Unsupervised discovery of tissue architecture in multiplexed imaging

Kim, Junbum; Rustam, Samir; Mosquera, Juan Miguel; Randell, Scott H.; Shaykhiev, Renat; Rendeiro, André F.; Elemento, Olivier

doi:10.1038/s41592-022-01657-2

Cited by 46 publications

(49 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Pixel-based tissue domain detection forms the basis of the top-down approach to spatial data analysis. Current methods of tissue domain detection are either based on a bottom-up approach, that is, building cellular neighborhoods using segmented single-cell data (20, 34, 59) and/or lack scalability across samples (8, 30, 55, 63). Here, we addressed this gap by developing MILWRM, an algorithm to detect spatial domains across samples through a top-down, pixel-based approach.…”

Section: Discussionmentioning

confidence: 99%

“…Cellular segmentation and annotation are the most challenging step in this kind of approach. There are various methods available for cellular segmentation (25, 40), annotation (39) and neighborhood analysis (20, 34, 59). Widely used lower resolution imaging data such as spatial transcriptomics (ST) and imaging mass spectrometry data are analyzed using cellular deconvolution algorithms to approximate singlecell composition.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Consensus tissue domain detection in spatial multi-omics data using MILWRM

Kaur

Heiser

McKinley

et al. 2023

Preprint

View full text Add to dashboard Cite

Spatially resolved molecular assays provide high dimensional genetic, transcriptomic, proteomic, and epigenetic information in situ and at various resolutions. Pairing these data across modalities with histological features enables powerful studies of tissue pathology in the context of an intact microenvironment and tissue structure. Increasing dimensions across molecular analytes and samples require new data science approaches to functionally annotate spatially resolved molecular data. A specific challenge is data-driven cross-sample domain detection that allows for analysis within and between consensus tissue compartments across high volumes of multiplex datasets stemming from tissue atlasing efforts. Here, we present MILWRM (multiplex image labeling with regional morphology) a Python package for rapid, multi-scale tissue domain detection and annotation. We demonstrate MILWRM's utility in identifying histologically distinct compartments in human colonic polyps and mouse brain slices through spatially- informed clustering in two different spatial data modalities. Additionally, we used tissue domains detected in human colonic polyps to elucidate molecular distinction between polyp subtypes. We also explored the ability of MILWRM to identify anatomical regions of mouse brain and their respective distinct molecular profiles.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Consensus tissue domain detection in spatial multi-omics data using MILWRM

Kaur

Heiser

McKinley

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…SpatialLDA [106] is a tumor microenvironment detection method that identifies associated topic or context of each cell based on cell type distribution of immediate spatial neighbors, which for tumor cells could be thought of as tumor microenvironments. UTAG [107] is a structural microanatomy annotation and analysis method that categorizes cells into anatomical structures across organs and diseases including cancers. Both SpatialLDA and UTAG infer larger-scale patterns or organization in tissue, which can be further interrogated to understand how cellular composition and interaction give rise to tissue structure capable of contributing to organ-specific physiology, overall organ architecture, or microenvironments in the tumor micro-environment which may condition clinically relevant outcomes.…”

Section: Novel Spatial Omics Methods Using Spatial Informationmentioning

confidence: 99%

Spatial omics technologies at multimodal and single cell/subcellular level

et al. 2022

View full text Add to dashboard Cite

Spatial omics technologies enable a deeper understanding of cellular organizations and interactions within a tissue of interest. These assays can identify specific compartments or regions in a tissue with differential transcript or protein abundance, delineate their interactions, and complement other methods in defining cellular phenotypes. A variety of spatial methodologies are being developed and commercialized; however, these techniques differ in spatial resolution, multiplexing capability, scale/throughput, and coverage. Here, we review the current and prospective landscape of single cell to subcellular resolution spatial omics technologies and analysis tools to provide a comprehensive picture for both research and clinical applications.

show abstract

“…In recent years, the representation of cells as a spatial graph 24,[43][44][45][46][47][48][49][50][51][52][53] has emerged as a promising way to discover new cellular phenotypes 45,49,53 , make slide-level predictions 24,44,47,52 , and hierarchically cluster cells into tissue microstructures 27,46,50,51 . However, for the most common imaging modality, Hematoxylin and Eosin (H&E) stained histology, these have only been applied to patch-level 27,46,48 , slide-level prediction 24,44,47,52 or graphs restricted to fixed regions 46,50 .…”

Section: Introductionmentioning

confidence: 99%

HAPPY: A deep learning pipeline for mapping cell-to-tissue graphs across placenta histology whole slide images

Vanea

Dzigurski

Rukins

et al. 2022

Preprint

View full text Add to dashboard Cite

The placenta is a heterogeneous and rapidly evolving organ. Digital pathology presents a novel set of approaches to histology analysis but has yet to be applied for the placenta. We present the ‘Histology Analysis Pipeline.PY’ (HAPPY), a hierarchical method for quantifying the variability of cells and micro-anatomical tissue structures across placenta histology whole slide images. In contrast to commonly used high-level patch features or segmentation approaches, HAPPY follows an interpretable biological hierarchy. It is an end-to-end pipeline able to represent cells and communities of cells within tissues at a single-cell resolution across whole slide images. Here we present a set of quantitative metrics from healthy term placentas as a baseline for future assessments of placenta health. HAPPY cell and tissue predictions closely replicate those from independent clinical experts and the placenta biology literature

show abstract

Unsupervised discovery of tissue architecture in multiplexed imaging

Cited by 46 publications

References 61 publications

Consensus tissue domain detection in spatial multi-omics data using MILWRM

Consensus tissue domain detection in spatial multi-omics data using MILWRM

Spatial omics technologies at multimodal and single cell/subcellular level

HAPPY: A deep learning pipeline for mapping cell-to-tissue graphs across placenta histology whole slide images

Contact Info

Product

Resources

About