IEEE Transactions on Medical Imaging (ieee-transactions-on-medical-imaging)

Journal positioning

IEEE Transactions on Medical Imaging (TMI), published jointly by several IEEE societies, is a flagship archival venue for methods in medical image formation, reconstruction, and analysis across modalities (MRI, CT, PET/SPECT, ultrasound, optical, and microscopy), including registration, segmentation, quantification, and machine learning for medical imaging. The defining expectation is a method whose contribution is specific to medical imaging and validated with appropriate data and proper technical and, where relevant, clinical evaluation — not a generic computer-vision method run on a medical dataset as an afterthought. Papers without medical-imaging specificity, or evaluated on tiny/unrepresentative data without rigorous protocol, are a poor fit. This skill is a fit / venue-selection / re-framing tool. It does not replace the journal's current official author information. Before submitting, re-check the live IEEE TMI author guidance and submission system.

When to trigger

• The author names TMI for an image reconstruction, registration, segmentation, or

imaging-ML manuscript and wants a fit/framing check.

• A method must be re-framed so the medical-imaging-specific contribution — the

physics, the modality, or the clinical task — is central, not a generic CV result.

• The author is unsure whether the contribution belongs in TMI (imaging methods) or

a broader biomedical/translation venue.

• The author needs TMI's dataset-and-validation bar and desk-reject heuristics.

Scope & topic fit

• Image formation and reconstruction: inverse problems and model-based or

learning-based reconstruction for MRI, CT, PET/SPECT, ultrasound, and optical imaging.

• Image analysis: segmentation, registration, detection, and quantification with a

medical-imaging-specific methodological advance.

• Machine learning for medical imaging when the method addresses imaging-specific

challenges (modality physics, limited/heterogeneous labels, domain shift, artifacts).

• Quantitative imaging, biomarker extraction, and motion/artifact correction tied to

a defined imaging or clinical task.

• Imaging-system and acquisition methods (sampling, hardware-aware reconstruction)

evaluated on realistic or measured data.

• Computational/physics models of image formation validated against acquired data.

Method & evidence bar

• The contribution must be imaging-specific: exploit modality physics, acquisition

model, or clinical task; a generic network applied to images does not clear the bar.

• Validation must use appropriate datasets with adequate size and diversity; report

data source, acquisition, ground-truth/reference standard, and any patient/ethics provenance.

• Evaluation must use task-appropriate metrics with statistics: reconstruction

fidelity, segmentation overlap/boundary error, registration accuracy, detection performance — with confidence intervals or significance where claimed.

• Compare against the right baselines (established imaging methods, not only one CV

model), with matched preprocessing and fair tuning; ablate key components.

• Address generalization and robustness: cross-site/scanner/protocol variation,

out-of-distribution behavior, and failure modes relevant to clinical use.

• Reproducibility: enough detail (and ideally code and data access per policy) to

reproduce the reported results.

Structure & house style

• IEEE double-column format; TMI publishes full-length Papers — match the

contribution to that archival scope and re-check current article types and length policy on the live guide.

• The introduction motivates the imaging/clinical gap and the methodological need,

then states the contribution; avoid framing it as a generic-CV improvement.

• Figures are load-bearing: example images with the relevant overlays, quantitative

comparison plots, and failure cases; include clinically meaningful visualizations.

• The methods section must specify the imaging model, data, and evaluation protocol

precisely enough to reproduce.

• A results section with quantitative tables across datasets and baselines is central.

Official-submission checklist

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

../../resources/official-source-map.md; start from the IEEE Author Center anchors, then cite the current TMI-specific page you checked.

• Search the live site for "IEEE Transactions on Medical Imaging information for

authors" and follow the current ScholarOne/IEEE version.

• Re-check article types, page/length limits and overlength policy, and the IEEE

double-column template.

• Confirm data/code-availability, human-subjects/ethics/IRB, and any de-identification

and reporting requirements.

• Re-check ORCID, competing-interests, funding, author-contribution, and AI-use

disclosure requirements, and IEEE open-access options.

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

instructions win.

Pre-submission self-check

• [ ] The contribution is medical-imaging-specific (physics/modality/clinical task), not a generic CV method on medical data.
• [ ] Validation uses appropriate, adequately sized and diverse datasets with documented reference standards.
• [ ] Metrics are task-appropriate and reported with statistics; baselines are the right imaging methods.
• [ ] Generalization across site/scanner/protocol and failure modes are addressed.
• [ ] Ethics/IRB and data provenance/de-identification are documented.
• [ ] Article type and length fit current TMI limits; methods are reproducible.

Common desk-reject triggers

• A generic computer-vision/deep-learning method with no medical-imaging-specific contribution.
• Evaluation on a tiny, single-site, or unrepresentative dataset with no rigorous protocol.
• Missing or inappropriate baselines; unfair comparisons or no ablation of the claimed novelty.
• No ethics/IRB statement or data provenance for human-subject imaging data.
• Clinical-utility claims with no clinically meaningful validation or appropriate reference standard.

Re-routing decision

• Broader biomedical engineering (devices, biosignals, non-imaging) → ieee-transactions-on-biomedical-engineering.
• Highest-significance clinical translation/impact story → nature-biomedical-engineering.
• Core contribution is a general signal-processing method → ieee-transactions-on-signal-processing.
• Surgical/medical-robotics contribution as the core → ieee-transactions-on-robotics.
• General computer-vision advance with no medical specificity → a computer-vision venue.

Output format

[Fit] High / Medium / Low (one-line reason)
[Target] IEEE Transactions on Medical Imaging
[Topic tags] <2–3 closest medical-imaging subtopics>
[Imaging-specific contribution] <what makes the method imaging-specific in one line>
[Method/evidence] <do the dataset + metrics + baselines clear TMI's validation bar?>
[Top risk] <the single most likely reason for rejection>
[Article type] Paper
[Official items to re-check] <article type / length / template / data-ethics / disclosures>
[Re-route suggestion] <if not a fit, a better-matched venue>