Related Experiment Video
Updated: Aug 5, 2026
Multimodal Cross-Device and Marker-Free Co-Registration of Preclinical Imaging Modalities
Published on: October 27, 2023
Beyond image alignment: Challenges and emerging solutions for MSI-histology registration in multimodal spatial omics
Jiadong Liu1, Chunping Tang2, Yang Ye2
1State Key Laboratory of Synthetic Biology, Frontiers Science Center for Synthetic Biology (Ministry of Education), School of Synthetic Biology and Biomanufacturing, Tianjin University, Tianjin, 300072, China; China-Serbia "Belt and Road" Joint Laboratory for Natural Products and Drug Discovery, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai, 201203, China.
Reliable registration is crucial for quantitative analysis in multimodal spatial omics. This review frames registration as analytical calibration, detailing methods for cross-modal comparability and transformation, ensuring accurate data integration.
Area of Science:
- Spatial omics
- Bioinformatics
- Computational pathology
Background:
- Multimodal spatial omics integrates diverse molecular and imaging data but faces registration challenges.
- Inherent differences in modalities (e.g., mass spectrometry, transcriptomics, proteomics, multiplexed imaging) and technical artifacts hinder accurate alignment.
- Visually plausible registration is insufficient for the quantitative reliability required in spatial omics analysis.
Purpose of the Study:
- To reframe registration in multimodal spatial omics as a critical analytical calibration step.
- To provide a comprehensive methodological overview of registration strategies and error sources.
- To establish a framework for validation and quality control in multimodal spatial omics data integration.
Main Methods:
- Examination of error sources impacting quantification and cross-section integration.
- Strategies for cross-modal comparability including ion images, embeddings, graph-based representations, and semantic segmentation.
- Review of similarity metrics (mutual information, structural, semantic consistency) and transformation frameworks (rigid, affine, nonrigid).
Main Results:
- Registration is advanced as an analytical calibration, not just preprocessing.
- Detailed analysis of error sources and their impact on quantitative outcomes.
- A structured approach to similarity modeling and transformation optimization for diverse spatial omics data.
Conclusions:
- Validation and quality control using geometric, structural, and analytical consistency are essential.
- A practical methodological map and evaluation framework are provided for multimodal spatial omics.
- Aligning registration methods with analytical goals enhances data integration and interpretation in high-throughput workflows.
