The HEADT Centre team is currently creating and implementing a comprehensive database with images from retracted scientific publications (“Image Integrity Database”, or short: “IIDB”). The aim is to produce a searchable database for researchers worldwide who want to learn more about the nature of inappropriate image manipulation. The database offers a test set of images for scholars who are working on the development of image manipulation detection algorithms or image analysis tools to facilitate an automated screening of images in publications.
The plan is to produce a structured database, which allows assessing images from retracted publications on an individual case level. For each image case there will be a full package of associated data that allows a better understanding of the complexity of such cases, such as links to RetractionWatch or PubPeer, links to the original article or to retraction notices, information about the authors and teams, the journals and the academic fields, and more.
Often evaluating image manipulation requires a broad investigation that goes beyond a computational analysis of pixels in an image, and involves an investigation into individual routines and practices, workflows and standards and community guidelines, as well as an evaluation of the larger ecology in which such cases occur. The database provides all available data to facilitate a more thorough understanding of the phenomenon.
One crucial element for the automated detection of image manipulations is to provide access to the original image data. Very often images in publications are only accessible in the form of pdf’s with reduced quality, which normally are the result of selection and sometimes even excessive editing.
The HEADT Centre team hopes to establish partnerships with publishers to include original image data wherever available and possible. For legal reasons the database may first be made accessible to restricted group of collaborators participating in further development, before it will be made accessible to a larger audience.
For more detailed information on the project or to keep in touch, please visit our blog again or simply leave a message.