DUET: Dual-Paradigm Adaptive Expert Triage with Single-cell Inductive Prior for Spatial Transcriptomics Prediction

2026-05-13

AI Summary

Current spatial transcriptomics imputation methods often oversimplify the task as a morphology-to-expression mapping, neglecting molecular consistency and struggling to effectively integrate large-scale single-cell data, thereby compromising both model expressiveness and biological fidelity. To address this, we propose DUET, a novel framework that introduces, for the first time, a dual-paradigm adaptive fusion mechanism. Guided by inductive priors from single-cell references, DUET concurrently performs parametric regression and memory-based retrieval, with a lightweight dynamic adapter modulating branch weights according to local spatial context. By incorporating single-cell references as molecular constraints and augmenting them with a structural refinement module, DUET substantially enhances the biological plausibility and robustness of predictions. Evaluated on three public datasets, DUET consistently achieves state-of-the-art performance, with ablation studies confirming the significant and coherent contributions of its individual components over existing approaches.

Abstract

Inferring spatially resolved gene expression from histology images offers a cost-effective complement to spatial transcriptomics (ST). However, existing methods reduce this task to a simple morphology-to-expression mapping, where visual similarity does not guarantee molecular consistency. Meanwhile, single-cell data has amassed rich resources far surpassing the scale of ST data, yet it remains underexplored in vision-omics modeling. Furthermore, current approaches commit to a monolithic paradigm with bottlenecks, unable to balance expressive flexibility with biological fidelity. To bridge these gaps, we propose DUET, a novel dual-paradigm framework that synergizes parametric prediction and memory-based retrieval under cellular inductive priors. DUET implements a parallel regression-retrieval paradigm, adaptively reconciling the outputs of its complementary pathways. To mitigate aleatoric vision ambiguity, we incorporate large-scale single-cell references to impose molecular states as biological constraints for faithful learning. Building upon structural refinement, we further design a lightweight adapter to dynamically assign branch preference across spatial contexts to achieve optimal performance. Extensive experiments on three public datasets across varied gene scales demonstrate that DUET achieves SOTA performance, with consistent gains contributed by each proposed component. Code is available at https://github.com/Junchao-Zhu/DUET

Problem

• spatial transcriptomics
• histology images
• single-cell data
• morphology-to-expression mapping
• biological fidelity

Innovation

• dual-paradigm learning
• spatial transcriptomics prediction
• single-cell inductive prior
• adaptive expert triage
• vision-omics integration