A Method for Detecting Distinctive Patterns of Real Patients in Generated Images

Generative diffusion models are a well-established method for generating high-quality images. However, there are studies that show that diffusion models are less privacy-friendly than generative models, such as generative adversarial networks and a growing family of their modifications. The discover...

Full description

Saved in:
Bibliographic Details
Main Author: V. A. Kovalev
Format: Article
Language:Russian
Published: Educational institution «Belarusian State University of Informatics and Radioelectronics» 2025-02-01
Series:Doklady Belorusskogo gosudarstvennogo universiteta informatiki i radioèlektroniki
Subjects:
Online Access:https://doklady.bsuir.by/jour/article/view/4061
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849772733013950464
author V. A. Kovalev
author_facet V. A. Kovalev
author_sort V. A. Kovalev
collection DOAJ
description Generative diffusion models are a well-established method for generating high-quality images. However, there are studies that show that diffusion models are less privacy-friendly than generative models, such as generative adversarial networks and a growing family of their modifications. The discovered vulnerabilities require in-depth study of various security aspects. This is especially important for sensitive areas such as medical image analysis tasks and their practical applications. The paper describes a method for detecting image patterns presented in generated images that can potentially be identified in real CT images of patients with pulmonary tuberculosis. The method includes the following main procedures: correlation of pairs of generated and real images to pre-select pairs that involve further analysis; calculation of correlation statistics using direct and inverse Fisher transforms; performing affine image registration and calculating pairwise similarity scores; nonlinear (elastic) image registration and recalculation of similarity scores to highlight the most similar/dissimilar image areas.
format Article
id doaj-art-2d615b99442d4ea2aceeb7fea9e36f00
institution DOAJ
issn 1729-7648
language Russian
publishDate 2025-02-01
publisher Educational institution «Belarusian State University of Informatics and Radioelectronics»
record_format Article
series Doklady Belorusskogo gosudarstvennogo universiteta informatiki i radioèlektroniki
spelling doaj-art-2d615b99442d4ea2aceeb7fea9e36f002025-08-20T03:02:15ZrusEducational institution «Belarusian State University of Informatics and Radioelectronics»Doklady Belorusskogo gosudarstvennogo universiteta informatiki i radioèlektroniki1729-76482025-02-01231475310.35596/1729-7648-2025-23-1-47-532049A Method for Detecting Distinctive Patterns of Real Patients in Generated ImagesV. A. Kovalev0The United Institute of Informatics Problems of the National Academy of Sciences of BelarusGenerative diffusion models are a well-established method for generating high-quality images. However, there are studies that show that diffusion models are less privacy-friendly than generative models, such as generative adversarial networks and a growing family of their modifications. The discovered vulnerabilities require in-depth study of various security aspects. This is especially important for sensitive areas such as medical image analysis tasks and their practical applications. The paper describes a method for detecting image patterns presented in generated images that can potentially be identified in real CT images of patients with pulmonary tuberculosis. The method includes the following main procedures: correlation of pairs of generated and real images to pre-select pairs that involve further analysis; calculation of correlation statistics using direct and inverse Fisher transforms; performing affine image registration and calculating pairwise similarity scores; nonlinear (elastic) image registration and recalculation of similarity scores to highlight the most similar/dissimilar image areas.https://doklady.bsuir.by/jour/article/view/4061diffusion generative modelscomputed tomographyprivacy preserving
spellingShingle V. A. Kovalev
A Method for Detecting Distinctive Patterns of Real Patients in Generated Images
Doklady Belorusskogo gosudarstvennogo universiteta informatiki i radioèlektroniki
diffusion generative models
computed tomography
privacy preserving
title A Method for Detecting Distinctive Patterns of Real Patients in Generated Images
title_full A Method for Detecting Distinctive Patterns of Real Patients in Generated Images
title_fullStr A Method for Detecting Distinctive Patterns of Real Patients in Generated Images
title_full_unstemmed A Method for Detecting Distinctive Patterns of Real Patients in Generated Images
title_short A Method for Detecting Distinctive Patterns of Real Patients in Generated Images
title_sort method for detecting distinctive patterns of real patients in generated images
topic diffusion generative models
computed tomography
privacy preserving
url https://doklady.bsuir.by/jour/article/view/4061
work_keys_str_mv AT vakovalev amethodfordetectingdistinctivepatternsofrealpatientsingeneratedimages
AT vakovalev methodfordetectingdistinctivepatternsofrealpatientsingeneratedimages