Detection and communication of possible inappropriate manipulation and fraudulent manipulation of image data is a sensitive issue. Therefore, due diligence of all involved is important. First, it is of the utmost importance that authors of a manuscript understand what image data manipulations are considered acceptable and do not engage in unacceptable or fraudulent image data manipulations. In this respect, it is recommended that all authors of a manuscript review images intended to support their manuscript against the original image data prior to submission of their manuscript for peer-review. In addition, it is recommended that authors report how image data were manipulated, even if the image manipulations are considered acceptable practice, or state that image data were not manipulated. Second, peer-reviewers need the appropriate experience to critically and constructively assess the quality and originality of image data associated with a given manuscript they peer-review. Third, journal editors hold responsibility to independently evaluate image data based on their own assessment and that provided by the peer-reviewers and should question authors and request additional information considerately as needed. This due diligence of all involved is important not only to publish accurate science, but also to avoid wrongful accusations (and associated consequences for the authors).
The Rockefeller University Press has defined two types of digital image–related misconduct: inappropriate manipulation and fraudulent manipulation. Inappropriate manipulation refers to adjustment of image data that violates the established guidelines but does not affect the interpretation of the data. Examples include adjustments of brightness/contrast to a gel image that completely eliminate the background (so the reader cannot tell how much of a gel is shown) or that obscure background smears or faint background bands. Another example is the splicing of images from different microscope fields into a single image that appears to be a single field. Fraudulent manipulation refers to adjustment of image data that does affect the interpretation of the data. Examples include deleting a band from a gel to “fix” a negative control that did not work or adding a band to a gel to indicate the presence of product that was actually not there.
The ease of image manipulation in powerful applications like Adobe Photoshop® may tempt authors to adjust or modify digital image files. Authors have been using these applications for more than 10 years. Many of the manipulations that are detected constitute inappropriate changes to the original data and may indicate that scientific misconduct has occurred. In more egregious cases, such manipulations may constitute fraud. For the purposes of this section of the document, fraud is defined as falsification or fabrication of image data; it is not meant to encompass the legal criteria of intent or harm to a third party who relied on the data.
Editors have a responsibility to set guidelines for authors on the proper handling of image data. Clear guidelines are important, because some level of image manipulation is accepted practice (e.g., image cropping or limited adjustment of brightness and contrast); authors must understand the boundary between acceptable and unacceptable manipulation.
Guidelines developed by The Rockefeller University Press have been published elsewhere.1 Examples of different types of manipulation and image manipulation cases are available.1, 2 Examples of guidelines from other publishers can be found here:
Journal of Cell Biology3
Proceedings of the National Academy of Sciences5
The Rockefeller University Press has established 4 basic guidelines:
These comprehensive guidelines were developed in 2002 by The Rockefeller University Press and are now used by many journals. An expanded set of ethical guidelines for the appropriate handling of scientific digital images is available on the website of the Office of Research Integrity7 and discussed in detail by Cromey (2010).8
After guidelines are established, editors have a responsibility to enforce them. To do so requires the establishment of definitions of misconduct, procedures for identifying misconduct, and policies for handling misconduct.9 Sample correspondence related to this topic is available on the CSE website.10
Examining image files. A simple “forensic” analysis of the images in a figure file can be accomplished by using the basic “Brightness/Contrast” slide bars in Photoshop to reveal inconsistencies in the pattern of background pixilation that are clues to manipulation or inappropriate adjustments to brightness and contrast. For color images, more sophisticated adjustments to contrast using the “levels” slides may be necessary to reveal inconsistencies. Informative examples are provided by Rossner and Yamada (2004; Figure 6 in their article)1 and by Cromey (2010; Figure 1 in his article).8
Obtaining original data. Authors’ reputations for impeccable research integrity among their scientific peers are vital for success in their careers. Authors will thus be concerned when the integrity of image data in a manuscript being peer reviewed or accepted for publication is questioned. It is important for an editor to reassure authors at this initial stage of investigation that only the presentation of the data is being questioned and not its scientific quality, which has been vetted by peer reviewers and academic editors. The letter requesting original data can even point out that often the inconsistencies revealed by image “forensics” are simply caused by the transfer of images from one computer application to another (e.g., from Microsoft Office PowerPoint® to Adobe Photoshop®) and that it is possible that no manual adjustments have been made by the authors. In addition, an editor could point out that it is in the authors’ interest to resolve the inconsistencies before the images are published online, because they may be questioned by a reader. Authors should also be assured that the inquiries at this stage are strictly confidential.
Handling misconduct. If a clear case of inappropriate manipulation is detected, the author should be requested to submit the figure in question with an accurate representation of the original image data. This approach applies only to adjustments for which there are clear solutions to remedy the problems; for example, lines need to be added to a gel image to indicate that lanes have been spliced out. In such cases, it is not necessary to request the original image data from the author. However, if there is any possibility that the manipulation may be fraudulent, the journal editor should be alerted, and the original image data should be obtained from the authors for comparison to the prepared figure. Although the Office of Research Integrity (ORI) guidelines for editors indicate that cases of “suspected” misconduct should be reported either to the ORI or to an author’s institution,11 journal editors should attempt to resolve the problem before a case is reported. This is because the vast majority of cases do not turn out to be fraudulent.2 One case of unfounded allegations of fraudulent image data manipulation has recently been reported in detail.12
If a comparison of the original data with the prepared figure indicates that images have been inappropriately but not fraudulently manipulated, the author should simply be asked to remake the figures with a more accurate representation of the original data.
If the comparison reveals that fraudulent manipulation has occurred, the first step is to revoke acceptance of the paper. At the Journal of Cell Biology, the conclusion that fraudulent manipulation has occurred must be agreed on by 4 people before such action is taken: the managing editor (a PhD scientist), the academic monitoring editor, the academic senior editor, and the academic editor-in-chief. Other journals are encouraged to adopt similar procedures.
A policy for reporting misconduct should be developed by each journal (refer to Sections 3.1, 3.2, and 3.3). Misconduct can be reported either to an author’s institution or to the ORI.13 The Journal of Cell Biology does not report digital image–related misconduct if the principal investigator takes responsibility for the action and indicates that measures have been taken to avoid image manipulation in the future.
Many institutions that receive Public Health Service (PHS) funding have an ombudsman for allegations of misconduct in science, whom a journal can contact if it decides to report misconduct to an author’s institution. Absent an ombudsman, every institution that receives PHS funding has an individual who has signed the PHS “Letter of Assurance,” which indicates that the institution will abide by the PHS code of conduct.
(Authorship: Michael Rossner took the lead in writing this section of the white paper on behalf of the CSE Editorial Policy Committee. Daniel Salsbury revised this section for the 2009 Update. Wim D’Haeze and Michael Roy revised this section for the 2012 Update. Members of the Editorial Policy Committee and the CSE Board of Directors reviewed and commented on it. This section was formally approved by the CSE Board of Directors on March 30, 2012.)