TIST

TIST integrates whole-genome scale spatial transcriptomics with matched histopathological images to extract Markov random field (MRF)-based cellular features, construct a TIST-net combining gene expression and spatial coordinates, identify spatial clusters via random-walk strategies, and apply neighborhood smoothing to reduce technical noise and enhance spatial gene expression patterns.

Key Features:

• Integration of Multi-Modal Data: Combines whole-genome scale gene expression from spatial transcriptomics with high-resolution cellular phenotypic information from matched histopathological images.
• Histopathological Feature Extraction: Extracts cellular phenotypic features from histopathological images using a Markov random field (MRF)-based method.
• TIST-net Construction: Integrates extracted image features, gene expression data, and spatial location information into a unified graph termed TIST-net.
• Spatial Cluster Identification: Identifies spatial clusters (SCs) within the TIST-net using a random walk-based strategy to group regions with similar expression profiles and phenotypic characteristics.
• Enhancement of Gene Expression Patterns: Applies neighborhood smoothing to refine and denoise spatial gene expression patterns.

Scientific Applications:

• Spatial organization analysis: Characterizing the spatial organization of gene expression within tissue samples.
• Cancer research: Resolving tumor microstructures and cellular interactions in cancer studies.
• Developmental biology: Investigating tissue patterning and cell-state spatial dynamics during development.
• Tissue engineering: Assessing engineered tissue architecture and cellular organization.

Methodology:

Computational steps explicitly include integration of transcriptomic data with histopathological images, MRF-based feature extraction from images, construction of the TIST-net by combining image features with gene expression and spatial coordinates, random walk-based spatial cluster identification, and neighborhood smoothing of gene expression patterns.