Stop the Picture Doctors

What’s behind paper retractions? (3)
by Adam Marcus and Ivan Oransky, Labtimes 04/2011

Photo: zettberlin / Source: PHOTOCASE

Banding together to fight image manipulation.

A picture is worth ... well, you know the rest. But a doctored image could result in the retraction of an entire journal article.

When it comes to scientific fraud, manipulated graphics have become one of the leading forms of fabrication. Much of what we cover on Retraction Watch involves plagiarism in its dictionary definition of the word “thievery”. But we have also noticed that a great many cases involve cooked-up images, from Western blots to cell cultures.

There have been recent cases in Cell, Circulation Research, Arteriosclerosis, Thrombosis and Vascular Biology, just to name a few. In another, a virologist was slapped with a ten-year publishing ban by the American Society for Microbiology after he was forced to retract 16 papers for image manipulation.

We’re not alone in noticing this pattern. According to the US Office of Research Integrity (ORI), which investigates scientists who commit fraud on the government’s dime, nearly 70 percent of its inquiries now involve manipulated graphics. That’s up from the low single-digits 20 years ago.

One reason for the surge in such tampering surely is the relative ease of meddling that today’s computer graphics software provides would-be cheaters. A few mouse clicks or key strokes and presto! Here’s a slide showing in situ hybridization that never occurred. Need to prettify a messy histopathology slide? Digital airbrush here, drop a few pixels there and voilà!

Fortunately, ORI sleuths have pretty impressive forensic skills be­cause they manage to find much of this stuff when they look for it.

The problem, however, is that if you don’t look for fraud, you aren’t likely to find it. And, with few exceptions, journals do not appear to be keeping an eye out for image manipulation.

Consider: Retraction Watch recently reported on the case of two researchers in China who plagiarised shamelessly – we’re willing to concede that not all plagiarism is so flagrant – from an earlier article, going so far as to steal several of the original author’s figures.

The theft of the images from a 1996 paper in Fuzzy Sets and Systems wasn’t fuzzy at all when you looked at the plagiarists’ 2009 work in Transportation Research Part E: Logistics and Transportation Review. That raised the question: What were the reviewers and editors doing when they screened the second manuscript? After all, even the abstracts were virtually identical, so a Medline search would have raised an immediate red flag.

Ah, you might say, but the papers appeared in different journals. So they did. But the two titles are owned by the same publishing house, Elsevier.

Now, Elsevier does, in fact, participate in the CrossRef group of publishers and journals that screen manuscripts for word theft (why that didn’t pick up the case in question, we don’t know). But no such system exists for reviewing images.

We were surprised to learn that major publishers like Elsevier don’t at least have an internal system for automatically screening submitted images for potential misuse.

It’s not impossible. Mike Rossner, executive editor of the Rockefeller University Press (RUP), has come up with what the Scholarly Publishing and Academic Resources Coalition has called a “leading-edge program to screen digital images in manuscripts accepted for publication at RUP for possible manipulation”. Rossner has written widely on that programme, which he is happy for others to use.

The reason for publishers doing in-house image comparisons seems to be the usual: time and cost. Wouldn’t avoiding retractions more than pay for itself in good will and confidence – from readers, authors, editors and the entire science community?

We should appeal to authors, too. As the editor of Autophagy put it in an editorial last year: “What were you thinking? Do not manipulate those data.”

(The authors run the blog Retraction Watch:

Last Changed: 03.05.2012