What is VerifEye, Research Square’s new tool for processing scientific images?
As I’ve previously discussed in more detail, image manipulation is increasingly common among figures in scientific manuscripts. Some recent studies, including our own, show that up to 25% of submitted manuscripts contain one or more figures with evidence of improper manipulation. These are not necessarily cases of fraud; in many cases, authors make mistakes or simply do not understand what is acceptable when creating figures. Either way, though, these manipulated figures can create distrust, muddy the scientific record, and lead to corrections or retractions.
How can the publishing industry respond?
Although automated efforts to screen images for manipulation are underway, these projects are in their infancy. The only way to catch improper image manipulation (or as much of it as you can catch) is to look. Currently, this occurs after publication for the vast majority of journals, when an intrepid researcher or journalist begins to scrutinize the figures from a new paper. Some journals, such as the Journal of Cell Biology, do take the time to screen incoming manuscripts, but this process is time-intensive and requires access to special image software.
Enter VerifEye — a simpler way to inspect scientific figures
For several years, Research Square has worked with major publisher partners to evaluate the integrity of submitted manuscripts, and we have detected a sizable portion of images manipulated in ways that can misrepresent the original research results or give the appearance of impropriety. While looking for ways to improve our image checks, we began developing an online software tool that makes the image screening process easier, and we are sharing that tool publicly now as VerifEye.
While an attentive human eye is still required, VerifEye has simplified image integrity checking for our team, and we believe it can do the same for busy editorial offices across the world.
The process
VerifEye exists completely online — no software to install and update. After contacting us to receive an account, the basic steps are as follows:
1. Upload the image files for your manuscript and name your project
2. Crop images, if necessary, to focus on the panels or portions you want to inspect
3. After processing is complete, select the image you want to evaluate from the thumbnails at the top of the screen
4. Scroll through the collection of filtered images, each based on a different algorithm designed to highlight latent features in the image, such as straight lines and sharp contrast changes
5. Make adjustments to brightness and contrast for any individual filtered image and securely save anything you’d like to look at later
The benefits
VerifEye can’t identify inappropriate manipulation by itself (yet…), but it can simplify the steps it takes to inspect a scientific figure more closely. No more expensive Photoshop licenses that require you to open each figure file separately and make your own adjustments. Instead, let VerifEye handle those steps rapidly so you can spend your valuable time on assessing images.
With VerifEye’s simpler process, we hope that more journals can start screening their figures, better safeguarding the integrity of the published scholarly record and their own reputations.
How to get started
VerifEye is currently available for free in an early testing phase, with limited accounts available. Please visit our homepage and let us know if you are interested in trying it out (or simply staying up to date on VerifEye’s progress). We hope to take feedback from our early trial users to improve the algorithms we use and explore automated image flagging, rolling those features out into a wider launch later in 2018.
Website: www.researchsquare.com/verifeye
Email: verifeye@researchsquare.com
