Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Segmentation-free integration of nuclei morphology and spatial transcriptomics for retinal images

Eduard Chelebian, Pratiti Dasgupta, Zainalabedin Samadi, Carolina Wählby, Amjad Askary

TL;DR

SEFI addresses segmentation-free integration of nuclei morphology with spatial transcriptomics by extracting self-supervised morphological features from DAPI images and integrating them with density-based gene expression maps. The method uses a ResNet18-SimCLR CNN for morphology, Points2Regions for gene maps, and hierarchical merging of k-means clusters to define niches. On developing retina with multiplexed smFISH across 33 genes, SEFI improves clustering robustness, particularly when gene panels are reduced, by leveraging morphology to compensate for missing transcripts. The resulting clusters correspond to distinct retinal regions and known cellular markers, demonstrating meaningful, biologically relevant niche discovery in imaging-based spatial transcriptomics.

Abstract

This study introduces SEFI (SEgmentation-Free Integration), a novel method for integrating morphological features of cell nuclei with spatial transcriptomics data. Cell segmentation poses a significant challenge in the analysis of spatial transcriptomics data, as tissue-specific structural complexities and densely packed cells in certain regions make it difficult to develop a universal approach. SEFI addresses this by utilizing self-supervised learning to extract morphological features from fluorescent nuclear staining images, enhancing the clustering of gene expression data without requiring segmentation. We demonstrate SEFI on spatially resolved gene expression profiles of the developing retina, acquired using multiplexed single molecule Fluorescence In Situ Hybridization (smFISH). SEFI is publicly available at https://github.com/eduardchelebian/sefi.

Segmentation-free integration of nuclei morphology and spatial transcriptomics for retinal images

TL;DR

SEFI addresses segmentation-free integration of nuclei morphology with spatial transcriptomics by extracting self-supervised morphological features from DAPI images and integrating them with density-based gene expression maps. The method uses a ResNet18-SimCLR CNN for morphology, Points2Regions for gene maps, and hierarchical merging of k-means clusters to define niches. On developing retina with multiplexed smFISH across 33 genes, SEFI improves clustering robustness, particularly when gene panels are reduced, by leveraging morphology to compensate for missing transcripts. The resulting clusters correspond to distinct retinal regions and known cellular markers, demonstrating meaningful, biologically relevant niche discovery in imaging-based spatial transcriptomics.

Abstract

This study introduces SEFI (SEgmentation-Free Integration), a novel method for integrating morphological features of cell nuclei with spatial transcriptomics data. Cell segmentation poses a significant challenge in the analysis of spatial transcriptomics data, as tissue-specific structural complexities and densely packed cells in certain regions make it difficult to develop a universal approach. SEFI addresses this by utilizing self-supervised learning to extract morphological features from fluorescent nuclear staining images, enhancing the clustering of gene expression data without requiring segmentation. We demonstrate SEFI on spatially resolved gene expression profiles of the developing retina, acquired using multiplexed single molecule Fluorescence In Situ Hybridization (smFISH). SEFI is publicly available at https://github.com/eduardchelebian/sefi.

Paper Structure

This paper contains 7 sections, 4 figures.

Table of Contents

  1. Introduction
  2. Method
  3. Experiments and results
  4. Data description
  5. Morphological features improve clustering
  6. Biological relevance of the clusters
  7. Original images

Figures (4)

  • Figure 1: Steps of SEFI, from left to right, starting with creating density maps for each gene, and morphological maps by learned feature extraction. The dimensions of the morphological maps are reduced by PCA. Finally, niches are detected by k-means pixel clustering and hierarchical cluster merging.
  • Figure 2: Morphological information can compensate for information lost when including fewer genetic markers for niche detection.
  • Figure 3: Results from applying the full SEFI pipeline on the 33 genes and the morphology
  • Figure 4: Original DAPI images for the three samples