INST-Align: Implicit Neural Alignment for Spatial Transcriptomics via Canonical Expression Fields

2026-04-13

AI Summary

This work addresses the challenge of aligning and integrating multi-slice spatial transcriptomics data, which is often confounded by non-rigid deformations and batch effects that conventional methods struggle to handle jointly. The authors propose an unsupervised pairwise framework that, for the first time, leverages implicit neural representations to construct a shared canonical expression field, coupled with a coordinate deformation network. Through spatial-feature joint matching and a two-stage training strategy, the approach enforces mutual constraints between alignment and embedding learning. Evaluated across nine datasets, the method achieves state-of-the-art performance, with average optimal transport (OT) and nearest-neighbor (NN) accuracies of 0.702 and 0.719, respectively, and reduces Chamfer distance by up to 94.9%, enabling biologically consistent 3D tissue reconstruction.

Technology Category

• Machine Learning: Matrix & Tensor Methods
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Computer Vision: Multi-modal Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Spatial transcriptomics (ST) measures mRNA expression while preserving spatial organization, but multi-slice analysis faces two coupled difficulties: large non-rigid deformations across slices and inter-slice batch effects when alignment and integration are treated independently. We present INST-Align, an unsupervised pairwise framework that couples a coordinate-based deformation network with a shared Canonical Expression Field, an implicit neural representation mapping spatial coordinates to expression embeddings, for joint alignment and reconstruction. A two-phase training strategy first establishes a stable canonical embedding space and then jointly optimizes deformation and spatial-feature matching, enabling mutually constrained alignment and representation learning. Cross-slice parameter sharing of the canonical field regularizes ambiguous correspondences and absorbs batch variation. Across nine datasets, INST-Align achieves state-of-the-art mean OT Accuracy (0.702), NN Accuracy (0.719), and Chamfer distance, with Chamfer reductions of up to 94.9\% on large-deformation sections relative to the strongest baseline. The framework also yields biologically meaningful spatial embeddings and coherent 3D tissue reconstruction. The code will be released after review phase.

Problem

• spatial transcriptomics
• non-rigid deformations
• batch effects
• multi-slice alignment
• spatial integration

Innovation

• Implicit Neural Representation
• Canonical Expression Field
• Unsupervised Alignment
• Spatial Transcriptomics
• Joint Optimization