03 · What You Need to Know
The Same Adjustment Can Clarify an Image or Distort It
Brightness and Contrast Are Display Operations With Scientific Consequences
Digital scientific images contain numerical intensity information. Display settings determine how those values are mapped into visible tones or colors.
Changing brightness or contrast can therefore make features easier to see without necessarily changing the underlying image file. But an extreme display adjustment can also suppress low-intensity features, saturate high-intensity regions, or exaggerate apparent differences.
The operation may sound cosmetic while its effect is evidentiary.
Global Adjustments Are Commonly Permitted When They Preserve Information
Scientific image-processing guidance generally distinguishes uniform adjustment of an entire image from selective adjustment of individual regions.
Cromey's widely cited guidance describes simple adjustments to an entire image as usually acceptable. Current Nature Portfolio image-integrity policies similarly permit processing such as brightness and contrast adjustment when it is applied appropriately across the entire image and equally to relevant controls.
The important qualification is that the adjustment should not cause data to disappear or create a misleading visual comparison.
Global adjustment
The same brightness or contrast transformation is applied across the image rather than selectively targeting particular features.
Selective adjustment
Only chosen regions or features are brightened, darkened, enhanced, suppressed, or otherwise processed, potentially changing their relationship to the rest of the image.
“Applied to the Whole Image” Does Not Mean Anything Goes
A global adjustment can still be excessive.
Suppose increasing contrast makes a faint secondary band disappear into the background. The software operation affected the whole image, but the displayed figure now conceals potentially relevant information.
Likewise, extreme brightness can saturate strong signals so that meaningful intensity differences are no longer visible.
The relevant question is not merely whether the slider was moved globally. Ask what information became invisible, exaggerated, or impossible to interpret as a result.
Do Not Process Experimental and Control Images to Different Evidentiary Standards
Comparative figures are particularly vulnerable to inconsistent processing.
Imagine increasing contrast aggressively in the experimental image so that a weak signal becomes striking while leaving the control image relatively dark. Readers may perceive a much larger difference than the underlying data support.
Nature Portfolio policies state that processing such as brightness and contrast should be applied equally to controls when comparison requires it. Scientific imaging guidance likewise emphasizes consistent acquisition and post-acquisition processing for images intended to be compared.
There can be technical situations in which different processing is scientifically necessary, but the reason and representation should be appropriate and disclosed where required.
Nonlinear Adjustments Require Particular Care
Not all contrast changes affect intensity values in the same way. Nonlinear operations such as gamma adjustments can alter relationships among displayed intensities differently from simple linear transformations.
Some journal policies require disclosure of nonlinear adjustments. Researchers should therefore understand what the software operation actually does rather than relying on the visual impression that the image merely looks clearer.
When quantitative information is derived from pixel intensities, image-processing decisions require even greater methodological care. Display processing and quantitative analysis should not be casually conflated.
Cropping Is Usually Acceptable When It Does Not Change Meaning
Cropping is one of the most common scientific figure operations. It can remove empty space, enlarge a region of interest, or focus attention on relevant structures.
Community-developed microscopy guidance explicitly recognizes cropping as permissible when it does not change the meaning conveyed by the image.
That qualification does most of the work.
A Crop Can Mislead by Removing Context
Suppose a micrograph contains 40 cells, only three of which display the predicted phenotype. Cropping tightly around those three cells and presenting the result as a representative image can create a very different impression from the full field.
No remaining pixel has been altered. The evidence has nevertheless been reframed through omission.
Similarly, cropping a gel so that an unexpected band disappears, removing a neighboring control, or excluding contextual features needed to interpret a specimen can make the resulting figure misleading.
This is why cropping belongs within the broader question of what counts as image manipulation in research, even though the operation does not necessarily modify the pixels that remain.
Cropping and Selecting a Representative Image Are Related but Different
A crop determines which portion of an image readers see. Image selection determines which image from the broader dataset readers see.
A researcher can therefore present a misleading figure without performing an aggressive crop at all. Selecting the most dramatic field from hundreds and labeling it “representative” can distort the apparent prevalence or consistency of a phenomenon.
The broader issue of selectively showing images or examples concerns how evidence is sampled for presentation, not merely how individual images are processed.
Cropped Gels and Blots Need Enough Information to Remain Interpretable
Journal policies may impose specific requirements for gels and blots. Nature Portfolio guidance, for example, states that cropped gels should retain important bands and requires clear indication when nonadjacent lanes have been rearranged.
Researchers should therefore avoid assuming that a crop acceptable for a conventional photograph is automatically appropriate for electrophoretic evidence.
The relevant standards depend partly on what information readers need to evaluate the experiment.
Never Use Brightness or Contrast to Erase an Unwanted Feature
One of the clearest red flags occurs when adjustment parameters are chosen specifically so that an inconvenient signal disappears.
A faint band, background feature, cell, lesion, spot, or other visual element may complicate the interpretation. If it exists in the original image and is scientifically relevant, adjusting the display until readers can no longer see it can make the figure inaccurately represent the underlying data.
Nature Portfolio's image-integrity guidance explicitly states that contrast should not be adjusted so that data disappear.
Do Not Use Cropping to Make Different Images Look Like One
Cropping can also facilitate undisclosed assembly. Researchers might crop pieces from separate images and place them together so seamlessly that the figure appears to represent one continuous field.
If images from different fields, times, samples, or experiments are juxtaposed, readers should not be misled about their provenance. The more specific issue of combining images from different experiments depends on whether the assembly accurately communicates where each component came from.
The Target Journal's Policy Matters
Image-processing standards are not completely uniform across journals and disciplines. A particular operation may be permitted with disclosure by one publication and restricted by another.
Researchers should therefore consult current instructions before finalizing figures, particularly for gels, blots, microscopy, medical images, and other image types for which publishers may have detailed integrity standards.
Journal permission, however, is not a license to mislead. The underlying requirement remains that the figure accurately represent the research evidence.
An Improper Adjustment Is Not Automatically a Formal Misconduct Finding
ORI emphasizes that detecting a discrepancy in an image does not itself establish falsification or research misconduct. Investigators need the original data and contextual evidence to determine what occurred.
Under the PHS framework, falsification requires manipulation, change, or omission that causes the research to be inaccurately represented, while a formal misconduct finding requires additional elements concerning accepted practices, culpability, and evidence.
An accidental over-adjustment, an inexperienced researcher's misunderstanding of a journal rule, and deliberate concealment of an unwanted signal may produce superficially similar figure problems while representing very different integrity situations.
Watch Out
Never set brightness, contrast, or crop boundaries by asking, “How can I make the effect look strongest?” Set them according to defensible visualization and scientific criteria. If changing the adjustment makes inconvenient evidence disappear, compare the figure with the original before proceeding.
