Skip to content

Paper 1 — Systematic Evidence Review Protocol

Protocol fixed: 2026-09-04

Status: prospective protocol. Search results and decisions will be recorded in EVIDENCE_REVIEW.md. The existing LITERATURE_REVIEW.md is seed material, not a completed systematic review.

Purpose

Determine whether the accessible evidence supports the provisional Paper 1 scope—rapid surrogate prediction of a specified three-dimensional radiation-protection field for 5–7.5 MeV conventional industrial LINACs in converted-X-ray mode—or requires that scope to be narrowed, expanded, or replaced.

This is a systematic evidence review for engineering method and scope selection, not a meta-analysis of one intervention. Reporting follows the applicable parts of PRISMA 2020 and PRISMA-S.

Review questions

  1. What industrial electron-accelerator and converted-X-ray systems, energies, converters, beam-delivery modes, and shielding geometries are documented?
  2. What evidence supports source, converter, shielding, photonuclear, and dose- quantity modelling for the provisional Paper 1 domain?
  3. What measurements and independent calculations are sufficiently specified to support PHITS verification or physical validation?
  4. What surrogate or reduced-order methods have predicted accelerator-shielding, photon-shielding, or spatial radiation-transport outputs?
  5. Does any located study combine the complete proposed contribution: an industrial conventional-LINAC electron-to-converter source, 5–7.5 MeV shielding domain, PHITS reference fields, whole-configuration test separation, and photonuclear assessment?
  6. Which evidence gaps and practical constraints should determine Paper 1's final scope and methodology?

Eligibility criteria

Include as direct scope evidence

A record must concern an industrial electron accelerator or industrial converted-X-ray facility and contribute evidence about at least one of:

  • conventional-LINAC source or beam delivery;
  • incident-electron energies overlapping or bracketing 5–7.5 MeV;
  • finite converter construction, spectrum, angular field, or residual electrons;
  • facility shielding, prompt radiation-protection fields, barriers, rooms, mazes, penetrations, or measurements;
  • photonuclear production or neutron transport near the applicable energies;
  • PHITS modelling, verification, validation, or uncertainty for a relevant transport component; or
  • a surrogate trained on Monte Carlo radiation-transport calculations.

Include as indirect methodological evidence

Evidence outside the complete Paper 1 system may be retained when it supplies a method needed by the study, including:

  • authoritative transport data, operational quantities, or safety guidance;
  • electron, photon, or photonuclear benchmarks;
  • Monte Carlo statistical-convergence methods;
  • surrogate evaluation, spatial-field prediction, uncertainty, or leakage- resistant partitioning; and
  • medical, research, or other accelerator studies with a transferable component method that is clearly separated from direct applicability.

Exclude from the Paper 1 evidence map

Exclude records that:

  • concern neither the physical system nor a transferable method;
  • discuss product absorbed dose without source, shielding, transport, or validation information relevant to this review;
  • are promotional summaries without traceable technical methods or data;
  • duplicate a more complete version of the same work;
  • provide no title or abstract sufficient for relevance screening; or
  • cannot support a claim beyond an inaccessible citation snippet.

Excluded records remain in the screening log with one primary reason.

Coverage

  • Publication dates: database inception through 2026-09-04.
  • Languages: English-language records and English metadata for non-English records. This language restriction will be reported as a limitation.
  • Document types: peer-reviewed papers, conference papers, technical reports, standards, official guidance, theses, datasets, and authoritative manuals.
  • Publication status: published and formally archived preprints; preprints will be identified and not treated as peer-reviewed evidence.

Information sources

Sources executable without institutional access

  1. OpenAlex metadata search.
  2. Crossref metadata search.
  3. INIS and IAEA publications.
  4. OSTI.GOV.
  5. PubMed where indexed.
  6. Publisher, standards-body, laboratory, and official government sites.
  7. Backward and forward citation searching from included records.

Sources requiring institutional or owner access

Before describing the review as complete, reproduce the searches in:

  1. Scopus;
  2. Web of Science Core Collection;
  3. IEEE Xplore; and
  4. INSPEC, if available.

Unavailability, access date, and the resulting coverage limitation will be reported rather than silently treated as zero results.

Search concept blocks

Database syntax may change, but no concept may be deleted without recording a protocol amendment.

S1 — Industrial converted-X-ray shielding

(industrial OR irradiation OR processing)
AND (electron accelerator OR electron linac OR LINAC)
AND (bremsstrahlung OR converted X-ray OR X-ray converter)
AND (shielding OR radiation protection OR barrier OR bunker OR maze OR dose)

S2 — Converter and source verification

(electron accelerator OR electron beam)
AND (bremsstrahlung converter OR X-ray converter OR target)
AND (5 MeV OR 6 MeV OR 7 MeV OR 7.5 MeV OR 8 MeV OR 10 MeV)
AND (spectrum OR angular distribution OR yield OR measurement OR Monte Carlo)

S3 — Photonuclear transport

(electron accelerator OR bremsstrahlung)
AND (photoneutron OR photonuclear OR neutron)
AND (PHITS OR MCNP OR Geant4 OR TRIPOLI OR Monte Carlo)
AND (target OR converter OR shielding)

S4 — Radiation-transport surrogates

(machine learning OR surrogate OR emulator OR reduced order OR neural operator)
AND (radiation transport OR Monte Carlo)
AND (shielding OR dose field OR photon OR electron accelerator)

S5 — Exact contribution intersection

(PHITS)
AND (industrial electron accelerator OR industrial X-ray LINAC)
AND (bremsstrahlung converter)
AND (shielding)
AND (machine learning OR surrogate)

Search execution and record keeping

For every source and query, record:

  • source name and interface;
  • exact query as executed;
  • search date;
  • filters and sort order;
  • number returned by the source;
  • number exported or inspectable;
  • export format or result URL; and
  • access limitations or errors.

Search results will be combined using DOI as the primary identifier, followed by normalized title and publication year. A journal article supersedes its conference abstract only when it contains the same study; both remain linked in the provenance record.

Screening

Screen in two passes:

  1. title and abstract; then
  2. full text or the fullest lawfully accessible technical record.

Each record receives one decision: include-direct, include-method, exclude, or awaiting-full-text. Exclusion uses one primary reason. Screening will not infer applicability from title keywords alone.

Because one researcher is performing the initial screening, all borderline and included records will receive a second consistency pass. The absence of an independent second screener will be reported as a limitation.

Data extraction

For each included record, extract when available:

  • bibliographic identity and document type;
  • accelerator architecture and application;
  • source mode and incident-electron energy;
  • converter materials, layers, dimensions, and beam conditions;
  • geometry, shielding materials, densities, and evaluation locations;
  • transported particles, code, version, data libraries, and physics settings;
  • quantity, unit, normalization, grid, and conversion coefficients;
  • measurements, detector response, uncertainties, and comparison method;
  • sampling unit, Dataset size, features, targets, and partitioning;
  • surrogate architecture, baselines, metrics, and test design;
  • principal finding relevant to Paper 1;
  • applicability to Paper 1; and
  • limitation preventing broader use.

Missing details will be recorded as missing, not inferred.

Quality and applicability assessment

Assess each source separately on:

  1. source and geometry specification;
  2. material and physics specification;
  3. quantity and normalization clarity;
  4. statistical or measurement uncertainty;
  5. verification or validation strength;
  6. reproducibility of the reported method; and
  7. direct applicability to Paper 1.

Each domain receives adequate, partial, inadequate, or not-applicable, with a short evidence-based reason. No aggregate score will turn an inapplicable study into direct evidence.

Synthesis and decision rule

Evidence will be synthesized by component rather than pooled statistically:

  • industrial system and scope;
  • converter and source;
  • shielding geometry and materials;
  • predicted quantity and normalization;
  • photonuclear treatment;
  • PHITS verification;
  • physical validation;
  • Dataset design; and
  • surrogate development and evaluation.

The provisional scope will be retained only if the evidence supports all necessary components and the exact-contribution search does not locate a study that already establishes the same combined contribution. Otherwise the scope or novelty statement will be revised before PHITS reference-Dataset generation.

Protocol amendments

Any change after 2026-09-04 must record the date, original rule, revised rule, reason, and affected screening records in EVIDENCE_REVIEW.md. Search expansion for synonyms or citation chasing is allowed when recorded; post hoc deletion of an inconvenient concept is not.