Not every change to a figure is dishonest. A microscopy image straight off the instrument may be too dark to read; adjusting it so readers can see what you saw is legitimate. But the same tools that make an image clearer can make it lie, and the difference between the two is one of the most misunderstood areas of research ethics. The principle is simple: you may improve how faithfully an image represents your data; you may never change what it shows.
What is acceptable
- Uniform adjustment of brightness, contrast or colour balance applied to the entire image.
- Cropping to focus on the relevant region — as long as it does not remove important context or hide a control.
- Resizing without distorting proportions.
The test for all of these: would the reader draw the same conclusion from the original file? If yes, and the whole image was treated the same way, you are fine.
What crosses the line
| Manipulation | Why it is misconduct |
|---|---|
| Splicing lanes or fields together | Presents images from different sources as one, implying they were run or captured together. |
| Cloning or copy-pasting regions | Duplicates data to fabricate a result or fill a gap. |
| Selective editing (erasing or adding features) | Removes inconvenient data or invents supporting data. |
| Non-uniform brightness/contrast | Adjusting only part of an image can make a faint band appear or disappear. |
Blots and gels have specific rules
Western blots and gels are where most image problems arise. Do not splice lanes to reorder or remove them without a clearly marked dividing line and a note in the legend. Show enough of each blot that a reader can judge it, and where a journal asks for the full, uncropped blots as a supplement, provide them. If bands were run on separate gels, say so.
Keep your raw data
Your single best protection is boring: keep the original, unprocessed image files, exactly as captured, with their metadata. If a question is ever raised, the raw files settle it in your favour. Record what adjustments you made and apply the same processing across a figure so it is defensible.
Journals are watching
This is not a theoretical risk. Many publishers now run figures through image-screening software that detects duplicated or altered regions, and trained reviewers routinely spot splices. Image manipulation is one of the fastest-growing causes of correction and retraction. The good news is that staying clean is easy: adjust the whole image or none of it, never add or remove features, and keep the originals.