
Productive scan mode	The MRI unit is actively engaged in scanning, producing images during ongoing examinations
Unproductive mode	When not actively generating images, the MRI unit may be in ready-to-scan, update, or idle modes
Ready to scan	When not capturing data between sequences or patients, the system enables immediate scanning on demand
Update	Performing software or hardware updates on the system
Idle	Power save mode minimizes power consumption when in off or low-power states. Upon operator console logout, settings offer two options provided by the manufacturer: - Restart: Exit the application software - Shutdown: Power off the computer, then restart it in the technical room

.

2.2. Standard and Optimized Protocols

The study focused on neuroradiology protocols including neurovascular imaging, brain tumor assessment, evaluation of neurodegenerative diseases, epilepsy, pituitary gland lesions and multiple sclerosis, spinal cord assessment, and musculoskeletal protocols, such as knee and lumbar spine assessment. Philips SmartSpeed AI-based software has been integrated into our MRI, merging Compressed SENSE (CS)—which combines compressed sensing with parallel imaging—and deep-learning reconstruction (CS AI) on raw k-space data at different undersampling rates. This approach reduces k-space sampling, markedly accelerating sequence times.

In collaboration with MRI clinical scientists, we restructured protocol sequences to include AI-accelerated imaging and multiband diffusion. Sequences unsuitable for optimization remained standard. Protocol parameters, including acceleration factors, are detailed in the Supplementary Material.

2.3. Phantoms

The ACR-MRI phantom (Orion J15935 Jm Specialty Parts, San Diego, CA, USA) and the knee coil phantom (Invivo Corporation, Gainesville, FL, USA) were used with 32-channel head coils. For the spinal cord and lumbar spine imaging, a dual phantom bottle (3000 cc and 5000 cc, Spectrasyn 4) was used in combination with a torso coil with 32 channels (Philips Medical Systems, Best, The Netherlands).

2.4. Power Measurements

Energy consumption was tracked via the DIRIS Digiware D-50 (Socomec SAS, 67235 Benfeld, France), excluding auxiliary cooling, with an accuracy of 0.2%. Data were logged in watts at 1 min intervals. To isolate the energy impact of each MRI sequence, 2 min breaks were used between sequences. Hourly consumption was validated using PACS timestamps.

2.5. Cost Data

We analyzed our hospital’s electricity costs from 2021 to 2023, focusing on winter periods provided by Électricité de France. Winter was chosen due to greater variations and the presence of peak hours. The time slots were off-peak (10 p.m. to 6 a.m.), peak (9 a.m. to 11 a.m. and 6 p.m. to 8 p.m.), and full hours (the rest of the time). For off-peak hours, the prices were 42.57 EUR/MWh in 2021, 54.10 EUR/MWh in 2022, and 20.34 EUR/MWh in 2023. For peak hours, the prices were 79.12 EUR/MWh in 2021, 85.22 EUR/MWh in 2022, and 835.75 EUR/MWh in 2023. For full hours, the prices were 65.75 EUR/MWh in 2021, 75.23 EUR/MWh in 2022, and 835.75 EUR/MWh in 2023.

2.6. Consumption Calculations and Statistical Analysis

Energy measurements were converted from watts to kilowatts (kW) for standardization and analyzed to compare energy use between standard and optimized MRI protocols, focusing on total and per-sequence consumption. Sequence duration adjustments allowed for accurate per-sequence energy calculations. Descriptive statistics in Excel were used for quantitative analysis. Data visualization was conducted in PyCharm IDE (version 2024.1) using Python libraries: NumPy, Pandas, and Matplotlib.

Figure 1 summarizes the tests conducted.

Determining the most realistic standard sequence:

Figure 2 shows an example of the daily electricity consumption profile of our MRI unit along with the daily hourly price variations observed during the winter season.

Based on an 8 a.m. to 8 p.m. schedule, with peak periods from 9–11 a.m. and 6–8 p.m., several models were developed. The first optimization scheduled the least energy-intensive protocols during peak hours and more energy-intensive ones outside these times.

The second optimization involved implementing these optimized protocols within the same operational window. Key constraints included mandatory 7 min, 1.5 kW breaks between protocols. Further constraints aimed to reduce antenna adjustments and implement body region-specific workflows to streamline technician operations. To optimize staff availability, protocols not involving contrast injection were scheduled at the beginning and end of the day.

3. Results

MRI idle mode: power consumption and savings.

Table 2. Parameters and price of idle mode (6:30 p.m. to 8:50 a.m.).

	Forgotten	Shutdown	Restart
Active power parameters (kW)
Minimum–maximum	[6.78–13.65]	[7.13–9.94]	[6.69–9.75]
Mean	7.62	7.49	6.81
Median	6.87	7.21	6.76
Standard deviation	2.06	0.83	0.38
Variance	4.23	0.68	0.14
Interquartile range	0.03	0.03	0.04
Mode	6.87	7.20	6.78
Sum	109.50	107.64	97.83
Total winter price (night standby) (EUR)
2021	6.2	5.4	5.3
2022	22.4	18.1	14.1
2023	46.5	43.2	37.5

presents the idle mode parameters and associated costs, while Figure 3 illustrates the power consumption profiles for each of the modes identified in our analysis: “Restart”, “Shutdown”, and “Forgotten”.

In the “Shutdown” mode, the system demonstrated excessive power usage for 92 min, averaging 9.9 kW. This level eventually dropped to a standby consumption of 7.2 kW, with no software updates being initiated.

The “Restart” mode caused a brief spike in power consumption to 9.4 kW lasting for 5 min, after which it stabilized at an average of 6.8 kW. During this mode, a scheduled update at 3:05 a.m. resulted in an average energy expenditure of 9.6 kW over a period of 10 min. The “Forgotten” mode was characterized by the system maintaining excessive power consumption for 1.5 h, peaking at 13.2 kW in “Ready-to-Scan” mode before transitioning to a standby state with an average power draw of 6.9 kW. A notable event during this mode involved an update at 3:05 a.m., which consumed an average of 9.7 kWh over 10 min. The “Shutdown” mode consumed a total of 107.6 kW. The “Restart” mode had a slightly lower consumption at 97.8 kW. The “Forgotten” mode was the most energy-intensive, with a total consumption of 109.5 kW.

Over the period, the annual costs associated with each idle mode varied due to fluctuations in electricity prices. The “Shutdown” mode incurred costs ranging from EUR 5.4 to EUR 43.2 per night according to annual prices. The “Restart” mode costs varied from EUR 5.3 to EUR 37.5, and in the “Forgotten” mode, the costs were higher, ranging from EUR 6.2 to EUR 46.5.

MRI scan mode: power consumption and savings.

Figure 4 displays the energy consumption of MRI protocols, comparing standard and optimized protocols. The energy consumption for standard protocols varied between 3.4 kW and 15 kWh per protocol. With the integration of AI-driven SmartSpeed acceleration and multiband diffusion, energy consumption for optimized protocols ranged from 2.3 kW to 10.6 kW per protocol. Specific energy consumption was lowest for musculoskeletal studies (knees and lumbar spine) and highest for neurodegenerative, epilepsy, and brain tumor protocols. Supplementary Data provide detailed contributions from each sequence.

The cost calculations for each year are presented by protocol, with the details provided in

Table 3. Energy cost comparison standard versus optimized protocols.

	Duration (s)	Consumption (kW/Protocol)	Off-Peak Hours Winter Price (EUR/Protocol)			Full Hours Winter Price (EUR/Protocol)			Peak Hours Winter Price (EUR/Protocol)
Years			2021	2022	2023	2021	2022	2023	2021	2022	2023
Standard protocols
Knees	420	3.4	0.15	0.19	0.07	0.22	0.29	2.86	0.27	0.29	2.86
Lumbar spine	630	4.9	0.21	0.27	0.10	0.32	0.42	4.10	0.39	0.42	4.10
Pituitary gland	708	5.6	0.24	0.31	0.11	0.37	0.48	4.71	0.45	0.48	4.71
Spinal cord	1029	8.24	0.35	0.45	0.17	0.54	0.70	6.89	0.65	0.70	6.89
Neurovascular	956	9.8	0.42	0.53	0.20	0.64	0.83	8.18	0.77	0.83	8.18
Multiple sclerosis	1046	10.7	0.45	0.58	0.22	0.70	0.91	8.91	0.84	0.91	8.91
Neurovascular and Supra trunks	1298	11.98	0.51	0.65	0.24	0.79	1.02	10.01	0.95	1.02	10.01
Brain tumor	1300	13.92	0.59	0.75	0.28	0.92	1.19	11.63	1.10	1.19	11.63
Epilepsy	1596	14.68	0.62	0.79	0.30	0.97	1.25	12.27	1.16	1.25	12.27
Neurodegenerative	1567	15.04	0.64	0.81	0.31	0.99	1.28	12.57	1.19	1.28	12.57
Optimized protocols
Knees	304	2.29	0.10	0.12	0.05	0.15	0.20	1.91	0.18	0.20	1.91
Lumbar spine	436	3.17	0.13	0.17	0.06	0.21	0.27	2.65	0.25	0.27	2.65
Pituitary gland	518	4.25	0.18	0.23	0.09	0.28	0.36	3.55	0.34	0.36	3.55
Spinal cord	580	4.71	0.20	0.25	0.10	0.31	0.40	3.94	0.37	0.40	3.94
Neurovascular	621	5.77	0.25	0.31	0.12	0.38	0.49	4.82	0.46	0.49	4.82
Multiple sclerosis	769	7.38	0.31	0.40	0.15	0.49	0.63	6.17	0.58	0.63	6.17
Neurovascular and supra trunks	963	7.96	0.34	0.43	0.16	0.52	0.68	6.65	0.63	0.68	6.65
Brain tumor	1021	10.31	0.44	0.56	0.21	0.68	0.88	8.62	0.82	0.88	8.62
Epilepsy	1200	10.55	0.45	0.57	0.21	0.69	0.90	8.82	0.83	0.90	8.82
Neurodegenerative	1150	10.61	0.45	0.57	0.22	0.70	0.90	8.87	0.84	0.90	8.87

. In the operational window encompassing peak and full hours, the cost of optimized protocols in 2021 was between EUR 0.15 and EUR 0.84 per scan, compared to EUR 0.22 and EUR 1.19 per scan for standard protocols. In 2022, costs for optimized protocols ranged from EUR 0.17 to EUR 0.90, while those for standard protocols were between EUR 0.42 and EUR 1.28 per scan. In 2023, standard protocols were priced between EUR 2.86 and EUR 12.57 per scan, whereas optimized protocols cost between EUR 1.94 and EUR 8.87 in electricity expenses.

Dual optimization: utilizing optimized protocols and MRI exam scheduling with winter electricity pricing.

Figure 5 shows the MRI exam timeline, and

Table 4. MRI exam program consumption and pricing over a theoretical day.

Years	2021		2022		2023
Protocols	Standard	Optimized	Standard	Optimized	Standard	Optimized
Protocols count	30	41	30	41	30	41
Daily Consumption (kW)
Protocols	285.95	234.53	285.95	234.53	285.95	234.53
Break	45.00	61.50	45.00	61.50	45.00	61.50
Total	330.95	296.03	330.95	296.03	330.95	296.03
Daily cost (EUR)
Protocols	19.46	15.96	22.00	18.05	238.98	196.01
Break	3.08	4.18	3.48	4.73	37.61	51.40
Total	22.53	20.15	25.48	22.78	276.59	247.41

lists energy consumption and costs.

The first optimization was to identify a sequence for the standard protocol, taking into account accommodates constraints related to changing antennas, hourly pricing, the need for breaks between protocols, and the absence of injections at the beginning and end of the day. Subsequently, within the same time window and the same constraints, the incorporation of SmartSpeed AI into our optimized imaging protocols enabled the completion of an additional five neurovascular and six knee protocols.

The total energy consumption for the MRI protocols was measured at 296.03 kW for 41 optimized protocols and 330.95 kW for 30 standard protocols. The operational window for optimized protocols includes more standby periods, amounting to 61.50 kW, compared to 45 kW for breaks in standard protocols. In the daily operational window, the cost analysis for 2021 and 2022 indicates that the total cost of running standard protocol sequences, including breaks, was EUR 22.53 and EUR 25.48, respectively. For optimized protocols, the corresponding costs were EUR 20.15 and EUR 22.78.

In 2023, the daily cost of running standard protocols was EUR 276.6, compared to EUR 247.4 for optimized protocols, resulting in a cost difference of EUR 29.18.

4. Discussion

Given the critical importance of MRI in healthcare and its significant consumption of resources, especially in the context of climate change and fluctuating costs for electricity and materials, it is imperative to refine practices to maintain healthcare quality. This optimization should target both productive (scan mode) and non-productive (idle time) operational phases.

An examination of two manufacturer-provided idle modes, “Shutdown” and “Restart”, showed that “Restart” was more energy-efficient, consuming 97.83 kWh per night compared to 107.6 kWh for “Shutdown”. Additionally, the analysis revealed that a “Forgotten” mode, which is inadvertently left unused by users, was the most energy-hungry, consuming 109.50 kWh per night. Achieving optimization requires staff awareness, especially during the machine’s idle phases. Failure to select energy-saving modes leads to unnecessary overconsumption. Although the energy loss may be small, it is still worth reducing.

An assessment of 10 neuroradiological and musculoskeletal protocols examined their energy and financial impact, with energy consumption ranging from 3.4 to 15 kWh.

The use of Philips SmartSpeed Intelligence AI-based software and multi-band diffusion provided faster, optimized alternatives with lower energy use, ranging from 2.3 to 10.6 kWh per protocol, resulting in savings of EUR 0.03 to 3.7 per protocol.

In order to substantiate these results, a set of protocols was developed, shaped by price fluctuations and taking into account various daily constraints.

From 8 a.m. to 8 p.m., optimized protocols allowed for the addition of five neurovascular and six knee protocols, and reducing daily energy consumption from 330.95 kWh to 296.03 kWh. Improved protocols and strategies increased productivity, enabling more MRI studies with lower energy use, yielding kWh savings and financial benefits.

Only a few studies have investigated the energy usage of MRI machines, focusing on annual evaluations of consumption factors, carbon footprint, and methods for reduction. Our design aims to implement strategies for this purpose.

The use of artificial intelligence to speed up image acquisition is increasing rapidly. However, the impact of AI on energy consumption has not yet been thoroughly reached.

On our scale, on working days, the implementation of AI-optimized protocols could save approximately 7927 kW/year per MRI. At the scale of the French fleet, comprising 1329 MRI units [20], annual savings could potentially reach 10,500–11,600 MWh/year. This amount of electricity is equivalent to the annual consumption of approximately 4740–5267 people in France. This consumption results in about 526.81 to 582.33 tons of CO₂ in France, which has relatively low-carbon production. The same consumption would result in carbon emissions of more than 4256.62 tons in the United States, and more than 6825 tons in China, where carbon emission rates vary significantly. Additionally, potential savings could be greater when considering the higher MRI density in these countries [21]. Figure 6 presents the annual extrapolations of potential energy, time, and CO₂ equivalent savings. For instance, Ibrahim et al. reported that each cardiac MRI booking slot saves 3.3 kWh of energy and avoids 1.4 kg of CO₂. In France, this energy saving would result in a reduction of 165 g of CO₂ [22].

Furthermore, there is limited data on the overall electricity consumption of MRI machines and their standby modes. Their survey highlighted the disparities among MRI models, which varied by manufacturers, magnetic field, and year of installation.

Non-productive consumption can vary. For example, Knott et al. [23] found that two 1.5T MRI machines (Siemens Advento 2013 and Magnetom Aera 2013) had the lowest power mode ranging from 13.9 to 14.5 kW. In contrast, the lowest power mode of our Philips 3T MRI (2022) was reduced to 6.76 kWh. Marine Cauz and Sven Rossier [24] recommended categorizing MRI machines by power consumption across different modes, based on data from 13 MRI units from GE Healthcare, Philips Healthcare, and Siemens Healthineers. For instance, our MRI was rated “Efficient” in idle mode (X ≤ 8 kW) and “Standard” in ready-to-scan mode (11 kW < X < 17 kW). Regulatory authorities should implement a standardized energy usage scale for MRI machines, similar to household appliance labels.

In 2023, electricity prices increased due to global sociopolitical factors, such as Europe’s post-COVID-19 economic recovery and the Ukraine conflict. The Lazard report [25] suggests that future electricity costs will be influenced by the sources of energy production. It points to a preference for more economical energy sources such as solar and onshore wind over nuclear and fossil fuels, which are extensively utilized in Europe.

The shift to variable energy sources is driving demand response, which combines third-party automation with consumer action, as we have implemented. This approach aligns with collective efforts toward carbon neutrality, as highlighted in the report by Réseau de Transport d’Électricité (RTE), France’s electricity transmission system operator, which indicates a requirement for 15 GW of flexibility by 2050 [26].

There were certain limitations to this study. Firstly, the data were directly linked to the MRI machine, excluding ancillary electricity consumption such as that of the cooling system. However, cooling and building HVAC (heating, ventilation, and air conditioning) loads outside of the MRI system itself often rise in summer (due to air conditioning) and can similarly fluctuate in winter (due to heating). Because our energy-measurement approach isolated the MRI scanner’s direct consumption, such facility-level seasonal variations would not be considered. This is noteworthy because previous research has shown that such consumption represents 44.5% of the total usage [12].

The fate of this thermal energy, which is integrated into the hospital’s central cooling network, remains unresolved, and no current initiatives for heat recycling have been explored.

Secondly, the cost analysis focused on a winter contract from one supplier, without accounting for global or seasonal variations in energy costs. This limitation may restrict the relevance and accuracy of the analysis in different countries and seasons. Our prices for 2023 are likely to have been adjusted, as European Network of Transmission System Operators for Electricity (E-NTSOE) [27] data from 2022 indicated a substantial increase in electricity costs to an average of 276 EUR/MWh. Our study focused on winter months primarily because electricity tariffs and demand peaks are most pronounced during this period, making it the season with the clearest cost differentials and the greatest potential for operational savings. During summer, while the inherent kWh consumption of the MRI would remain effectively the same per protocol, the pricing structure may be less volatile in France, thus altering the cost-saving potential from scheduling. However, in some regions, warmer temperatures can elevate a facility’s overall HVAC loads, which could influence total site energy usage. In the United States, electricity pricing differs substantially from France/Europe, both in absolute cost and in structure. Commercial electricity rates in the US commonly average between 0.10 and 0.15 USD/kWh but vary by seasonal demand patterns [28]. Furthermore, it should be noted that electricity tariffs can fluctuate significantly from year to year.

Another limitation is the use of phantoms without considering the Specific Absorption Rate (SAR) deposited, which varies for each patient based on factors such as body size.

Finally, the findings of our study were derived from a single imaging installation and technology type, specifically focusing on one MRI machine and its accompanying AI from Philips. For future research, it would be beneficial to investigate machines from other manufacturers to broaden the understanding and applicability of our findings. Our acceleration factors were set in advance in collaboration with engineers. Since energy consumption is linked to operational time, identifying the optimal trade-off between scan duration reduction and image quality could lead to greater savings.

The use of parallel imaging is also a way to reduce the sequence time and consequently the electrical consumption of an examination: Thus, by moving from a 16- to a 64-channel coil, we will be able to increase the acceleration factor and reduce the number of phase encodings, while maintaining a good Signal/Noise.

Future research could evaluate AI-driven imaging systems from different vendors, assessing image quality to confirm their consistency, reliability, and any variations between technologies; additionally, incorporating other intelligent solutions that work in post-acquisition image processing to enhance sharpness and reduce noise and ensure compatibility with any make or model of MRI scanner, such as Subtle MR [29].

There is an opportunity for further optimization through AI programming, with the goal of improving the overall efficiency in examination scheduling and execution processes.

5. Conclusions

Our analysis integrates energy management insights with cost-saving measures, emphasizing eco-friendly approaches to reduce electricity consumption that yield financial and sustainable healthcare benefits. By identifying energy-saving opportunities, we promote scalable, sustainable practices across healthcare facilities to maintain patient care amidst energy challenges. Our findings advocate for a strategic framework to adopt sustainable practices across the healthcare sector, emphasizing the dual benefits of environmental responsibility and economic efficiency. They underscore the critical role of resource optimization in sustaining patient care levels, even in the face of potential energy crises and economic challenges.

This approach encourages industry-wide implementation of sustainable practices, ensuring resilience and sustainability of healthcare.