Microbeam Radiation Therapy (MRT) is a drastically novel approach of radiosurgery, challenging the understanding of how normal tissues respond to radiation therapy, where X-rays generated by a synchrotron light source are collimated into wafers of high-dose (hundreds of Gray), thin (50 µm) parallel microbeams. Recent work showed that MRT dose prescription must consider a dual approach: separating normal tissue dose constraints tied to the valley dose prescription and anti-tumor effects, depending on the prescribed number of ports. Here, we consolidate this assumption by increasing the number of MRT incidences to irradiate 9L-bearing rats with a cumulated valley dose of 10 Gy. 9L-bearing rats were irradiated by eight ports of conventional Broad Beam (BB) or MRT. Animals were followed up clinically by MR imaging, histopathology and survival analysis. MRI follow-up revealed that MRT significantly amplified antitumor effect and tumor tissue damage while improving the clinical score of animals. Ten Gy, delivered by either BB or MRT, increased median survival time to 15 and 39.5 days after irradiation, respectively. Neither complete ablation of brain tumors nor long-term survival was achieved, but multiport MRT (eight ports) showed an outstanding dose equivalence factor (×2.9), which needs to be tested in a clinical environment.

Purpose: Both local tumor control and distant metastasis are important indicators of the efficacy of radiation therapy treatment. Synchrotron microbeam radiation therapy (MRT), spatially fractionated radiation delivered at ultrahigh dose rates, shows remarkable normal tissue sparing with excellent local control in some models. Some MRT regimens trigger an antitumor immune response that contributes not only to the local but also to systemic treatment efficacy. Despite recent advances in the treatment of primary breast cancer, metastatic disease is still the major cause of treatment failure in the clinic. Here, in an aggressive preclinical triple-negative breast cancer model, we compared local tumor response and metastasis following different MRT treatment programs. Methods and Materials: 4T1.2 mouse mammary tumors were treated with 300 Gy peak/7 Gy valley dose MRT and/or 8 Gy broad beam (BB) radiation, all delivered as daily fractionated programs (3 consecutive daily sessions of either MRT or BB or 1 MRT combined with 2 BB sessions, the first or last of the 3 fractions). The mice were euthanized on day 9 post last irradiation, when unirradiated control animals reached an ethical endpoint. Primary tumors were collected to evaluate immune cell prevalence, while lungs, spinal cords, and locoregional lymph nodes were collected to measure metastatic burden. In parallel, local tumor growth and survival were monitored. Results: The combined MRT/BB treatment shifted the balance between pro- and antitumorigenic macrophages toward the accumulation of antitumorigenic macrophages in the tumor. Monitoring of the tumor volume and animal health indicated the benefit of the combined MRT/BB treatment for local control and treatment tolerance, while animal survival was only marginally longer for one combined schedule. The metastatic burden was similar for all 4 treatment schedules. Conclusions: The addition of a single MRT to BB treatment improved the primary tumor response. This provides a basis for future experiments incorporating adjuvant immunotherapy or chemotherapy to improve local and systemic treatment outcomes.

BACKGROUND:In clinical radiotherapy, the patient remains static during treatment and only the source is dynamically manipulated. In synchrotron radiotherapy, the beam is fixed, and is horizontally wide and vertically small, requiring the patient to be moved through the beam to ensure full target coverage, while shaping the field to conform to the target. No clinical system exists that performs both dynamic motion of the patient and dynamic shaping of the beam. PURPOSE:We developed and tested a new dynamic treatment delivery system capable of delivering conformal fields with a robotic patient positioning system for use on the Imaging and Medical Beamline (IMBL) at the Australian Nuclear Science and Technology Organisation, Australian Synchrotron. METHODS:An industrial robotic manipulator was modified to enable dynamic radiotherapy treatments on IMBL. The robot, combined with a carbon-fiber treatment couch-top and a recently developed dynamic collimator, formed the basis of the new treatment delivery system. To synchronize the motions of the robot and collimator, a real-time, hardware-based event-handling system was utilized. To test the system, a ball bearing in a medical physics phantom was treated with circular fields ranging from 5 to 40 mm in diameter and at treatment speeds from 2 to 50 mm  s - 1 ${\rm s}^{-1}$ . The position of the ball bearing was compared to the center of the circular fields and the positional and temporal accuracy of the treatment delivery system was assessed, and appropriate treatment margins for the system were determined. RESULTS:The vertical position of the ball bearing varied with treatment delivery speed ( - 1.06 to 0.93 mm $-1.06 \, {\rm to}\, 0.93 \,\mathrm{mm}$ ) while the horizontal position remained consistent ( - 0.05 to 0.09 mm $-0.05 \,{\rm to}\, 0.09 \,\mathrm{mm}$ ). The time-delay between the robot and the collimator remained consistent ( - 35.5 ms to 18.5 ms $-35.5 \,\mathrm{ms}\,{\rm to}\, 18.5 \,\mathrm{ms}$ ) at treatment speeds above 2 mms - 1 $2 \,\mathrm{mm s}^{-1}$ . Data at 2 mm s - 1 $2 \,\mathrm{mm}\mathrm{s}^{-1}$ was right at the edge of both the robot capabilities and the analysis technique, and had larger variations in timing ( 0.0 ms to 57.9 ms $0.0 \,\mathrm{ms}\,{\rm to}\, 57.9 \,\mathrm{ms}$ ). Horizontal margins of 0.51 mm $0.51 \,\mathrm{mm}$ and vertical margins of up to 2.3 mm $2.3 \,\mathrm{mm}$ were calculated for the treatment delivery system. CONCLUSIONS:We have implemented the first robotic treatment delivery system for synchrotron radiotherapy treatments. The largest errors were observed in the direction of motion of the patient through the beam and with future improvements, can be reduced. The system was both accurate and repeatable and is ready to support future treatments on IMBL.

Purpose: Even with modern immobilisation devices, some amount of intrafraction patient motion is likely to occur during stereotactic radiosurgery (SRS) delivery. The aim of this work was to investigate how robustness of plans to intrafraction motion is affected by plan geometry and complexity. Methods: In 2018, the Trans-Tasman Radiation Oncology Group conducted a multiple-target SRS international planning challenge, the data from which was utilised in this study. Patient geometry included five intracranial targets with a prescription of 20 Gy. A previously validated in-house algorithm was used to simulate realistic intrafraction patient motion for these plans. Three scenario types were simulated: translational intrafraction motion; rotational motion; and simultaneous rotational and translational motion. Dosimetric impact was assessed using: dose covering 98 % of planning target volume, dose covering 99 % of gross tumour volume (GTV D99%), volume of normal brain receiving 12 Gy and maximum dose covering 0.03 cc brainstem. Results: GTV D99% was reduced by up to 70 %, with the strongest correlations between planning factors and robustness to intrafraction motion found for plan complexity. Despite only moderate correlation strength at r = 0.4, lower complexity plans had, on average, 5 % - 9 % less intrafraction motion scenarios with failing targets compared to the highest complexity plans. Conclusions: SRS plans with lower complexity, in particular larger mean multi-leaf collimator (MLC) gap and MLC aperture irregularity, were shown to improve plan robustness to intrafraction patient motion.

Background/objectives: Brain cancer remains difficult to treat, with survival statistics stagnant for decades. The resistance of glioblastoma brain tumours can greatly challenge the effectiveness of conventional cancer radiotherapy. However, high dose rate radiotherapy has unique effects that allow for normal tissue sparing whilst maintaining tumour control. The addition of targeted radiosensitisers, such as the chemotherapeutic drug methotrexate (MTX) or the high-Z halogenated pyrimidine drug iododeoxyuridine (IUdR), can improve radiotherapy outcomes. Combining these radiosensitiser agents with ultra-high dose rate (UHDR) synchrotron X-rays can bear synergistic effects to enhance the efficacy of these multi-modal UHDR therapies, providing a means to overcome the radioresistance of brain cancer. Methods: Here, we use controlled in vitro assays following treatment, including a clonogenic assay to determine long-term cell survival and γH2AX immunofluorescent confocal microscopy to quantify double-strand DNA breaks (DSBs). Results: We find significant enhancement for highly synergistic combinations of IUdR+MTX with synchrotron X-rays. Cell survival results demonstrate 5.4 times increased 9L gliosarcoma cell killing when these agents are combined with UHDR synchrotron X-rays compared with conventional X-rays alone at the same 5 Gy dose. The underlying mechanisms are unveiled using γH2AX imaging and reveal significant increases in DSBs and dying cells following exposure to UHDR radiation. Conclusions: Our results demonstrate that highly synergistic combination treatments using UHDR synchrotron radiation can yield significantly improved brain cancer killing compared with conventional radiotherapy. We anticipate that these additive, multi-modal combination therapies will provide options for more targeted and effective use of radiotherapies for the future treatment of brain cancer.

We review current dosimetry practices for non-standard radiotherapy sources. We classify radiotherapy sources as established, variations of established, or novel. Our review concentrates on novel sources including ultra-high dose-rate (FLASH) sources and some that have yet to be used for clinical radiotherapy. Factors which differentiate non-standard sources include dose-rate, temporal pulse structure, spatial fractionation, focussing, the presence of magnetic fields, and energy range. For the most part we exclude techniques which use materials inside the tumor to modify the dose. Dosimetry techniques include ionization chambers, film, alanine, calorimetry, and solid-state detectors. We review dosimetry only, neglecting other issues such as beam monitoring, patient delivery systems and treatment planning.

Objective. This study aims to characterize radiological properties of selected additive manufacturing (AM) materials utilizing both material extrusion and vat photopolymerization technologies. Monochromatic synchrotron x-ray images and synchrotron treatment beam dosimetry were acquired at the hutch 3B and 2B of the Australian Synchrotron-Imaging and Medical Beamline. Approach. Eight energies from 30 keV up to 65 keV were used to acquire the attenuation coefficients of the AM materials. Comparison of theoretical, and experimental attenuation data of AM materials and standard solid water for MV linac was performed. Broad-beam dosimetry experiment through attenuated dose measurement and a Geant4 Monte Carlo simulation were done for the studied materials to investigate its attenuation properties specific for a 4 tesla wiggler field with varying synchrotron radiation beam qualities. Main results. Polylactic acid (PLA) plus matches attenuation coefficients of both soft tissue and brain tissue, while acrylonitrile butadiene styrene, Acrylonitrile styrene acrylate, and Draft resin have close equivalence to adipose tissue. Lastly, PLA, co-polyester plus, thermoplastic polyurethane, and White resins are promising substitute materials for breast tissue. For broad-beam experiment and simulation, many of the studied materials were able to simulate RMI457 Solid Water and bolus within ±10% for the three synchrotron beam qualities. These results are useful in fabricating phantoms for synchrotron and other related medical radiation applications such as orthovoltage treatments. Significance and conclusion. These 3D printing materials were studied as potential substitutes for selected tissues such as breast tissue, adipose tissue, soft-tissue, and brain tissue useful in fabricating 3D printed phantoms for synchrotron imaging, therapy, and orthovoltage applications. Fabricating customizable heterogeneous anthropomorphic phantoms (e.g. breast, head, thorax) and pre-clinical animal phantoms (e.g. rodents, canine) for synchrotron imaging and radiotherapy using AM can be done based on the results of this study.

Background/Objectives: Brain cancer is notoriously resistant to traditional treatments, including radiotherapy. Microbeam radiation therapy (MRT), arrays of ultra-fast synchrotron X-ray beams tens of micrometres wide (called peaks) and spaced hundreds of micrometres apart (valleys), is an effective alternative to conventional treatments. MRT’s advantage is that normal tissues can be spared from harm whilst maintaining tumour control. Combining MRT with targeted radiosensitisers, such as nanoparticles, chemotherapeutic drugs, and halogenated pyrimidine drugs, can further improve radiotherapy by enhancing radiation damage. However, the underlying mechanisms of MRT are still being understood, which is essential to ensuring the reliable and successful use of MRT. Methods: An in vitro study was performed using γH2AX imaging, and quantification was performed via confocal microscopy and a clonogenic cell survival assay. Results: We show that methotrexate chemotherapeutics and iododeoxyuridine enhance MRT cell-killing and thulium oxide nanoparticles (TmNPs) broaden MRT peaks, and using γH2AX immunofluorescent confocal microscopy to quantify DNA damage, we further our knowledge of MRT mechanisms. γH2AX images verify the biological responses of cells aligning with the physical collimation of MRT, and we can accurately measure MRT microbeam characteristics bio-dosimetrically. The peak-to-valley dose ratio (PVDR), the ratio of the peak dose to the valley dose that characterises an MRT field, was accurately measured biologically using γH2AX imaging, despite studies previously finding this challenging. Conclusions: The measurement of biological PVDR has been performed for the first time with high-Z radiosensitisers, including nanoparticles, and several novel radiosensitiser-enhanced MRT mechanisms were discovered. Our results deepen our understanding of MRT with radiosensitisers, and can contribute to its accurate and future successful use in treating cancer.

DNA damage occurs in all living cells. γ-H2AX imaging by fluorescent microscopy is widely used across disciplines in the analysis of double-strand break (DSB) DNA damage. Here we demonstrate a method for the quantitative analysis of such DBSs. Ionising radiation, well known to induce DSBs, is used in this demonstration, and additional DBSs are induced if high-Z nanoparticles are present during irradiation. As a deliberate test of the methodology, cells are exposed to a spatially fractionated ionising radiation field, characterised by regions of high and low absorbed radiation dose that are only ever qualitatively verified biologically via γ-H2AX imaging. Here we validate our bio-dosimetric quantification method using γ-H2AX assays in the assessment of DSB enhancement. Our method reliably quantifies DSB enhancement in cells when exposed to either a spatially contiguous or fractionated irradiation fields. Using the γ-H2AX assay, we deduce the biological dose response, and for the first time, demonstrate equivalence to the independently measured physical absorbed dose. Using our novel method, we are also able quantify the nanoparticle DSB enhancement at the cellular level, which is not possible using physical dose measurement techniques. Our method therefore provides a new paradigm in γ-H2AX image quantification of DSBs, as well as an independently validated bio-dosimetry technique.

Objective. This study aims to design, manufacture, and test 3D printed quality assurance (QA) dosimetry phantoms for synchrotron brain cancer radiation therapy at the Australian synchrotron. Approach. Fabricated 3D printed phantoms from simple slab phantoms, a preclinical rat phantom, and an anthropomorphic head phantom were fabricated and characterized. Attenuation measurements of various polymers, ceramics and metals were acquired using synchrotron monochromatic micro-computed tomography (CT) imaging. Polylactic acid plus, VeroClear, Durable resin, and tricalcium phosphate were used in constructing the phantoms. Furthermore, 3D printed bone equivalent materials were compared relative to ICRU bone and hemihydrate plaster. Homogeneous and heterogeneous rat phantoms were designed and fabricated using tissue-equivalent materials. Geometric accuracy, CT imaging, and consistency were considered. Moreover, synchrotron broad-beam x-rays were delivered using a 3 Tesla superconducting multipole wiggler field for four sets of synchrotron radiation beam qualities. Dose measurements were acquired using a PinPoint ionization chamber and compared relative to a water phantom and a RMI457 Solid Water phantom. Experimental depth doses were compared relative to calculated doses using a Geant4 Monte Carlo simulation. Main results. Polylactic acid (PLA+) shows to have a good match with the attenuation coefficient of ICRU water, while both tricalcium phosphate and hydroxyapatite have good attenuation similarity with ICRU bone cortical. PLA+ material can be used as substitute to RMI457 slabs for reference dosimetry with a maximum difference of 1.84%. Percent depth dose measurement also shows that PLA+ has the best match with water and RMI457 within ±2.2% and ±1.6%, respectively. Overall, PLA+ phantoms match with RMI457 phantoms within ±3%. Significance and conclusion. The fabricated phantoms are excellent tissue equivalent equipment for synchrotron radiation dosimetry QA measurement. Both the rat and the anthropomorphic head phantoms are useful in synchrotron brain cancer radiotherapy dosimetry, experiments, and future clinical translation of synchrotron radiotherapy and imaging.

Purpose: This study aimed to analyse correlations between planning factors including plan geometry and plan complexity with robustness to patient setup errors. Methods: Multiple-target brain stereotactic radiosurgery (SRS) plans were obtained through the Trans-Tasman Radiation Oncology Group (TROG) international treatment planning challenge (2018). The challenge dataset consisted of five intra-cranial targets with a 20 Gy prescription. Setup error was simulated using an in-house tool. Dose to targets was assessed via dose covering 99 % (D99 %) of gross tumour volume (GTV) and 98 % of planning target volume (PTV). Dose to organs at risk was assessed using volume of normal brain receiving 12 Gy and maximum dose covering 0.03 cc of brainstem. Plan complexity was assessed via edge metric, modulation complexity score, mean multi-leaf collimator (MLC) gap, mean MLC speed and plan modulation. Results: Even for small (0.5 mm/degrees) errors, GTV D99 % was reduced by up to 20 %. The strongest correlation was found between lower complexity plans (larger mean MLC gap and lower edge metric) and higher robustness to setup error. Lower complexity plans had 1 %-20 % fewer targets/scenarios with GTV D99 % falling below the specified tolerance threshold. These complexity metrics correlated with 100 % isodose volume sphericity and dose conformity, though similar conformity was achievable with a range of complexities. Conclusions: A higher level of importance should be directed towards plan complexity when considering plan robustness. It is recommended when planning multi-target SRS, larger MLC gaps and lower MLC aperture irregularity be considered during plan optimisation due to higher robustness should patient positioning errors occur.

Research utilizing synchrotron radiation for radiotherapy applications is becoming increasingly complex as it pushes towards canine clinical trials [1]–[3]. Recently, several beamline developments at the Imaging and Medical Beamline (IMBL) have taken place towards supporting the forthcoming canine clinical trials. A high speed dynamic collimator has been developed and tested on IMBL for shaping the synchrotron beam as a target is moved through the beam. A 6-Axis Kuka robot has also been configured to allow a target to be positioned and treated with respect to the synchrotron beam. In this work, we describe the synchronisation of these two systems when combined to deliver conformal radiotherapy treatments on IMBL for large animals. A proof-of-concept treatment is demonstrated whereby a medical physics phantom (ball bearings encased in an acryllic cube) is aligned and treated with the robot and dynamic collimator. A full characterisation of the systems capabilities is to follow such that appropriate treatment margins can be generated when using this configuration. More importantly, the ability to dynamically deliver conformal fields on the synchrotron beamline for large animals will ensure that the canine clinical trails can utilise conformal fields for the treatment of cancers whilst avoiding nearby critical structures.

Microbeam radiation therapy (MRT) utilizes coplanar synchrotron radiation beamlets and is a proposed treatment approach for several tumor diagnoses that currently have poor clinical treatment outcomes, such as gliosarcomas. Monte Carlo (MC) simulations are one of the most used methods at the Imaging and Medical Beamline, Australian Synchrotron to calculate the dose in MRT preclinical studies. The steep dose gradients associated with the 50μm-wide coplanar beamlets present a significant challenge for precise MC simulation of the dose deposition of an MRT irradiation treatment field in a short time frame. The long computation times inhibit the ability to perform dose optimization in treatment planning or apply online image-adaptive radiotherapy techniques to MRT. Much research has been conducted on fast dose estimation methods for clinically available treatments. However, such methods, including GPU Monte Carlo implementations and machine learning (ML) models, are unavailable for novel and emerging cancer radiotherapy options such as MRT. In this work, the successful application of a fast and accurate ML dose prediction model for a preclinical MRT rodent study is presented for the first time. The ML model predicts the peak doses in the path of the microbeams and the valley doses between them, delivered to the tumor target in rat patients. A CT imaging dataset is used to generate digital phantoms for each patient. Augmented variations of the digital phantoms are used to simulate with Geant4 the energy depositions of an MRT beam inside the phantoms with 15% (high-noise) and 2% (low-noise) statistical uncertainty. The high-noise MC simulation data are used to train the ML model to predict the energy depositions in the digital phantoms. The low-noise MC simulations data are used to test the predictive power of the ML model. The predictions of the ML model show an agreement within 3% with low-noise MC simulations for at least 77.6% of all predicted voxels (at least 95.9% of voxels containing tumor) in the case of the valley dose prediction and for at least 93.9% of all predicted voxels (100.0% of voxels containing tumor) in the case of the peak dose prediction. The successful use of high-noise MC simulations for the training, which are much faster to produce, accelerates the production of the training data of the ML model and encourages transfer of the ML model to different treatment modalities for other future applications in novel radiation cancer therapies.

This work describes the creation and experimental validation of DoseMRT, a new software package, and its associated workflow for dose calculations in synchrotron-generated broad beam and microbeam radiation treatment fields. The DoseMRT software package allows users to import CT DICOM datasets into Geant4 for Monte Carlo dose calculations. It also provides basic treatment planning capabilities, simplifying the complexity of performing Geant4 simulations and making our Monte Carlo dose calculation algorithm accessible to a broader range of users. To demonstrate the new package, dose calculations are validated against experimental measurements performed in homogeneous water tank phantoms and the anatomically complex Alderson Radiotherapy Phantom for both broad-beam and microbeam configurations. Additionally, DoseMRT is successfully utilised as the primary method for patient-specific treatment prescription in an in vivo experiment involving tumour-bearing rats at the Imaging and Medical Beamline of the Australian Synchrotron.

Synchrotron Microbeam Radiation Therapy (MRT) is an innovative technique that spatially segments the synchrotron radiation field for cancer treatment. A microbeam peak dose is often hundreds of times the dose in the valley (the sub-millimeter region between the peaks of the microbeams). Peak and valley doses vary with increasing depth in tissue which effects tumor dose coverage. It remains to be seen whether the peak or valley is the primary factor in MRT cancer control. This study investigates how unilateral MRT doses can be modulated using a bolus, and identifies the valley dose as a primary factor in MRT cancer control. Fischer rats bearing 9 L gliosarcoma tumors were irradiated with MRT at the Imaging and Medical Beam Line of the Australian Synchrotron. MRT valley doses of 8–15 Gy (250–1040 Gy peak doses) were used to treat tumors with and without a 5 mm dose-modulating bolus. Long-term survival depended on the valley dose primarily (92% correlation), and the use of the bolus reduced the variance in animal survival and improved to the mean survival of rats treated with MRT by 47% and 18% using 15 Gy and 8 Gy valley doses, respectively.

BACKGROUND:Novel synchrotron radiotherapy techniques are currently limited to using prefabricated beam-limiting blocks for field definition. For large experiments, a single square tungsten block is often used for every treatment since conformal blocks are both patient and field specific, and require long lead times for fabrication. Future synchrotron radiotherapy treatments would benefit from a dynamic collimator system.PURPOSE:We developed and tested a novel collimator design for use on the Imaging and Medical Beamline (IMBL) at the ANSTO Australian Synchrotron.METHODS:The maximum usable beam size on IMBL is 50-mm wide by 3-mm tall. Given the beam shape, targets must be vertically scanned through the synchrotron beam to cover the target volume. To shape the beam, a novel collimator design was developed, consisting of two semi-circular leaves made from 4-mm thick tungsten sheets, with each leaf capable of both vertical and horizontal movement. A software model was created to optimize motor trajectories and generate deliverable treatment fields. A series of geometric field shapes and clinical target volumes were delivered using the collimator and imaged with a digital imaging detector. Four similarity metrics (volumetric similarity, DICE, and the average and maximum Hausdorff distances) were used to measure differences between the input and planned fields, and the planned and delivered fields.RESULTS:Differences between input and planned fields increased with delivery speed, and were worse for rectangular and square fields compared to circular fields. However, the differences between planned and delivered fields were small, where the maximum average deviation between the fields was 0.25 mm (one pixel). Field repeatability was consistent with no difference (σ = 0 for all metrics) observed in consecutively delivered fields.CONCLUSIONS:We have successfully built and demonstrated a novel collimator for synchrotron radiotherapy applications on IMBL. Several design improvements have been highlighted and will be addressed in future revisions the collimator. However, in its current state, the collimator enables dynamically delivered conformal treatment fields to be utilized on IMBL, and is ready to support the forthcoming canine treatments on IMBL.

Microbeam radiotherapy (MRT), a high dose rate radiotherapy technique using spatial dose fractionation at the micrometre range, has shown a high therapeutic efficacy in vivo in different tumour entities, including lung cancer. We have conducted a toxicity study for the spinal cord as organ of risk during irradiation of a target in the thoracic cavity. In young adult rats, the lower thoracic spinal cord was irradiated over a length of 2 cm with an array of quasi-parallel microbeams of 50 µm width, spaced at a centre-to-centre distance of 400 µm, with MRT peak doses up to 800 Gy. No acute or subacute adverse effects were observed within the first week after irradiation up to MRT peak doses of 400 Gy. No significant differences were seen between irradiated animals and non-irradiated controls in motor function and sensitivity, open field test and somatosensory evoked potentials (SSEP). After irradiation with MRT peak doses of 450–800 Gy, dose-dependent neurologic signs occurred. Provided that long-term studies do not reveal significant morbidity due to late toxicity, an MRT dose of 400 Gy can be considered safe for the spinal cord in the tested beam geometry and field size.

Fast and accurate dose predictions are one of the bottlenecks in treatment planning for microbeam radiation therapy (MRT). In this paper, we propose a machine learning (ML) model based on a 3D U-Net. Our approach predicts separately the large doses of the narrow high intensity synchrotron microbeams and the lower valley doses between them. For this purpose, a concept of macro peak doses and macro valley doses is introduced, describing the respective doses not on a microscopic level but as macroscopic quantities in larger voxels. The ML model is trained to mimic full Monte Carlo (MC) data. Complex physical effects such as polarization are therefore automatically taking into account by the model. The macro dose distribution approach described in this study allows for superimposing single microbeam predictions to a beam array field making it an interesting candidate for treatment planning. It is shown that the proposed approach can overcome a main obstacle with microbeam dose predictions by predicting a full microbeam irradiation field in less than a minute while maintaining reasonable accuracy.