Enhancing Access to Cancer Diagnostics with Drone Delivery of PET Isotopes: The Significance of Weather and Clinical Workflows

Highlights

What are the main findings?

• Simulations using real-world data showed that long-range drone transport substantially reduces delivery times for PET radioisotopes, improving isotope preservation and enabling significant annual cost savings. However, actual flights have yet to be realized.
• Weather is a critical operational constraint. In order to reap the potential benefits, cargo drones must be designed with the ability to travel faster than 200 km/h and be wind and rain-resistant.

What are the implications of the main findings?

• Drone-enabled PET logistics should be developed to increase capacity and regional access.
• Adoption of drone transport should prompt reconsideration of current practices of PET production and clinical scanning routines.

Abstract

The short half-life of positron emission tomography (PET) radioisotopes makes transport time a critical factor in medical logistics. While drones have demonstrated advantages in short-range medical deliveries, the feasibility and benefits of long-distance drone transport remain largely unexplored. In a comparative simulation-based modelling framework, this study explores whether long-range drone transport (117–376 km) can improve delivery performance of fluorodeoxyglucose-18 ([¹⁸F]FDG) PET isotopes compared with two existing ground-only routes (146 km and 348 km) and two combined car–airplane routes (532 km and 546 km). Simulated transport times, radioactive decay losses, and economic implications were estimated using drone speeds of 150, 200, and 250 km/h. Hourly weather data from 2023–2024 were incorporated to model flight feasibility and weather-related no-fly conditions. Time savings were translated into preserved radioactive activity and analyzed together with break-even transport costs. A drone speed of 150 km/h provided limited benefit, whereas speeds of 200–250 km/h preserved activity corresponding to a reduction from the current total use of 118 GBq to 72 and 65 GBq, respectively. Weather constraints reduced feasible winter flights by up to 30%. Estimated break-even drone costs ranged from EUR 3–18/km and increased to EUR 14–20/km when accounting for preserved isotopes, corresponding to annual economic gains of EUR 1.0–1.7 million. These results suggest that long-range drone transport could reduce isotope losses and improve diagnostic capacity, although feasibility depends on drone costs, weather resilience, and integration into clinical logistics systems.

1. Introduction

Drones are increasingly tested and used in the healthcare field for emergency response, delivery of blood products, vaccines, and medicines, and assistance for elderly populations [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20]. During the COVID-19 pandemic, they maintained essential deliveries during lockdowns [17,21,22,23]. More advanced proposals include their use in organ transport, though clinical evidence remains limited [24,25].

Despite their potential, most drone healthcare projects rely on public or pilot funding and remain economically unproven. Prior studies largely focus on short-range, last-mile deliveries with small payloads, limiting claims of a logistics revolution.

Context is critical. In low-infrastructure regions, drones can provide essential access [7]. Similar needs exist in high-income countries with dispersed populations [11,26,27,28]. However, widespread adoption hinges on sustainable payment models and users’ willingness to pay [29,30]. Technological progress is enabling longer-range drones with greater payloads. However, sustainable business models must demonstrate measurable healthcare value. Within value-based care, higher drone costs must be justified by better outcomes or service quality. Drones may also inspire broader system change by shifting how logistics and services are organized.

The most commonly encountered aerodynamic effect of wind on small UAVs is reduced ground speed under headwind conditions, which directly constrains operational range and can compromise mission feasibility. Reliability is key, however. Predictable service is equally as critical as speed, particularly in the field of healthcare. Weather, wind, rain, and temperature extremes can significantly restrict drone operations [31,32,33,34,35,36]. Achieving dependable services requires resilient drone design and contingency planning. Important lessons can be derived from the study of Oakey et al. [36], documenting differing required wind tolerances across differing geographical locations ranging from 14 m/s to more than 23 m/s, and the studies of Bauranov et al. [37] and Choi et al. [38], in which the authors illustrated the impact of object avoidance from differing topography, buildings, and infrastructures at take-off and landing sites.

Beyond wind-induced performance penalties, in-flight icing also represents a critical safety risk for small fixed-wing drones operating in cold and humid environments [39,40,41]. Ice accretion substantially alters the local flow field, degrading lift, increasing drag, and impairing propulsive efficiency and control authority. Empirical evidence from a previous study conducted in Oslo indicates that meteorological conditions conducive to icing may occur on approximately 5–10% of winter days [39]. Consequently, routine winter operations in Norway and comparable climatic regions should be assumed to require ice protection systems (IPS), encompassing propellers, lifting surfaces, airspeed sensors, and potentially control surfaces. IPS implementation introduces substantial mass and energy penalties, which in turn reduce payload capacity and effective range, creating a trade-off between operational safety and transport efficiency. Despite the aforementioned constraints, research on new solutions is surging [32,40,41]. While route planning and mission timing can be adapted to minimize exposure to icing-prone atmospheric layers, the effectiveness of such operational mitigations is contingent on the accuracy and spatial resolution of icing forecasts. Continued advances in high-resolution meteorological modelling and onboard sensing are therefore pivotal to enabling reliable, year-round drone operations in cold-climate healthcare logistics.

May Drones Improve the Transport of PET Radioisotopes?

Positron emission tomography (PET) is a high-cost tool in medical imaging with increasing applications in oncology, cardiology, rheumatology, neuropsychiatry, and other diseases/conditions [42]. The principle of PET (positron emission tomography) imaging is that biologically active radiotracers emit positrons, which are detected by the PET scanner. A commonly used tracer is 18F-fluorodeoxyglucose (FDG), in which a glucose analogue is labelled with the radioactive isotope fluorine-18 (¹⁸F), which has a physical half-life of approximately 110 min. This isotope is used in the current study; however, multiple PET tracers and isotopes exist, all with different costs and half-lives. As many cancer cells exhibit increased glucose metabolism compared to healthy tissues (the Warburg effect), they accumulate FDG at a higher rate, resulting in areas of increased signal intensity (“hot spots”) on PET images, enabling the detection of primary tumors and metastases that may not yet be visible (or findings that are inconclusive) on CT or MRI. PET images are combined with CT or MRI to provide precise anatomical localization of metabolically active lesions that may appear morphologically normal on structural imaging alone. The same metabolic imaging principle is applied in the evaluation of infection, inflammation, neurodegenerative disorders, and other disease processes [43,44].

PET radioisotopes have a fixed physical half-life, and the time between production and delivery to PET facilities impacts the amount of activity available for clinical use. The interval between production and delivery to PET facilities is therefore critically important.

To ensure clinical value and maintain adequate image quality, the administered activity must remain above a certain threshold. While extending the acquisition time can partly compensate for reduced activity, there is a limit below which diagnostic information becomes insufficient. Consequently, PET radioisotopes present unique supply chain challenges compared to standard medical supplies.

The development of cost-effective digital PET scanners represents a significant advancement in medical diagnostics. Although the accelerating technological advancements reduce the cost of digital PET scanners, large disparities in access to imaging persist in low-income countries and across geographically isolated regions [45,46,47,48,49,50,51].

Radioisotope production is costly, and their short half-lives in relation to transport from the cyclotron to imaging centers present logistical constraints. For PET centers without local cyclotrons, service quality hinges on reliable and timely transport. Road congestion or airline delays exacerbate decay. Although some improvements have been achieved through service chain optimization [52], challenges remain.

While clinical PET imaging per se is not strictly time-sensitive, radiotracer decay begins immediately after production. Most scans occur during the daytime, requiring morning delivery, which restricts scheduling flexibility and complicates contingency plans for weather-related drone cancellations.

Drones suitable for long-distance PET transport must combine vertical take-off and landing (VTOL) and fixed-wing flight to achieve the desired speed. They must also ensure temperature control (typically 15–25 °C) to maintain radiopharmaceutical integrity. Payloads must therefore accommodate heavy shielding (15–30 kg), often needed for ¹⁸F-FDG.

A number of research groups have examined the economic and sustainability aspects of drone transport in both healthcare and parcel logistics [17,33,53,54,55,56,57,58,59,60,61,62,63,64], with most focusing on how drones can reduce delivery times by avoiding traffic congestion through lightweight, short-range, last-mile deliveries. Many of these studies were conducted in immature operational settings, i.e., relying on manual piloting and temporary landing infrastructure.

Economic studies are methodologically inconsistent and incomplete, as life cycle assessments (LCAs) of drones, particularly regarding battery manufacturing, composite materials, end-of-life disposal, and costs for regulatory solutions, have not been included [52,53,60], and a disconnect remains between technical feasibility and financial sustainability. While technical feasibility is increasingly demonstrated, economic viability remains a dominant bottleneck. Cornell et al. [63] demonstrated that labor costs account for up to 95% of delivery costs under current regulatory regimes (one operator per drone + visual observer), making drone delivery non-competitive with conventional vehicles. Grote et al. [62] estimated that drone costs would need to drop to ~19% of current levels to become competitive in multimodal healthcare logistics in the UK, an assumption that may remain unrealistic even under future automation scenarios. In contrast, Haidari et al. [17] determined cost savings in LMIC settings, suggesting that contextual factors (road quality, distance, labor cost, and geography) critically mediate economic performance, thus revealing a structural contradiction in the literature; drones may appear cost-effective in low-infrastructure, high-friction environments, but they are economically inferior in high-income, well-developed logistics systems. How will partial automation (e.g., remote supervision of multiple drones) alter the cost structure under realistic regulatory timelines?

There remains limited causal evidence linking drones to improved health outcomes. Although many study authors report improved response times and potential survival benefits [53], robust causal evidence of clinical impact remains sparse. Campbell et al. [60] explicitly note the lack of systematic evaluation of patient health outcomes and that most studies emphasize surrogate metrics (delivery time, distance, cost, and feasibility) rather than clinical endpoints (mortality, morbidity, transfusion delays avoided, and vaccine coverage improvements). Gauba et al. [55] demonstrated the technical feasibility of blood transport without degradation but did not quantify system-level effects on clinical outcomes (e.g., reductions in time-to-transfusion or mortality in trauma settings).

Simplified modelling assumptions are used in many studies. Under full life-cycle accounting, do healthcare delivery drones outperform existing solutions if battery lifespan, regulatory systems, and infrastructures are included in the cost assessments? In multiple reviews [53,60,64], the authors identify regulation, privacy, safety, and public acceptance as key barriers; however, these factors are typically treated as exogenous constraints, not endogenous design variables, although regulations directly shape cost structures [63], making governance a first-order determinant of economic feasibility.

Another unresolved issue is related to system-level integration research. The authors of most studies related to the transport of blood, defibrillators, medicine, and vaccines have paid little attention to interoperability with hospital integration into existing logistics workflows and workforce reconfiguration and institutional change management. Translation into sustainable institutional adoption has received little attention to date.

The existing literature indicates that economic assessments of drone solutions frequently overlook key costs by excluding critical components such as public infrastructure expenditures, airspace surveillance and traffic-management requirements, and broader environmental impacts. The authors of most drone logistics studies in health services focus on short-range and last-mile purposes, with real-world data on large-scale, long-range operations being scarce. Extrapolating from short-range parcel models is inappropriate. For example, costs for a 10-mile, 1 kg drone delivery of ~USD 13 cannot be scaled linearly to 100 miles or 50 kg [64]. In this study, we simulate data for scenarios requiring drones with payload capacities exceeding 50 kg, flight ranges of several hundred kilometers, and long-term regular flights. Break-even analyses were used to determine the km-based cost thresholds at which drone transport of PET isotopes becomes economically viable compared to existing car/airplane logistics.

In our analysis, we also treat activity preserved through reduced radioactive decay as a source of economic value. Furthermore, international regulations, particularly IAEA guidelines, govern the handling of radioactive materials and impose stringent requirements for packaging, labelling, and documentation, thereby adding to logistical complexity [65]. This factor could not be modelled in this study.

Decay and clinical value depend on isotope half-life. For example, ¹¹C (t_½ = 20 min) requires near-instant delivery, whereas ¹⁸F (t½ = 109 min) loses ~50% activity after 109 min, making efficient transport of [¹⁸F] FDG particularly valuable.

This study was organized with a modular design, whereby transport feasibility impacts economic viability and increases the availability of PET isotope capacity (Figure 1).

Research questions:

• How much PET activity of ¹⁸FDG can be preserved during drone transport compared with current car/air transport across varying distances and drone speeds?
• How do varying weather conditions affect drone reliability?
• Can clinical routines be optimized to increase the value of drone transport?

2. Materials and Methods

The Norwegian Medical Cyclotron Centre (NMCC) (https://syklotronsenteret.no/en/, accessed on 12 February 2026) delivers PET radioisotopes to several regions in Norway and to all hospitals in the Health Region South-East with PET scanners. This coverage constitutes approximately 50% of Norwegian hospitals and more than 60% of the Norwegian population, in addition to Stavanger University Hospital in Health Region West and Aalesund Central Hospital in the Central Norway Regional Health Authority, located in central Norway.

2.1. Models Used in This Study

Drone solutions in four existing models (Figure 2) were compared:

• a: Car transport to Central Hospital Kristiansand (KRS) (348 km road, 281 km air).
• b: Car transport to Elverum Hospital (ELV) (146 km road, 117 km air).
• c: Car + plane transport to Stavanger University Hospital (SVG) (546 km road, 302 km air).
• d: Car + plane transport to Central Hospital Aalesund (AES) (532 km road, 376 km air).

Current departures occur at 5:15 a.m. to align with clinic schedules. Cancellations in ground transport due to snow/ice are rare (recorded once in 2023 and 2024) and were not modelled, potentially biasing the results, particularly in winter, with ground transport favored.

Drone solutions versus current transport routes match existing destination times with flights starting at 6 a.m. Haversine distances were used to estimate direct drone flight lengths, modelling drone speeds of 150, 200, and 250 km/h. Theoretical transport times were calculated as distance divided by drone speed. It will be necessary for drones developed in the future to comply with strict rules for the temperature regulation of cargo, as radiopharmaceuticals must be stored during transport within a defined temperature interval, with that for ¹⁸FDG being 15–25 °C. In a previous study conducted at the OUS hospital [66], take-off from the hospital depended significantly on local wind conditions and wind directions, in addition to the flight direction out of Oslo (Figure 3). In a previous unpublished pilot study at OUS, times for landing and take-off varied between 2 and 8 min. Therefore, an additional ten minutes was employed to compensate for the take-off and landing of all flights. No similar data for the receiving hospitals were available. However, all are located in lowlands without mountains and no adjacent high buildings; thus, a similar compensation strategy was applied.

The existing costs applied in the study are procurement costs to external transport providers, thereby absorbing total transport costs for the NMCC. Drawing on data from the NMCC and the receiving hospitals, in our analysis, it is assumed that existing clinical logistics components, such as packaging, radiopharmaceutical release, loading, receiving procedures, and handovers, could be transferred to the corresponding process for drones, as the landing sites will be located close to the current cargo central. Accordingly, the simulated drone transport times were compared to the ground transport time, which is currently measured from car departure to arrival.

2.2. Weather and Drone Flight Viability

To assess flight times and flight feasibility during varying weather conditions, hourly historical weather data from the Norwegian Meteorological Institute [67], covering the complete years 2023 and 2024, were used. Data were measured from ground weather stations located approximately every 50 km along the differing drone routings as a proxy for flight weather conditions. Parameters included wind speed, wind gusts, direction, temperature, and precipitation. Meteorological data for each flight segment were matched to the corresponding expected hour of drone passage. Wind impact on flight time was calculated using the trigonometric projection of wind angle relative to the flight path.

A standardized set of universally accepted quantitative weather limits (e.g., strict wind and rain thresholds) could not be identified in the published literature; rather, the operational thresholds vary according to drone design, payload characteristics, and mission profile. Based on the studies of Gao et al. [31] and Oakey et al. [36], no-fly days for the main analyses were defined with the following parameters:

• Wind gusts > 15 m/s
• Temperature < −20 °C
• Precipitation > 10 mm/h.

In addition, we conducted simulations for no-fly days of wind-gust scenarios of 40, 50, and 60 km/h.

As stated in the Introduction, icing may represent a significant meteorological impact on drone flights. Data for such conditions were not available for this study.

A stepwise approach was used to assess the number of days in the year and per month during which these criteria would render drone flights infeasible due to weather conditions (no-fly days). On certain days, multiple weather variables exceeded tolerance limits, whereas on others, only a single variable was affected. Resulting times were used to estimate [¹⁸F] decay.

2.3. The Economics of Drone Transport for PET Radioisotopes

Cost-effectiveness was assessed as follows:

Net Gain = Revenue from Preserved Activity − (Current Transport Cost − Drone Cost)

The current total costs of existing NMCC transport solutions were used as benchmarks. As these services are procured from external providers, these costs were not decomposed into fixed and variable components; instead, the observed costs reflect the market rates for the complete transport service. Variations across existing routes are primarily driven by differences in transport distance, which affect driver time, vehicle utilization, and energy consumption, in addition to being driven by modal differences (e.g., ground versus air transport). As peer-reviewed cost analyses of drone-based logistics solutions that closely match the operational concept evaluated in this study were not identified, a detailed bottom-up costing of drone operations was considered to provide limited additional value relative to the study’s main objectives. Accordingly, the total costs of current transport solutions were compared with total drone transport costs derived from assumed cost-per-kilometer values. Drone cost thresholds (EUR/km) were then used to identify break-even levels across the different operational models and flight speeds considered. Lastly, potential revenue gains from reduced radioactive decay were estimated based on clinical applicability and dosing requirements.

2.4. Revenue Potential from Preserved Isotope Activity

While drone logistics do not lower the production cost of PET radioisotopes, they can reduce radioactive decay by shortening transport times, increasing the usable amount of the radioisotope upon arrival, potentially generating additional revenue, provided the saved activity can be applied effectively. The economic benefit depends on several factors, including cyclotron batch structure, the type of radioisotope used, pricing strategies, and the ability to reallocate or expand clinical scanning schedules.

A key consideration is the relationship between the cost of drone transport and the added value from preserved activity as a function of reduced transport time. A practical approach that can guide whether drone deployment is economically justifiable for a specific route is to compare drone transport costs directly to the monetary value of saved radioactivity.

Assume a radioisotope with half-life Ԏ_1/2, a drone transport distance of d km, flight velocity v (km/h), and a theoretical drone cost c/km flight (EUR).

Then, drone transport time can be calculated as follows:

(1)

T_d = d/v

and total drone cost as follows:

(2)

C_d = c × d.

The decay constant λ for a radioisotope quantifies the rate at which the isotope decays, i.e., how quickly a radioactive substance will lose its radioactivity, and is given by the following:

(3)

λ = ln (2)/Ԏ_1/2.

Ln (2) is the natural logarithm of 2 and applies for all radioisotopes, whereas Ԏ_1/2 is the halftime for the actual isotope (109.7 min for the ¹⁸F isotope in this study).

Assume a comparison of two differing transport times, T_d1 = t and a longer T_d2 = t + δ t. If Ro is the radioactivity at the start of transport, for T_d1, the remaining radioactivity R₁ for Ԏ₁ is as follows:

(4)

R₁ (t) = Ro × $e^{-\lambda \mathrm{t}}$

For the longer transport time T_d2 = (t + δ t),

(5)

R₂ (t + δ t) = Ro × $e^{-\lambda (\mathrm{t}+\delta t)}=\mathrm{R}\mathrm{o} e^{-\lambda \left(\mathrm{t}\right) }\times e^{-\lambda \left(\delta t\right)}$

Then, the relative increase in the remaining radioactivity of transport 1 compared to the longer transport 2 will be as follows:

(6)

R₁/R₂ $={\displaystyle \frac{\mathrm{R}\mathrm{o} \times e^{-\lambda \mathrm{t}}}{{\mathrm{R}\times e}^{-\lambda \left(\mathrm{t}\right)}\times e^{-\lambda \left(\delta t\right)}}=e^{\lambda \left(\delta t\right)}}$

A stepwise incremental cost-efficient ratio (ICER) [68] may then be constructed as the ratio comparing the increased cost per km drone flight to the additional radioactive benefit (R1/R2) (Figure 4).

2.5. Drone Transport Times

All drone flights began at 6 a.m., aligning with current arrival schedules. Drones were assumed to combine VTOL and fixed-wing capacities. Simulations assessed speeds of 150, 200, and 250 km/h, speeds currently achievable for many commercially available drones in the EASA category C3 or above.

2.6. Patient Throughput and Isotope Utilization

In our analysis, we employed a simplified model that assumed an administered activity of 250 MBq per patient and 30 min injection intervals, reflecting common clinical practice.

Two approaches were analyzed:

• Extend scanning schedules using increased arrival activity.
• Maintain current arrival activity but reduce starting activity, reallocating the surplus to other imaging centers.

3. Results

3.1. Current Transport Characteristics

The total costs of the current transport models, including car drivers, vehicle, energy, and, for SVG and AES, flight costs, are summarized in

Table 1. Descriptive data on existing isotope transport.

	KRS	ELV	SVG	AES
Isotope Activity at the Start of Transport (GBq)	40	26	26	26
Current Transport Costs (Euro)	820	435	5222	5222
Departure Time from the Cyclotron Centre	05:15	05:15	05:15	05:15
Arrival Time at the Imaging Centre	09:30	07:00	07:30	07:30
Total Transport Time (hours)	04:15	01:45	02:15	02:15
Typical Isotope Activity at Arrival (GBq)	7.9	9.1	9.1	9.1

. Accordingly, the current transport costs presented in Table 1 absorb all current delivery costs.

The information presented in the table demonstrates substantial variation in current transport costs across the different routes.

3.2. Simulated Time Gains

Time savings depended on drone speed and route distance. As an example, at 150 km/h, flights to KRS and ELV saved 1.4 and 0.5 half-lives of ¹⁸F, respectively. At 250 km/h, savings increased to 1.8 and 0.7 half-lives, though with diminishing returns (Figure 5).

3.3. Weather-Adjusted Drone Flight Times and No-Fly Days

Most mean drone speeds, including wind impacts, outperformed existing solutions, with the exception of the AES route at 150 km/h.

Figure 6 illustrates the simulation speeds and temperature and precipitation data for the KRS model, which exhibits the largest time potential, and the AES route, where gains were only obtained by mean velocities above 200 km/h. Some of the large deviations in transport times seen in Figure 6 occurred on days in which large wind gusts resulted in no-fly days and were not used when calculating maximum flight times among the differing routes.

Wind gusts emerged as the major decisive weather factor for no-fly days, with substantial differences across the routes. The consequences of various drone wind gust tolerances are illustrated in Figure 7 for a gust tolerance of 40–60 km/h.

The information presented in

Table 2. Average percent of no-fly days per year and maximum no-fly percent in a month related to no-fly limits of wind gusts of 40, 50, and 60 km/h (flight days per year: 250).

Route	Variable	40 km/h	50 km/h	60 km/h
SVG	Average no-fly/year	12%	7%	4%
	Highest Month	33%	33%	24%
KRS	Average no-fly/year	19%	8%	3%
	Highest Month	41%	23%	14%
ELV	Average no-fly/year	10%	2%	0%
	Highest Month	29%	10%	0%
AES	Average no-fly/year	35%	19%	12%
	Highest Month	62%	43%	33%

illustrates the occurrence of no-fly days as a percentage of days for the four routings, illustrating that a gust tolerance of 40 km/h. would not be sustainable for any of the routes.

As Figure 7 illustrates, differing no-fly wind-gust limits had significantly varying impacts across the routes. As seen in Table 2, most routes would be infeasible with a 40 km/h gust tolerance, resulting in no-fly days for 35% of the year and up to 65% in the most challenging month. In contrast, all routes would experience substantially fewer annual no-fly days with a tolerance above 60 km/h.

Based on the simulated transport times using the no-fly limit of 15 m/s, we estimated the corresponding time savings and reductions in radioactive decay for all routes (

Table 3. Statistical data of simulated time for drone speeds of 150, 200, and 250 km/h and savings and isotope activity for the four routings.

		Saved Transport Time			Saved PET Half-Life
Route		150 km/h	200 km/h	250 km/h	150 km/h	200 km/h	250 km/h
KRS	Max	165.2	184.1	196.4	1.52	1.69	1.80
	Min	106.6	152.0	176.1	0.98	1.39	1.62
	Mean	143.5	171.2	187.9	1.32	1.57	1.72
	SD	6.2	3.5	2.2
	50% Perc.	151.8	176.3	191.3	1.39	1.62	1.75
	95% Perc.	161.1	181.6	194.8	1.48	1.67	1.79
ELV	Max	66.8	75.0	80.3	0.61	0.69	0.74
	Min	43.6	62.3	72.3	0.40	0.57	0.66
	Mean	57.8	69.7	76.8	0.53	0.64	0.70
	SD	2.7	1.5	1.0
	50% Perc.	57.5	69.5	76.7	0.53	0.64	0.70
	95% Perc.	61.8	72.0	78.3	0.57	0.66	0.72
SVG	Max	41.8	60.8	73.4	0.38	0.56	0.67
	Min	−24.5	24.3	50.2	−0.22	0.22	0.46
	Mean	14.7	44.7	62.7	0.14	0.41	0.58
	SD	4.9	2.2	1.4
	50% Perc.	14.6	44.7	62.7	0.13	0.41	0.58
	95% Perc.	21.1	48.3	65.1	0.19	0.44	0.60
AES	Max	12.4	38.6	55.5	0.11	0.35	0.51
	Min	−67.4	−4.7	28.3	−0.62	−0.04	0.26
	Mean	−16.7	21.5	44.3	−0.15	0.20	0.41
	SD	5.9	3.3	2.1
	50% Perc.	−12.7	24.2	48.5	−0.12	0.22	0.43
	95% Perc.	−4.2	29.1	49.6	−0.04	0.27	0.46

). The best case for transport time improvements was long-distance KRS ground transport, with the long-haul air transport of AES demonstrating the weakest performance.

3.4. Route Paths and Transport Viability

The long-haul routes, SVG and AES, exhibited the greatest variability in geographic terrain and climate for transport procedures. Using AES as a case, Figure 8 illustrates adaptive routing with a 19 km longer path, mitigating no-fly challenges.

3.5. PET Activity Savings

Compared to current solutions, shorter transport times allow for reduced starting activity or increased output. The results presented in

Table 4. Estimated PET isotope savings (GBq) with shorter drone transport times, assuming current activity (number of scans) at arrival.

Transport	Current Starting GBq	Starting GBq Required with 150 km/h	Starting GBq Required with 200 km/h	Starting GBq Required with 250 km/h	Saved GBq with 150 km/h	Saved GBq with 200 km/h	Saved GBq with 250 km/h
KRS	40	16.1	13.5	12.1	23.9	26.5	27.9
ELV	26	17.6	16.3	15.6	8.4	9.7	10.4
SVG	26	23.6	19.5	17.4	2.4	6.5	8.6
AES	26	28.9	22.6	19.6	−2.9	3.4	6.4
ALL	118	86	72	65	32	46	53
Percent saved isotope capacity					27.0%	39.1%	45.2%

illustrate the estimated isotope savings (GBq) by route and drone speed. Figure 9 illustrates two approaches for utilizing saved activity, one with extended scanning schedules and one using parallel sessions.

By using the current starting activities as shown in Table 1 for the four transport modes, Table 4 illustrates the amount of PET isotopes preserved on arrival, i.e., increased remaining isotope activity.

Figure 9 illustrates how different patient throughput configurations influence the effective utilization of available PET isotope activity. Upon arrival at the imaging facility, the isotope container is transported to the radio pharmacy or dispensing unit, where patient-specific doses are prepared. Tracer injection is typically performed with the patient in a quiet, reclined position, often in a dedicated uptake room adjacent to the scanner, using secure intravenous access; saline flushing is applied to ensure complete and safe administration of the radiotracer. For standard whole-body ¹⁸F-FDG PET examinations, injected activities commonly range from 150 to 400 MBq, depending on patient characteristics and scanner sensitivity. Following injection, an uptake period of approximately 30–60 min is observed to allow tracer distribution, after which PET image acquisition is performed over 20–60 min, frequently combined with CT or MRI for anatomical co-registration. Modern 3D PET systems with higher detection efficiency enable lower administered activities (typically ~150–250 MBq), thereby reducing patient radiation exposure while maintaining diagnostic image quality and improving overall tracer utilization efficiency.

Patients are scheduled consecutively throughout the day, and due to continuous radioactive decay, later examinations require higher fractions of the initially delivered activity to achieve the same injected dose.

By using the drone speed of 150 km/h on the KRS route, 23.9 GBq of PET isotope would be saved. Assuming 16 patients are scanned sequentially on a single scanner, as illustrated in Figure 9 (Model a), the theoretical activity consumption per patient is 1.49 MBq. In contrast, when two scanners are available and operated in parallel, enabling higher patient throughput (Model b), up to 20 patients can be examined with an average activity consumption of only 0.68 MBq per patient. While such gains are contingent on the availability of parallel scanning capacity, the figure serves primarily to illustrate the substantial impact of patient flow organization and scanner throughput on overall isotope utilization efficiency.

The two differing approaches illustrate the outcomes if current departure activity is maintained. The utilization of the saved isotope is related to the local scanning capacity. With only one scanner at KRS, six more patients could be scanned with today’s delivery using an extended-day schedule. Depending on the scanner capacity, a substantially greater number of patients could be scanned with two scanners, using almost 10 MBq less activity per patient, provided other resources such as personnel and reimbursement are in place.

Another strategy could involve reducing departure activity and distributing surplus PET isotopes to other imaging centers as new centers are established.

3.6. Economic Advantages

Transport Costs

The break-even drone transport costs varied by route. Those of the KRS and ELV routes must remain below EUR 3/km, whereas the SVG/AES routes could tolerate up to EUR 17–18/km due to the current reliance on high-cost air transport (Figure 10).

3.7. Potential Increased Revenue from Saved Isotopes

The results presented in Table 4 illustrate how the use of drones reduces decay-related losses during transport, thereby lowering the required initial PET activity per shipment and potentially freeing capacity for additional doses. This surplus activity could be reallocated for redistribution or sale to other imaging centers. Isotope prices vary substantially across suppliers and isotope types, making exact revenue calculations difficult. A corresponding estimate of the potential income effects is presented in

Table 5. Estimated increase in income (EUR) for amplified delivery capacity compared to existing solutions, assuming 250 transport days per year.

Route	Drone Speed 150 Km/h	Drone Speed 200 Km/h	Drone Speed 250 Km/h
KRS/Transport	2490	2760	2901
ELV/Transport	1352	1556	1671
SVG/Transport	378	1035	1373
AES/Transport	−458	537	1027
All/Transports	3762	5888	6971
All/Year	940,559	1,471,964	1,742,863

.

The estimates are based on the current mix of isotope deliveries (with differing substances and isotopes having different prices) and assume an average dose price of EUR 525 (range EUR 350–700). It is further assumed that the increase in usable activity translates proportionally into increased billable doses and, consequently, higher income. While PET isotopes are not traded in a fully liberalized market, public hospitals remunerate suppliers based on delivered doses. This pricing structure implies that improved delivery efficiency can increase revenues for producers and service providers, helping to offset higher production and transport costs associated with drone-based logistics.

3.8. Break-Even Analysis

By combining the transport-related and PET-related savings, the analysis modelled break-even scenarios across drone operating costs ranging from EUR 1 to 20/km. For these cost estimates, total drone expenditures were assumed to be fully absorbed within the kilometer rate, including platform acquisition, certification, infrastructure, airspace services, staffing, insurance, and operational overheads. Figure 11 illustrates the resulting route-specific cost thresholds at which drone operations outperform existing logistics models.

4. Discussion

4.1. Principle Findings

Our findings illustrate that drone transport can significantly reduce delivery times for PET radioisotopes across long distances, enhancing isotope preservation and enabling substantial annual cost savings. Our work demonstrates the potential impact and provides a framework for evaluating early feasibility in sandbox exercises. Future test flights may provide real-life data to demonstrate a reduction in isotope loss.

However, the magnitude of these benefits depends on several factors: transport distance, drone speed, isotope half-lives, clinical workflow integration, available drone fleet and infrastructure, and, last but not least, meteorological conditions. Weather emerged as a key barrier to consistent drone operations, suggesting up to 30–60% of flights could be disrupted during some winter months, depending on wind gust tolerance. In particular, routes exposed to coastal or mountainous conditions are vulnerable, aligning with prior research indicating that wind gusts, precipitation, and temperature extremes limit drone reliability [31,36,66,69,70,71,72,73].

Operational adaptations may possibly mitigate these challenges. Fallback to ground transport based on high-resolution weather forecasts, seasonal route suspension, flexible scheduling, or reconfiguring route paths are examples of possible actions (Figure 7).

4.2. Drone Design and Technical Considerations

Drone feasibility for PET logistics requires designs capable of carrying ≥ 50 kg to accommodate shielded, temperature-controlled containers (15–25 °C). Long-range missions necessitate hybrid VTOL-fixed wing systems that balance vertical mobility with cruising efficiency.

Battery-powered drones lack range and remain unsuitable for long-distance flights in cold climates due to limited energy density and cold-weather degradation. Despite the fact that recharging infrastructure has been proposed [62,74,75,76], recharging or battery exchange would consume valuable time and remain impractical for PET transport over remote areas with no infrastructure. Although fuel-powered systems involve higher operating costs and emissions, they offer superior range and will be necessary for the distances and payloads studied until battery energy densities improve [62,74,75,76].

It should be acknowledged that sustained drone speeds of 200–250 km/h over long distances while carrying payloads suitable for PET isotope transport are not feasible with current technology, the increasing regulations for such larger drones operating in public space are developing and may spark further technological development [77]. Nevertheless, both high-speed drones and platforms designed for heavy payloads or extended range are under active development. Importantly, however, high payload capacity and long operational range represent partially conflicting design objectives, reflecting fundamental trade-offs in energy efficiency, aerodynamics, and propulsion systems [78,79]. To support such industrial development, it is essential to document scalable and sustainable use cases that are economically viable for both operators and customers. In this regard, the operational concept explored in this study may contribute by illustrating a potentially viable demand scenario and application context for future high-performance drone systems.

4.3. Organizational and Economic Implications

Drones have the potential to catalyze broader changes in PET logistics. Faster delivery may enable increased capacity through extended scanning hours or parallel scanner usage, as illustrated in Figure 9, maximizing clinical throughput. Alternatively, shorter transport durations could support isotope reallocation to underserved sites, enhancing regional imaging access.

There is a historical analogy to be drawn from the evolution of theragnostics, once limited to iodine treatments for thyroid cancer, now a cornerstone of precision oncology [80]. Similarly, integrating drone logistics may reconfigure how PET services are designed and delivered. Adopting drones may prompt reconsideration of current practices, such as current single-batch isotope productions. Drones allow for staggered deliveries, reducing waste and increasing responsiveness to clinical demand. Furthermore, combining drone logistics with next-generation whole-body PET scanners, requiring lower doses, could defer the need for costly cyclotron expansion.

However, unidirectional flights present inefficiencies. Economically viable implementation may depend on backhaul use, transporting other medical goods or diagnostics. Integrating PET deliveries with other services, via mixed on-demand and scheduled flights, will require redesigning ground logistics, drone ports, and digital infrastructure [62,74,75,76].

The cost-effectiveness of drones is highly context specific. Extensive literature exists on optimizing drone routing strategies [81,82,83,84,85,86]. While last-mile drone deliveries have shown potential cost savings [62,87,88,89,90], this evidence does not extend to long-haul healthcare logistics, however. Our simulations highlight that drone solutions are more economically viable on routes currently reliant on expensive air transport (e.g., AES), with reduced viability for routes with low-cost road delivery (e.g., KRS).

Cost optimization remains essential and must balance speed, regulatory compliance, and operational expense to meet stakeholder expectations. The evolving U-space regulatory framework in Europe will be critical to ensure safe and cost-efficient integration [91].

Overall, drone logistics for PET require further empirical validation, full cost modelling (including infrastructure and staffing), and robust regulatory alignment.

4.4. Limitations

Several limitations of this study should be acknowledged.

First, there are currently no publicly available real-world operator data for routine drone-based transport of PET isotopes against which to benchmark the modelled performance and cost estimates. Although regulatory frameworks for beyond-visual-line-of-sight (BVLOS) operations are evolving, substantial additional costs are likely to arise from future infrastructure requirements, including drone ports and landing platforms, airspace management and surveillance systems, certification processes, and compliance with public aviation regulations. At present, the authors were unable to identify peer-reviewed studies providing empirically grounded cost estimates for these elements. To avoid introducing speculative or poorly substantiated assumptions, in our analysis, we compare ground transport costs with aggregate drone delivery costs, implicitly assuming that the latter internalize all UAV-related expenditures, including platform acquisition, certification, infrastructure, airspace services, staffing, insurance, and operational overhead. The authors of future studies should seek to validate and refine these estimates using real-world pricing from operational drone service providers, including explicit cost components for airspace usage, drone port access, and regulatory compliance.

Second, the weather inputs applied in this study were derived from ground-level meteorological observations. While sufficient for indicative scenario modelling, this approach does not capture vertical wind profiles, localized turbulence, or icing risk at flight altitudes. Integration with aviation-grade meteorological data and forecasting systems will be essential to support operationally robust flight planning and to improve the fidelity of performance and reliability estimates.

Weather-related delays in ground transport were ignored as such issues occurred only once in the two years, 2023 and 2024. This condition favors ground solutions, with the assumptions related to gains from drones remaining valid.

Third, the routing assumptions employed in this study are idealized. Although the modelled results capture the direct impact of wind on flight speed, adverse weather conditions may necessitate dynamic rerouting to avoid no-fly zones, hazardous meteorological cells, or controlled airspace. In such cases, longer and less direct flight paths may be operationally preferable to mission cancellation, leading to deviations from the great-circle (Haversine) distances assumed herein.

Similarly, future airspace management constraints and deconfliction requirements may further limit route optimality, introducing additional detours and time penalties not captured in the current simulations. In addition, transport of radioactive materials is subject to regulations of dangerous goods, which will most likely influence flight planning and costs [92,93].

Lastly, local infrastructure effects, including wind shear and turbulence in the vicinity of urban landing and take-off sites, rooftop installations, and confined terminal areas, were not explicitly modelled. Based on very similar geographical topography and information from the receiving hospitals, similar wind conditions at the receiving hospitals were assumed. Despite obtaining weather data from these locations, uncertainties in this assumption should be acknowledged. Such micro-scale aerodynamic effects may have a substantial influence on safety margins, energy consumption, and operational reliability, particularly in dense urban environments.

4.5. Future Directions

The authors of future studies should explore hybrid logistics models that dynamically switch between drone and ground transport based on weather forecasts or clinical urgency. Multi-sectoral partnerships may also enable shared drone services, increasing utilization and cost-efficiency beyond PET transport.

Comprehensive modelling of drone logistics costs, including infrastructure, airspace access, and staffing, is critical. Standardizing these inputs would improve comparability across studies and strengthen the evidence base for drone integration in healthcare logistics.

5. Conclusions

This study evaluated the feasibility and potential benefits of long-distance drone transport for PET radioisotope logistics by comparing simulated drone operations with existing ground and combined ground–air transport solutions. Using weather-resolved modeling and route-based comparisons, the analysis quantified transport time savings, preserved radioactive activity, and economic implications for drone delivery distances of 117–376 km.

The results indicate that drone transport can significantly reduce isotope decay losses when operating at speeds of 200–250 km/h, preserving 46–53 GBq of activity, i.e., reducing the total current use of 118 GBq to 72–65, respectively. At these speeds, break-even operating costs ranged from EUR 3–18/km, increasing to EUR 14–20/km when accounting for the economic value of preserved isotopes, corresponding to estimated annual gains of EUR 1.0–1.7 million. However, weather constraints remain a key operational factor, with up to 30% of winter flights potentially restricted.

These findings suggest that long-range drone transport could improve delivery efficiency and increase diagnostic capacity by reducing radioactive decay losses and enabling more flexible distribution of PET isotopes. Realizing these benefits will require integration of drone operations into existing clinical logistics systems, as well as further advances in weather resilience, regulatory frameworks, and cost-efficient drone technologies. Rather than simply replacing current transport methods, drones could enable a broader transformation in medical logistics. With system-wide planning and stakeholder coordination, drone logistics may become a cornerstone of a more flexible, decentralized, and responsive healthcare infrastructure.