According to Retraction Watch, over 250 papers regarding COVID-19 published since the start of the pandemic have since been retracted. While retractions are an acknowledged necessity for academic journals, the growing justifications demand attention. Much focus is concentrated on checking the text of manuscripts, but in the digital age, images are often more important when conveying results to readers. Here, Dr. Dror Kolodkin-Gal, founder of automated image integrity checking software Proofiger Ltd., discusses the importance of image integrity and common sources of issues.
According to the STM Global Brief 2021 – Economics and Market Size report, the global science, technology and medical publishing market was worth 26.5 billion USD in 2020. Scientific research papers are seen as the gold standard for increasing our understanding of the world. Given the high regard in which academia is held, along with the growing size of the publishing market, the importance of maintaining the integrity of the industry is unquestionable. It is vital for the industry to ensure that findings and methodology are kept to rigorous standards to prevent reader misinterpretation and maintain credibility.
Retractions and amendments have always been a part of academic publishing, especially considering the recognised notion that science is self-correcting. However, according to the previously mentioned report, roughly 48,000 scholarly peer-reviewed journals are published yearly. Furthermore, the recent surge in misinformation, particularly involving COVID-19 and vaccines, reaching public opinion as well as steady retraction rates, has highlighted an important issue – that the review process must be modernised.
Common Sources of Image Integrity Issues
Scientific image integrity covers a broad scope, with different journals holding unique guidelines for ensuring integrity. Recent developments, such as the joint guidelines issued by representatives from eight publishers in late 2021, offered detailed information for specific scenarios. The three categories listed covered scenarios where some images have been altered or beautified whilst avoiding affecting the research’s conclusions, to level three which involved “severe manipulation”. Each category contains examples of issues, as well as actions for journal editors to take. According to the International Journal of Cancer, a Wiley published journal, the most frequently found errors are duplicated panels deriving from copy and paste errors, magnification errors in microscopic images, and inappropriate splicing of gel sections together. The prevalence of duplication errors in life sciences manuscripts can be attributed to a variety of factors. Notably, many authors from different locations can now contribute to the same study and work on the same manuscript, opening the door for duplication errors through human error and miscommunication. Accidental image duplications can be missed by researchers, editors and peer reviewers because identifying overlapping sections of complex images is difficult for the untrained human eye. Taking many microscopy images – with or without using different levels of magnification – can lead to small overlaps within the same samples. Technically, it is difficult to completely avoid overlaps, and it’s even harder to detect these issues. The acknowledgement of purposeful manipulations as part of fraudulent research is also unavoidable. This is an increasingly worrying issue in the academic industry, with most publishers being the target of systemic academic fraud via paper mills. This is the illegal and unethical manufacture and sale of articles, some of which are completely fabricated, by companies to researchers who claim the articles are the fruit of their own research. Paper mills increase the pressure on the scientific community. When such papers make it through the peer-review process, it undermines the credibility of real research. Publishers and researchers need the tools to help make sure credible researchers are recognised for their contribution to the field. Paper mill articles are difficult to detect in isolation, but comparisons of multiple papers from different sources make it much easier. The review process must adapt to cover unusual similarities or the use of stock images across several papers. As already mentioned, the joint efforts of publishers are a start, but a wider collaboration between publishers and investigators is necessary to ensure awareness and effective action across the whole academic sector.