Young Investigators Symposium

Short Abstract Preview

Strategies for the delivery of spatially fractionated radiotherapy using conventional and FLASH-capable sources

Purpose: To investigate and compare strategies for delivering novel, spatially micro-fractionated treatments including microbeam/minibeam radiation therapy (MRT/MBRT), GRID and ultra-high dose rate (FLASH) RT.

Methods: A modular, cost-effective multi-slit collimator (MSC) was fabricated to facilitate MRT/MBRT delivery on the small-animal radiation research platform (SARRP). Two 3D-printed tungsten GRID collimators were also procured to facilitate cross-modality comparison. 2D dose distributions, acquired using radiochromic films, allowed for calculation of key performance metrics including dose-rates and peak-to-valley dose ratios (PVDR) at various depths. MSC treatments comprised a single-beam irradiation of a realistic 3D-printed mouse phantom and 360° arc treatment within a 6cm3 square solid-water phantom. The 1.6cmx1.6cmx1.0cm (microGRID) and 5.0cmx5.0cmx1.0cm (miniGRID) collimators were statically placed on the square phantom at minimum and maximum achievable SSD, respectively. A first-generation converter flange and parallel-slit MSC for the TRIUMF ARIEL beamline have been designed and simulated with multi-physics and Monte Carlo software to determine peak outputs and thermal stresses.

Results: The SARRP-mounted MSC produced 9x200μm microbeams while proximal GRID collimation produced a more regular 2D-array of beamlets. The microGRID collimator boasted the largest PVDR, followed by the MSC and miniGRID collimator. Calculated maximum dose rates for the ARIEL FLASH x-ray source were found to be 80Gy/s and 19Gy/s for open-field and 300μm microbeams, respectively, at maximum beam current (10mA).

Conclusion: Experimental minibeam therapies have been implemented on the SARRP. The theoretical performance for a FLASH-capable x-ray source on the ARIEL beamline was evaluated to drive ongoing optimization within practical operational limits.

Long Abstract Preview

Strategies for the delivery of spatially fractionated radiotherapy using conventional and FLASH-capable sources


The present work explores technical implementation of two novel therapies with the capacity to increase the therapeutic efficacy of radiation therapy (RT) over conventional treatment modalities. Both spatially fractionated microbeam/minibeam (MRT/MBRT) and GRID RT as well as ultra-high dose rate (FLASH) RT serve to exploit a differential response between normal and cancerous tissue. The development and evaluation of various techniques for delivering these new therapies in pre-clinical contexts may serve to accelerate their development and eventual translation into clinical practice.


Normal-tissue toxicity and dose constraints to organs at risk (OAR) continue to limit curative potential of RT. In a bid to combat these limitations, emergent modalities capable of reducing radiation-induced toxicity without compromising overall treatment effectiveness have garnered a rapid-growing interest within the physics and oncological community. Spatially fractionated RT, in the form of MRT/MBRT and GRID (2D minibeam array) RT, as well as FLASH RT stand as notable examples, having each demonstrated an extraordinary capacity for normal tissue sparing and selective tumoricidal effects1-4.

In cases where OAR tolerances would otherwise limit the efficacy of conventional RT, both MRT and FLASH RT offer significant utility. The former is characterized by its use of microplanar x-ray beams [<100-μm wide (MRT) or >200-μm wide (MBRT)] to deliver ablative doses within single fractions1,2, whereas the latter is defined by the use of dose rates >30-40Gy/s, depending on target site3,4. While the therapies are common in their apparent ability to enhance the differential response between normal and tumor tissues, they are believed to be driven by different, and as of yet uncertain, biological mechanisms. To elucidate the details of the underlying basis for the increased therapeutic ratio and ultimately allow safe exploitation of the supposed benefits in the clinic, there is a need to develop alternative, and more accessible platforms for translational research.

To address such a demand, the delivery of MRT on small-animal irradiators, such as the image-guided small animal radiation research platform (SARRP), is proving attractive among active research groups5,6. FLASH RT, on the other hand, remains a comparatively underdeveloped area of research, such that there are very few experimental irradiation platforms yet suited to the task of delivering ultra-high dose rate delivery. The present work thus aims to investigate and compare strategies for delivering novel spatially micro-fractionated treatments (MRT, micro-GRID) on conventional and FLASH-capable kilovoltage x-ray sources.

Materials and Methods

A)Microbeam and micro-GRID RT

A portable, cost-effective and modular multi-slit collimator (MSC) was fabricated in order to facilitate MRT/MBRT delivery on commercial SARRP systems. The design of the collimator was previously informed through Monte Carlo (MC) simulations and experiments using a parallel-slit MSC prototype6. This second-generation MSC functions to produce an array of up to 16x200-μm microbeams, separated 300μm on-center, and may be mounted directly onto the SARRP gantry/nozzle assembly. Two 3D-printed tungsten GRID collimators, of varying size and divergence, were also procured to facilitate cross-modality comparison. The smaller, 1.6cmx1.6cmx1.0cm (microGRID) collimator boasted 400μm square apertures and 287.5-μm thick septa; by comparison, the larger, 5.0cmx5.0cmx1.0cm Wolfmet® 3D (miniGRID) collimator had 1200-μm wide hexagonal apertures and 300-μm thick septa.

The effectiveness of each collimation scheme was evaluated using EBT3 Gafchromic films. Key performance metrics, which included the peak and valley dose rates as well as the peak-to-valley dose ratio (PVDR), were evaluated at various depths. Typically, a PVDR>10 is achieved such that valley doses remain <10% of the peak entrance dose.

Four treatment configurations were evaluated, each leveraging the SARRP 220kVp therapy beam (13mA, large focal-spot). MSC treatments comprised a single-beam irradiation of an anatomically correct 3D-printed mouse phantom7 and 360° arc treatment of a 6cm3 square phantom (Figure1a) with doses scored using films oriented parallel to the incident beam direction. GRID irradiations utilized the same beam parameters, however the collimators were placed directly on top of the square phantom, at source-to-collimator distances (SCD) best suited to match their in-built divergence. The highly divergent (SCDideal=10.5cm), microGRID collimator was used along with the SARRP 10cmx10cm treatment nozzle and positioned at the minimum achievable SCD, 30cm (Figure1b). By contrast, the less divergent (SCDideal3m) miniGRID collimator was irradiated with an open field at the maximum possible SCD, 80cm (Figure1c). For each GRID irradiation, films were placed at various depths (0cm,0.3cm,0.9cm,1.8cm,2.7cm). In all cases, suitable irradiation times were estimated from output and depth-dose data so that doses remained within the film’s optimal sensitivity range of 0.2-10Gy.


The TRIUMF ARIEL electron beamline is presently being functionalized to deliver FLASH RT/MRT. To this end, a photon converter flange and parallel-slit MSC (with 300μm slits, spaced 1000μm on-center) were designed for the 300keV section of the ARIEL e-linac with future work aimed at moving towards more clinically-relevant higher energy sections of the beamline (10-30MeV). ANSYS® multi-physics and TOPAS MC software have been used to simulate FLASH irradiations with varying beamline parameters in an attempt to maximize central-axis (CAX) dose rates within acceptable physical and thermal constraints. Dose was scored in a biological target positioned 1cm downstream of a tantalum target, of variable thickness. Parameters to optimize included the electron beam size, beam current, flange thickness and MSC geometry.

Results and Discussion

A)Microbeam and micro-GRID RT

Film (2D) dose distribution data and derived dose profiles (Figure2) help to illustrate the relative effectiveness of each collimation scheme.

Depth dose data and derived PVDR values (Figure3) suggested that using the microGRID collimator, in conjunction with the 10cmx10cm treatment nozzle, offered the most desirable dose distribution characteristics. This is evidenced through the low valley doses and exceptionally high PVDR values achieved (58.9 on surface) while still benefitting from the comparatively high dose rates received at shorter SSD (Figure3g-i). Furthermore, the depth dose data reveal a unique linear PVDR trend, clearly driven by the peak dose behaviour, which could potentially improve sparing within superficial structures.

The SARRP-mounted MSC produced a parallel array of 9x200-μm wide microbeams with an approximate field size of 4mmx8mm (Figure2a). This contrasts with the expected 8mmx8mm array of 16 uniform beamlets which points to a difference between the effected and intended slit divergence of the MSC. Nevertheless, there is significant utility derived from using a gantry-mounted collimator to deliver non-static irradiations. The single-beam MRT irradiation of the mouse phantom (Figure3a-c) boasted the second-highest surface PVDR (12.9), albeit with a stronger depth dependence that may have resulted from increased phantom scatter and beam broadening at larger SSD.

Use of the large, low-divergence, miniGRID collimator yielded superior uniformity across the sampled dose region (Figure2e,f). On the other hand, the increased ratio of aperture width to septal width and the significant beam divergence at large SSD promoted a sizeable increase in valley dose and consequent reduction in PVDR (6.2 on surface) (Figure3j-l).


A first-generation photon converter flange for the ARIEL beamline was designed (Figure4) with ongoing iterations aimed at improving heat loading and CAX dose rates for both open-field and MRT/MBRT FLASH irradiations.

Calculated peak dose rates for the simulated beam were 62Gy/s and 19Gy/s for open-field and MRT configurations, respectively (Figure5). These values correspond to an electron beam size of 1cm (2σ), however smaller beam sizes, were found to increase CAX dose rates, up to 80Gy/s for a 5mm beam-spot size (Figure5e).

The current open-field dose rates are suitable for FLASH irradiation as they lie above the ~40Gy/s threshold defined in the literature4,5, but further optimization is required to ensure FLASH dose rates are achieved at reduced beam power (Figure4d). Minimizing the aluminum flange and tantalum target thickness maximized the dose rate, but the consequent reduction in thermal mass resulted in substantially increased temperatures for critical structures. The maximum PVDR achieved in the FLASH-MRT simulations was 7.7, though this value is strongly dependent on the choice of parameters such as slit size, septal thickness and beam spot size, which remain to be finalized. 


Experimental microbeam and GRID therapies have been implemented on the SARRP system using various delivery strategies and collimator designs. The best overall performance was observed with the 3D-printed microGRID collimator, though this required a static collimation setup, in contrast with the gantry-mounted MSC. The theoretical performance of a prototype FLASH-RT photon converter on the TRIUMF ARIEL beamline was also evaluated with a focus on optimizing performance within practical operational limits of the system. Dose rates of up to 80Gy/s for the open beam were obtained and ongoing efforts are being aimed at functionalizing the beamline to deliver FLASH-MRT at dose rates above 40Gy/s.


  1. D.N. Slatkin et al. (1995) Med.Sci. 92, 8783-8787.

  2. F.A. Dilmanian et al. (2007) Exp.Hematol.35, 69-77.

  3. V. Favaudon et al. (2014) Sci.Trans.Med. 6, 245ra93-245ra93.

  4. P. Montay-Gruel et al. (2018) Radiother.Oncol. 129, 582-588.

  5. Y. Prezado et al. (2017) Sci.Rep. 7, 17295.

  6. N.M. Esplen et al. (2018) Phys.Med.Biol. 63. 175004

  7. N.M. Esplen et al. (2019) Med.Phys. 46, 1030-1036.