HAICON 2026

AI FOR SCIENCE 8-11 JUNE 2026, MUNICH, GERMANY

A 3D Foundation Model for Generalizable Biological Structure Segmentation in Tissue Clearing Images Authors: Qi Liu1,2,3,4, Ying Chen1,2,3,4, Markus Elsner1,6, Ali Ertürk1,2,4,6,7 #
1 Institute for Stroke and Dementia Research, Klinikum der Universität München, Ludwig-Maximilians University Munich, Munich, Germany 2 Institute for Intelligent Biotechnologies (iBIO), Helmholtz Center Munich, Neuherberg, Germany 3 Faculty of Medicine, Ludwig-Maximilians University Munich, Munich, Germany 4 Munich Cluster for Systems Neurology (SyNergy), Munich, Germany 5 Munich Medical Research School (MMRS), Munich, Germany 6 Deep Piction GmbH, Munich, Germany 7 School of Medicine, Koç University, İstanbul, Turkey These authors contributed equally: Qi Liu, Ying Chen #Correspondence: ali.erturk@helmholtz-munich.de

Abstract Tissue clearing combined with light-sheet microscopy (LSM) enables the 3D visualization and analysis of intricate cellular and subcellular structures across tissues and organisms, characterized by high contrast and super-resolution capabilities. However, segmenting diverse biological structures in 3D LSM images remains a major challenge due to significant variations in morphology, artifacts, signal-to-noise ratio, and surrounding tissue context. Conventional supervised learning-based segmentation models typically require extensive voxel-wise annotations and the training structure-specific models, thereby limiting their generalizability and scalability across datasets and applications. Foundation models (FMs) are expected to mitigate this limitation. FMs trained on large-scale data are anticipated to achieve zero-shot or few-shot generalization. Despite their success in other domains, their application to LSM data remains underexplored. In this study, we propose a foundation model tailored for comprehensive segmentation of diverse biological structures in tissue-cleared mouse LSM dataset and evaluate its domain generalization capability. We construct a large-scale dataset containing more than 9 biological structures and 50,000 patches of size 300³ and perform self-supervised learning (SSL) to learn robust representations from diverse LSM data. The model is evaluated across extensive 3D segmentation tasks, including out-of-distribution datasets. Our findings demonstrate that a self-supervised pretrained foundation model enables effective cross-structure transfer for 3D image segmentation and indicate its strong ability to generalize to previously unseen biological structures. In several downstream tasks, it outperforms state-of-the-art task-specific 3D segmentation models. Overall, this study underscores the potential of FMs in LSM image domain and demonstrates their capability as a unified approach for segmentation across diverse biological domains.