IEEE Transactions on Medical Imaging Submission Guide

Official TMI pages explain the 10-page rule, checklist, prior-review disclosure, file requirements, and Author Portal workflow. They do not tell you whether your paper reads as TMI-ready medical-imaging methodology or as a clinical application paper, MICCAI-style conference extension, or general machine-learning paper. This guide focuses on that pre-upload judgment.

If you want the quick pre-upload call, run an IEEE Transactions on Medical Imaging submission readiness check before opening the IEEE Author Portal.

Source: IEEE TMI journal home, Submission Checklist, TMI author instructions, 2026 IEEE APC list, Clarivate JCR 2024, accessed May 2026.

IEEE TMI's initial-submission page limit is the central editorial constraint:

Verbatim from the Submission Checklist: Initial submissions are limited to 10 pages (10-point font size or larger), including references.

The 10 pages include:

• Title and authors

• Abstract

• Main text

• Figures and tables

• References (this is the part authors most often miss)

The cap is enforced. TMI states that initial regular or special-issue papers over 10 pages are returned without review. Challenge papers are different and have a 14-page initial limit.

The strategic implication: write the paper for the 10-page envelope from the start. Conference papers from CVPR, MICCAI, or NeurIPS often include long related-work sections that need to be compressed for IEEE TMI. The fix is to consolidate related-work into a focused 1-1.5 page survey, move detailed proofs and additional experiments to supplementary material, and keep references tightly relevant rather than encyclopedic.

A new Editor-in-Chief term began on January 1, 2025. Verify the current Editor-in-Chief on the journal's editorial-team page before quoting any name in a cover letter. The current editorial direction emphasizes biomedical imaging, deep learning for medical imaging, and AI-augmented imaging methods: recent IEEE TMI issues are heavily weighted toward deep-learning methodology, generative models for medical imaging, and AI/ML methods.

The practical consequence: the 2025-2026 editorial culture rewards papers that contribute methodological advances at the intersection of deep learning and medical imaging. Pure-classical-imaging papers (ultrasound physics, MRI physics, CT reconstruction without learning components) still publish, but the high-traffic editorial direction favors AI/ML-augmented work.

Recent editorial signals include:

• A Special Issue on Large Multimodal and World Models for Medical Imaging (deadline August 1, 2026), reflecting strong interest in foundation-model approaches

• Increasing volume of diffusion-model papers (image generation, super-resolution, reconstruction)

• Continued interest in classical imaging (ultrasound, optical) when methodology is genuinely novel

Editors specifically screen for three operational signals before sending a paper into review:

1. Medical-imaging methodological novelty. TMI publishes methodology, not pure clinical-application studies. A paper that applies an existing deep-learning method to a new disease cohort fits clinical journals (Radiology, Journal of Magnetic Resonance Imaging) better. A paper that introduces a novel reconstruction algorithm, generative model, or methodological framework with experimental validation fits TMI.

If the contribution is a general signal-processing method rather than an imaging-specific method, compare it with the IEEE Transactions on Signal Processing submission guide before treating TMI as the natural IEEE venue.

2. Comparison to state-of-the-art baselines on standardized benchmarks. The journal expects rigorous comparison to recent baselines using established benchmarks (e.g., fastMRI for MRI reconstruction, AAPM challenges for CT reconstruction). Papers that report results on bespoke datasets without standardized comparison face a harder editorial read.

3. Reproducibility signals. Code availability, model release, and data availability are increasingly expected. The journal does not yet enforce mandatory code release, but the current editorial direction emphasizes reproducibility, and papers without code-availability commitment may face additional reviewer scrutiny.

Recent papers, with DOIs:

• "3D MedDiffusion: A 3D Medical Latent Diffusion Model for Controllable and High-Quality Medical Image Generation" (December 2025, Vol 44 Issue 12, pages 4960-4972), 10.1109/TMI.2025.3585372. 3D medical latent diffusion model with patch-volume autoencoder for medical-image generation across CT, MRI, and segmentation tasks.

• "Likelihood-Scheduled Score-Based Generative Modeling for Fully 3D PET Image Reconstruction" (2025), 10.1109/TMI.2025.3576483. Score-based generative modeling for 3D PET reconstruction.

• "MRI Super-Resolution with Partial Diffusion Models" (March 2025, Vol 44 Issue 3, pages 1194-1205), 10.1109/TMI.2024.3483109. Accelerated diffusion model for MRI super-resolution.

The pattern across recent TMI papers: each is a methodological contribution at the deep-learning / medical-imaging intersection, with experimental validation on standardized medical-imaging benchmarks, and clear comparison to state-of-the-art baselines. Diffusion models, foundation models, and generative-modeling approaches dominate the 2025-2026 editorial direction.

For initial submission via the IEEE Author Portal:

1. Manuscript within 10 pages (10pt font including references)

1. Title page with all authors and affiliations, including ORCID identifiers

1. Abstract within standard length

1. Cover letter only when needed for prior review or conference-paper extension

1. Suggested reviewers and exclusions as needed

1. Supplementary material for proofs, additional experiments, and model details that don't fit in 10 pages

1. Code-availability statement (recommended; reviewers increasingly expect it)

1. Conflict-of-interest disclosure for all authors

1. Author contributions statement following IEEE EMBS author-role guidance

1. Funding statement disclosing grants, sponsor support, and any clinical-trial funding

1. Ethics statement for human-subjects data, IRB approval, or sensitive medical-imaging datasets

1. Data availability statement with references to public medical-imaging benchmarks where used

1. Conference predecessor PDF plus a difference-explanation document where the submission extends MICCAI, CVPR, or other conference work

A IEEE TMI submission readiness check before upload can flag whether the manuscript fits the 10-page rule, whether the methodological contribution is visible in the introduction, and whether reproducibility signals (code, baselines, benchmarks) are present.

TMI's flow follows the IEEE EMBS/SPS co-sponsor policies and what TMI authors report through community channels. Treat as planning ranges, not promises.

• Day 0: IEEE Author Portal upload. The portal accepts the package, runs the 10-page limit check, and routes to an Associate Editor in the matching imaging-modality area.

• Days 1 to 14: Administrative review and editor assignment. Editorial staff verify the 10-page limit, IEEE format, conference-extension declarations, and prior-review disclosure; the Associate Editor evaluates methodological novelty.

• Days 14 to 30: Reviewer invitations. TMI typically invites two to three reviewers with deep-learning and medical-imaging expertise. Finding reviewers for novel modalities (foundation models, diffusion) can extend the timeline.

• Days 30 to 120: Peer review. Reviewer reports return on a 8 to 12 week cadence; methods-heavy papers extend the timeline because reviewers verify benchmarks and baseline comparisons.

• Days 90 to 150: First editorial decision. Major revision is the most common outcome for papers that pass desk review.

• Days 150 to 360: Revision rounds and acceptance. Single-revision acceptances run roughly 10 months; multi-round revisions push closer to 12 to 14 months. Online publication typically follows acceptance within weeks.

The sources above define the mechanics; the harder question is whether this draft earns review. The review tells you whether your paper clears the IEEE Transactions on Medical Imaging fit check before upload, especially around manuscript exceeds the 10-page initial-submission cap (including references), application-focused paper without methodological novelty in the introduction's contribution statement, and insufficient comparison to current 2024-2026 baselines on established benchmarks. Paid Manusights reviews include a 60-day money-back guarantee, and we do not train models on submitted manuscripts.

Across medical-imaging manuscripts targeting IEEE Transactions on Medical Imaging, three recurring decision risks matter most across submissions that TMI associate editors filter out at the desk-screen stage. (Per IEEE TMI submission instructions, initial regular submissions are limited to 10 pages including references, challenge papers have a 14-page initial limit, and prior-review or conference-extension history must be disclosed.) Use the three checks below before you open IEEE Author Portal upload slot.

A Manusights review checks whether your paper clears the IEEE TMI-specific readiness checks that official IEEE instructions cannot evaluate from a generic Author Portal checklist. Paid Manusights reviews include a 60-day money-back guarantee; submitted manuscripts are not used for model training.

Manuscript exceeds the 10-page initial-submission cap (including references)

Across TMI-targeted manuscripts, we consistently see authors arriving from CVPR (8-page main + supplementary appendix), MICCAI (8-10 page papers + supplementary), or NeurIPS (9-page main + extended appendix) submit longer manuscripts and discover the strict 10-page cap at editorial review.

Since January 2022, IEEE TMI no longer accepts supplementary documents that extend the manuscript with additional text and figures, so the 10-page cap (at 10-point font in IEEE double-column single-spaced format, references included) is the entire envelope: methods, results, discussion, and references must all fit. The ScholarOne submission portal portal returns over-cap manuscripts without review.

The fix is to compress the related-work section to a focused 1-page positioning, move detailed proofs and bonus experiments to a public repository (referenced in the methods section but not extending the page count), keep the references tightly scoped to the 30-40 most-relevant prior works, and reserve the 10 pages for the methodological contribution + benchmark comparison + ablation studies.

Check whether your IEEE TMI manuscript fits the 10-page envelope →

Application-focused paper without methodological novelty in the introduction's contribution statement

We frequently see TMI manuscripts apply established deep-learning architectures (U-Net, nnU-Net, ViT, diffusion models) to new medical-imaging tasks (a new disease cohort, a new modality, a new clinical workflow) without introducing the methodological advance that TMI's editorial direction requires.

The 2025 editorial direction specifically emphasizes methodology development (diffusion models for inverse problems, foundation models for cross-modality, generative models for imaging) over application-only deployment.

Manuscripts where the introduction's contribution statement reads "we apply [established method] to [new task] and report [performance numbers]" without a methodological component get returned with the suggestion that the work fits a clinical-radiology journal (Radiology, European Radiology) or an applications-focused venue (Medical Image Analysis for some application classes) rather than TMI.

The fix is to surface the methodological novelty in the introduction's contribution statement (a new loss function, a novel training strategy, an architectural modification with theoretical justification, a new evaluation framework), and if no methodological novelty is present, redirect to an application-focused venue.

Check whether your IEEE TMI contribution statement names medical-imaging methodology →

Insufficient comparison to current 2024-2026 baselines on established benchmarks

The third recurring pattern in TMI-targeted manuscripts is comparison tables in the methods section that cite older 2020-2022 baselines or use bespoke datasets without standardized references, when TMI associate editors expect comparison against current 2024-2026 foundation-model methods on established benchmarks (fastMRI for MRI reconstruction, BraTS for brain tumor segmentation, AAPM challenges for low-dose CT, ACDC for cardiac segmentation, ISIC for dermatology).

TMI reviewers actively track the current foundation-model literature in medical imaging (MedSAM, BiomedCLIP, RadFM, recent diffusion-model papers) and comparison tables anchored on older baselines look strategically incomplete.

The fix is to include at least 2 baselines from 2024-2026 in the main comparison table on at least one established benchmark, run ablation studies that isolate the proposed methodological contribution from the architectural baseline, and report results following the benchmark's standard splits and metrics (DICE, HD95, PSNR, SSIM where appropriate) rather than custom evaluation protocols.

Check whether your IEEE TMI baseline and benchmark package is current enough →

Check whether your IEEE TMI manuscript is submission-ready →

Submission caps: IEEE TMI enforces a strict 10-page cap for Regular Articles (Challenge Papers get 14 pages) including references at 10-point font in IEEE double-column single-spaced format, which typically corresponds to 8,000 to 10,000 words main text with 6 to 10 main figures integrated inline. Since January 2022, supplementary documents that extend the manuscript with additional text and figures are no longer accepted.

The ScholarOne submission portal portal enforces the page limit and format compliance at upload; papers over 10 pages are returned without review per the IEEE TMI Instructions to Authors.

• Is IEEE Transactions on Medical Imaging a good journal?

• Medical Image Analysis Submission Guide

• Nature Methods Submission Guide

Last verified: May 2026 against IEEE TMI editorial pages and recent papers covering diffusion models, generative methods, and foundation-model approaches in medical imaging.