Abstract Vision foundation models (FMs) are accelerating the development of digital pathology algorithms and transforming biomedical research. These models learn, in a self-supervised manner, to represent histological features in highly heterogeneous tiles extracted from whole-slide images (WSIs) of real-world patient samples. The performance of these FMs is significantly influenced by the size, diversity, and balance of the pre-training data. Yet, data selection has been primarily guided by expert knowledge at the WSI level, focusing on factors such as disease classification and tissue types, while largely overlooking the granular details available at the tile level. In this paper, we investigate the potential of unsupervised automatic data curation at the tile-level, taking into account 350 million tiles. Specifically, we apply hierarchical clustering trees to pre-extracted tile embeddings, allowing us to sample balanced datasets uniformly across the embedding space of the pretrained FM. We further show that these datasets are subject to a trade-off between size and balance, potentially compromising the quality of representations learned by FMs. We propose tailored batch sampling strategies to mitigate this effect. We demonstrate the effectiveness of our method through improved performance on a diverse range of clinically relevant downstream tasks.

Authors Boqi Chen, Cédric Vincent-Cuaz, Lydia A Schoenpflug, Manuel Madeira, Lisa Fournier, Vaishnavi Subramanian, Sonali Andani, Samuel Ruiperez-Campillo, Julia E Vogt, Raphaëlle Luisier, Dorina Thanou, Viktor H Koelzer, Pascal Frossard, Gabriele Campanella, Gunnar Rätsch

Submitted Medical Image Computing and Computer Assisted Intervention (MICCAI) 2025

Link DOI

Abstract Multiplexed protein imaging offers valuable insights into interactions between tumours and their surrounding tumour microenvironment, but its widespread use is limited by cost, time and tissue availability. Here we present HistoPlexer, a deep learning framework that generates spatially resolved protein multiplexes directly from standard haematoxylin and eosin (H&E) histopathology images. HistoPlexer jointly predicts multiple tumour and immune markers using a conditional generative adversarial architecture with custom loss functions designed to ensure pixel- and embedding-level similarity while mitigating slice-to-slice variations. A comprehensive evaluation of metastatic melanoma samples demonstrates that HistoPlexer-generated protein maps closely resemble real maps, as validated by expert assessment. They preserve crucial biological relationships by capturing spatial co-localization patterns among proteins. The spatial distribution of immune infiltration from HistoPlexer-generated protein multiplex enables stratification of tumours into immune subtypes. In an independent cohort, integration of HistoPlexer-derived features into predictive models enhances performance in survival prediction and immune subtype classification compared to models using H&E features alone. To assess broader applicability, we benchmarked HistoPlexer on publicly available pixel-aligned datasets from different cancer types. In all settings, HistoPlexer consistently outperformed baseline methods, demonstrating robustness across diverse tissue types and imaging conditions. By enabling whole-slide protein multiplex generation from routine H&E images, HistoPlexer offers a cost- and time-efficient approach to tumour microenvironment characterization with strong potential to advance precision oncology.

Authors Sonali Andani, Boqi Chen, Joanna Ficek-Pascual, Simon Heinke, Ruben Casanova, Bernard Friedrich Hild, Bettina Sobottka, Bernd Bodenmiller, Viktor H Koelzer, Gunnar Rätsch

Submitted Nature Machine Intelligence

Link DOI

Abstract Single-source domain generalization (SDG) aims to learn a model from a single source domain that can generalize well on unseen target domains. This is an important task in computer vision, particularly relevant to medical imaging where domain shifts are common. In this work, we consider a challenging yet practical setting: SDG for cross-modality medical image segmentation. We combine causality-inspired theoretical insights on learning domain-invariant representations with recent advancements in diffusion-based augmentation to improve generalization across diverse imaging modalities. Guided by the “intervention-augmentation equivariant” principle, we use controlled diffusion models (DMs) to simulate diverse imaging styles while preserving the content, leveraging rich generative priors in large-scale pretrained DMs to comprehensively perturb the multidimensional style variable. Extensive experiments on challenging cross-modality segmentation tasks demonstrate that our approach consistently outperforms state-of-the-art SDG methods across three distinct anatomies and imaging modalities. The source code is available at https://github.com/ratschlab/ICMSeg.

Authors Boqi Chen, Yuanzhi Zhu, Yunke Ao, Sebastiano Caprara, Reto Sutter, Gunnar Rätsch, Ender Konukoglu, Anna Susmelj

Submitted The IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2025

Link

Abstract Simulating the complex interactions between soft tissues and rigid anatomy is critical for applications in surgical training, planning, and robotic-assisted interventions. Traditional Finite Element Method (FEM)-based simulations, while accurate, are computationally expensive and impractical for real-time scenarios. Learning-based approaches have shown promise in accelerating predictions but have fallen short in modeling soft-rigid interactions effectively. We introduce MIXPINN, a physics-informed Graph Neural Network (GNN) framework for mixed-material simulations, explicitly capturing soft-rigid interactions using graph-based augmentations. Our approach integrates Virtual Nodes (VNs) and Virtual Edges (VEs) to enhance rigid body constraint satisfaction while preserving computational efficiency. By leveraging a graph-based representation of biomechanical structures, MIXPINN learns high-fidelity deformations from FEM-generated data and achieves real-time inference with sub-millimeter accuracy. We validate our method in a realistic clinical scenario, demonstrating superior performance compared to baseline GNN models and traditional FEM methods. Our results show that MIXPINN reduces computational cost by an order of magnitude while maintaining high physical accuracy, making it a viable solution for real-time surgical simulation and robotic-assisted procedures

Authors Xintian Yuan, Yunke Ao, Boqi Chen, Philipp Fuernstahl

Submitted The IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2025

Link

Authors Sonali Andani, Boqi Chen, Joanna Ficek-Pascual, Simon Heinke, Ruben Casanova, Bettina Sobottka, Bernd Bodenmiller, Tumor Profiler Consortium, Viktor H Kölzer, Gunnar Rätsch

Submitted medRxiv

Abstract Deep learning models can achieve high accuracy when trained on large amounts of labeled data. However, real-world scenarios often involve several challenges: Training data may become available in installments, may originate from multiple different domains, and may not contain labels for training. Certain settings, for instance medical applications, often involve further restrictions that prohibit retention of previously seen data due to privacy regulations. In this work, to address such challenges, we study unsupervised segmentation in continual learning scenarios that involve domain shift. To that end, we introduce GarDA (Generative Appearance Replay for continual Domain Adaptation), a generative-replay based approach that can adapt a segmentation model sequentially to new domains with unlabeled data. In contrast to single-step unsupervised domain adaptation (UDA), continual adaptation to a sequence of domains enables leveraging and consolidation of information from multiple domains. Unlike previous approaches in incremental UDA, our method does not require access to previously seen data, making it applicable in many practical scenarios. We evaluate GarDA on three datasets with different organs and modalities, where it substantially outperforms existing techniques. Our code is available at: https://github.com/histocartography/generative-appearance-replay.

Authors Boqi Chen, Kevin Thandiackal, Pushpak Pati, Orcun Goksel

Submitted Medical Image Analysis

Link DOI

Abstract Multiple Instance Learning (MIL) methods have become increasingly popular for classifying gigapixel-sized Whole-Slide Images (WSIs) in digital pathology. Most MIL methods operate at a single WSI magnification, by processing all the tissue patches. Such a formulation induces high computational requirements and constrains the contextualization of the WSI-level representation to a single scale. Certain MIL methods extend to multiple scales, but they are computationally more demanding. In this paper, inspired by the pathological diagnostic process, we propose ZoomMIL, a method that learns to perform multi-level zooming in an end-to-end manner. ZoomMIL builds WSI representations by aggregating tissue-context information from multiple magnifications. The proposed method outperforms the state-of-the-art MIL methods in WSI classification on two large datasets, while significantly reducing computational demands with regard to Floating-Point Operations (FLOPs) and processing time by 40–50 . Our code is available at: https://github.com/histocartography/zoommil.

Authors Kevin Thandiackal, Boqi Chen, Pushpak Pati, Guillaume Jaume, Drew FK Williamson, Maria Gabrani, Orcun Goksel

Submitted European Conference on Computer Vision (ECCV) 2022

Link