arxiv: 2605.02605 · v1 · submitted 2026-05-04 · ⚛️ physics.med-ph

Recognition: unknown

Mitigation of Boundary Sampling Artifacts in Phase Space Generation for Electron FLASH Radiotherapy

Rafael Carballeira (1) , Rongxiao Zhang (1 , 3) , Kevin J. Willy (1) , Hayley Cash (4) , David J. Gladstone (1 , 2) ((1) Thayer School of Engineering , Dartmouth College

show 11 more authors

Hanover New Hampshire (2) Dartmouth Cancer Center Lebanon (3) School of Medicine University of Missouri Columbia Missouri (4) Medical University of South Carolina Charleston South Carolina)

Authors on Pith no claims yet

Pith reviewed 2026-05-08 02:04 UTC · model grok-4.3

classification ⚛️ physics.med-ph

keywords phase spaceMonte CarloGeant4electron FLASHboundary artifactsapplicatorR50DTA

0 comments

The pith

A 1 mm downstream offset in phase space scoring planes fully eliminates boundary sampling artifacts in Geant4 simulations for electron FLASH radiotherapy.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that scoring phase space files exactly at the applicator-air interface in Monte Carlo simulations creates artifacts that distort dose distributions for electron FLASH radiotherapy. These distortions produce R50 shifts up to 2.2 mm and distance-to-agreement values of 4-6 mm, exceeding clinical criteria and worsening for smaller apertures. Offsetting the scoring plane 1 mm downstream resolves the artifacts completely, restoring DTA to 2.0 mm or better, matching or exceeding linac-exit references. The root cause is identified as degenerate behavior in the fUseSafety step-limitation algorithm when safety distance reaches zero at material boundaries, which disrupts secondary electron equilibration and bremsstrahlung production. A smaller 0.1 mm offset improves the primary spectrum but leaves the bremsstrahlung tail incomplete.

Core claim

Positioning phase space scoring planes coincident with the applicator-air interface in Geant4-based simulations of a 9 MeV Mobetron UHDR model produces boundary sampling artifacts across twelve clinical aperture sizes. These artifacts arise because the fUseSafety step-limitation algorithm becomes degenerate when safety equals zero at exact material boundaries, resulting in incomplete secondary electron equilibration and suppressed bremsstrahlung production. This manifests as proximal R50 shifts up to 2.2 mm and DTA values of 4-6 mm. Scoring 1 mm downstream fully resolves the artifacts to achieve mean DTA within 2.0 mm, equivalent to or better than linac-exit references, while angular data at

What carries the argument

The fUseSafety step-limitation algorithm in Geant4, which degenerates at zero safety distance when a scoring plane is placed exactly at a material boundary, causing incomplete secondary particle transport.

Load-bearing premise

The artifacts originate exclusively from the fUseSafety algorithm at zero safety distance and a 1 mm offset introduces only negligible perturbation without new biases.

What would settle it

Running the same 9 MeV simulations with the 1 mm offset and comparing DTA and R50 values against a reference using a different step-limiting algorithm or finer voxel grid to check if artifacts reappear or new discrepancies emerge.

Figures

Figures reproduced from arXiv: 2605.02605 by 2) ((1) Thayer School of Engineering, (2) Dartmouth Cancer Center, 3), (3) School of Medicine, (4) Medical University of South Carolina, Charleston, Columbia, Dartmouth College, David J. Gladstone (1, Hanover, Hayley Cash (4), Kevin J. Willy (1), Lebanon, Missouri, New Hampshire, Rafael Carballeira (1), Rongxiao Zhang (1, South Carolina), University of Missouri.

**Figure 4.** Figure 4: With the scoring plane at the exact material view at source ↗

**Figure 4.** Figure 4: Boundary sampling artifact behavior for the representative A6I3 configuration. (A) Identification of the artifact: the PHSP scored at the exact material-air interface (397 mm, red) produces a 2.2 mm proximal R50 shift and premature PDD falloff relative to measured data (black), while a 1 mm downstream offset (398 mm, blue) restores agreement. (B) Offset distance sensitivity: 0.1 mm offset (green) partially… view at source ↗

read the original abstract

Applicator-specific phase space (PHSP) files recorded at the aperture exit reduce Monte Carlo dose calculation time by 30-50% for electron FLASH radiotherapy. However, positioning PHSP scoring planes coincident with the applicator-air interface introduces boundary sampling artifacts. This study characterizes these artifacts in Geant4-based simulations and demonstrates their mitigation. PHSP files were generated using GAMOS 6.2.0 for a 9 MeV Mobetron UHDR model across twelve clinical aperture configurations (2.5-10 cm diameter). Three scoring plane positions were evaluated relative to the physical aperture exit: coincident with the interface, 0.1 mm downstream, and 1 mm downstream. Scoring at the exact interface produced proximal R50 shifts of up to 2.2 mm and Distance-to-Agreement (DTA) values of 4-6 mm, exceeding clinical acceptance criteria. Artifact severity scaled inversely with aperture diameter, with the smallest configurations most severely affected. A 0.1 mm offset partially restored the primary electron energy spectrum but failed to recover the bremsstrahlung tail. A 1 mm offset fully resolved all artifacts, achieving mean DTA values within 2.0 mm, equivalent to or better than linac-exit references. These artifacts arise from the degenerate behavior of the fUseSafety step-limitation algorithm when safety equals zero at exact material boundaries, producing incomplete secondary electron equilibration and suppressed bremsstrahlung production. Angular distribution analysis revealed a near-forward particle pileup in the 0 mm PHSP and a deficit of large-angle secondaries recovered by the 1 mm offset. A 1 mm downstream offset fully mitigates these artifacts while introducing negligible perturbation to primary beam characteristics. This requirement applies to any Geant4-based framework (including TOPAS and GATE) scoring PHSP files at material exit surfaces.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

A 1 mm downstream offset in phase space scoring planes removes the boundary artifacts in Geant4 electron FLASH simulations, but the paper stops short of confirming the exact mechanism.

read the letter

The main thing to know is that scoring phase space files exactly at the applicator-air interface in Geant4 produces measurable distortions in the electron spectrum and dose for 9 MeV FLASH beams, and shifting the plane 1 mm downstream restores acceptable R50 and DTA values across the tested apertures. The work shows this clearly through direct comparisons of three plane positions on twelve clinical apertures, with the zero-offset case hitting 2.2 mm R50 shifts and 4-6 mm DTA while the 1 mm case brings mean DTA under 2 mm, matching or beating linac-exit references. Smaller apertures suffer more, which matches intuition about edge effects. The angular pileup and bremsstrahlung tail recovery at 1 mm give a concrete picture of what is missing at the boundary. This is a practical, targeted fix that anyone generating applicator PHSP files in GAMOS, TOPAS, or GATE should note. The authors also flag that the problem is general to Geant4-based codes when safety distance hits zero at material boundaries. On the downside, the manuscript supplies no particle counts, no statistical uncertainties on the reported shifts, and no physics list or production cut details. The attribution to fUseSafety degeneracy is plausible from the symptoms but rests on inference rather than a direct test, such as forcing non-zero safety at the boundary or swapping step-limit algorithms. No measurements against physical beams are shown, so the clinical size of the bias remains unanchored. The 1 mm offset is treated as negligible after internal checks, but an external reference beam would strengthen that. Readers doing Monte Carlo work on electron FLASH will get immediate value from the offset recommendation and the aperture-size dependence. The paper is coherent on its own terms and reports a reproducible simulation result, so it is worth sending for peer review even though the evidence is simulation-only and the mechanism claim could be tightened.

Referee Report

2 major / 2 minor

Summary. The manuscript investigates boundary sampling artifacts when generating applicator-specific phase space (PHSP) files for Geant4-based Monte Carlo simulations of electron FLASH radiotherapy. Using GAMOS 6.2.0 for a 9 MeV Mobetron UHDR model across twelve apertures (2.5–10 cm), it compares PHSP scoring planes at the exact applicator-air interface, 0.1 mm downstream, and 1 mm downstream. Exact-interface scoring produces proximal R50 shifts up to 2.2 mm and DTA values of 4–6 mm; a 1 mm offset resolves these to mean DTA ≤ 2.0 mm while recovering the bremsstrahlung tail and large-angle secondaries, which the authors attribute to degeneracy in the fUseSafety step-limitation algorithm when safety distance equals zero at material boundaries.

Significance. If the reported mitigation holds under broader validation, the work supplies a concrete, low-overhead guideline for reliable PHSP generation in electron FLASH RT, enabling the 30–50 % reduction in dose-calculation time without compromising dosimetric accuracy. The systematic comparison across multiple clinical apertures and the quantitative DTA/R50 metrics provide practical evidence that a 1 mm offset is sufficient and largely perturbation-free for this class of simulations.

major comments (2)

[Results and Discussion] Results/Discussion: The central attribution of artifacts exclusively to fUseSafety degeneracy at safety=0 is not directly tested. No controlled simulation is presented that forces non-zero safety (via custom step limits, physics-list modification, or alternative boundary handling) while scoring at the 0 mm plane to isolate the mechanism from possible boundary-crossing or production-cut effects.
[Methods and Results] Methods and Results: Reported R50 shifts (up to 2.2 mm) and DTA values (4–6 mm) lack statistical uncertainties, primary particle counts, or physics-list details, preventing assessment of whether the observed differences exceed simulation noise and supporting the claim that the 1 mm offset introduces only negligible primary-beam perturbation.

minor comments (2)

[Methods] The manuscript should explicitly state the Geant4 version, physics list (e.g., Livermore or Penelope), production cuts, and total number of simulated primaries for each configuration to allow reproducibility.
[Figures and Results] Figure captions and text should clarify whether DTA values are mean or maximum and specify the reference dose distribution used for comparison (linac-exit PHSP or measurement).

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their thorough review and valuable feedback on our manuscript investigating boundary sampling artifacts in phase space generation for electron FLASH radiotherapy. We have carefully considered each major comment and provide detailed responses below, including commitments to revisions where appropriate.

read point-by-point responses

Referee: [Results and Discussion] Results/Discussion: The central attribution of artifacts exclusively to fUseSafety degeneracy at safety=0 is not directly tested. No controlled simulation is presented that forces non-zero safety (via custom step limits, physics-list modification, or alternative boundary handling) while scoring at the 0 mm plane to isolate the mechanism from possible boundary-crossing or production-cut effects.

Authors: We agree that the manuscript does not include a direct controlled simulation to isolate the fUseSafety degeneracy as the sole cause. Our conclusion is drawn from the specific patterns in the data, including the angular distribution showing forward pileup and deficit of large-angle secondaries at the 0 mm plane, the incomplete recovery of the bremsstrahlung tail with the 0.1 mm offset, and the full resolution at 1 mm, which are consistent with the known behavior of Geant4's fUseSafety algorithm when the safety distance is set to zero at a material boundary. We have cross-referenced this with Geant4 documentation and source code. While we cannot exclude contributions from other boundary-related effects without additional targeted simulations, the evidence strongly supports the proposed mechanism. In the revised manuscript, we will update the discussion section to describe this attribution as well-supported by the observed phenomena and Geant4 internals rather than exclusively proven, and we will note the value of future controlled experiments to further confirm the mechanism. revision: yes
Referee: [Methods and Results] Methods and Results: Reported R50 shifts (up to 2.2 mm) and DTA values (4–6 mm) lack statistical uncertainties, primary particle counts, or physics-list details, preventing assessment of whether the observed differences exceed simulation noise and supporting the claim that the 1 mm offset introduces only negligible primary-beam perturbation.

Authors: We acknowledge that the current version of the manuscript does not explicitly report the statistical uncertainties, primary particle counts, or detailed physics-list specifications. These details were omitted for brevity but are available from our simulation setup. The primary particle counts were chosen to ensure low statistical noise in the dose distributions, and the observed shifts and DTA values are significantly larger than the expected Monte Carlo uncertainties for these configurations. The physics list employed is the standard electromagnetic physics list in GAMOS 6.2.0, which utilizes Geant4's Livermore low-energy models. In the revised manuscript, we will add a dedicated subsection in the Methods describing the simulation statistics and physics list, and we will include error bars or uncertainty estimates on the R50 and DTA metrics in the Results to facilitate evaluation of the significance. This will also strengthen the claim regarding the negligible perturbation from the 1 mm offset by showing that primary beam characteristics remain consistent within uncertainties. revision: yes

Circularity Check

0 steps flagged

No circularity; empirical simulation results are self-contained

full rationale

The paper reports outcomes from direct Geant4/GAMOS Monte Carlo runs comparing PHSP scoring planes at 0 mm, 0.1 mm, and 1 mm offsets, using quantitative metrics (proximal R50 shifts, DTA values, energy spectra, angular distributions). No equations, fitted parameters, predictions, or derivations appear; the attribution to fUseSafety degeneracy is an interpretive statement grounded in the observed simulation differences rather than a reduction to prior definitions or self-citations. The work is self-contained against external benchmarks via explicit comparative simulations.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard Geant4 Monte Carlo assumptions for particle transport at boundaries; no free parameters, new entities, or ad-hoc axioms beyond domain-standard physics modeling are introduced.

axioms (1)

domain assumption Geant4 fUseSafety step-limitation algorithm exhibits degenerate behavior when safety distance equals zero at exact material boundaries, leading to incomplete secondary electron equilibration and suppressed bremsstrahlung.
Invoked directly to explain the origin of the observed proximal R50 shifts and DTA violations.

pith-pipeline@v0.9.0 · 5723 in / 1420 out tokens · 79688 ms · 2026-05-08T02:04:15.955670+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

13 extracted references · 12 canonical work pages

[1]

Vincent Favaudon et al., Ultrahigh dose-rate FLASH irradiation increases the differential response between normal and tumor tissue in mice.Sci. Transl. Med.6,245ra93- 245ra93(2014).DOI:10.1126/scitranslmed.3008973

work page doi:10.1126/scitranslmed.3008973 2014
[2]

Wilson JD, Hammond EM, Higgins GS and Petersson K (2020) Ultra-High Dose Rate (FLASH) Radiotherapy: Silver Bullet or Fool's Gold?. Front. Oncol. 9:1563. doi: 10.3389/fonc.2019.01563

work page doi:10.3389/fonc.2019.01563 2020
[3]

and Siebers, J.V

Chetty, I.J., Curran, B., Cygler, J.E., DeMarco, J.J., Ezzell, G., Faddegon, B.A., Kawrakow, I., Keall, P .J., Liu, H., Ma, C.-.-M.C., Rogers, D.W.O., Seuntjens, J., Sheikh-Bagheri, D. and Siebers, J.V. (2007), Report of the AAPM Task Group No. 105: Issues associated with clinical implementation of Monte Carlo-based photon and electron external beam treat...

work page doi:10.1118/1.2795842s 2007
[4]

S. Agostinelli et al., Geant4—a simulation toolkit, Nuclear Instruments and Methods in Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment, Volume 506, Issue 3, 2003, Pages 250-303, ISSN 0168-9002, https://doi.org/10.1016/S0168- 9002(03)01368-8

work page doi:10.1016/s0168- 2003
[5]

Arce, J.I

P . Arce, J.I. Lagares, L. Harkness, D. Pérez-Astudillo, M. Cañadas, P . Rato, M. de Prado, Y . Abreu, G. de Lorenzo, M. Kolstein, A. Díaz, Nucl. Instrum. Methods Phys. Res. A 735 (2014) 304–313. https://doi.org/10.1016/j.nima.2013.09.036

work page doi:10.1016/j.nima.2013.09.036 2014
[6]

First Monte Carlo beam model for ultra-high dose rate radiotherapy with a compact electron LINAC

Dai T, Sloop AM, Rahman MR, et al. First Monte Carlo beam model for ultra-high dose rate radiotherapy with a compact electron LINAC. Med Phys. 2024;51:5109–5118. https://doi.org/10.1002/mp.17031

work page doi:10.1002/mp.17031 2024
[7]

(2000), Accurate condensed history Monte Carlo simulation of electron transport

Kawrakow, I. (2000), Accurate condensed history Monte Carlo simulation of electron transport. I. EGSnrc, the new EGS4 version. Med. Phys., 27: 485-498. https://doi.org/10.1118/1.598917

work page doi:10.1118/1.598917 2000
[8]

Capote, et al., IAEA phase-space database for external beam radiotherapy, IAEA-TECDOC- 1540, 2006

R. Capote, et al., IAEA phase-space database for external beam radiotherapy, IAEA-TECDOC- 1540, 2006. https://www-pub.iaea.org/MTCD/publications/PDF/te_1540_web.pdf

2006
[9]

and Verhaegen, F

Poon, E. and Verhaegen, F. (2005), Accuracy of the photon and electron physics in GEANT4 for radiotherapy applications. Med. Phys., 32: 1696-1711. https://doi.org/10.1118/1.1895796

work page doi:10.1118/1.1895796 2005
[10]

Presta: The Parameter Reduced Electron-Step Transport Algorithm for Electron Monte Carlo Transport

Bielajew, Alex F., and D.W.O. Rogers. “Presta: The Parameter Reduced Electron-Step Transport Algorithm for Electron Monte Carlo Transport.” Nuclear Instruments and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms, vol. 18, no. 1-6, 1986, pp. 165– 181, https://doi.org/10.1016/s0168-583x(86)80027-1. Accessed 29 Jan. 2023

work page doi:10.1016/s0168-583x(86)80027-1 1986
[11]

Parameterization of multiple Bragg curves for scanning proton beams using simultaneous fitting of multiple curves

Josep Sempau, et al. Monte Carlo Simulation of Electron Beams from an Accelerator Head Using PENELOPE. Vol. 46, no. 4, 1 Apr. 2001, pp. 1163–1186, https://doi.org/10.1088/0031- 9155/46/4/318

work page doi:10.1088/0031- 2001
[12]

Development and verification of an electron Monte Carlo engine for applications in intraoperative radiation therapy

Rank L, Lysakovski P , Major G, et al. Development and verification of an electron Monte Carlo engine for applications in intraoperative radiation therapy. Med Phys. 2024;51:6348–6364. https://doi.org/10.1002/mp.17180

work page doi:10.1002/mp.17180 2024
[13]

and Rogers, D.W.O

Almond, P .R., Biggs, P .J., Coursey, B.M., Hanson, W.F., Huq, M.S., Nath, R. and Rogers, D.W.O. (1999), AAPM's TG-51 protocol for clinical reference dosimetry of high-energy photon and electron beams. Med. Phys., 26: 1847-1870. https://doi.org/10.1118/1.598691

work page doi:10.1118/1.598691 1999