AI software helps bust image fraud in academic papers

According to The Register, Taylor & Francis and the American Association for Cancer Research are using an artificial intelligence program called Proofig to detect fraud in academic paper submissions. According to the company's website, Proofig uses artificial intelligence, computer vision and image processing to review image integrity in scientific publications.

According to The Register, AACR used Proofig to screen over 1,300 papers for publication. There were over 200 papers that required author contact to clear up issues.

Many journals need help detecting image duplication fraud in Western blots, which are a specific style ofprotein detection imagery consisting of line segments of different width. There are subtle differences in a Western blotting's appearance that can translate to dramatically different conclusions about test results, and many cases of academic fraud have seen unscrupulous researchers duplicate, crop, stretch, and rotation Western blottings to make it appear like they have more data than they really do. It can be difficult for humans to detect duplicate images, which is why some firms are trying to automate the process.

According to The Register, Proofig's and Image Twin's solutions have limitations. Human expertise is needed to interpret detection results. Proofig is expensive due to its intensive process of analyzing 120 images for an individual. The journals can't analyze every paper at the submission stage because of the high cost and requirement for manual oversight. They have reserved it for later in the process.

A publication's reputation can be damaged by academic fraud. Recent revelations about image fraud in Alzheimer's research and a large volume of academic papers make the field ripe for computer vision tools that can assist humans with fraud detection. The effectiveness and how widely they are adopted is still being developed.