DDPM-Based Histopathology Data Augmentation for Blood Vessel Segmentation
Authors
Tanczos, T.; Benešová, V.; Pavlovičová, J.; Vajsová, A.
Year
2026
Published
Proceedings of the 19th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2026). Setúbal: Science and Technology Publications, Lda, 2026. p. 316-323. BIODEVICES, BIOIMAGING, BIOINFORMATICS. vol. 2. ISSN 2184-4305. ISBN 978-989-758-802-0.
Type
Proceedings paper
Departments
Annotation
Deep learning approaches for histopathological image analysis require large, well-annotated datasets, which are difficult to obtain due to the need for domain expertise and privacy constraints. This work addresses dataset limitations through synthetic data augmentation using Denoising Diffusion Probabilistic Models (DDPMs). We propose a unified framework that explores two complementary strategies: fully synthetic image generation and partially synthetic generation via inpainting. Our approach systematically compares pixel-space and latent-space diffusion models on our in-house histopathology datasets consisting of post-transplant heart tissue biopsies from the Institute for Clinical and Experimental Medicine (IKEM). We evaluate the quality of synthetic data using standard metrics (KID, FID, LPIPS) as well as expert pathologist assessment. The synthetic samples are subsequently employed for augmentation in segmentation pipelines to assess improvements in detecting underrepresented structures, particularly blood vessels. Our experiments demonstrate that synthetic data augmentation enhances recall (a substantial gain in medical imaging where sensitivity to fine structures is critical) in vessel detection (from 0.34 to 0.42) while maintaining overall segmentation capability, though with some trade-offs in precision. These findings indicate that systematic diffusion-based augmentation represents a promising approach to addressing class imbalance in histopathological datasets, particularly for rare anatomical structures.
Using Diffusion Models to Synthesize Patches of Histological Images Based on Nuclear Atypia Score
Authors
Bohumel, S.; Kollár, M.; Váczlavová, E.; Benešová, V.
Year
2026
Published
Lecture Notes in Networks and Systems. Springer Science and Business Media Deutschland GmbH, 2026. p. 372-383. 1676 LNNS. vol. 2. ISSN 2367-3370. ISBN 978-3-032-07988-6.
Type
Proceedings paper
Departments
Annotation
Neural networks and deep learning are widely used for solv-ing tasks in multiple domains, including computer vision. In the field of medical image processing, these approaches can bring efficient and fast diagnosis. However, there is a challenge associated with the lack of annotated training data needed to train the models. The collection and especially the annotation of such data can be time-consuming and expen-sive because of the required knowledge of a domain expert. In this work, we explore the usage of generative models for histological data synthesis that could complement existing training sets and possibly improve the performance of deep learning models. The main area of our research is the synthesis of tissue patc hesbasedonthenuclearatypiascoreusedinbreastcancerdiagnostics.Nuclearatypiaisusuallymanifestedbyenlargedcellsandirregularshapes.Wetakeadvantageofdiffusionprob-abilisticmodelsusedtosynthesisepatchesofhistopathologicaltissueswithaspecificatypiascore.Finally,throughevaluation,wedemonstratethestrengthsandweaknessesofsuchpatch-leveldataenhancementforthetaskofnuclearatypiascoring.