Back

High-throughput histopathology for complex in vitro models

Harter, M. F.; D'Arcangelo, E.; Aubert, J.; Lavickova, B.; Havnar, C.; Stoll, B.; Cubela, I.; Filip, A. M.; Gaspa-Toneu, L.; Scherer, J.; Karagkiozi, E.; Dupre, J.-S.; Lopez-Sandoval, R.; Hsia, R.; Norona, L.; Brancati, G.; Okuda, R.; Camp, G.; Shamir, E.; Garcia-Cordero, J. L.; Pereiro, I.; Gjorevski, N.

2025-04-06 pathology
10.1101/2025.04.01.646522 bioRxiv
Show abstract

Human complex in vitro models (CIVMs) have demonstrated remarkable potential to study tissue development, physiology and disease at high-throughput. To effectively employ these miniaturized systems in translational preclinical research, their in-depth benchmarking is pivotal. Histology has been the core of tissue characterization for centuries and the foundation of spatial phenotyping. However, standard histology workflows are inherently low-throughput and centered on large tissue pieces. This does not match the high sample volumes and small sample sizes in CIVM research. Here, we introduce a holistic histo-workflow, utilizing 3D-printed histomolds that facilitate co-planar embedding of CIVMs at high-throughput, resulting in up to 48 samples in one section. We developed a variety of model-specific histomold designs that enable spatially controlled histological sectioning and downstream analyses. We describe these workflows, including mold generation, highplex staining and image analysis, and exemplify their application to histological analyses of various CIVMs. Altogether, the histomolds introduced here afford opportunities for CIVM processing and analysis, while significantly reducing labor and reagent resources, thereby democratizing high-throughput CIVM in histopathology.

Published in Cell Reports Methods (predicted rank #2) · training set

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.