By Jason Riley, PhD
This October, the NICHD began its "Research Ethics Series," organized by the stellar Brenda Hanning, with a workshop titled "What's in a picture?" The inaugural session was led by the National Institutes of Health's very own star Dr. Kenneth Yamada of NIDCR. For his credentials—which make some impressive reading—see his CV at: http://www.nidcr.nih.gov/Research/NIDCRLaboratories/CellDevelopmental/KennethYamada.htm
During the workshop, Dr. Yamada discussed a similar subject to his article in The Catalyst about image manipulation in scientific publications (http://www.nih.gov/catalyst/2004/04.05.01/index.html), a subject close to his heart as a journal editor. The selection of Dr. Yamada as someone to guide fellows through ethical issues was solidified not only by an incredible publication history as a scientist, but also by his presence on many boards and participation in other such services at the NIH, such as involvement in director-appointing panels. His impressive service record gives you a feel for his well-known integrity.
Dr. Yamada opened the session with a few alarming numbers. In one journal's investigation, forensic imaging was used to determine if images had been "fixed." While the journal found that many images were indeed enhanced in a way that did not change the interpretation of the data, 1 percent of the images had been modified to the point of deception. Those manuscripts therefore had to be rejected. If we consider that not all journals complete such a thorough investigation, we can apply the 1 percent image manipulation rate to the eighteen million articles in Pubmed, giving us a quarter of a million deceptive articles. This assumes that forensic imaging can catch all image modifications.
What are our obligations?
What should we as scientists do? Dr. Yamada continued the workshop with some quick tips: First, we shouldn't cheat by massaging our images. As someone who works in the field of image reconstruction, I can say that I am often skeptical of images that look "too good'" in a paper. Second, when serving as a co-author, we have a responsibility (to the best of our ability) to verify that data isn't deceptive in a manuscript. There will be times when we have to rely on trust, but if the data is in the form of an image, we can ask to see the original. Third, as a reviewer, we can flag suspicious images as something the journal should check with a forensic imager or ask for the original data from the author to validate the image before accepting the article.
The Bottom Line Dos and Don'ts:
DO keep all of your original data. DON'T modify an image for publication without recording what you did, and DO explicitly state in your manuscript any image manipulations/enhancements you performed, no matter how trivial. If you state what you did and why you enhanced an image, you are not misleading anyone.