Simon Rallison
Director of Publications, The Physiological Society

https://doi.org/10.36866/pn.95.9

A 1937 photograph of Stalin shows him, half-smiling in greatcoat and military cap, on the banks of the Moscow-Volga Canal. To his left stands the slight figure of Nikolai Yezhov, head of the NKVD. In later copies of the photo, though, Stalin’s companion has vanished. After Yezhov was secretly executed in February 1940 (in an execution cell of his own design), all traces of him were removed from the official record by censors. Nikolai Yezhov was airbrushed out of Soviet history.

While the deliberate alteration of photographs is almost as old as photography itself, the potential for less detectable manipulation was hugely increased in the late 1980s and early 1990s by the arrival of digital cameras followed by software programs like Adobe’s Photoshop and Illustrator. By the early 2000s, the editors of scientific journals knew that they could have a problem on their hands.

Good science requires reliable data. Fraudulently changing or omitting data or results so that the outcome of research is not accurately represented in the record is falsification or fabrication, serious forms of research misconduct and breaches of publishing ethics. At the less serious end of the spectrum are manipulations of image data that violate recognized guidelines without affecting the interpretation of the data.

The potential problems that digital image manipulation raised for journals, particularly those in cell and molecular biology, were first drawn to the attention of the wider community by Mike Rossner, the then Editor-in-Chief of Journal of Cell Biology. Rossner and Yamada’s manifesto, set out in a 2004 editorial ‘What’s in a picture? The temptation of image manipulation’ (J Cell Biol 166, 11–15), is recommended reading for authors, including those of The Society’s own journals (it’s cited in the Author Information).

The general rule in publishing images is that the final image must remain consistent with the original data, accurately representing the information as captured, and must conform to community standards. While manipulation of an image is acceptable in some circumstances, for instance the adjusting of contrast or brightness, any adjustment must be applied equally to the entire image and must be described in the paper.

Unacceptable manipulations generally include adding to, altering, moving or removing a specific feature of an image. The instructions for The Society’s journals exclude using software to edit images such as micrographs or photographs of gel arrays, as these are primary data. However, it is acceptable to size and crop the images.

Although there are also technologies for detecting image manipulation, in many cases there are tell-tale signs that can be picked up by eye, for instance by spotting irregularities in the background signal or in the outer margins of images. Or the background signal is not uniform, with noticeable variations in pixilation, colour or texture. Or sometimes it is possible to see duplication of background features like dust particles.

Gels and blots are subject to a recognized set of rules. For instance if gels, blots or fields are rearranged, then dividing lines should indicate this. Pasting-in or rearranging individual bands is simply a non-starter.

While there are no figures for the prevalence of undisclosed image manipulation in physiology, it is estimated that in cell biology 25% of papers accepted for publication have one or more figures that have been ‘inappropriately modified’. Journal of Cell Biology, Nature and some other journals check all figures for manipulation on acceptance for publication.

Despite the efforts of the journals, not all authors are aware of the prohibitions on image manipulation. They may naively be hoping to make their point more clearly or even just trying to make their photographs more pleasing to look at. Because of this, journals try not to be heavy-handed in dealing with suspected cases of image manipulation. Authors are generally given the opportunity to correct an unsatisfactory figure, either by replacing it with another example from their dataset or where fitting by explaining in the text the manipulation it has undergone. Transparency is always at a premium in issues of publishing ethics. The author should be prepared to provide the journal with the original, unedited data and images, plus details of what software was used in acquiring and processing the image and what was done with it. These measures almost always resolve the issue.

Science has been fortunate in being allowed to police itself to a very large degree. Maintaining high standards of integrity is a small price to pay for this. Institutions, societies, journals and editors can play their part but it is up to individual scientists to keep themselves informed of ethical issues and to question their own actions.

To quote Rossner and Yamada:

‘Just because the tools exist to clean up sloppy work digitally, that is no excuse to do sloppy work… If you would have redone an experiment to generate a presentation-quality image in the days before the digital age, you should probably redo it now.’