An Adaptive Proton FLASH Therapy Using Modularized Pin Ridge Filter

Ahmal Jawad Zafar; David Yu; Jun Zhou; Pretesh R. Patel; Tian Sibo; Xiaofeng Yang; Zachary Diamond

REVIEW 3 major objections 5 minor 3 references

An Adaptive Proton FLASH Therapy Using Modularized Pin Ridge Filter

T0 review · 3 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Adaptive proton FLASH replans can recycle 65–91% of ridge-filter pins and restore coverage to within 2.3 percentage points of the original plan.

desk verdict Genuinely practical idea for adaptive proton FLASH pin reuse, with three supportive cases, but the manuscript's internal inconsistencies and missing no-recycling baseline keep this at major-revision status. read the letter →

arxiv 2502.01011 v1 pith:J3L542DM submitted 2025-02-03 physics.med-ph

Ahmal Jawad Zafar , Xiaofeng Yang , Zachary Diamond , Tian Sibo , David Yu , Pretesh R. Patel , Jun Zhou This is my paper · ORCID

classification physics.med-ph

keywords adaptiveprotontherapyFLASHradiotherapypinridgefilterpencilbeamdirectionrecyclingleast-squaresregressionliverSBRTtreatmentplanadaptationeffectivenessmodel

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper tries to establish that adaptive replanning of proton FLASH therapy can be done by reusing most of the physical ridge-filter pins from the original treatment plan instead of manufacturing new filters. The method identifies pencil beam directions whose spot-energy pattern changed little when the patient's anatomy changed, then iteratively freezes a subset of those directions with their beam weights intact while reoptimizing the rest. Across three liver stereotactic body radiotherapy cases, the adaptive plans restored target coverage that applying the unmodified original plan to the new CT lost, reaching V100 within 2.3 percentage points of the original plan (and slightly better in one case) while recycling 64.7–91.2% of ridge pins. The modeled FLASH effect, scored with a 40 Gy/s dose-rate threshold and 1 Gy dose threshold, was preserved in the recycled plans. If correct, this makes adaptive FLASH replanning faster and cheaper by turning a full re-manufacturing problem into a reuse problem.

What carries the argument

The central object is the modularized pin ridge filter (pRF), a stack of pyramid-shaped steps made from 1–6 mm cuboid modules with 5 mm water-equivalent thickness each, which spreads a 250 MeV single-energy beam into a spread-out Bragg peak. The argument is carried by the PBD recycling loop: pencil beam directions whose maximum or minimum spot energy shifts by $\Delta E \ge 5$ MeV are tagged new/changed, and the rest become recycling candidates; an iterative least-squares regression $\min_\beta \| \mathbf{Y}\beta - \mathbf{X}\|_2$ on normalized spot ion MUs selects the two PBDs per iteration whose spot-weight distribution best matches the adaptive beamset, freezes their weights, and places them in a background plan while the remaining PBDs are reoptimized. The transfer criterion is that a small normalized spot-MU error means the physical pin geometry can be reused without reoptimizing that PBD's weight.

What would settle it

Reoptimize each adaptive pRF plan from scratch on the re-CT with no frozen PBDs, and compare CTV V100, homogeneity index, and OAR DVHs against the recycled-pin plans; if the recycled plans are worse than the fully reoptimized ones by more than the roughly 2 percentage points already seen, or if an independent phantom delivery with film and ion chambers shows the frozen-weight recycled pins do not reproduce the planned dose, the recycling criterion is not sufficient.

Watch

Extended reading notes

Core claim

The central claim is that adaptive FLASH replanning can be reduced to a pin-recycling problem: a plan adapted to new anatomy can be built by freezing most of the original single-energy 250 MeV pencil beam directions and reoptimizing only the remainder, provided the frozen PBDs are chosen by a least-squares criterion on normalized spot MU distributions. This recovers dosimetric quality lost when the original pin ridge filter plan is simply recalculated on the re-CT: V100 went from 89.2% to 91.4%, 60.2% to 91.7%, and 69.9% to 98.8% across the three cases, against benchmarks of 93.7%, 93.5%, and 97.3% for the original plan on the planning CT. The recycled fraction was 91.2%, 64.7%, and 71% of PBDs respectively, and the modeled FLASH effect remained comparable to the original plan.

Load-bearing premise

The load-bearing premise is that a pencil beam direction whose spot-energy limits change by less than 5 MeV and whose normalized spot-MU distribution has small least-squares error can be transferred to the new anatomy with its beam weight frozen, without clinically meaningful dosimetric degradation.

Editorial extensions

If this is right

Adaptive pRF plans can be produced without full hardware re-manufacture; 65–91% of the existing ridge pins carry over to the new plan.
Target coverage on the re-CT is restored to within 2.3 percentage points of the original pCT plan, and in one case slightly exceeds it.
Organs at risk remain within protocol tolerance in the recycled plans, apart from one duodenum V0.5cc value that rises 2.03 Gy above the initial plan in case 3.
The modeled FLASH effect is preserved: for example, PTV-GTV mean dose is reduced by 35.5% in the adaptive plan versus 35% in the initial plan for case 1.
The recycling rate tracks anatomical change, with the smallest CTV change recycling over 90% of pins and larger or shifted CTVs recycling roughly 65–71%.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

If the least-squares transfer criterion generalizes, the same recycling loop could be tested in other hypofractionated sites where anatomy changes between fractions, such as lung or pancreas; the paper only demonstrates liver SBRT.
A sensitivity study that varies the 5 MeV energy-change threshold and the two-PBD-per-iteration selection rate could reveal the maximum recyclable fraction before dosimetric degradation, and whether the roughly 2 percentage point V100 gap is a floor or an artifact of these choices.
The clinical benefit still rests on the FLASH effectiveness model's parameters; direct measurement of normal-tissue sparing with recycled pRFs would be needed before the modeled 33% dose reduction is relied on.
The least-squares error used for pin selection could double as an online plan-quality early warning, triggering adaptation automatically when anatomical change passes a threshold.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear Colleague,

The headline: this paper has a genuinely practical idea—recycling modular pin ridge filter pins in adaptive proton FLASH replanning—and its three liver cases do show that the adapted plan restores target coverage that the initial plan loses on the new CT. But the manuscript is not ready as written. The numbers have several internal inconsistencies, and the paper never measures the cost of freezing pins relative to a fully reoptimized plan.

What is new: the iterative least-squares selection of pencil beam directions (PBDs) to retain from the original plan, while reoptimizing the rest, is a sensible extension of the authors' prior modular pRF work. The recycling rates (65–91%) would be a real logistical win for adaptive FLASH, where reprinting ridge filters is currently slow and expensive. The motivation is clearly laid out, and Table 1 supports the qualitative claim that reusing most pins yields a plan comparable to the original on pCT.

Soft spots, in rough order of importance. First, there is no baseline of a fully reoptimized adaptive pRF plan (no recycling) on the re-CT. The reported V100 values could reflect the reoptimization of the remaining PBDs alone; without that baseline, the dosimetric benefit of the recycling step itself is unmeasured. The selection criterion in Eq. (5) is spot-MU similarity, not dose impact—anatomical changes shift water-equivalent path lengths, so a pin with similar spot weights can still deliver a shifted SOBP. A proper comparison would show whether the frozen pins actually hold up or merely get compensated for. Second, the manuscript contains multiple numeric errors: the text claims initial plans 'ensuring 98% of the CTV received the full prescription dose' but Table 1 lists V100 of 93.7% and 93.5%; the Discussion swaps cases 2 and 3 in the largest V100 drop (case 2, not case 3, drops 33.3 pp); and case 2's CTV volume is given as 21.35 cc in the Discussion but 25.2 cc in the methods. These are likely typos, but there are enough of them that a careful reader loses trust in the exact numbers. Third, the algorithm's stopping criterion is vague ('until sufficient selected PBDs are recycled'), and the iterative reoptimization is not spelled out precisely enough to reproduce. Fourth, the claim of 'maintaining all clinical goals' is overstated—case 3's duodenum V0.5cc reportedly exceeds the BR001 tolerance.

The stress-test note is fair on the missing baseline; it doesn't kill the practical claim, but it does mean you can't yet say recycling is harmless. A reviewer should ask for the full-reoptimization comparison and for spot-level dose validation of frozen pins.

Who is this for? Anyone working on proton FLASH planning, particularly adaptive workflows. It deserves a serious referee, but with major revision. I'd recommend accept-with-revisions only after the baseline and numeric fixes are in.

Best.

Referee Report

3 major / 5 minor

Summary. The paper proposes an adaptive proton FLASH therapy workflow that recycles modularized pin ridge filter (pRF) pins from an initial plan when replanning on a new CT. Pencil beam directions (PBDs) are classified by energy change, and an iterative least-squares regression selects a subset of PBDs whose normalized spot-MU distributions are most similar to the initial plan; these PBDs are frozen in the background while the remaining PBDs are reoptimized. The method is tested on three liver SBRT cases, reporting that V100 values on re-CT improve from 89.2%, 60.2%, and 69.9% (initial pRF applied to re-CT) to 91.4%, 91.7%, and 98.8% (ADP-pRF), with 91.2%, 64.7%, and 71% of pins recycled, while preserving a modeled FLASH effect.

Significance. If the recycling claim holds, the method would substantially reduce the cost and turnaround time of adaptive FLASH replanning by reusing most hardware. The study is a planning study with direct dosimetric comparisons, and the authors correctly identify that initial pRFs degrade on re-CT, motivating adaptation. The FLASH-effectiveness evaluation is quantitative and the dosimetric metrics are standard. However, the central claim that recycling pins preserves plan quality is currently not isolated from the effect of reoptimization, and there are inconsistencies between stated clinical goals and the reported numbers. These issues need to be resolved before the conclusions can be considered supported.

major comments (3)

[Section 2.4, Section 3.1.1, Table 1] The stated clinical goal is V100 ≥ 95% for the CTV, but Table 1 shows initial pRF plans on pCT with V100 = 93.7% and 93.5% for cases 1 and 2, i.e., below the goal. Section 3.1.1 also states that the initial pRF plan ensured '98% of the CTV received the full prescription dose,' which is inconsistent with the Table 1 values. The abstract's claim of 'maintaining all clinical goals' is therefore not supported: the ADP-pRF V100 values for cases 1 and 2 (91.4% and 91.7%) also fall below the 95% threshold. Please clarify which dose specification is actually used and reconcile the text with the table.
[Section 2.2(c), Eq. (5), Section 3.1.2] The paper never compares the proposed ADP-pRF plan (with recycled pins) against a fully reoptimized adaptive plan on re-CT with no recycling. The selection criterion in Eq. (5) is a normalized spot-MU similarity measure, not a dosimetric measure; it does not quantify the dosimetric impact of freezing a given PBD's pin on the re-CT anatomy, where water-equivalent path lengths may have changed. Without a no-recycling baseline, the reported V100 values and recycling rates do not establish that recycling is harmless—the quality recovery could be entirely due to reoptimization of the remaining PBDs. Please provide a baseline plan with 0% recycling and, ideally, a curve of plan quality versus fraction of recycled pins to quantify the true cost of recycling.
[Section 3.2, Figure 4] The FLASH-effect evaluation compares the ADP-pRF FLASH-1Gy plan on re-CT to 'the initial IMPT plan designed on pCT' rather than to a conventional IMPT plan on re-CT. Because the anatomical changes between pCT and re-CT are substantial (e.g., case 2 V100 drops to 60.2% when the initial pRF is evaluated on re-CT), the reported reductions in OAR doses may be confounded by anatomical differences rather than reflecting the FLASH effect of the adaptive plan. To support the claim that the FLASH effect is preserved, the FLASH-1Gy reductions should be recomputed against a conventional IMPT plan optimized on re-CT.

minor comments (5)

[Introduction] Typo: 'FAST-01 trail' should be 'FAST-01 trial'.
[Section 3.1.2, Table 1] The organ name is misspelled as 'Deudenum' in both the text and Table 1; it should be 'Duodenum'.
[Section 3.1.2] The reference 'Figure 1(d)' is incorrect; the ADP-pRF dose distribution is shown in Figure 3(d), not Figure 1(d).
[Section 2.2(c), Eq. (5)] In Eq. (5), the vector X is used but not explicitly defined; the text defines y_ij and Y, but X should be defined as the normalized spot-MU vector of the initial pCT plan for the corresponding PBD.
[Section 2.2(a)] The energy-change threshold uses ΔE_max or ΔE_min ≥ 5 MeV, but it is not clear whether these differences are absolute values and whether the threshold applies to the maximum or minimum spot energy within each PBD or to some other per-PBD energy metric; please clarify the definition.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular core: adaptive pRF quality is dose-computed and independent; minor same-group citations only.

full rationale

The paper's central claim (adaptive proton FLASH plans on re-CT can recycle a large fraction of modular pRF pins while keeping target coverage and modeled FLASH sparing comparable to the initial pRF plan on pCT) is supported by an independent dosimetric evaluation: the ADP-pRF plans are constructed in RayStation on the re-CT and scored by V100, HI, and OAR metrics, with no quantity fit to itself. The recycling selector, Eq. (5), minimizes normalized spot-MU least-squares error; this is a screening mechanism for choosing which PBDs to freeze, not the dosimetric endpoint, and the final reported V100 values are not determined by Eq. (5). Thus the claim does not reduce by construction to its inputs. The FLASH-effectiveness parameters (40 Gy/s, 200 ms, 0.67, D0 = 1 Gy) are taken from the same group's prior work (refs 31, 32), and the modular pRF construction is inherited from ref 26; these self-citations are real inputs but are not load-bearing for the adaptive comparison, which would stand or fall on the dose engine results. The main weakness is an omitted control: the paper never compares ADP-pRF against a fully reoptimized no-recycling adaptive plan on re-CT, so the marginal dosimetric cost of freezing recycled pins is not separately quantified. That is a validation gap, not a circularity: nothing in the derivation defines the conclusion in terms of the selection metric. A second weakness is that the FLASH-sparing numbers are model outputs (a 0.67 effectiveness factor applied where thresholds are met), not measured biological endpoints, but the initial-vs-ADP FLASH comparison is still a legitimate model-based comparison. Score 2 reflects only the presence of minor same-group citations and the absence of a no-recycling baseline; no circular step was found.

Assumptions & free parameters 10 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the planning framework from refs 26, 32, 34 (same group) and the FLASH effectiveness model with parameters taken from prior work. The new recycling algorithm itself contributes no new physical entities but relies on several hand-set thresholds and an unverified assumption that frozen PBDs remain valid.

free parameters (10)

MU threshold for spot reduction = 300 MU
Set to guarantee deliverability of the IMPT-pRF plan at 500 nA and 0.5 ms minimum spot duration; chosen by hand, not fit to data.
Energy layer spacing in IMPT-DS plan = 5 mm WET
Set to match unit module WET; influences step thickness and resolution. Chosen in the planning process, not fitted.
PBD change threshold = 5 MeV
Threshold classifying PBDs as changed or unchanged based on maximum or minimum spot energy difference. Hand-set criterion.
PBDs recycled per iteration = 2
Two PBDs with least square error retrieved per iteration; arbitrary choice affecting convergence speed and plan quality.
FLASH dose threshold D0 = 1 Gy
Used in FLASH effectiveness evaluation; chosen from the stated 1 to 25 Gy range, not fitted to the data.
FLASH dose rate threshold = 40 Gy/s
Taken from previously reported values (ref 32); not fit in this paper.
FLASH persistence time = 200 ms
Taken from previously reported values (ref 32); not fit in this paper.
FLASH effective factor = 0.67
Implies 33% dose reduction when FLASH effect is active; from prior literature, not fit here.
Unit module widths = 1 to 6 mm in 1 mm increments
Design choice for reusable modules; bottom step width L1 fixed at 6 mm.
Step width resolution = 1 mm
Matches the TPS grid resolution and the streamlined pRF design from prior work (ref 26).

assumptions (4)

domain assumption FLASH effectiveness model correctly predicts normal tissue sparing from dose and dose rate thresholds
The FLASH effect evaluation in Section 2.3 uses the model from ref 31 with parameters from ref 32; biological effectiveness is not measured.
domain assumption Translation from IMPT-DS plan to pRF hardware preserves dosimetric accuracy
The pRF design framework from ref 26 is assumed valid; the current paper does not validate hardware or delivery.
domain assumption Recycled PBDs retain their spot weight distribution when transferred to the background plan
Section 2.2(c) states structural integrity of recycled PBDs is preserved; this underpins the least squares selection but is not independently verified in the adapted plan.
domain assumption RayStation TPS dose calculation accurately models single-energy proton delivery with range shifters and pRFs
All dosimetric results are computed in silico in RayStation 10B; no measurement validation is provided.

how reviews work

0 comments

Cite this review

Pith. "Pith review of An Adaptive Proton FLASH Therapy Using Modularized Pin Ridge Filter." pith.science (2026). https://pith.science/paper/J3L542DM

@misc{pith2026250201011,
  author       = {Pith},
  title        = {Pith review of: An Adaptive Proton FLASH Therapy Using Modularized Pin Ridge Filter},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J3L542DM}},
  note         = {Machine review of arXiv:2502.01011}
}

abstract

In this paper, we proposed a method to optimize adaptive proton FLASH therapy (ADP FLASH) using modularized pin ridge filters (pRFs) by recycling module pins from the initial plan while reducing pRF adjustments in adaptive FLASH planning. Initially, single energy (250 MeV) FLASH pRF plans were created using pencil beam directions (PBDs) from initial IMPT plans on the planning CT (pCT). PBDs are classified as new/changed ($\Delta$E > > 5 MeV) or unchanged by comparing spot maps for targets between pCT and re-CT. We used an iterative least square regression model to identify recyclable PBDs with minimal relative changes to spot MU weighting. Two PBDs with the least square error were retrieved per iteration and added to the background plan, and the remaining PBDs were reoptimized for the adaptive plan in subsequent iterations. The method was validated on three liver SBRT cases (50 Gy in 5 fractions) by comparing various dosimetric parameters across initial pRF plans on pCT, reCT and the ADP FLASH pRF plans on reCT. V100 for initial pRF plans on pCT, reCT, and ADP FLASH pRF plans for the three cases were as follows: (93.7%, 89.2%, 91.4%), (93.5%, 60.2%, 91.7%), (97.3%, 69.9%, 98.8%). We observe a decline in plan quality when applying the initial pRF to the reCT, whereas the ADP FLASH pRF approach restores quality comparable to the initial pRF on the pCT. FLASH effect of the initial pRF and ADP pRF plans were evaluated with a dose and dose rate threshold of 1Gy and 40Gy/s, respectively, using the FLASH effectiveness model. The proposed method recycled 91.2%, 71%, and 64.7% of PBDs from initial pRF plans for the three cases while maintaining all clinical goals and preserving FLASH effects across all cases.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

3 extracted references · 3 canonical work pages

[1]

Introduction: Radiotherapy has been a pivotal part of cancer treatment for many decades. However, from the beginning the use of high-energy radiation to target cancerous cells, while avoiding damage to healthy tissues has been one of the leading challenges in the field of radiotherapy. FLASH radiotherapy (FLASH-RT), has shown promising results at an ultra...

work page
[2]

# determines the water equivalent thickness (WETs), 𝑇!, of step 𝑖 (where 𝑖>1) in a ridge pin as described in equation (1). 𝑇!= 𝑅!−𝑅!

Methods and Materials: This section presents the comprehensive methodology adopted for this study, including the development of the Flash planning framework and the modularization of the pin-RF configuration. We detail the approach used to ensure effective FLASH planning and the specific parameters involved in optimizing the design. Additionally, we descr...

work page
[3]

An Integrated Biological Optimization framework for proton SBRT FLASH treatment planning allows dose, dose rate, and LET optimization using patient-specific ridge filters

Results 3.1 Adaptation Process: The total number of ridge pins in the initial pRF plan on pCT for cases 1, 2, and 3 were 52, 69, and 73, respectively. For each case, the initial pRF plan, designed on pCT, served as a benchmark for comparison with the ADP-pRF plan. The analysis of this recycling PBD method was conducted by comparing three scenarios. First,...

work page Pith review arXiv 2014

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.

[1] [1]

Introduction: Radiotherapy has been a pivotal part of cancer treatment for many decades. However, from the beginning the use of high-energy radiation to target cancerous cells, while avoiding damage to healthy tissues has been one of the leading challenges in the field of radiotherapy. FLASH radiotherapy (FLASH-RT), has shown promising results at an ultra...

work page

[2] [2]

# determines the water equivalent thickness (WETs), 𝑇!, of step 𝑖 (where 𝑖>1) in a ridge pin as described in equation (1). 𝑇!= 𝑅!−𝑅!

Methods and Materials: This section presents the comprehensive methodology adopted for this study, including the development of the Flash planning framework and the modularization of the pin-RF configuration. We detail the approach used to ensure effective FLASH planning and the specific parameters involved in optimizing the design. Additionally, we descr...

work page

[3] [3]

An Integrated Biological Optimization framework for proton SBRT FLASH treatment planning allows dose, dose rate, and LET optimization using patient-specific ridge filters

Results 3.1 Adaptation Process: The total number of ridge pins in the initial pRF plan on pCT for cases 1, 2, and 3 were 52, 69, and 73, respectively. For each case, the initial pRF plan, designed on pCT, served as a benchmark for comparison with the ADP-pRF plan. The analysis of this recycling PBD method was conducted by comparing three scenarios. First,...

work page Pith review arXiv 2014