Imprimir Republish

Good practices

Publishers team up to identify altered images

Major publishers of scientific journals worldwide have created a working group to create new standards and promote technologies capable of detecting manipulated or duplicate images in papers. While the text of an article can be quickly scanned by anti-plagiarism software, there are not yet any tools considered efficient at searching for doctored images. There are companies that provide services of this nature, but most journals still use the human eye to identify alterations, which are common. A manual analysis conducted in 2016 found problems in 4% of 20,000 biomedical articles.

The main objective of the working group is to establish minimum performance requirements for tampered-image-detection software. Another goal is to create general guidelines for dealing with different types of alteration, which can result from misunderstandings, attempts to enhance certain colors or contrast, or fraud. The working group includes representatives from companies such as Elsevier, Wiley, Springer Nature, Embo Press, and Taylor & Francis, and was created by the International Association of Scientific, Technical, and Medical Publishers (STM), based in the UK.

Developing software capable of supporting the peer-review process is seen as essential to identifying a type of fraud that is very difficult for reviewers to detect, where the same image is reused in papers by different research groups, published in different journals. “It is like a kind of industrialized cheating,” Catriona Fennell, head of publishing services at Elsevier and a member of the STM working group, told Nature. A recent analysis by microbiologist Elizabeth Bik, a fraud expert, found more than 400 articles with images so similar that they suggested a common origin—the suspicion is that they were written by a company that sells papers on demand. Software alone is not enough to detect this type of fraud. Publishers also need a database of all images published in their scientific literature in order to detect possible reuse. This approach has long been used for text. Since 2010, publishers have been depositing all of the articles they publish on a centralized database called CrossCheck, so that they can screen submitted manuscripts for plagiarism. “We are going to need the same collaboration for images,” says Fennell.