A graphical abstract is often the first part of your paper a browsing reader sees. It must state one finding clearly at thumbnail size, in the exact format the journal specifies.
What a graphical abstract is for
A graphical abstract is a single image that gives a reader the take-home message of your paper at a glance. Elsevier's graphical abstract guidance notes that it appears in online search results, contents lists and on the article page, but typically not in the PDF or print version.
So you are designing for someone scanning a list of titles, often outside your specialty, who has not read your abstract and sees your image small. That is why a graphical abstract is not a shrunken copy of your best figure. A figure presents evidence for a claim, with axes, error bars and a legend. A graphical abstract states the claim itself.
Can you reuse a figure?
Publishers disagree, so check before you plan:
- Elsevier says some authors use a concluding figure from the article, while many now design a separate image.
- Cell Press asks for an image distinct from any model figures in the paper, with no data items of any type.
- ACS tells authors to avoid a graphic that already appears in the manuscript.
If your journal says nothing, design a new image. It is rarely more work than cleaning up a crowded multi-panel figure.
Graphical abstract or visual abstract?
Elsevier uses the two terms interchangeably. In medicine, though, "visual abstract" often means the format Annals of Surgery introduced in July 2016: panels summarising the study question, key outcomes with numbers and units, and the citation, built for social media. If a journal asks for a visual abstract, check whether it supplies a template. The difference matters, because one format may forbid data and the other is built around it.
Read the journal's requirements before you draw anything
Specifications vary between publishers and sometimes between journals from the same publisher. Three examples show how far apart they are:
| Publisher guidance | Shape and size | File types | Text |
|---|---|---|---|
| Elsevier | Landscape, at least 1328 × 531 pixels (width × height) at 300 dpi; larger images keep the same ratio | TIFF, EPS, PDF or MS Office | Times, Arial, Courier or Symbol, large enough to survive reduction |
| Cell and Cell Reports Medicine | Exact square, 1200 × 1200 pixels at 300 dpi | TIFF, PDF or JPG | Simple labels, text used sparingly; Arial, 8–12 pt in the graphical abstract guide |
| ACS | Fits within 3.25 × 1.75 inches (about 8.25 × 4.45 cm) at actual size | TIF at 300 dpi for colour and 1200 dpi for black and white, or EPS | Sans serif such as Helvetica, preferably 8 pt and no smaller than 6 pt |
Those three are basic-science and chemistry examples, useful mainly because they publish their numbers. Many clinical journals publish no graphical abstract specification at all: they say nothing, point you at a general figure page, or supply their own visual abstract template. Silence is not an absence of rules. PLOS Medicine's figure guidelines govern any image you submit: TIFF or EPS only, 789–2,250 pixels wide at 300–600 dpi, and Arial, Times or Symbol at 8–12 point. If your journal has no graphical abstract page, build to its figure rules and ask the editorial office what shape they want.
The stage at which you need one varies too. The ACS guidelines for table of contents graphics say a graphic must be included with each manuscript. Cell lists the graphical abstract among its final file requirements. Elsevier sends authors to each journal's guide for authors. A layout designed for a wide rectangle rarely works as a square, so if you move to another publisher, expect to rebuild the image rather than resize it.
Before you open any software, note these six things from the guide for authors:
- Shape and minimum pixel dimensions
- Accepted file types
- Font and minimum text size
- Whether data, numbers or reused figures are allowed
- Whether the file is due at submission, revision or acceptance
- Any rule on third-party images or generative AI
Treat the graphical abstract as one item in your submission package. Our manuscript submission checklist covers the others.
How to make a graphical abstract in six steps
1. Write the take-home message as one sentence
Write the finding as a single sentence you could say aloud to a colleague. It is the same discipline as stating your finding in a cover letter to the editor. For clinical research, a workable pattern is population, intervention or exposure, comparison and outcome: In [population], [intervention] was associated with [outcome] compared with [comparator].
Then test the sentence against your results. An observational study should not produce an image that implies causation, and an effect found in one subgroup cannot be drawn as a general result. Cell asks authors to leave out speculative features unless their speculative nature is made visually clear, and Andrew Ibrahim's visual abstract primer, written from his work as the first Creative Director at Annals of Surgery, says the image "should simply reflect the actual article itself." Reviewers notice a bold image attached to careful results, so the advice on avoiding overclaiming applies here too.
2. Choose a reading direction
Elsevier and Cell both ask for a clear start and end, reading from top to bottom or left to right. Left to right suits a wide rectangle; a square gives you room to read from top to bottom.
A structure that works for most biomedical papers has three zones:
- Context: who or what was studied, such as the population, tissue, cell type or setting. Cell specifically asks for a visual indication of biological context.
- What was done: the intervention, exposure, comparison or mechanism.
- What was found: the outcome, shown as a direction or change rather than a table.
3. Sketch it at the final shape
Draw a box in the journal's proportions and sketch in pencil: rectangles for zones, arrows for the flow, one word per element. Work in grey, and add colour last to point at the new finding. Cell warns that heavily saturated primary colours can distract. Do not rely on red versus green alone to carry meaning, because some readers cannot tell them apart.
If the sketch needs more than three zones or arrows that cross, the message is too big. Go back to step 1.
4. Build it at the journal's dimensions
Set the canvas to the required size before placing anything. Presentation software is enough for most designs; ACS suggests thinking of the graphic as a single PowerPoint slide that tells the story of your article. A vector editor such as Inkscape, which is free and open source, gives finer control over lines and text.
Keep the style consistent: one font family, two or three colours, the same line weight on every arrow, and icons from a single visual style. As you place each image, record its source and licence.
5. Cut the text and test at display size
Delete every word that an icon or arrow already conveys. Then view the image at the size readers will see: Elsevier scales graphical abstracts to fit a 500 × 200 pixel window, and to 200 pixels high in contents lists. Enlarge or remove any label you cannot read.
ACS suggests two more checks worth doing for any journal: print the graphic to see whether lines and text stay crisp, and ask someone who has not read the manuscript what the paper is about. If their answer does not match your step 1 sentence, revise.
6. Export and check the file
Export in an accepted format, then open the file's properties and confirm the actual pixel dimensions rather than trusting the export settings. Never enlarge a small image to reach a pixel minimum; it will look blurred. Rebuild it at the correct size instead.
Check the publisher's handling rules as well. Elsevier asks for the graphical abstract as a separate file, with no heading reading "graphical abstract" and no unnecessary white space. ACS asks for fonts in EPS files to be embedded or converted to outlines, and for the graphic to be labelled "For Table of Contents Only" on the last page of the manuscript.
Where images and icons can safely come from
The safest image is one you drew yourself. A few journals require it: the ACS guidelines, for example, ask for entirely original, unpublished artwork created by one of the coauthors. Most medical journals are less absolute and instead make third-party material your problem to clear, which in practice means keeping evidence of every licence and permission.
Where journals allow third-party material, use sources with clear terms and meet them:
| Source | Terms to check |
|---|---|
| Servier Medical Art | Licensed CC BY 4.0: reuse and adaptation allowed, with credit |
| NIH BioArt Source | Free for any use, but each entry has its own licence and attribution terms |
| Bioicons | Each icon carries its own licence, such as CC0, CC BY, CC BY-SA or MIT |
| BioRender | The Free plan is marked "No commercial use or publication"; journal use needs a paid plan, and BioRender's terms require you to follow the citation guidelines shown in the application when you publish |
Creative Commons licence codes tell you what is allowed. BY means you must give credit. SA means adaptations must be shared under the same terms. ND means no adaptations, so avoid ND images if you plan to modify them. NC limits use to noncommercial purposes, which is the code that catches out medical authors most often: PLOS Medicine, for example, publishes every figure under CC BY, a licence that permits commercial reuse, so an NC icon cannot lawfully travel into that article. Check the licence your article will carry before you place NC material.
Leave out:
- Pictures found through a general web search. Being online does not make an image free to use.
- Figures from published papers, including your own earlier ones. Cell asks for copies of permissions for any figure reprinted or adapted from a previous publication, so clear the reuse before you build the image, not after a revision request. Your own published figure is not automatically yours to reuse: check the licence you signed.
- Logos and trademarked products. The ACS guidelines exclude trademarked items such as company logos.
- Identifiable patient images. The ICMJE Recommendations say identifying information should not appear in photographs unless it is essential for scientific purposes and the patient gives written informed consent. A generic icon usually makes the point.
Generative AI images
Elsevier's generative AI policies for journals state that general-purpose generative AI image tools must not be used to create graphical abstracts, and encourage dedicated scientific illustration tools with well-defined licensing terms. Other publishers set their own rules, so check your journal's current policy before using any AI image tool.
How much text a graphical abstract should carry
Use labels, not sentences. Two decisions follow from the publisher rules in the table above:
- Numbers depend on the format. Cell's guidelines exclude data items, while a visual abstract is built around key outcomes with units. Follow the format your journal asks for.
- Words have their own place. Some journals ask for a text summary separately. Cell, for example, asks for three to four highlights of no more than 85 characters each. Put the sentences there, not in the picture.
If your image only makes sense after reading the text inside it, you have made a slide. Cut words until the arrows and icons carry the message.
Before you upload
- Shape, pixel dimensions and file type match the guide for authors, with no upscaling
- One message, with a clear start and end
- Every label readable when the image is viewed at display size
- No "graphical abstract" heading or unnecessary white space
- No data, numbers or reused figures unless the journal allows them
- Nothing in the image claims more than your results show
- Every image original or licensed, with credits and permissions recorded
- No identifiable patient information, logos or trademarked products
- Any AI use checked against the journal's current policy
- Someone who has not read the paper can say what it found
The one thing worth extra time
Spend your time on the sentence in step 1. A clean layout cannot rescue a message that is vague or larger than the evidence, while a precise sentence almost designs itself into three zones and two arrows.
If you are preparing a paper for a Directive Publications journal, start with the author guidance collected on our For Authors page and check the journal's own page for any additional file requirements. When your files are ready, you can submit your manuscript directly.
Frequently asked questions
What is the difference between a graphical abstract and a regular figure?
A figure presents evidence inside the paper, usually data with axes, legends and statistics, and it is read alongside the text. A graphical abstract is a single image that states the paper's main message for someone browsing search results or a contents list. Cell Press asks for an image distinct from any model figures or diagrams in the paper itself, and ACS advises authors to avoid a graphic that already appears in the manuscript text, so check your journal before adapting an existing figure.
What free tools can I use to create a graphical abstract?
Free presentation software such as LibreOffice Impress is enough for most designs if you set the page to the journal's proportions, and Inkscape is a free, open-source vector editor if you want finer control. For icons, Servier Medical Art is licensed CC BY 4.0, NIH BioArt Source images are free for any use under each entry's licence, and Bioicons shows a licence for every icon. BioRender's Free plan is marked no commercial use or publication, so you need a paid plan and its citation before using it for a journal.
Can I use images or icons from the internet in a graphical abstract?
Only if the licence allows it and you meet its terms, such as giving credit for CC BY material. Finding an image through a web search is not permission to use it, and figures from published papers need permission from the rights holder. Check the licence your own article will carry too, because a noncommercial icon cannot go into an article published under CC BY. Some journals are stricter still and require entirely original artwork created by one of the authors.
Is a graphical abstract required for submission?
It depends on the journal. Many clinical journals treat it as optional or invite one after acceptance, while Cell lists the graphical abstract among its final file requirements and Elsevier directs authors to each journal's guide for authors. Check the current guidelines for your specific journal, including the stage at which the file is due, and ask the editorial office if the guide is silent.
How much text should a graphical abstract include?
As little as possible: short labels rather than sentences. ACS asks authors to limit text to labels and avoid long phrases, and Elsevier requires any text to be part of the image file with no separate synopsis. If the image cannot be understood without reading a paragraph inside it, simplify the design rather than shrinking the font.