Accurate phenotyping of cells in the tumor microenvironment is essential for understanding cancer biology but typically requires precise cell segmentation, limiting scalability. Here, we introduce Contrastive Learning Enabled Accurate Registration of Immune and Tumor cells (CLEAR-IT), a self-supervised framework that learns cell-level features from multiplexed images using only cell locations. CLEAR-IT encoders achieve strong linear evaluation performance, improve substantially with hyperparameter optimization, and maintain high accuracy across imaging modalities and with up to 90% fewer labels. When substituted for handcrafted features in a state-of-the-art classifier, CLEAR-IT features yield higher performance, and their combination enables comparable accuracy with less than half of the labeled data otherwise required. The learned representations also support prognostic modeling: using annotations from a single patient, CLEAR-IT-based phenotyping identifies survival-associated tissue features that generalize across two cohorts and modalities. CLEAR-IT provides a segmentation-light, label-efficient approach for scalable cell phenotyping and enhances existing workflows in digital pathology and tumor microenvironment analysis.

Modeling the blood–brain tumor barrier (BBTB) in vitro remains a major challenge due to the structural and functional complexity of the brain microvasculature and its dynamic interactions with glioma cells. Here, we present 3D microvascular structures fabricated by two-photon polymerization (2PP) that mimic capillary architecture and enable multicellular models for studying the BBTB. Immunofluorescence and scanning electron microscopy confirm that these structures support homogenous colonization by both human umbilical vein endothelial cells (HUVECs) and human cerebral microvascular endothelial cells (hCMEC/D3), forming tubular endothelial monolayers with polarized nuclear morphology and alignment, comparable to in vivo conditions. Additionally, endothelial cells show increased expression of cytoskeletal (tubulin, F-actin) and barrier markers (ZO-1, CD31) compared to 2D cultures. The engineered model responds to TNF-α stimulation and supports co- and tri-cultures with pericytes and glioma cells. Incorporation of glioma cells leads to reduced CD31 and elevated PLVAP expression, indicating barrier destabilization. The µPCs are also integrated into commercially available microfluidic chips via in-chip 2PP, enabling stable perfusion and providing access to both luminal and abluminal sides of the endothelium. In summary, our model provides a biomimetic and adaptable platform for studying endothelial integrity, tumor-vascular crosstalk, and broad applicability in barrier biology studies.