Abstract
Diffusion MRI (dMRI) provides white-matter microstructural information, but its associations with marker-specific histopathology are often assessed by global or regional scalar summaries, which may obscure focal, boundary-localized, or region-specific patterns. We propose a new dMRI-to-IHC spatial proxy mapping task under weak MRI–histology registration. Given dMRI and auxiliary structural priors, the goal is to generate inspectable tissue-scale virtual immunomarker proxy maps that preserve regional marker distributions and support fixed-protocol quantification. By retaining spatial heterogeneity and regional MRI–histology correspondence, this formulation moves beyond direct scalar readout regression, but remains challenging due to cross-domain gaps, weak alignment, and scale breaks in cascaded generation. To address these challenges, we propose TriSCoV-Net, a cross-scale verified framework that combines dMRI-conditioned proxy generation, PLI-assisted conditional super-resolution, support-domain constraint, and training-time cross-scale fine-tuning. Using a fixed ImageJ DAB workflow, we select 16.8 μm/pixel as the primary output and 4.2 μm/pixel as an auxiliary scale. On ODBB under LOSO evaluation, TriSCoV-Net improves image fidelity and fixed-protocol quantitative agreement over representative baselines.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/CMMCA2026_010.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=iLfEQ0c733
BibTex
@InProceedings{ZhuHon_TriSCoVNet_MICCAISAT2026,
author = { Zhu, Hongjie AND Zhou, Huayuan AND Tang, Hao AND Liu, Hanyu AND Li, Chao},
title = { { TriSCoV-Net: Cross-Scale Verified Virtual Immunomarker Proxy Generation from Diffusion Magnetic Resonance Imaging } },
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}
}