Dosimetric Characterization of a Rotating Anode X-Ray Tube for FLASH Radiotherapy Research

Abstract

Purpose:

Most current research toward ultra-high dose rate (FLASH) radiation is conducted with advanced proton and electron accelerators, which are of limited accessibility to basic laboratory research. An economical alternative to charged particle accelerators is to employ high-capacity rotating anode x-ray tubes to produce kilovoltage x-rays at FLASH dose rates at short source-to-surface distances (SSD). This work describes a comprehensive dosimetric evaluation of a rotating anode x-ray tube for potential application in laboratory FLASH study.

Methods and Materials:

A commercially available high-capacity fluoroscopy x-ray tube with 75 kW input power was implemented as a potential FLASH irradiator. Radiochromic EBT3 film and thermoluminescent dosimeters (TLDs) were used to investigate the effects of SSD and field size on dose rates and depth-dose characteristics in kV-compatible solid water phantoms. Custom 3D printed accessories were developed to enable reproducible phantom setup at very short SSD. Open and collimated radiation fields were assessed.

Results:

Despite the lower x-ray energy and short SSD used, FLASH dose rates above 40 Gy/s were achieved for targets up to 10-mm depth in solid water. Maximum surface dose rates of 96 Gy/s were measured in the open field at 47 mm SSD. A non-uniform high-to-low dose gradient was observed in the planar dose distribution, characteristic of anode heel effects. With added collimation, beams up to 10-mm diameter with reasonable uniformity can be produced. Typical 80%–20% penumbra in the collimated x-ray FLASH beams were less than 1 mm at 5-mm depth in phantom. Ramp-up times at the maximum input current were less than one millisecond.

Conclusion:

Our dosimetric characterization demonstrates that rotating anode x-ray tube technology is capable of producing radiation beams in support of preclinical FLASH radiobiology research.

Introduction

FLASH radiotherapy is a novel therapeutic technique in which radiation dose is delivered at ultra-high dose rates, at 2 to 3 orders of magnitude higher than that used in conventional radiotherapy^1,2. Preclinical studies have shown that FLASH irradiation results in reduced normal tissue toxicities, while achieving tumor control probability similar to conventional dose-rate irradiation ^3,4. The biological effects of FLASH irradiation, known as FLASH effects, have been observed in in vivo models at specific radiation conditions of dose and dose rate^5–8. Mean dose rates higher than 40 Gy/s have generally been considered as a minimum threshold to induce FLASH effects, however the absolute dose appears to be tissue and endpoint specific. The prospect of widening the therapeutic ratio through FLASH has led to rapid development of technology ⁹ and research programs to accelerate translation ^10,11. The first human patient with a cutaneous T-cell lymphoma was treated with FLASH electron therapy in 2018, and the first-in-human clinical trial of FLASH proton therapy on the safe operation for pain relief of symptomatic bone metastases in extremities was recently completed (FAST-01)^12–14. Despite the significant enthusiasm for clinical translation, the underlying mechanisms of FLASH-RT remain to be elucidated ².

Equipment accessibility and cost are major factors limiting broad laboratory research toward the underlying mechanisms of the FLASH effect. The majority of present FLASH irradiators are complex machines that produce high-energy charged particles, commonly 4 – 10 MeV electrons produced by linear accelerators or 70 – 230 MeV cyclotron-accelerated protons ⁴. Recent works have explored x-ray tube technology to accelerate laboratory discovery and translational research. Bazalova-Carter and Esplen (2019) examined the use of 160 kVp stationary-anode x-ray tubes for this purpose ¹⁵. At the maximum input power (i.e. 6 kW), superficial dose rates above 100 Gy/s were achievable in unfiltered x-ray beams. However, the dose rates decreased rapidly below 40 Gy/s beyond the initial 2-mm depth in a plastic water phantom, suggestive of a high dose contribution from low-energy x-ray and electron scatter from the target assembly. An alternative approach by Rezaee et al (2020) explored the use of rotating anode x-ray tubes for FLASH applications, which are capable of significantly higher input power and heat-loading capacity compared to stationary anode x-ray tubes ¹⁶. Through a Monte Carlo study, Rezaee demonstrated that dose rates above 40 Gy/s could be achieved at 20 mm depth in water and above 80 Gy/s for depths less than 5 mm with 100 kW input power and 0.025 mm Cu filtration. Rezaee also proposed a parallel-opposed arrangement of dual rotating anode x-ray sources to achieve better dose rate uniformity, and dose rates above 100 Gy/s, throughout a 2-cm thick target.

Given adequate input power, rotating anode x-ray tubes are a more economical option for FLASH radiation research, both in terms of equipment and shielding costs. FLASH effects have previously been observed in vivo from synchrotron and linear accelerator-generated x-rays, supporting the use of x-ray tubes to induce FLASH effects ^17,18. Kilovoltage x-rays from rotating anode x-ray tubes thus hold promise as a viable means to produce FLASH-capable irradiation for the laboratory. In support of the Monte Carlo simulations published by Rezaee et al. (2020), we have characterized dosimetric properties of a high-powered rotating anode x-ray tube for potential applications in preclinical FLASH radiotherapy research.

Materials and Methods

FLASH x-ray system and dosimetry

A high-capacity rotating anode x-ray tube, the RAD-44 model (Varex Imaging Co., Salt Lake City, UT), with a 75 kW generator (CPI International, Palo Alto, CA) refurbished by Gulmay, Inc. (Suwanee, GA), was implemented for x-ray FLASH radiation study (Figure 1). This system features a 16-degree tungsten-rhenium-molybdenum anode with 0.7 mm aluminum (Al) inherent filtration and additional 0.025 mm copper (Cu) filtration to limit low-energy x-rays and contamination from electrons. The x-ray tube operates at a peak voltage of 150 kV and maximum input current of 500 mA. The measured HVL with these parameters was 0.17 mm Cu. In this study, the x-ray source was operated in radiographic mode to deliver FLASH x-rays in a single pulse of 10 – 500 ms. Per NCRP-99 and reported performance for other models of radiographic and fluoroscopic tubes with rotating anode technology ^19,20, the exposure times from our three-phase x-ray generator are accurate to within 5% ²¹.

Figure 1:

Photograph and schematic of the RAD44-based x-ray FLASH system. (A) x-ray tube installed on aluminum scaffolding, with polycarbonate z-translation stage. (B) the flange of the x-ray tube. The exit window of the tube lies at the center of a conical opening in the tube, lined with lead to mitigate scatter out of field. (C) Cross section diagram of the x-ray tube flange (to-scale). The distance between the focal spot and the outer edge of the flange is 56 mm. The flange is 34.5 mm at its widest and 18.8 mm deep. The exit window on the x-ray tube is 16.7 mm in length.

All radiation exposures were performed using the larger, nominal 2-mm focal spot. Confirmation of the focal spot size was performed using the pinhole technique for comparison with manufacturer specifications ²². A 0.5-mm diameter pinhole collimator was used to expose Gafchromic™ EBT3 film (Ashland Global, Wilmington, DE) at an extended SSD of 456 mm, 300 mm beyond the exterior surface of the x-ray tube. The SSD was chosen to achieve desirable intensity signal on EBT3 films without excessive heat burden to the x-ray tube. The FWHM was scaled based on the magnification factor of 8.44 to determine the focal spot dimensions. Additional measurements were made to assess output linearity over the full dynamic range of tube input currents and exposure times, from as low as 10 mA and 1 ms using a PTW30013 ionization chamber (PTW Freiburg, Germany) in air. An extended SSD (66 mm) was used for ion chamber measurements to limit potential charge collection deficiencies from ion recombination effects. The magnitude of ion recombination effects was measured using the 2-voltage technique recommended in TG-61 ²³. End effects from tube ramp up time were quantified using the graphical extrapolation method ²³.

Beam profiles were measured using calibrated EBT3 films, which have been established to have minimal dose-rate dependencies in FLASH research ²⁴. Film calibration irradiations were performed at the University of Wisconsin Medical Radiation Research Center (UWMRRC) using a NIST-traceable 120 kVp x-ray beam with 3.0 mm Al and 0.1 mm Cu filtration. This source was chosen for its similar x-ray spectrum, average energy, and anode material to the RAD44 x-ray, as inferred by comparison of the measured and calculated HVL in a previous work ^16,25. EBT3 film cutouts were irradiated in incremental doses up to 10 Gy and returned to our institution for digitization using an Epson 11000XL flatbed scanner. Scanning of all films occurred at a minimum of 24 h after irradiation. A scanning resolution of 300 dpi was used, with no compression or color correction. The resulting calibration curve was applied to films exposed at 150 kVp using the FLASH x-ray system to determine the delivered dose in each experimental condition. For each film irradiated using the FLASH x-ray tube, the average and standard deviation of dose rate were calculated from a region of interest defined within the high-dose irradiated area. All films used in this study originated from the same batch. Dose uncertainty from the film calibration irradiations was certified as 3.6% at the 95% confidence interval by the accredited dosimetry calibration laboratory.

TLD microcubes (TLD-100; LiF:Mg, Ti; 1 cubic-mm) were employed for secondary independent output measurement²⁶. Prior works have demonstrated dose-rate independent response of TLDs from x-ray and electron radiation at dose rates up to 10⁹ Gy/s ²⁴. TLDs were shipped to our laboratory from UWMRRC for irradiation and returned for remote readout. TLD calibration irradiations were performed using the same NIST-traceable 120 kVp x-ray beam and filtration used for film calibration. TLD processing and analysis has been described in detail elsewhere ²⁷. Uncertainty from the TLD dose calibration was estimated at 5% at the 95% confidence interval.

Dose rate was quantified from all detectors given as the measured dose divided by the requested exposure time. Accurate film and TLD positioning were ensured using a custom plastic docking system (Fig. 2) 3D printed with an Ultimaker S3 (Ultimaker, Utrecht, Netherlands) system. The docking system was designed with three major components: a mounting plate, a solid water tray, and an adaptor. The mounting plate consists of a flat plate bolted onto the surface of the x-ray tube as a rigid point of attachment. Two solid water trays were designed to allow setup of solid water stacks both inside the ‘flange’ of the x-ray tube housing at very short SSD (47 mm) and outside the ‘flange’ at an extended SSD (60 mm) (Fig. 1B–C). The solid water trays were coupled to the mounting plate using an adaptor piece, to hold a custom collimator and the solid water phantom at a fixed location relative to the x-ray focal spot and allowed film measurements at known SSDs of 47-mm or 60-mm, respectively, with an estimated uncertainty of 0.5 mm. The dose rate measurement uncertainty from phantom setup was estimated at 2.9% (standard deviation) from triplicate film measurements at both assessed SSDs.

Figure 2:

3D printed docking tool to reproducibly align solid water phantoms with the x-ray FLASH beam. A tray (left; blue) holds up to 2cm depth of solid water. Trays are designed for two sizes of solid water pieces, for positioning at short SSD (top row) or at the x-ray tube surface (bottom row). The trays feature notches that index into the flange adaptor pieces (middle). The adaptors are designed to rigidly hold a collimator in a fixed position over the solid water phantoms. Lastly, the adaptors are rigidly docked onto the x-ray tube using a mounting plate (right) bolted onto the exterior of the x-ray tube.

Dosimetric characterization

Percent depth dose (PDD) and beam profiles were measured in CIRS kV-compatible solid water phantoms (CIRS, Inc., Norfolk VA) using calibrated EBT3 film and TLD microcubes, irradiated in an open field at the two SSDs of 47- and 60-mm. Each EBT3 film and TLD microcube irradiation was performed individually to limit confounding attenuation effects from overlying dosimeter material. Each measurement included 2-cm of solid water distal to the detector to maintain consistent backscatter material. Film cutouts were placed horizontally between the solid water slabs to measure beam profiles and PDD in 5-mm increments up to 20-mm depth. Beam profiles were sampled from EBT3 films parallel and orthogonal to the cathode-anode direction at the phantom surface. Dose-rate-area histograms (DRAH) were calculated for each phantom depth and used to estimate volumetric dose rate distribution. This information was utilized to determine the field size for each biological sample where FLASH dose rates (>40 Gy/s) can be achieved in entire sample volume, in addition to the quantification of dose rate non-uniformity in the volume. Depth dose rate along the central axis was also measured up to 10mm depth in phantom using TLDs loaded into a 1.5 mm diameter, 1 mm depth wells machined into the kV solid water. TLDs were loaded and recovered from the phantoms using vacuum tweezers to minimize surface contamination.

Collimated x-ray beams were characterized using circular 2-mm, 5-mm, and 10-mm diameter collimators drilled from 3-mm sheets of lead. The openings of the collimators were centered in the middle of the open beam selected for FLASH irradiation. A fixed 2-mm air gap was maintained between the collimation and phantom surface. Beam outputs were measured as a function of field size using EBT3 film at the surface of the 20 mm thick solid water phantom, at both 47- and 60-mm SSD, and compared with TLD readings. TLD positions were aligned to collimation with an estimated uncertainty of 0.5 mm using the phantom setup tool described in the previous section. Beam profiles were measured using EBT3 film positioned perpendicular to the beam axis. Profiles were sampled at the phantom surface and in 5-mm depth increments up to 20 mm in solid water at 47-mm SSD for the highest achievable dose rates. Flatness, symmetry, and beam penumbra were quantified from each measurement. In addition, changes in beam penumbra and PDD were quantified as a function of phantom depth from the exposed films.

Results

FLASH-capable x-ray tube characteristics

The pinhole measurements showed the large focal spot of the RAD44 to be 2.31 mm parallel to the cathode-anode direction and 1.56 mm in the perpendicular direction (Figure 3A–B). These dimensions are within the maximum acceptable limits described in NCRP 99 for a nominal 2-mm focal spot ²¹. The output of the FLASH x-ray tube was highly linear with tube current and exposure time (R² > 0.99; Figure 3C–D). Output linearity with exposure time confirms constant dose rate within the exposure, allowing the delivery of desired total dose by adjusting the exposure duration. Output linearity with tube current allows the direct adjustment of dose rate with current specification. Through simple control of the exposure duration and tube current, the x-ray tube can be used to deliver a wide range of doses at different dose rates. In addition to the linear response, the ion recombination correction (P_ion) calculated from the 2-voltage technique remained below the suggested limit of 1.05 recommended in AAPM TG-61. Charge collection inefficiencies were not significant for our measurements performed at extended SSD of 66 mm. Ramp-up times were minimal, measured at 0.13 ± 0.01 ms using the graphical extrapolation in the end effect method.

Figure 3:

Characteristics of the RAD44 x-ray tube. (A) The focal spot was measured with EBT3 using the pinhole technique. (B) Intensity profiles of the focal spot image were wider along the cathode-anode (parallel; 2.31 mm) direction than perpendicular (1.56 mm). (C) Measured end effects are sub-ms (0.13 ± 0.01 ms), indicating negligible ramp-up times during delivery. (D) Output measurements were highly linear with both requested exposure time and input current. Note that error bars may be smaller than the plot markers.

Open field dosimetry

Figures 4A and 4B show the cross-beam dose distributions at the surface of the water phantom at 47 mm and 60 mm SSD, respectively. The field size at 47 mm SSD was 20×20 mm² and increased to 26×26 mm² at 60 mm SSD due to beam divergence. A prominent in-field dose gradient was observed from uncompensated anode heel effects along the inline (cathode-to-anode) direction (Fig. 4C). At the phantom surface at 47 mm SSD, the maximum dose rate of 96.5 ± 4.5 Gy/s was measured at a 6-mm offset from the beam central axis, toward the cathode-side of the field. The surface dose rate decreased to 75.7 ± 3.5 Gy/s at the same 6-mm offset toward the anode-side of the field. With depth in solid water (Fig. 4D), the beam intensity at 47 mm SSD falls to 63.7 Gy/s at 5-mm depth (66% of the surface dose rate) and 46.32 Gy/s (48% of surface dose rate) at 10-mm depths in solid water at 47mm SSD.

Figure 4:

Open field EBT3 film dosimetry at the SSD of 47 mm (A) and extended SSD of 60 mm (B) at the surface of a solid water phantom. Due to the limited beam filtration used to maintain high dose rates, anode heel effects result in a non-uniform beam. The beam FWHM at 47 mm SSD is 20 mm and increases to 26 mm at 60 mm SSD. Dose maps are set to the same spatial scale for comparison. (C) Inline and crossline profiles at the surface of the phantom for both SSDs. The open field is highly sloped in the inline direction, with the high dose regions near the cathode-side of the beam. At extended SSD, the dose rate decreases but the slope of the in-field dose gradient also decreases. (D) Depth dose curves for both SSD setups indicate a steep dose falloff with depth in solid water. The PDD was sampled at the center of the open field, which is at the crossing of perpendicular dashed lines in panels A and B. TLD and film PDD curves agree to within 3%. The error bars represent the uncertainties from film and TLD calibration and positioning setup. Error bars, if not visible, are smaller than the plot markers.

At 60mm SSD, the in-field dose rates decreased from inverse square effects and ranged from a maximum of 46.9 ± 2.2 Gy/s (7-mm from central axis) to 36.9 ± 1.7 Gy/s at the phantom surface. The depth-dose gradient became shallower at 60mm SSD, falling to 33.8 Gy/s (72% of surface dose) and 27.2 Gy/s (58% of surface dose) at 5-mm and 10-mm depths, respectively. The SSD of the rotating anode system can be an additional adjustable parameter to control dose rate.

DRAH for open fields at 47- and 60-mm SSD as a function of depth are shown in Figure 5A–B. At 47-mm SSD, the entire radiation field (i.e., 400 mm²) exceeds 40 Gy/s at the phantom surface, decreasing to 88% (387 mm²) at 5-mm, 42% (204 mm²) at 10-mm, and 0% beyond 10-mm (Fig. 5C). In contrast, only 33% of the field (i.e., 226 mm²) exceeds 40 Gy/s at the phantom surface at 60-mm SSD, decreasing to 0% at 5-mm depth and beyond. While SSD may be used as an adjustable parameter to magnify the radiation field or relax geometric constraints from the x-ray tube flange, the reduction of dose rate from inverse square effects can restrict the usable portion of the beam to superficial targets.

Figure 5:

Dose-rate area histograms as a function of depth for open-field x-ray FLASH irradiations at 47-mm SSD (A) and 60mm SSD (B). Increasing the SSD can be used to increase the field size and uniformity of the radiation field at the cost of reduced dose rate. In this example, despite the larger overall field size at the SSD of 60mm, the usable surface field size, where FLASH dose rates are achieved, decreases by nearly a factor of two (C). Error bars represent the uncertainties from the film calibration and positioning setup. Error bars, if not visible, are smaller than the plot markers.

Collimated field dosimetry

Figure 6A–I shows the effect of field size on the delivered dose rate at 47 mm SSD. Inline and crossline beam profiles as a function of depth in kV solid water and field size are shown for collimated x-ray FLASH beams with diameters of 2-mm (Fig. 6A–B), 5-mm (Fig. 6C–D) and 10-mm (Fig. 6E–F). The mean dose-rates increased with field size, ranging from 59.7 ± 3.2 Gy/s to 92.2 ± 5.0 Gy/s for 2-mm and 10-mm diameter fields, respectively. Anode heel effects were discernible in the 10-mm diameter beams, but minimal in the smaller fields. EBT3 film and TLD measurements at the central axis on the phantom surface agreed to within 5% (Fig. 6G). PDDs were comparable between all field sizes (Fig. 6H). Beam penumbra widths, i.e., the separation between the 80% and 20% dose falloff, increased monotonically with depth (Fig. 6I). Penumbra widths at shallow depths (≤ 5 mm) were less than 1 mm for all field sizes. The penumbra increases with field dimension due to the greater collimator opening at the short SSD. The penumbra for the 10-mm collimated beam has a noticeably higher variation between inline and crossline samples due to more prominent anode-heel effects, which causes larger error bars as shown in Figure 6I. A summary of measured dosimetric parameters at the phantom surface including flatness, symmetry, penumbra, output, and beam FWHM as a function of field size is shown in Table 1.

Figure 6:

Crossline profiles for the collimated beams as a function of phantom depth for (A) 2-mm, (B) 5-mm, and (C) 10-mm diameter fields at 47 mm SSD. Inline profiles for the collimated beams as a function of phantom depth for (D) 2-mm, (E) 5-mm, and (F) 10-mm diameter fields at 47 mm SSD. Data shows widening of the FWHM and penumbra with depth and field size. The dose rate increases with field size (G). The PDD did not significantly change with field size (H). Penumbra widths increased with field size (I). The error bars represent the standard deviation calculated from two independent measurements of each parameter.

Table 1:

Collimated beam metrics from EBT3 film measurements at the surface of a kV solid water phantom. Uncertainties are estimated from the film and TLD calibration and positioning setup. TLD measurements were on average 2.7% lower than EBT3 measurements.

SSD	Nominal Field Size (mm diam.)	EBT3 Avg. Dose Rate (Gy/s)	TLD Dose Rate (Gy/s)	Relative difference btw TLD and EBT3	FWHM (mm)	80%-20% Penumbra width (mm)	Flatness	Symmetry
47 mm	2	59.71 ± 3.17	59.50 ± 3.44	−0.35%	1.91 ± 0.03	0.17 ± 0.02	0.02	1.01
	5	77.99 ± 4.19	75.27 ± 4.35	−3.49%	5.42 ± 0.08	0.36 ± 0.04	0.02	1.03
	10	92.15 ± 4.95	87.45 ± 5.05	−5.11%	10.80 ± 0.16	0.74 ± 0.08	0.04	1.06
60 mm	2	38.54 ± 2.06	37.53 ± 2.17	−2.62%	1.91 ± 0.03	0.23 ± 0.03	0.04	1.00
	5	43.45 ± 2.36	42.32 ± 2.44	−2.60%	4.95 ± 0.08	0.36 ± 0.04	0.03	1.04
	10	44.90 ± 2.36	43.93 ± 2.54	−2.17%	11.22 ± 0.17	0.32 ± 0.04	0.02	1.04

Discussion

FLASH dose-rate irradiation has immense potential to enhance the therapeutic ratio of radiotherapy. However, clinical translation necessitates comprehensive laboratory studies to elucidate the still-unclear biological, biochemical, and biophysical mechanisms of the FLASH effect. It would be ideal to extend the capabilities of self-shielded orthovoltage x-ray (160 – 320 kVp) cabinet systems, commonly used for laboratory radiation research, to support preclinical FLASH irradiation of small animals and biological samples ²⁸. Unfortunately, low bremsstrahlung yield and limited heat capacity have rendered the current stationary anode technology inadequate to support the requisite dose rates for in vivo FLASH irradiation. Input power and heat capacity limits would need to be increased by more than one order of magnitude to achieve FLASH dose rates with the currently implemented stationary anode technology¹⁶. For kV x-rays, these requirements are currently achievable only with the rotating anode technology employed in fluoroscopic and tomographic imaging systems. Even with a generator of sufficient high power, the requisite for dose rates above 40 Gy/s dictates that x-ray FLASH irradiations need to be conducted at very close distances between the source and subject with minimal filtration. At these conditions, the characteristics of the x-ray beam are susceptible to several important factors including inverse-square effect, heel effect, finite size of focal spot, and field size. Our dosimetric investigation in this work characterizes these factors in a high-capacity radiographic imaging system, a RAD-44 model x-ray tube, which has been suggested as a FLASH x-ray irradiator for preclinical laboratory research¹⁶.

The RAD-44 x-ray source with a 75-kW generator can produce 150 kVp x-rays at dose rates up to 96.5 Gy/s at the surface of a solid water phantom located at 47 mm SSD. At this close distance to focal spot, the dose gradient is 6.8% per mm in the first 5-mm thickness of solid water phantom. The high depth-dose gradient limits potential FLASH irradiations with this system to superficial samples with a thickness up to 10 mm, as shown in Figures 4 and 5. In-plane dose non-uniformities were also present in the beam due to anode heel effects, where any additional compensating filtration would be detrimental to the dose rate. SSD may be considered an adjustable parameter to reduce the steepness of the in-plane and depth-dose gradient at the expense of rapid decrease of dose rate from inverse-square effects, which may not be sufficiently high for FLASH effect research. At 60 mm SSD, for example, the depth-dose gradient reduced to 2.7% per mm while the surface dose rate decreased to 43.9 Gy/s in a 10-mm diameter field (Table 1). Although the overall field size has increased by 70% in comparison to the beam at 47-mm SSD, the usable field size for FLASH irradiation (>40 Gy/s) has reduced to 250 mm². Higher power x-ray tubes and generators can prospectively increase dose rate, which allows FLASH irradiation at larger SSDs. For example, a prospective power increase from 75 to 150 kW increases dose rates by a factor of two times, which can achieve surface dose rate of 40 Gy/s at 88 mm SSD. A parallel-opposed x-ray system can potentially overcome the non-uniform depth dose resulting from the single x-ray tube utilized in this work ¹⁶. Prior modeling studies demonstrated uniform depth-dose can be achieved for a 2-cm thick sample, and anode heel effects would be compensated by the composite dose distributions from mirrored beams of two opposing sources.

Beam sharpness and narrow penumbra widths are distinctive characteristics of kV x-ray beams compared to higher-energy charged particle beams ²⁹. Most x-ray fields evaluated in this study, from 2 – 10 mm in diameter, had very narrow penumbra widths of less than 0.5 mm at shallow depths in phantom. In contrast, the lateral dose spreads of MeV electron beams and passively scattered proton beams would range from a few mm to 1 cm, which may not be appropriate for the preclinical FLASH research on small animals ^30,31. FLASH effects in the penumbra region where dose and dose rates reduce to 20% of intended dose have not yet been evaluated in current preclinical FLASH studies. Given adequate input power, a kV x-ray beam can be prospectively attenuated at the field edge to broaden the penumbra and emulate that of a charged particle beam. kV x-ray pencil beams with modified penumbra may then be advantageous to support research on the local-regional nature of FLASH radiotherapy, such as evaluation of the biological impact of ultra-high dose rates in pencil beam scanning, grid, and mini-beam radiotherapy techniques.

Preliminary studies have begun to utilize this kV x-ray FLASH platform for in vivo and in vitro experiments. The geometric constraints of this system necessitate careful animal placement to reproducibly irradiate targets at short SSD, and limit achievable targets to those at superficial depths. Although experiments remain in progress, this kV FLASH system can be utilized for the irradiation of some in vivo models such as murine skin, flank tumor, murine and rodent eye^32–33. Successful use of the rotating anode x-ray tube for FLASH research has required the development of robust immobilization devices to elevate and mechanically register samples into the x-ray tube, ensuring that SSD requirements are met and that samples are placed in a consistent region of the non-uniform field.

Conclusion

Rotating anode x-ray tubes, such as the RAD-44 utilized in this work, are capable of producing x-rays at dose rates above 40 Gy/s that have been reported to induce FLASH effects. The depth-dose characteristics of the described 150 kVp beam with sharp lateral penumbra are suitable for the irradiation of superficial targets in both in vitro and in vivo studies. Tube current, exposure time, and SSD are adjustable parameters that allow flexibility in the delivered dose and dose rates. This work serves as an important initial step in the development of a FLASH-capable laboratory irradiator for basic and translational laboratory research on the FLASH effects.