Key Considerations for Treatment Planning System Development in Electron and Proton FLASH Radiotherapy

Abstract

The global cancer burden continues to increase worldwide. Among the various treatment options, radiotherapy (RT), which employs high-energy ionizing radiation to destroy cancer cells, is one of the primary modalities for cancer. However, increasing the absorbed dose to the target volume also increases the risk of damage to surrounding healthy tissues. This radiation-induced toxicity to normal tissues limits the desirable dosage that can be delivered to the tumor, thereby constraining the effectiveness of radiation therapy in achieving tumor control. FLASH radiotherapy (FLASH-RT) has emerged as a promising technique due to its biological advantages. FLASH-RT involves the delivery of radiation at an ultra-high dose rate (≥40 Gy/s). Unlike conventional RT, FLASH-RT achieves comparable tumor control rates while significantly reducing damage to surrounding normal tissues, a phenomenon known as the FLASH effect. Although the mechanism behind the FLASH effect is not fully understood, this approach shows considerable promise for future cancer treatment. The development of specialized treatment planning systems (TPS) becomes imperative to facilitate the clinical implementation of FLASH-RT from experimental studies. These systems must account for the unique characteristics of FLASH-RT, including ultra-high dose rate delivery and its distinctive radiobiological effects. Critical reassessment and optimization of treatment planning protocols are essential to fully leverage the therapeutic potential of the FLASH effect. This review examines key considerations for the TPS development of electron and proton FLASH-RT, including electron and proton FLASH techniques, biological models, crucial beam parameters, and dosimetry, providing essential insights for optimizing TPS and advancing the clinical implementation of this promising therapeutic modality.

1. Introduction

In contemporary society, cancer represents a significant threat to public health. It ranks as either the leading or second leading cause of mortality before age 70 in 112 out of 183 countries [1]. As a primary treatment modality for cancer, radiotherapy (RT) has gained increasing recognition within the medical community. This therapeutic approach employs high-energy ionizing radiation to destroy cancer cells. However, during treatment, both the tumor and surrounding normal tissue are inevitably exposed to radiation. It can cause damage in healthy tissue, limiting the therapeutic dose delivered to the tumor. To address this challenge, FLASH radiotherapy (FLASH-RT) is emerging as one of the most promising breakthroughs in RT [2,3].

FLASH-RT employs ultra-high dose-rate (UHDR) irradiation, typically defined as radiation delivered at dose rates ≥ 40 Gy/s. This treatment method can maintain tumor control rates comparable to those of conventional radiotherapy (Conv-RT) while substantially reducing damage to surrounding normal tissues, a phenomenon known as the FLASH effect.

The FLASH effect was first observed in 1959 [4]. It was described by the enhanced survival of cells exposed to UHDR radiation. Based on this initial observation, recent studies [5,6,7,8,9] demonstrated that UHDR beams could significantly reduce damage to normal tissues in murine models. Furthermore, Bourhis et al. [10] conducted a groundbreaking study using FLASH-RT on a patient with cutaneous lymphoma. The treatment achieved excellent tumor growth control with only mild side effects, confirming the feasibility and safety of FLASH-RT in humans.

These extensive preclinical studies convincingly demonstrated the therapeutic potential of the FLASH-RT. The transition of FLASH-RT from experimental settings to clinical applications is crucial. This transition necessitates the development of specialized treatment planning systems (TPS). However, existing TPS frameworks require reassessment to accommodate the unique physical, temporal, and biological characteristics of FLASH-RT.

To develop TPS for electron FLASH-RT (eFLASH-RT), current research focuses on addressing clinical challenges arising from its inherent physical properties. Because the photon beams lack the skin-dose build-up zone protection effect, electron beams are susceptible to excessive irradiation of normal tissues in front of tumors during superficial tumor treatment [11]. Although their physical characteristics limit their application to deep-located tumors, electron beams remain well suited for precise and efficient treatment options for superficial or near-surface tumors. It is important to note that the development of very high-energy electron (VHEE) beams can overcome conventional range limitations. It creates new opportunities for treating deep-seated tumors.

Regarding the technical implementation of eFLASH-RT TPS, Kyuhak et al. [12] integrated an electron Monte Carlo algorithm into a commercial TPS. They successfully developed the FLEX system and validated its accuracy in generating UHDR electron dose distributions. Dai et al. [13] developed a Monte Carlo electron beam model specifically for the Mobetron device. This model demonstrated advantages in dose distribution calculations during preclinical studies. However, incorporating dose-rate constraints into the optimization framework has not yet been achieved. Rahman et al. [14] enhanced the Monte Carlo beam model for eFLASH-RT by utilizing clinical accessories and geometric structures. Zhang et al. [15] systematically investigated the currently known FLASH parameters for VHEE radiotherapy and established a benchmark for its FLASH dose-rate performance.

Due to the depth-dose characteristics of the Bragg peak, proton FLASH-RT (pFLASH-RT) offers significant advantages for treating deep-seated tumors. By precisely controlling proton energy deposition, high-dose zones within the target area can be created while simultaneously ensuring maximal protection of surrounding normal tissues. Furthermore, combining it with the FLASH effect further safeguards normal tissues.

Current pFLASH-RT TPS research primarily focuses on optimization of dose-rate constraints. Wase et al. [16] have indicated that the spatial and temporal distribution of spots in pencil beam scanning (PBS) treatment plans is a critical factor in dose-rate distribution. The optimal spot sequence is defined as the one that maximizes the volume receiving the FLASH dose rate. They applied the Traveling Salesman Problem algorithm to optimize spot orders in PBS. This approach improved the computational efficiency of FLASH coverage evaluation by approximately three orders of magnitude compared with conventional sorting strategies. Other teams [17,18] used genetic algorithms to explore global optimal solutions. To illustrate how spot sequencing influences dose-rate metrics in PBS-FLASH delivery, Figure 1 shows a simple example with five spots irradiated under different scanning orders. Although the total delivered dose remains identical, changing the temporal sequence alters the pulse structure and thereby affects several dose-rate quantities.

In the context of dose rate quantification, Kang et al. [19] developed a multi-metric evaluation framework encompassing dose-averaged dose rate (DADR), average dose rate (ADR), and dose threshold dose rate (DTDR). DADR, ADR, DTDR, and effective field dose rate (EFDR) represent different methodological approaches for dose-rate assessment, as summarized in

Table 1. A summary of different dose rate quantification methods.

Dose Rate Type	Function	Description
Mean Dose Rate	$\dot{D}={\displaystyle \frac{D}{T}}$	The average dose rate over the total irradiation time T, where D is the total dose delivered during the time T.
Instantaneous Dose Rate	$\dot{D}\left(t\right)={\displaystyle \frac{dD\left(t\right)}{dt}}$	$D\left(t\right)$ denotes the cumulative dose at time t, ${\displaystyle \frac{dD\left(t\right)}{dt}}$ is the time derivative of the dose function.
DADR	${\dot{D}}_j^{DADR}={\displaystyle \sum _{i=1}^N}{\displaystyle \frac{D_{j,i}}{\sum _{i=1}^N{D}_{j.i}}}{\dot{D}}_{j.i}$ ${\dot{D}}_{j.i}={\dot{D}}_{max}{e}^{-{\displaystyle \frac{{\left(r_j-r_i^c\right)}^2}{{\sigma }^2}}}$	i denotes a spot, j represents a voxelized region in the target, and $D_{j.i}$, ${\dot{D}}_{j.i}$  is the dose deposited/dose rate by the i-th spot to the j-th voxel. ${\dot{D}}_{max}$  is the max dose rate at the spot center. $r_j$  denotes the position of j-th voxel,  $r_i^c$  denotes the position of the i-th spot center, and σ is the spot sigma.
ADR	${\dot{D}}_j^{ADR}={\displaystyle \frac{D_j-2d*}{T_j}}$ $\begin{array}{ll}{d}_j\left(t_0\right)& =d*\\ d_j\left(t_1\right)& =D_j-d*\\ T_j & =t_1-t_0\end{array}$	(Dj − 2d*) is the total dose deposited in voxel j during the irradiation $T_j$, d* is a preset dose-threshold that determines the irradiation start time $t_0$ and the end time $t_1$.
DTDR	${\dot{D}}_j^{DTDR}=min\left({\dot{D}}_{j,i}\right),if D_{j,i}>d*,i=\mathrm{1,2}\dots n$	d* is a preset dose-threshold.  ${\dot{D}}_{j.i}$  is the i-th spot dose rate in the j-th voxel.
EFDR	$\dot{D\left(x,y,z\right)}=D\left(x,y,z\right)/T$ $D\left(x,y,z\right)={\displaystyle \sum _n}{D}_0\left(x_n,y_n,x,y,z\right){\displaystyle \frac{M{U}_n}{M{U}_{n0}}}$$T={\displaystyle \sum _n}\left[T_0\left(x_n,y_n\right){\displaystyle \frac{M{U}_n}{M{U}_{n0}}}{\displaystyle \frac{I_0}{I_n}}+T_{slew}\right]+{\displaystyle \sum _m}{T}_{energyswitch,m}$	$D_0\left(x_n,y_n,x,y,z\right)$ indicate the dose at a point ($x,y,z$) in a water phan-tom from spot n with position ($x_n,y_n$) and $M{U}_{n0}$. $T_0\left(x_n,y_n\right)$ is spot delivery time, $I_0$ is cyclotron output current. $T_{slew}$ is slew time and $T_{energyswitch,m}$ is the energy switching time for the mth energy layer.

. Each of these approaches is characterized by its distinct emphasis on specific aspects of the evaluation process. Wei et al. [20] addressed multidimensional constraints, including target volume uniformity, dose limits for critical organs, and FLASH dose-rate coverage.

Photon FLASH-RT faces more severe physical implementation challenges than charged particle beams. Because X-ray generation depends on the electron–target interactions, its inherently low conversion efficiency limits the availability of devices capable of producing UHDR X-rays. It This limitation directly constrains the development of related TPS. Consequently, this review focuses on key considerations in TPS development for both proton and electron technologies.

The development of TPS for proton and electron FLASH-RT remains in its early stages. Current pFLASH-RT TPS has only achieved basic modeling of dose rates, while eFLASH-RT TPS focuses on UHDR dose calculation verification. Neither has established a complete clinical application framework. Consequently, numerous critical factors remain to be considered in the development of TPS for proton and electron FLASH-RT. The present review first summarizes the methods by which proton and electron systems achieve UHDR delivery. It then proposes several additional key factors that must be addressed to advance the clinical application of proton or electron FLASH-RT TPS. These additional key factors include dosimetric validation, precise biological dose estimation, and integration of the physical beam parameters, as shown in Figure 2. UHDR realization methods refer to the beam delivery strategies and accelerator configurations used to achieve UHDR. The interplay between FLASH-specific radiobiology and physical beam parameters necessitates advanced modeling. Such modeling will refine dose delivery strategies and validate therapeutic outcomes. Meanwhile, dosimetry will ensure the accuracy of the physical inputs to the TPS biological model. These adjustments will ensure the accurate implementation of the FLASH effect in clinical practice.

2. UHDR Realization: Proton and Electron Methods in FLASH-RT

2.1. Proton

In conventional proton radiotherapy, the limitations of passive scattering technology in clinical applications have prompted research teams to explore PBS–based approaches for FLASH delivery. However, conventional PBS technology has limitations in pFLASH-RT, particularly the requirement for multiple proton energy layers and long switching times between them. These shortcomings compromise the achievable dose rate [11].

To address this limitation, numerous research teams [20,21,22] have studied transmission-beam irradiation schemes. They utilized the region at the front end of the Bragg peak of the highest-energy protons for irradiation, ensuring that the Bragg peak remained entirely outside the patient’s body. This method ensures a stable FLASH dose rate by eliminating the energy-switching step. Combined with a multi-field configuration, it can produce dose distributions analogous to those of conformal irradiation. Despite its comparatively diminished conformal effect relative to conventional PBS, the coverage of UHDR is demonstrably enhanced. Nevertheless, this technique still exhibits the inherent disadvantage of exposing normal tissues behind the tumor to high-dose irradiation for some patients.

Building on this, a technical approach based on single-energy Bragg peak intensity modulation has been developed. This approach utilizes range shifters and range compensators [23,24,25,26] to pull back the ranges of the highest-energy proton beams, enabling distal conformity to the target. It is then combined with multi-field to optimize the dose distribution. Research has demonstrated that this technique maintains comparable coverage of the transmission beam dose rate while markedly enhancing conformity. This approach can enhance tumor treatment efficacy while better sparing critical organs. Furthermore, the advantages associated with the Bragg peak, when combined with the FLASH effect, may provide additional therapeutic benefits compared with transmission beam schemes.

Subsequent research endeavors encompass the implementation of ridge filters [22,27,28,29,30,31,32,33,34,35]. Both dynamic and static ridge filters have been demonstrated to produce single-energy spread-out Bragg peaks (SOBPs), further enhancing conformity. Because this approach modulates a single-energy proton beam without requiring energy-layer switching, delivery time can be significantly reduced. However, this method continues to confront substantial challenges. For static ridge filters, positioning errors and manufacturing precision must be carefully considered. Dynamic ridge filters expand the dose field via mechanical motion, providing greater tolerance to positioning errors. However, the motor system’s speed of movement and positioning accuracy pose implementation challenges. The common problems with these technical solutions are the requirement for ultra-high-hardware precision. Range shifters, range compensators, and ridge filters all require extremely high manufacturing precision and positioning accuracy. Furthermore, the mechanical components of dynamic ridge filters necessitate more stringent control over movement and positioning accuracy. The current manufacturing precision and positioning accuracy of devices may cause shifts in the Bragg peak position, potentially leading to insufficient target dose or the risk of organ overdose. Consequently, there is a compelling need to continue refining these standards to ensure optimal treatment accuracy.

Notably, these methods need to utilize multi-field techniques to ensure the implementation of conformal effects. However, in their studies, the time required for the gantry head movement was not included in the total treatment time. It is a potential factor that could affect treatment efficacy in proton FLASH-RT.

2.2. Electron

High-energy electrons (HEE) are effective in treating superficial tumors effectively. In contrast, VHEE has shown promising dosimetric performance in treating deep-located deep-seated tumors.

At present, several devices are capable of generating HEE beams at UHDR. Rahman et al. [36] validated the feasibility of intensity-modulated passive scattering electron UHDR beams based on a modified medical linear accelerator. However, multi-leaf collimators used for conformal treatments in Conv-RT face challenges similar to those of dynamic ridge filters in eFLASH-RT, as they require highly precise motion and positioning within very short time intervals. Moreover, due to inherent limitations in gantry rotation speed, existing technologies struggle to generate multiple fields within an exceedingly brief period. As a result, HEE electron beams can only be applied to single-field applications. Consequently, developing intensity modulation and multi-field techniques for electron beams will be a pivotal direction for HEE FLASH-RT.

VHEE with UHDR devices remain at the stage of theoretical design and simulation verification stage, with related research primarily focused on dosimetric characteristic analysis [37,38,39,40]. Giuliano et al. [37] demonstrated the technical feasibility of selecting key parameters for VHEE devices using on a C-band linear accelerator prototype. Next, Bohlen et al. [38] evaluated the dosimetric performance of a 50–250 MeV VHEE radiotherapy system, indicating that HEE beams can be a lightweight alternative to proton therapy, with comparable dose-planning quality. Finally, Bedford et al. [40] noted that increasing beam energy reduces electron scattering and produces a sharper penumbra, but simultaneously causes the depth-dose curves to flatten. This characteristic is less advantageous than proton technology in normal tissue protection.

Despite this limitation, the sharp penumbra of VHEE beam remains beneficial for achieving highly conformal dose distribution plans. However, implementing multi-field techniques remains technically challenging. The clinical translation of VHEE requires advancements in accelerator engineering, particularly the development of VHEE accelerator systems capable of producing UHDR beams. In parallel, incorporation of intensity modulation and multi-field configuration strategies is necessary to realize their feasibility and dose-delivery advantages for deep-seated tumor treatment.

3. Biological Model: Physical-to-Biological Conversion Tool in FLASH-RT TPS

The FLASH effect maintains tumor control rates comparable to those of Conv-RT while minimizing damage to surrounding normal tissues. Novel biophysical models and computational methods are needed to estimate radiation-induced biological effects better. These FLASH-related advantages require re-evaluating and optimizing existing tumor treatment plans. The benefits of the FLASH effect can be fully exploited through careful optimization, thereby maximizing therapeutic efficacy and minimizing adverse reactions.

The unique characteristics of FLASH-RT primarily include UHDR delivery and its associated radiobiological effects. These characteristics make the development and application of new biological models particularly critical. These models will deepen our understanding of the mechanisms underlying the FLASH effect. Additionally, they will provide valuable guidance for developing and optimizing TPS. This optimization ensures that TPS meets the specific design and application requirements of FLASH-RT.

The current treatment planning evaluation system lacks biological dose constraints, with physical dose-rate distribution serving as the sole optimization objective. During the development of transmission proton beam therapy plans [20] and single-energy Bragg peak intensity-modulated plans [25,26], it was hypothesized that the relative biological effectiveness (RBE) of proton FLASH irradiation in normal tissues would remain equivalent to that of conventional dose-rate proton irradiation. However, this approach does not fully account for the FLASH effect. This optimization strategy neglects appropriate weighting of biological effects. It may result in predictions of tumor control probability and normal tissue complication probability that are not aligned with actual treatment outcomes.

To address the existing gaps in biological effective dose (BED) modeling, recent studies have sought to quantify the clinical impact of the FLASH effect by modifying biological effect models. Jones et al. [41] proposed modifying the traditional BED model to account for short exposure times and FLASH effect. They introduced a cubic-root formulation for dose rates while retaining monoexponential repair kinetics. This approach incorporates the impact of increasing dose rates on the intrinsic α/β ratio. However, BEDs cannot be directly compared when different α/β ratios are used [42]. To address this limitation, the concept of ‘excess BED’ was developed to describe the same biological system when exposed to UHDR.

The excess BED represents the total BED minus the normal tissue tolerance BED. Figure 3 illustrates the dose-rate dependence on excess BED. Experimental observations revealed that excess BED increases proportionally with escalating dose rates within the conventional dose-rate range (0.01–100 Gy/h). However, a marked reduction in excess BED was observed in the high dose-rate regime (10³–10⁵ Gy/h). Although these dose rates did not exceed the UHDR range, the trend suggests that the excess BED would continue to decrease under UHDR conditions. Finally, Jones et al. [41] indicated that the excess BED model could provide isoeffective dose estimation to guide RT with different dose rates. However, they emphasized the need for further experimental validation to define and validate the model. The excess BED model provided a promising framework for converting FLASH-RT plans according to varying dose rates. Nevertheless, the model is still in its early stages, and a systematic evaluation of various physical and biological parameters is required for further exploration.

To enable integration into TPS, the excess BED model relies on several implicit assumptions. The biological response is governed by a short irradiation timescale, such that sublethal damage repair during delivery is negligible under UHDR conditions. Oxygen tension is assumed to be quasi-static during conventional irradiation but dynamically reduced during FLASH delivery due to rapid radiochemical oxygen consumption. Under these assumptions, dose rate becomes a biologically relevant variable rather than a purely physical descriptor. From a TPS perspective, the excess BED formulation provides a computable biological surrogate that can be incorporated either as an additional optimization objective or as a constraint on excess BED in normal tissues, while maintaining equivalent BED in the target volume.

Moreover, Hu et al. [43] employed two-dimensional (2D) reaction-diffusion equations to model the heterogeneity of oxygen distribution in capillaries and tissues. Meanwhile, they coupled it with a modified linear-quadratic (LQ) model to characterize the surviving fraction under different total doses and dose rates. They demonstrated that, under UHDR radiation, exceeding a minimum threshold in total dose resulted in a significant reduction in damage to surrounding normal tissues. Jones et al. [44] also reported related findings about total-dose threshold in their study. Hu [43] further noted that the surviving fraction displayed a “plateau effect” under UHDR radiation. This plateau indicated that the survival fraction remained relatively constant within a specific dose range. This phenomenon was influenced by factors such as oxygen tension and dose rate.

From a TPS perspective, such biological modeling directly informs optimization constraints by defining acceptable dose windows, minimum total dose thresholds required to trigger the FLASH effect, and temporal delivery conditions under which oxygen depletion is maintained. These parameters can be translated into TPS objectives and constraints to optimize FLASH treatment planning. While current models remain largely conceptual and 2D, they provide a mechanistic basis for integrating FLASH-specific biological response into TPS optimization. Future work should extend these models to three dimensions and validate them experimentally to enable clinically robust biological cost functions for FLASH treatment planning.

Shiraishi et al. [45] employed an integrated microdosimetric-kinetic model for UHDR irradiation. This model demonstrated the ability to predict the fraction of surviving cells under these treatment conditions. It is noteworthy that Michelle [46] demonstrated through a study correlating proton RBE differences with potential physical parameters that average proton energy, survival endpoints, and cell type influence proton RBE. This finding underscores the need to incorporate biological constraints into proton FLASH-RT TPS. As demonstrated in related studies [47], microdosimetry has been shown to serve as an effective tool for RBE estimation. Integrating microdosimetry parameters into proton FLASH-RT TPS through the synergistic modeling of physical and biological constraints has the potential to markedly enhance treatment plan quality. Bohlen et al. [48] proposed a formalism based on the LQ and LQ-L models. Using this approach, they quantified the minimal NT required by the FLASH effect to compensate for hypofractionation. Incorporating microdosimetric parameters into proton FLASH-RT TPS allows RBE to be treated as a plan-dependent biological variable rather than a fixed constant, providing a framework for biologically informed dose optimization under UHDR conditions.

Taken together, current FLASH biological models suggest that treatment planning should move beyond only physical dose-rate optimization toward biologically informed objectives. Key parameters, including oxygen tension, irradiation time scale, dose thresholds, and effective RBE modulation, can be abstracted into TPS-compatible constraints and cost functions. Such integration provides a practical pathway for translating mechanistic FLASH models into robust and clinically usable optimization strategies.

4. Beam Parameters: Multi-Parametric Framework for FLASH-RT TPS

As mentioned above, dose rate is a crucial determinant of the FLASH effect and can be adjusted in TPS to align with biological models [49]. However, accumulating evidence [5,43,44,50,51,52,53,54,55,56,57,58,59,60,61,62,63] indicates that a single dose rate is insufficient to characterize FLASH irradiation fully. Instead, multiple delivery-related parameters significantly influence the FLASH effect, including the mean dose rate [5,51,52], the instantaneous dose rate of individual pulses [53,54], the total delivered dose [43,44,55,56,57], and the temporal structure of irradiation (encompassing pulse structure and overall delivery time) [50,58,59,60].

Multiple studies suggested that the sparing effect on healthy tissues becomes significant when mean dose rates exceed 40 Gy/s, and optimal effects are observed with dose rates greater than 100 Gy/s. However, contradictory findings [64,65] highlighted the need for further research to elucidate the relationship between mean dose rate and the FLASH effect. Understanding this relationship is crucial for optimizing clinical applications. Meanwhile, Sampayan et al. [54] highlighted the critical role of instantaneous dose rates. Similarly, Sunnerberg et al. [53] demonstrated the influence of instantaneous dose rates on hydrogen peroxide production and oxygen consumption. These processes play crucial roles in the mechanisms of the FLASH effect. These findings collectively emphasize the needed to consider both mean and instantaneous dose rates in TPS to describe and optimize FLASH-RT accurately.

Acknowledging the fundamental discrepancies in the definitions of average and instantaneous dose rates in proton and electron FLASH-RT is imperative. For instance, the average dose rate for electrons can be defined as the ratio of the total dose to the irradiation time. In contrast, the instantaneous dose rate corresponds to the dose rate within a single pulse [66]. However, this definition does not account for the characteristic of voxel dose rate in proton PBS technology, which is influenced by the beam scanning mode. For proton PBS, the existing dose-rate assessment methods include DADR [67], EFDR [68], and ADR [69]. These definitions accentuate disparate dimensions, and using divergent calculation methodologies within a single plan may yield substantial discrepancies. In the context of dose distribution optimization in FLASH-RT TPS, it is imperative to incorporate both average dose rate and instantaneous dose rate into constraint conditions concurrently. This approach requires careful consideration of the dose-rate coverage required to elicit optimal FLASH effects, thereby facilitating improved treatment outcomes.

Meanwhile, preclinical studies [55,56] demonstrated dose-dependent normal tissue sparing with UHDR irradiation. Krieger et al. [57] noted that the threshold total dose strongly influences the FLASH effect. Jones et al. [44] demonstrated that lung injury was significantly reduced under FLASH conditions only when the total dose exceeded 10 Gy. Meanwhile, Cooper [46] reported superior FLASH-RT efficacy at 20–30 Gy, whereas Saade [45] observed optimal therapeutic effects at 30 Gy vs. 40 Gy. These studies indicated that the optimal dose ranges for the manifestation of the FLASH effect may be species-specific and tissue-dependent. The theoretical analysis [43,44] also suggested different dose thresholds and ranges depending on α/β ratios. These findings collectively demonstrate the necessity of considering comprehensive total dose thresholds and ranges in TPS for optimal FLASH-RT implementation.

Consequently, within the FLASH-RT TPS, it is imperative to optimize physical dose distribution and incorporate dose threshold constraints that initiate the FLASH effect. Research has demonstrated that dose thresholds may vary considerably across organs and tumor types, underscoring the need for systematic preclinical validation. In the context of proton FLASH-RT, the notion of DTDR [19] has been proposed to quantify the dose-dose rate joint criterion required for FLASH effect initiation. However, as with the aforementioned dose rate assessment methods, calculation results across different definition frameworks still exhibit significant differences, and the mechanisms underlying these differences’ impact on the clinical efficacy of FLASH-RT remain unclear. Consequently, it is imperative to establish dose rate assessment standards through further systematic research to ensure the reproducibility of treatment outcomes.

Notably, FLASH-RT relies on an accelerator that delivers exceptionally high-dose pulses over a very short duration. Therefore, the beam′s time structure is a critical consideration. Ruan et al. [58] found that the pulse number and frequency influence crypt survival in the abdominal tissue of mice under FLASH irradiation. Almeida et al. [59] reported that variations in irradiation times affect memory preservation in the whole brains of mice. Furthermore, Karsch et al. [60] observed that macro pulsing, which extends treatment duration in the UHDR regime, diminishes the protective effects of FLASH-RT. These studies demonstrate that the time structure is a critical determinant in the FLASH effect. Significantly, the time structure of particle therapy using PBS (protons and heavy ions) differs from that of conventional electron and photon beams. The differential effects of these different time structures on the FLASH effect require further investigation to optimize the time structure and enhance the FLASH effect.

Zayas [61] demonstrated that consistent, reproducible FLASH effects can be achieved when identical physical beam parameters are applied. Therefore, implementing FLASH-RT in TPS requires extensive integration of multiple critical parameters. These parameters exhibit complex interdependencies in the FLASH effect, emphasizing the need for sophisticated treatment planning platforms capable of multi-parametric optimization. Such comprehensive parameter integration is fundamental for maximizing therapeutic ratios and enabling robust clinical implementation of FLASH-RT.

To provide reference ranges for potential FLASH thresholds, we compiled representative dose and mean dose-rate conditions from published preclinical studies using both electron and proton beams [5,8,9,58,64,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86] (

Table 2. Summary of representative in vivo FLASH-RT studies, showing particle type, delivered dose range, mean dose-rate range, and the corresponding biological and functional experimental endpoints.

Particle Type	Dose Range (Gy)	Mean Dose Rate Range (Gy/s)	Experimental Endpoints
Electrons	10–30	30–5.6 × 106	Cognitive function, novel object recognition, astrocyte activation, microvascular integrity
Protons	5–25	100–269	Cognitive performance, memory, neuroinflammatory response
Electrons	5–16	35–940	Crypt regeneration, acute intestinal injury, lymphocyte depletion (spleen/heart), vascular collapse
Protons	12–18	78–110	Crypt regeneration, fibrosis

and Figure 4). These data summarize the experimentally observed parameter spaces associated with normal-tissue sparing across different organs and particle types. Because reported FLASH conditions vary widely between studies—owing to differences in pulse structure, beam modality, biological endpoints, and experimental design—the ranges presented here should not be interpreted as strict thresholds. Instead, they serve as an evidence-based parameter map that highlights where FLASH-associated biological benefits have been observed to occur.

5. Dosimetry: Validation Tools in FLASH TPS

To ensure the accuracy of TPS dose calculation algorithms, dosimetric measurements must be performed for validation. These validations typically involve comparing measured phantom doses with those calculated by the TPS.

Table 3. A summary of phantoms and detectors in UHDR proton and electron TPS validation research.

Phantom	Detector	Gy/s Measurable Range	Dose per Pulse	Temporal Resolution	Saturation Limit	Suitability for FLASH	DADR/DTDR Verification	Reference
Solid water	Radiochromic film	Extremely wide	–	-	–	Proton/ Electron	Suitable for mean dose rate in DADR/DTDR verification	[12,13,30,70,87]
Water phantom, PMMA	Bragg peak chamber	21 Gy/s (99.5% saturation)	0.90 mGy (99.5% saturation)	lon collection time 67 μS	Limited	Not suitable	Not suitable	[33,88]
		42 Gy/s (99.0% saturation)	1.80 mGy (99.0% saturation)
Water phantom	Advanced Markus Chamber	187 Gy/s (99.5% saturation)	2.78 mGy (99.5% saturation)	lon collection time 22 μS	Limited	Proton/ Electron	Suitable for mean dose rate in DADR/DTDR verification	[34]
		375 Gy/s (99.0% saturation)	5.56 mGy (99.0% saturation)

provides a concise overview of the phantoms and detectors utilized to verify dose distributions in proton and electron FLASH-RT treatment plans. During these validation processes, During these validation processes, dose calculations were first performed in the TPS using phantom geometries. Experimental conditions were then configured to match the TPS settings as closely as possible, ensuring consistency of relevant variables. Dose distributions were measured using detectors, and the consistency between the two dose distributions was assessed to confirm the accuracy of the TPS calculations. The categories of phantoms used exhibited no substantial difference between Conv-RT and FLASH-RT. Radiochromic film is a widely used tool for measuring dose distribution, characterized by its excellent dose independence and spatial resolution. Furthermore, Bragg peak chambers, and Advanced Markus chambers were utilized to verify the dose and dose-rate distributions of proton FLASH-RT. These measurements are designed to ensure that any discrepancies remain within clinically acceptable limits.

In addition to summarizing phantom–detector combinations, Table 3 has been expanded to include key dosimetric parameters that determine the suitability of each system for FLASH-RT verification. These parameters include the measurable dose-rate range (Gy/s), dose-per-pulse limits, temporal resolution, and detector saturation behavior. In the context of ionization chambers, the reported temporal resolution corresponds to the ion collection time, which defines the minimum time scale over which rapid dose-rate variations can be resolved. Saturation limits are reported because they directly constrain a detector’s ability to quantify instantaneous dose-rate metrics such as DADR and DTDR. Furthermore, the table explicitly indicates whether each detector is applicable to proton and/or electron FLASH delivery and whether it is capable of verifying dose-rate–dependent metrics required by TPS algorithms that incorporate time-structured beam delivery.

In addition, Table 3 also highlights the current limitations of dosimetry tools when applied to UHDR beams. Verification of TPS outputs under FLASH conditions requires detectors that can simultaneously and accurately capture total dose, dose rate, and temporal structure at dose rates far exceeding those encountered in Conv-RT. Integrating detectors such as radiochromic film, while robust for absolute dose measurement, are inherently limited to mean dose-rate verification and cannot resolve sub-millisecond temporal features of FLASH delivery. Moreover, the extreme dose-per-pulse and instantaneous dose-rate conditions characteristic of FLASH introduce well-documented challenges, including signal loss, saturation effects [89], and irreversible detector damage due to rapid energy deposition in some cases [90]. Consequently, ionization chambers developed for conventional electron linear accelerators exhibit fundamental limitations for real-time dosimetry and instantaneous dose-rate verification under UHDR conditions.

Furthermore, the methods employed for assessing dose rate in proton FLASH-RT using the PBS approach differ from those applied in electron FLASH-RT. Ionization chambers designed for electron linear accelerators may be unsuitable for accurately assessing the dose rate in proton FLASH-RT. Most studies employing two-dimensional ion chamber arrays for patient-specific quality assurance measurements have demonstrated the capacity of these detectors to perform time-resolved measurements. However, they exhibit spatial and temporal resolution limitations, and there is a possibility of saturation at high dose rates or high dose values [91].

As a result, exploring new detectors and innovative dose measurement methods is essential. Dosimetry in FLASH-RT must address additional factors beyond those considered in Conv-RT. These include the dose-rate independence of the detector, spatial resolution, and temporal resolution [66,92]. The following sections will discuss current dosimetry approaches used in UHDR, focusing on real-time dose monitoring and the distribution of dose and dose rate

5.1. Real-Time Dose Monitoring

Real-time dose monitoring involves continuously tracking and recording the instantaneous dose rate and time structure. These approaches can ensure the stability of accelerator output. Consequently, real-time dose monitoring demands detectors with exceptional dose-rate independence and high temporal resolution.

Firstly, luminescent dosimeters generate optical signals proportional to absorbed dose using radiation-induced photons. These have demonstrated outstanding dose-rate independence and temporal resolution [92]. As a result, these properties facilitate their implementation for real-time dosimetric monitoring in FLASH-RT applications [93,94,95,96].

Meanwhile, Cherenkov radiation [97,98,99] has emerged as a prominent research focus in luminescent dosimetry studies. Once the generation threshold for Cherenkov light is exceeded, the emitted photon yield is approximately proportional to the absorbed dose. A study [100] has demonstrated that Cherenkov light is generated instantaneously, with a response time of approximately 10⁻¹² s. This approach demonstrates great potential in evaluating the pulse time structure of electron FLASH-RT.

Other detectors, such as ionization chambers, face significant charge recombination challenges in UHDR environments. They require substantial correction factors [89,100,101,102,103,104]. However, recent innovations have demonstrated promising solutions. Zou et al. [105] developed a high-resolution position-sensitive transmission ionization chamber with a multichannel electrometer. They demonstrated its feasibility for proton pencil beam monitoring under FLASH conditions. Yang [106] characterized a newly designed 2D strip segmented ionization chamber array with high spatial and temporal resolution. These position-sensitive ionization chambers have been demonstrated to effectively monitor dose rate changes at different locations, which is beneficial for accurately assessing the spatial and temporal distribution of PBS at different spots and the resulting changes in voxel dose rates. Researchers [101,107,108] have also investigated reducing electrode distance. Gómez et al. [101] demonstrated that ultra-thin parallel plate chambers (UTIC) achieve over 99% charge collection efficiency under specific UHDR conditions. These findings support the feasibility of ion chambers for FLASH-RT dosimetry and highlight their high potential for future UHDR applications.

Beam Current Transformers (BCTs) have been identified as a promising solution for real-time monitoring of electron beams, with the potential to avoid interference and saturation effects [109]. A BCT consists of conductive windings wrapped around a ring-shaped ferromagnetic core, generating a voltage proportional to the beam current through electromagnetic induction [110]. A key advantage of BCTs in UHDR beam monitoring is its capacity to verify beam parameters such as pulse number, pulse width, and pulse frequency [111].

Ensuring real-time verification of machine output stability will further advance the clinical application of FLASH-RT. It will support consistency between TPS input parameters and actual equipment settings.

5.2. Dose/Dose-Rate Distribution

The dosimetry used for verification must meet stringent criteria to ensure the accuracy of dose calculation algorithms within TPS. These criteria include excellent energy independence, high spatial resolution, and tissue equivalence. Such dosimetric performance is essential to ensure the accuracy of the total dose and mean dose rate used in the TPS biological model.

The diamond detector offers superior spatial resolution and tissue equivalence, with a stable and reliable response within a specific energy range [112,113]. Recently, researchers have designed diamond detectors specifically for UHDR measurements. Studies [114] with commercially available micro-diamond detectors demonstrated excellent performance in ultra-high dose-per-pulse (UHDPP) conditions. In addition, the diamond Schottky diode detector was also proven effective for dose measurement under UHDR conditions. Marinelli et al. [115] successfully employed a diamond Schottky diode detector to measure doses, demonstrating the feasibility of UHDR dosimetry with this technology. Subsequent research confirmed the suitability of the diamond Schottky diode detector for UHDPP and UHDR conditions [116,117]. This advancement represents a significant improvement in the temporal resolution available for the time-structure measurement using charge-based dosimeters.

Concurrently, Yogo et al. [118] performed luminescence imaging of radiation in water. They used a charge-coupled device (CCD) camera under UHDR proton irradiation. Their images revealed the clear visibility of water luminescent under UHDR conditions. Furthermore, the light intensity was linearly correlated with the absorbed dose and exhibited no discernible dependence on dose rate. With appropriate image processing, such luminescence imaging techniques can be used to reconstruct three-dimensional dose distributions.

In addition, chemical dosimeters, such as alanine [119] and Gafchromic films [92,95,120], have demonstrated excellent dose independence and spatial resolution. These chemical dosimeters are therefore commonly used as auxiliary tools to ensure measurement accuracy in dosimetry. Several studies [121,122,123] have shown that radiochromic films could function at dose rates up to 1.5 × 10¹⁰ Gy/s, remaining independent of dose rate effects. While post-irradiation chemical reactions limit real-time monitoring capabilities, ongoing research [124,125,126] is exploring solutions for enhancing real-time applications. Spruijt et al. [91] combined radiochromic films with high-speed cameras to achieve time-resolved thin-film dosimetry for pFLASH-RT and demonstrated the feasibility of effectively assessing changes in voxel dose rates for PBS.

The introduction of UHDR and short irradiation times in FLASH-RT presents challenges. Under these conditions, conventional dose monitoring methods may be insufficient. Consequently, research into dosimetry in the context of FLASH-RT is crucial. Such studies are essential to ensure the precision of TPS dose calculation algorithms and to verify that accelerator parameters used as TPS inputs correspond to the actual equipment settings.

6. Conclusions

Developing and optimizing specialized TPS tailored for FLASH-RT are essential for translating this promising technology into clinical practice. Key considerations include methods to achieve UHDR, ensuring accurate dosimetry, developing robust biological models, understanding critical parameters affecting the FLASH effect, and integrating advanced computational methods. These considerations aim to maximize therapeutic benefits while minimizing potential risks.

Research on the FLASH effect has shown that this phenomenon is observed only when the total dose exceeds a threshold of 8–10 Gy. These studies indicate that both the dose rate and the total dose must meet specific criteria. The instantaneous dose rate and time structure are equally important physical parameters. Although the FLASH effect has been validated in animal models, its underlying mechanisms remain unclear and require further investigation to explore potential biological models and clinical applications.

In this context, investigating biological models in conjunction with key physical and biological parameters is essential. These models enable clinicians to accurately evaluate treatment efficacy and normal tissue risks, providing a scientific foundation for personalized treatment strategies.

Moreover, FLASH-RT TPS faces challenges in dosimetry. Due to its extremely short radiation time and UHDR, traditional dosimetry methods are insufficient. Therefore, developing new dose-distribution and real-time dose-monitoring technologies integrated into TPS quality assurance and quality control workflows is critical. Current technologies include novel ionization chambers, luminescent dosimeters (which may become gold standards for real-time dose monitoring), chemical dosimeters as auxiliary tools to ensure measurement accuracy, and diamond detectors. These technologies are continuously being refined to provide more accurate measurements, ensuring the safety and efficacy of FLASH-RT TPS in clinical applications.

In conclusion, FLASH-RT represents a significant advancement in RT, potentially enhancing tumor control while minimizing damage to normal tissues. Translating this technology from experimental research to clinical applications requires developing specialized TPS, refining dosimetry techniques, and integrating advanced biological models. These efforts are essential to fully realizing the therapeutic potential of FLASH-RT.