Division of Research

Managing and Verifying Scientific Image Integrity

Breadcrumb

Managing and Verifying Scientific Image Integrity

In life sciences and technical research, complex visual data - including microscopy, Western blot gels, flow cytometry, and photographic panels - requires careful handling. Unintentional errors can compromise data integrity and reproducibility, and also cause delays or retractions during journal peer review.

Additionally, image errors in published literature are increasingly being identified and flagged by independent image integrity sleuths. When these discrepancies are identified, they are often directed to institutional research integrity departments to be formally assessed according to internal policies, including the Research Misconduct policies. 

Automated image integrity tools use Artificial Intelligence (AI) to help researchers pre-screen visual figures. These tools identify potential image irregularities before manuscript submission or grant reporting. 

Important Notice

There is currently no institutional-wide license for standalone image integrity software. However, Brown University is currently evaluating the usage and experience of AI tools for our research community in order to determine whether to acquire a license. 

Please note that Brown University does not have access to the images you upload. Image Integrity tools are a researcher support tool, and are not used as a monitoring or compliance system. Their purpose is to help researchers proactively identify and address potential image concerns before submission or publication.

What Do Automated Image Integrity Tools Detect?

Automated screening tools extract individual sub-images and panels within complex figures to flag potential issues, including:

  • Internal Duplication & Reuse: Unintentional duplication of panels, lanes, or regions within a single manuscript or notebook entry.
  • Transformations: Rotated, flipped, stretched, or contrast-adjusted instances of identical visual data.
  • Manipulation & Editing: Signs of splicing, local cloning/copy-pasting, or background filling.
  • Cross-Publication Plagiarism: Comparisons of image features against large indexes of millions of open-access published figures (e.g., PubMed).
  • AI Synthetic Content: Emerging markers indicative of artificially generated or manipulated imagery.

Overview of Popular Tools for Researchers

1. Proofig AI

  • Focus: Pre-submission manuscript screening and electronic lab notebook validation.
  • Key Capabilities: Performs automated feature extraction on complex figures (microscopy, blots, graphs) to cross-reference sub-images against each other and external literature. It provides side-by-side forensic comparisons and confidence scores for potential duplications or edits.

Note: A complimentary trial access option for individual LabArchives notebook users is currently available through the end of 2026. Follow the step-by-step instructions provided in the official setup guide: LabArchives & Proofig AI Setup Instructions

2. ImageTwin

  • Focus: AI-powered image analysis for researchers, editors, and peer-reviewers.
    Imagetwin
  • Key Capabilities: Designed to flag full and partial duplications across an extensive database of published literature.Offers built-in forensic viewing tools (e.g., brightness/contrast adjustment, keypoint matching overlays) to help authors visually verify flagged regions.

Best Practices for Researchers & Lab Leaders

Automated tools serve as an initial visual "first pass," but they do not replace rigorous raw data management.

  • Always Maintain Raw Files: Store uncropped, unadjusted raw data files securely on lab computers and, ideally, attach them directly inside the electronic or paper lab notebooks alongside final publication figures.
  • Verify Flagged Items Manually: Automated software can flag legitimate scientific comparisons (such as zoomed-in regions of interest properly noted in figure legends). Always evaluate automated reports in context.

Resources and Contact Information

Proofig

Best Practices for Image Analysis

To support responsible image handling across disciplines, Proofig provides guidance for analyzing different types of scientific images:

For additional information about Proofig AI’s features and capabilities, visit the Proofig.com FAQ page. 

If you need technical assistance or have questions regarding data privacy and usage, please contact contact@proofig.com 

ImageTwin

For additional details regarding software capabilities, user guides, or setting up individual accounts, visit ImageTwin.ai.

For technical support, account inquiries, or data privacy questions, contact tech@imagetwin.ai or contact@imagetwin.ai.