radiology

Radiology (radiology)

Journal positioning

Radiology is the flagship journal of the Radiological Society of North America (RSNA),

publishing original research across diagnostic and interventional imaging — imaging

physics and technique, diagnostic accuracy, image-guided intervention, and imaging

artificial intelligence — with a strong emphasis on rigorous design, adequate sample

size, and clinical relevance. The defining expectation is a **methodologically sound

imaging study with a clinically meaningful question and an appropriate reference

standard**, not a small retrospective series or an AI model evaluated on a single

internal dataset. This skill is a fit / venue-selection / re-framing aid; it is

not clinical or regulatory advice and does not replace the journal's current

instructions. Before submitting, re-check the live Radiology author instructions.

When to trigger

• The author names Radiology for a diagnostic-imaging, imaging-physics, interventional,

or imaging-AI study and wants a fit/framing check.

• An imaging study must be re-framed around a clinically meaningful diagnostic or

outcome question with a valid reference standard.

• The author is choosing between Radiology, a subspecialty imaging journal, and a

general clinical journal.

• The author needs the journal's diagnostic-accuracy reporting and reproducibility

expectations (STARD, CLAIM for AI).

Scope & topic fit

• Diagnostic-accuracy studies across modalities (CT, MRI, ultrasound, PET, radiography)

with an appropriate reference standard.

• Imaging physics, acquisition, reconstruction, and quantitative-imaging biomarker

development and validation.

• Image-guided and interventional procedures with outcome data.
• Artificial intelligence and machine learning for imaging, with rigorous training/

validation/test design and external validation.

• Prognostic and screening imaging studies with clinically meaningful endpoints.

Method & evidence bar

• Diagnostic-accuracy studies need an adequate, representative sample, a valid and

independent reference standard, and reporting per STARD; spectrum and verification

bias must be addressed.

• Sample size and statistical power must be justified; reader studies require adequate

readers and inter-/intra-reader agreement analysis.

• AI/ML studies require clearly separated training/validation/test data, external/

multi-site validation, and reporting per CLAIM; performance must be benchmarked

against a clinically relevant baseline (e.g., radiologists or standard of care).

• Quantitative-imaging claims need repeatability/reproducibility evidence and, where

relevant, multi-vendor/multi-site generalizability.

• Retrospective designs must address selection bias and confounding; prospective and

multi-center evidence strengthens fit.

Structure & house style

• RSNA format with a structured abstract and a short "key results" / summary statement;

re-check current article types (Original Research, etc.) and limits on the live guide.

• A STARD (or CLAIM for AI) flow diagram and completed checklist are expected where

applicable.

• Figures are central and must be high-quality, de-identified images with clear

annotations; report acquisition parameters.

• Methods must give enough acquisition, analysis, and (for AI) model and data detail

to allow reproduction; data/code sharing strengthens the submission.

Official-submission checklist

• Before giving submission-ready advice, read ../../resources/source-basis.md and

../../resources/official-source-map.md; start from the ICMJE/EQUATOR and RSNA

anchors, then cite the current Radiology page you checked.

• Search the live site for "Radiology RSNA instructions for authors" and follow the

current version.

• Re-check article types, abstract/summary format, and word/figure limits.
• Confirm the STARD (diagnostic) or CLAIM (AI) checklist, and prospective registration

where the study design requires it.

• Re-check IRB/ethics and consent, patient-image de-identification and consent, ICMJE

authorship and conflict-of-interest disclosure, funding, data/code availability, and

AI-use disclosure.

• If the live official instructions conflict with this skill, the official instructions

win.

Pre-submission self-check

• [ ] The study asks a clinically meaningful imaging question with a valid, independent reference standard.
• [ ] Sample size/power is justified; reader studies report inter-/intra-reader agreement.
• [ ] AI/ML work separates train/validation/test data and includes external/multi-site validation (CLAIM).
• [ ] Diagnostic-accuracy reporting follows STARD with a flow diagram; spectrum/verification bias addressed.
• [ ] Images are de-identified, high-quality, and annotated; acquisition parameters reported.
• [ ] IRB/consent, disclosures, and a data/code-availability statement are prepared.

Common desk-reject triggers

• Small, single-center retrospective series with no reference-standard rigor or limited generalizability.
• AI models evaluated only on internal data, with no external validation or clinical baseline.
• Diagnostic-accuracy studies with verification or spectrum bias and no STARD reporting.
• Quantitative-imaging claims with no repeatability/reproducibility evidence.
• Pure technical/phantom work with no clinical relevance, better suited to a physics or subspecialty journal.

Re-routing decision

• Subspecialty imaging focus (neuro/cardiac/abdominal) → a dedicated subspecialty imaging journal.
• Imaging-AI methodological advance over clinical validation → a medical-imaging methods venue (e.g., ieee-transactions-on-medical-imaging in the engineering bundle).
• Cardiology/neurology clinical outcome dominant over imaging method → jama-cardiology / jama-neurology / stroke.
• Oncology imaging with a clinical-oncology endpoint → jama-oncology / annals-of-oncology.
• Broad, practice-changing significance → general medicine (jama / NEJM in the natural-science bundle).

Output format

[Fit] High / Medium / Low (one-line reason)
[Target] Radiology (RSNA)
[Imaging tags] <modality + task, e.g. MRI diagnostic accuracy, imaging AI>
[Design / reporting guideline] <diagnostic-STARD / AI-CLAIM / interventional-outcome>
[Method/evidence] <reference standard, sample size, external validation>
[Top risk] <the single most likely reason for rejection>
[Official items to re-check] <article type / STARD or CLAIM / registration / de-identification / disclosures>
[Re-route suggestion] <if not a fit, a better-matched venue>