Abstract
Cancer progression is shaped by spatially organized communication among tumor, stromal, immune and vascular-associated cells. H&E histopathology captures tumor architecture and local tissue morphology, while spatial transcriptomics (ST) provides spatially resolved molecular states. Because ST is commonly acquired as discrete two-dimensional sections, directed ligand-receptor communication inference in continuous three-dimensional tissue space remains limited. We propose OncoRelay3D, a multimodal framework that reconstructs 3D tumor cell-expression fields and infers mechanism-aware directed cell-cell communication. Prov-GigaPath extracts local morphology embeddings from cell-centered H&E crops to regularize cross-slice correspondence, while ST expression remains the primary signal for intermediate cell-location and expression reconstruction. Original and predicted sections are assembled into a continuous 3D cell-expression field, where directed ligand-receptor-specific cell graphs are modeled with 3D geometry, pathway annotation, mechanism type and triplet directional context. Experiments on serial-section Open-ST data from metastatic cervical lymph nodes of patients with head and neck squamous cell carcinoma show improved interpolation and downstream 3D communication rationality. Cancer analysis on the Open-ST HNSCC sample summarizes normalized pathway scores at the full-section and tumor-associated patch levels and visualizes directed 3D communication within the selected tumor-containing patch, identifying tumor-microenvironment programs involving ECM remodeling, antigen presentation and immune-checkpoint signaling.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/CMMCA2026_012.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=qNqfnZ5y6G
BibTex
@InProceedings{LiZhi_OncoRelay3D_MICCAISAT2026,
author = { Li, Zhitong AND Jiang, Yixuan AND Liu, Ziming AND Wu, Chenjie AND Que, Ningfeng AND Li, Chao},
title = { { OncoRelay3D: Foundation-Model-Assisted Reconstruction of 3D Tumor Cell-Expression Fields for Mechanism-Aware Directed Cell-Cell Communication } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17278},
month = {pending},
page = {pending}
}