Published online 9 October 2009 | Nature | doi:10.1038/news.2009.991

News

Science journals crack down on image manipulation

US figures show that incidents have jumped in the past two years.

Hands holding a DNA Sequencing SlideThere is a fine line between beautification of scientific images and fraud.Getty

More science journals are taking action to tackle the growing problem of falsified and manipulated images in papers submitted to them for publication.

At a meeting on plagiarism in London last week, Virginia Barbour, chief editor of PLoS Medicine, a peer-reviewed journal published by the Public Library of Science (PLoS), which is headquartered in San Francisco, California, said that the problem of image manipulation has "crept up" on journal editors since the advent of software such as Photoshop.

"Everything is submitted electronically, which makes manipulating images much easier to do," she told Nature.

At the conference, Barbour presented data from a one-year pilot study that began in 2008, in which all of the images in papers provisionally accepted for publication in the journals PLoS Biology and PLoS Medicine were checked for manipulation.

“Everything is submitted electronically, which makes manipulating images much easier to do.”

Virginia Barbour
PLoS Medicine

In a two-month period, PLoS Biology found problems with 5 images in 3 out of a total of 35 papers checked. Barbour says that they saw similar levels of problem pictures during the rest of the year. Over the whole year, PLoS Medicine also encountered 5 problem images in 3 papers, this time out of a total of 13 that they checked.

In one case, researchers had duplicated a panel in a western blot used to detect proteins. In another, scientists had spliced together two gel images into one, without indicating that this had been done.

Checks were made on the original data and images, which were requested from the authors. Discrepancies in the images can be detected by playing around with contrast controls in Photoshop and by looking at the background of the images.

Barbour says that authors gave "satisfactory explanations" for all of the manipulations, and that none of the problem papers was rejected. She adds that she was "reassured" that the study did not find any serious cases of misconduct, but that any attempt to clean up and change images without making clear this has been done is falsifying data. "There is a culture within universities that it is ok to fiddle with images and this has to be addressed," she says.

Barbour adds she was "shocked" at researchers' "appalling" systems for organizing and filing their data. The pilot study found that authors of 25% of the problem papers had difficulty finding their original data.

She says that PLoS Medicine will continue to check images for all provisionally accepted papers, and that the images of one in four papers, selected at random, will be checked for PLoS Biology. PLoS is now considering whether it should roll out checks to its other journals, which include PLoS Genetics and PLoS Pathogens.

The full picture

Others agree that image manipulation is a growing issue. In its most recent figures, published in September, the Office of Research Integrity of the US Department of Health and Human Services found that, in 2007–08, 68% of all of its opened cases of research misconduct involved falsified images (see graph below). The ORI, based in Rockville, Maryland, opens around 20 cases per year. The new figures represent a significant rise since 2005–06, when just over 40% of open cases involved image manipulation. In 1989–90, under 5% of cases opened by the ORI involved questionable images. The ORI, which releases quarterly newsletters on its activities, is currently preparing one totally devoted to image issues.

John Dahlberg, director of the ORI's Division of Investigative Oversight, told Nature that image manipulation "has escalated over the past two years" and that it is a "growing problem that needs to be dealt with".

ORI cases of falsified imagesClick for a larger version.ORI

Procedures for checking images, and standards for what is regarded as acceptable or not, vary between journals, he says. "There needs to be a unified approach."

For example, in early 2006, the Nature Publishing Group officially brought in a policy to check images in a minimum of two manuscripts provisionally accepted for publication per issue of Nature and its sister research journals. The journal Science began checking the images in every provisionally accepted paper in 2005.

Katrina Kelner, managing editor of Science 's research journals, says that they do not see many problem images.

Editors and authors are becoming more aware of the problem, says Linda Miller, executive editor of Nature. But sometimes, "people don't understand the line between beautification of an image and fraud". 

Commenting is now closed.