Paper

Forensic Analysis of Synthetically Generated Western Blot Images

Sara Mandelli1 , Davide Cozzolino6 , Edoardo Cannas1 , João Phillipe Cardenuto3 , Daniel Moreira4 , Paolo Bestagini1 , Walter Scheirer5 , Anderson Rocha3 , Luisa Verdoliva2 , Stefano Tubaro1 , Edward Delp6

1Politecnico di Milano, Italy, 2University Federico II of Naples, Italy, 3Universidade Estadual de Campinas, Campinas, São Paulo, Brazil, 4Loyola University Chicago, Chicago, IL, USA, 5University of Notre Dame, Notre Dame, IN, USA, 6Purdue University, West Lafayette, IN, USA

Data DOI:10.1109/ACCESS.2022.3179116
Preview Image

Abstract

The widespread diffusion of synthetically generated content poses a significant threat to scientific integrity. In this paper, we present a forensic analysis of synthetically generated Western blot images—an essential element in biomedical research. We introduce explainable, low-level artifact‐based techniques that can detect synthetic manipulations and attribute images to the specific generative model used to create them. Evaluations on datasets demonstrate that the proposed methods reliably distinguish between genuine and synthetic images.

For further details, please refer to the full publication in IEEE Access.


Citation

Mandelli S, et al. (2022) Forensic Analysis of Synthetically Generated Western Blot Images. IEEE Access, 10, 59919–59932. DOI: 10.1109/ACCESS.2022.3179116
1
2
3
4
5
6
7
8
9
10
@article{Mandelli2022,
  title={Forensic Analysis of Synthetically Generated Western Blot Images},
  author={Mandelli, Sara and Cozzolino, Davide and Cannas, Edoardo D and Cardenuto, Joao P and Moreira, Daniel and Bestagini, Paolo and Scheirer, Walter J and Rocha, Anderson and Verdoliva, Luisa and Tubaro, Stefano and Delp, Edward J},
  journal={IEEE Access},
  volume={10},
  pages={59919--59932},
  year={2022},
  publisher={IEEE},
  doi={10.1109/ACCESS.2022.3179116}
}

Data: Dataset