Strengthening Scientific Integrity: Digital Forensics for Biomedical Research Imaging

• João Phillipe Cardenuto UNICAMP
• Daniel Moreira Loyola University Chicago
• Anderson Rocha UNICAMP

Resumo

To combat the increasing number of misconduct cases in science, this Ph.D. research addressed the challenge of scientific integrity with a pioneering investigation into digital forensic analysis specifically tailored for biomedical research imaging. This work conducted extensive research into key manipulation types—copy-move forgery, image reuse, and AI-generated content—developing novel, fully explainable, and auditable computational detection methods for each. In a commitment to transparency and to promote research in the area, these techniques are provided as open-source resources. Besides the isolated techniques for each type of image forged, a central contribution is the development of an end-to-end system, created through collaboration with international forensic experts and the U.S. Office of Research Integrity (ORI). This system automates the analysis of scientific publications, starting from PDF documents and ending by identifying figures with potential integrity concerns.

Referências

E. M. Bik, A. Casadevall, and F. C. Fang, “The prevalence of inappropriate image duplication in biomedical research publications,” mBio, vol. 7, no. 3, Jul. 2016. [Online]. DOI: 10.1128/mbio.00809-16

E. M. Bucci, “Automatic detection of image manipulations in the biomedical literature,” Cell Death & Disease, vol. 9, no. 3, Mar. 2018. [Online]. DOI: 10.1038/s41419-018-0430-3

NCBI Resource Coordinators, “Pubmed: the database,” National Center for Biotechnology Information [Internet], 2005, available from: [link]. Accessed on June 2024.

D. E. Acuna, P. S. Brookes, and K. P. Kording, “Bioscience-scale automated detection of figure element reuse,” bioRxiv, Feb. 2018, available at DOI: 10.1101/269415 (Access March 2025).

M. Rossner and K. M. Yamada, “What's in a picture? the temptation of image manipulation,” Journal of Cell Biology, vol. 166, no. 1, pp. 11–15, Jul. 2004. [Online]. DOI: 10.1083/jcb.200406019

J. Christopher, “Systematic fabrication of scientific images revealed,” FEBS Letters, vol. 592, no. 18, pp. 3027–3029, Sep. 2018. [Online]. DOI: 10.1002/1873-3468.13201

D. Chawla, “A single ‘paper mill’ appears to have churned out 400 papers, sleuths find,” Science, Feb. 2020. [Online]. DOI: 10.1126/science.abb4930

E. Bik, “The stock photo paper mill,” Science Integrity Digest [Internet], 2020, available at [link]. (Accessed March 2025).

H. Else and R. V. Noorden, “The fight against fake-paper factories that churn out sham science,” Nature, vol. 591, no. 7851, pp. 516–519, Mar. 2021. [Online]. DOI: 10.1038/d41586-021-00733-5

R. Van Noorden, “How big is science’s fake-paper problem?” Nature, vol. 623, no. 7987, p. 466–467, Nov. 2023. [Online]. DOI: 10.1038/d41586-023-03464-x

C. Qi, J. Zhang, and P. Luo, “Emerging concern of scientific fraud: Deep learning and image manipulation,” BioRxiv [Preprint], Nov. 2020, available from DOI: 10.1101/2020.11.24.395319.

H. Farid, “Exposing digital forgeries in scientific images,” in Proceeding of the 8th workshop on Multimedia and security - MM&Sec '06. ACM Press, 2006. [Online]. DOI: 10.1145/1161366.1161374

M. Rossner, “A false sense of security,” Journal of Cell Biology, vol. 183, no. 4, pp. 573–574, Nov. 2008. [Online]. DOI: 10.1083/jcb.200810172

P. S. Brookes, “Misconduct detection — evolving methods & lessons from 15 years of scientific image sleuthing,” Journal of Law, Medicine & Ethics, p. 1–11, 2025.

D. Moreira, J. P. Cardenuto, R. Shao et al., “Sila: a system for scientific image analysis,” Scientific Reports, vol. 12, no. 1, Oct. 2022. [Online]. DOI: 10.1038/s41598-022-21535-3

J. P. Cardenuto, D. Moreira, and A. Rocha, “UPM - DATASET,” Jul. 2024, available at [link] (Accessed March 2025). [Online]. Available: [link]

J. A. Byrne and J. Christopher, “Digital magic, or the dark arts of the 21stcentury—how can journals and peer reviewers detect manuscripts and publications from paper mills?” FEBS Letters, vol. 594, no. 4, p. 583–589, Feb. 2020. [Online]. DOI: 10.1002/1873-3468.13747

M. E. Tipping and C. M. Bishop, “Probabilistic principal component analysis,” Journal of the Royal Statistical Society Series B: Statistical Methodology, vol. 61, no. 3, pp. 611–622, 01 2002. [Online]. DOI: 10.1111/1467-9868.00196

J. P. Cardenuto, S. Mandelli, D. Moreira, P. Bestagini, E. Delp, and A. Rocha, “Explainable artifacts for synthetic western blot source attribution,” in 2024 IEEE International Workshop on Information Forensics and Security (WIFS), 2024, pp. 1–6. [Online]. DOI: 10.1109/WIFS61860.2024.10810680

J. P. Cardenuto and A. Rocha, “Benchmarking scientific image forgery detectors,” Science and Engineering Ethics, vol. 28, no. 4, Aug. 2022. [Online]. DOI: 10.1007/s11948-022-00391-4

S. Mandelli, D. Cozzolino, E. D. Cannas et al., “Forensic analysis of synthetically generated western blot images,” IEEE Access, vol. 10, pp. 59 919–59 932, 2022. [Online]. DOI: 10.1109/ACCESS.2022.3179116

J. P. Cardenuto, D. Moreira, and A. Rocha, “Unveiling scientific articles from paper mills with provenance analysis,” PLOS ONE, vol. 19, no. 10, p. e0312666, Oct. 2024. [Online]. DOI: 10.1371/journal.pone.0312666

J. P. Cardenuto, J. Yang, R. Padilha et al., “The age of synthetic realities: Challenges and opportunities,” APSIPA Transactions on Signal and Information Processing, vol. 12, no. 1, 2023. [Online]. DOI: 10.1561/116.00000138

J. P. Cardenuto and A. Rocha, “Recod.ai scientific image integrity dataset (rsiid),” Aug. 2022, available at [link] (Accessed March 2025). [Online]. Available: [link]

Google, “Conheça os vencedores do prêmio lara 2021, o programa de bolsas de pesquisa do google,” 2021, available at [link] (Accessed March 2025).