Image Integrity

From The Embassy of Good Science

Image Integrity

What is this about?

Digital image manipulation is very easy. You might be tempted to make an image more convincing, but simultaneously, no researcher with integrity wants to misrepresent their data. Image manipulation can be classified as scientific misconduct. It can be hard to find the ethical lines of what is and what is not allowed. Also, some images might look suspicious to you as a reviewer or journal editor. Luckily, comprehensive guidelines and tools exist.

Why is this important?

Images often serve as primary data (e.g. cell biology). In other instances, they are key in making an article attractive to read or serve comprehensive purposes. Accordingly, images are often included in article abstracts. The information they carry is thus a vital part of research and should remain identical to what is observed in the experiment.[1]


  1. Mike Rossner, Kenneth M. Yamada; What's in a picture? The temptation of image manipulation . J Cell Biol 5 July 2004; 166 (1): 11–15. doi: https://doi-org.vu-nl.idm.oclc.org/10.1083/jcb.200406019

For whom is this important?

What are the best practices?

The Catholic University of Leuven (KU Leuven) has a dedicated webpage on image integrity. They identified some of the most important sources and tools on the subject (available here, accessed on 24-04-2020). As their page is brief, a more elaborate description of what it contains, and additional sources, follows below.

Rossner & Yamada (2004)[1] wrote a prominent article arguing for a standard for image integrity. Working as Editors for The Journals of Cell Biology, they noticed the discrepancies between guidelines on image integrity journals gave to their authors (if any). To have a comprehensive overview, they developed their own guidelines for the Journal of Cell biology. They write that, for every aspect of the guideline, the main question is: “Is the image that results from this adjustment still an accurate representation of the original data?”[1] (p. 5). Whenever the answer is ‘no’, researchers should provide a detailed description of the adjustments, its purpose and the original image on request. If not, their actions might be regarded as misconduct.

A step-by-step translation of the guideline is available on the website of American Journal Experts (access here, accessed on 24-04-2020) and on the KU Leuven webpage. A similar guideline, and additional editorials on the subject, are given by the journal Nature on their editorial policies page (available here, accessed on 24-04-2020).

The Center for Ethics and Values in the Sciences, of the University of Alabama in Birmingham, created a website for both students and researchers with much material regarding image integrity (available here, accessed on 24-04-2020). They provide guidelines with more in depth explanations and illustration videos, but also educational material such as case studies, discussion hand outs and a quiz.

The Office of Research Integrity provides a tutorial on how to use ‘action sets’ in photoshop (available here, accessed on 24-04-2020). These actions sets allow you to document the changes you make to an image and ‘slide’ (i.e. going back and forward) between all the steps you made. The process of the image you manipulated will hereby be completely transparent if you provide the ‘action set’ combine with a copy of the original image.

For those reviewing papers, a free open source program, called InspectJ, is available on GitHub to identify cloning, stitching, patching and erased objects within an image. An advanced version also provides histogram equalization and gamma correction for improved image inspections (both available here, accessed on 24-04-2020)

  1. 1.0 1.1 Mike Rossner, Kenneth M. Yamada; What's in a picture? The temptation of image manipulation . J Cell Biol 5 July 2004; 166 (1): 11–15. doi: https://doi-org.vu-nl.idm.oclc.org/10.1083/jcb.200406019

Other information

Who
Virtues & Values
Cookies help us deliver our services. By using our services, you agree to our use of cookies.
5.1.6