While the scientific research community is aware that there can be issues with image manipulation and duplication, few researchers understand how frequently these issues can occur. If researchers have published multiple papers with multiple images, it is likely that some of the papers contain image issues. Publishing research later found to have image issues can be hugely detrimental to a researcher’s career. As such, it is important that scientists know how issues can occur, the potential consequences of submitting manipulated or duplicated images, and how artificial intelligence (AI) can help researchers find issues pre-emptively.

Where images go wrong

There are many forms of image integrity issues. Image duplication refers to reusing the same image in different parts of the paper without stating it. The image may be used the same way twice, or it may have been altered by changing its orientation, size, or scale. The image may have also been flipped or cropped during duplication. In other cases, researchers may overlap two parts of the same image.

The integrity of the image source may also come into question. For example, as well as duplicating the image in the same article, researchers might, intentionally or unintentionally, self-plagiarise and use an image from an older piece of research. This can happen when a large paper is split into two smaller publications. 

Usually, it is not a researcher’s intention to include images with these issues, simply making an innocent mistake. Researchers may conduct research over many years, collecting hundreds to thousands of images of specimens. Keeping so many images organized is difficult, which can lead to mistakes. Scientific work is also often collaborative, with scientists from different organizations working on the same project. Without effective communication between collaborators, it can be easy to inadvertently use duplicate images throughout a research paper. 

Consequences

Image integrity issues can go unnoticed because it is difficult to manually review all the images in an article. The consequences of not detecting these issues, however, can be severely detrimental to a researcher’s reputation. When conducting new research, for example, many researchers will write a grant request to gain funding for the project. If the grant authority finds an image manipulation that the researcher failed to detect before submission, it is likely that the request will be denied and the researcher will struggle to access funding elsewhere, halting the research. 

Similarly, if a researcher does gain funding and conduct research, but submits for publication without detecting mistakes, they risk rejection. Publishers do not have to disclose the reason for rejection, so researchers will know that it’ll become more difficult to publish in the future but won’t know how to improve the likelihood of publication.

When reviewing research articles, publishers will also check images manually. The lack of accuracy means that some papers might be published with image manipulations and duplications, which may cause costly issues in the future in two ways. First, other researchers may want to base their research on an existing paper. If the original paper contains inaccurate data, any data in new research will also be incorrect, wasting funding. Similarly, any researchers basing their experimental procedures on an existing paper that contains errors will struggle to replicate the results in the original and not know why, leading to more wasted time, materials, and funding.

Alternatively, if someone detects an issue post-publication and reports it in web blogs, the publisher must open an investigation to determine if the allegation is true and, if so, how the issue occurred. Investigations can take up to two years, putting pressure on the researcher and significantly reducing their chances of winning funding, conducting research, or publishing elsewhere, meaning that no matter the outcome of the investigation, researchers must work hard to rebuild their reputation. 

Solving the problem

Researchers should not leave image integrity to chance. Checking images manually is time-consuming and inaccurate, and missing any issues before publication can result in costly investigations and reputational damage. Additionally, submitting an article to be published by a renowned scientific journal can cost thousands of dollars, so instead of leaving image integrity to chance, researchers can now access technologies that enable them to proactively check images. Some software uses AI to automatically scan every image in a research paper and detect the images that appear to be duplicated or manipulated, checking an entire paper in one to two minutes. Each image can be checked against itself and against the rest of the images to detect any anomalies that researchers should amend before publication. 

With data issues and duplication being some of the main reasons for retraction, integrating image proofing technology allows researchers to check images as quickly and efficiently as they currently check for grammar and plagiarism before submitting a paper. Detecting and resolving any image manipulations and duplications before sending to publishers reduces the risk of papers printing mistakes, ensuring the integrity of research, reducing the likelihood of costly retractions and building the reputation of the researcher.