Most universities already provide text-similarity checking through tools like iThenticate, but image integrity, covering figure duplication, manipulation, and AI-generated visuals, is typically absent from institutional QA workflows. Proofig AI fills that gap. It can be added alongside existing tools as a standalone solution or integrated into a coordinated quality assurance hub through PubShield.
Why Image Integrity Is the Missing Layer in University QA
University research offices have invested heavily in text-based quality assurance. Tools like iThenticate screen for text similarity, Grammarly or similar platforms catch language issues, and reference managers help organize citations. These tools serve their purposes well. But none of them examine the scientific figures that form the evidentiary backbone of most research manuscripts.
This gap matters. According to image integrity analyst Jana Christopher in an interview with UKRIO, roughly 20 to 35% of submitted life-science manuscripts are flagged for image-related issues during editorial screening. These issues range from accidental duplications introduced during figure preparation to more serious concerns about manipulation or reuse.
The consequences extend beyond individual papers. A peer-reviewed analysis of top-20 world-class universities documented retraction patterns across 2010 to 2019, with image manipulation among the leading causes of retraction. And Nature has reported that approximately 0.2% of articles published in 2022 are now retracted, roughly triple the rate from a decade earlier. For institutions, each retraction carries significant financial and reputational consequences, including grant losses, legal costs, and damage to future funding prospects.
The point is not that text-based tools are inadequate for their own purpose. They are essential. The point is that image integrity represents a distinct quality assurance layer that most institutional workflows do not yet address.
What Does Proofig AI Detect?
Proofig AI’s image integrity capabilities are designed specifically for scientific figures. The platform covers five core detection areas:
| Detection Capability | What It Screens For |
|---|---|
| Image Manipulation Detection | Cloning, editing, deletion, splicing, and content-aware edits within a single sub-image |
| Duplication & Reuse Analysis | Scaling, rotation, flipping, full and partial overlap within a manuscript |
| AI-Generated Image Identification | Synthetic images across microscopy, histology, Western blots, gels, medical scans, and more |
| Image Plagiarism (PubMed Source) | Reuse of sub-images from published manuscripts, checked against tens of millions of images |
| My Database Self-Plagiarism Control | Comparison with a personalized repository of prior research to prevent inadvertent self-reuse |
The platform covers all major scientific image types, including confocal, light, and electron microscopy, Western blots and gels, FACS, histology, cell plates, animal imaging, and medical scans.
Critically, Proofig AI surfaces suspected issues for human review. It does not determine guilt, confirm misconduct, or make final judgments. This distinction matters for institutional adoption, where flagged findings must be handled through formal research integrity processes, not automated verdicts.
For institutional review staff concerned about alert fatigue, the tool’s false-positive rates are notably low: 0.0093% for microscopy (tested on 200,000 real images) and 0.0020% for Western blots (tested on 250,000 real images). These rates mean the tool will not overwhelm integrity offices with false alarms.
How Proofig AI Fits into an Existing Institutional QA Stack
Proofig AI is designed to complement, not replace, existing tools. A university that already uses iThenticate for text similarity can add Proofig AI to cover the image layer without disrupting current workflows.
Here is how a typical institutional QA stack looks before and after adding Proofig AI:
| Integrity Layer | Before Proofig AI | After Adding Proofig AI |
|---|---|---|
| Text similarity & plagiarism | iThenticate / Turnitin | iThenticate / Turnitin (unchanged) |
| Grammar & language | Grammarly or equivalent | Grammarly or equivalent (unchanged) |
| Image duplication & manipulation | Not covered | Proofig AI |
| AI-generated image detection | Not covered | Proofig AI |
| Image plagiarism (published literature) | Not covered | Proofig AI |
| Reference integrity | Manual or not checked | Available via PubShield (RefGuard) |
| AI-generated text detection | Varies by institution | Available via PubShield (Pangram) |
| Data reporting & compliance | Manual or not checked | Available via PubShield (DataSeer) |
Two Integration Pathways
Standalone Proofig AI. Researchers or administrators upload manuscripts as PDFs or image files. Proofig scans all figures automatically, flagged items are reviewed using built-in forensic tools, and a report is exported for internal records or submission. This pathway sits alongside any existing text-checking workflow with no changes required to current processes. Learn how Proofig AI screens scientific images.
PubShield as a coordinated hub. For institutions that want a single submission point for all integrity checks, PubShield brings image integrity (Proofig AI), text similarity (iThenticate/Turnitin), AI-generated text detection (Pangram), reference integrity (RefGuard), and data compliance readiness (DataSeer) into one dashboard. This eliminates scattered logins and fragmented workflows. For institutions using electronic lab notebooks, the LabArchives partnership supports bringing integrity practices closer to the point of research creation. The Aries Systems integration enables workflow connectivity for publishers.
It is important to note that iThenticate, Pangram, DataSeer, and RefGuard are PubShield ecosystem partners. Their capabilities exist within PubShield, not as standalone Proofig AI features.
Why Pre-Submission Screening Protects Researchers and Institutions
A common concern when institutions consider image screening is whether it signals distrust of researchers. The opposite is true. Pre-submission screening is a quality assurance service that protects researchers, their labs, and the institution.
Many image integrity issues are unintentional. They are introduced during figure preparation, by collaborators, students, or through shared data-handling workflows. Even researchers who uphold the highest ethical standards may unknowingly submit problematic images because of errors introduced by team members handling data at various stages. A peer-reviewed study at the University of Split found that researchers detected only 32.4% of image duplications when reviewing figures manually, demonstrating the limits of unaided human review.
Pre-submission screening shifts the discovery of potential issues from post-publication, where consequences are severe, to pre-submission, where issues can be corrected quietly and constructively. Nature has reported on the personal and career stress researchers face when retracting papers, often due to honest mistakes that pre-submission screening could have caught.
For institutions, the benefits extend further. Campus-wide adoption supports responsible conduct of research (RCR) programs, provides audit-ready reporting for research integrity offices, and demonstrates proactive commitment to research quality. SSO and secure licensing options, along with central dashboards and analytics, make institutional deployment practical.
Third-Party Validation
Proofig AI’s adoption by leading publishers provides independent validation of its capabilities. The Science family of journals adopted Proofig AI across all six journals in 2024, after a multi-month pilot demonstrated effectiveness at detecting problematic figures pre-publication. MDPI signed a multi-year deal, with Milos Cuculovic, Head of Technology Innovation at MDPI, noting: “From the experts, it seemed Proofig AI was by far the best choice, and here we are: further ensuring research integrity and guaranteeing the high quality of research output published by MDPI.”
Practical Next Steps for Research Integrity Officers
For administrators evaluating image integrity QA for their institution, here is a concrete path forward:
- Audit your current QA stack. Identify which integrity layers (text, image, reference, AI-generated content) are covered and which are not.
- Assess the gap. If image integrity is not addressed, your institution faces the same risks that lead 20 to 35% of submitted manuscripts to be flagged at the editorial stage.
- Evaluate standalone vs. hub integration. Determine whether adding Proofig AI as a standalone image-integrity layer meets your needs, or whether a coordinated approach through PubShield better fits your institutional workflow.
- Pilot with a department or graduate program. Start with a high-volume research group to assess workflow fit and review the types of issues surfaced.
- Integrate into RCR training. Use screening results (anonymized) to educate researchers about common figure-preparation errors and best practices.
Adding image integrity QA now addresses risk at the point where it is most preventable, before submission, not after publication. The tools exist. The integration pathways are practical. The question for institutions is no longer whether to screen for image integrity, but how quickly they can close the gap.
