Label-Free Deep Profiling of the Tumor Microenvironment

Area of Science:

• Biomedical Imaging
• Computational Biology
• Cancer Research

Background:

• Label-free nonlinear microscopy offers nonperturbative visualization of cellular and tissue structures.
• Understanding the tumor microenvironment (TME) is crucial for cancer diagnosis and treatment.
• Current methods often lack the resolution to analyze cellular and extracellular vesicle (EV) interactions within the native TME.

Purpose of the Study:

• To develop a computational pipeline for quantitative analysis of the TME using label-free nonlinear microscopy.
• To enable single-cell, single-EV, and cell-to-EV analysis within cancerous tissue.
• To characterize the architectural and metabolic complexity of the TME.

Main Methods:

• A multiclass pixelwise segmentation neural network was employed to segment individual cells (tumor and stromal) and EVs.
• Label-free nonlinear microscopy images were analyzed to assess metabolic status and molecular structure.
• A computational pipeline was developed for quantitative analysis of cell-EV neighborhoods.

Main Results:

• The pipeline successfully segmented individual cells and EVs for detailed analysis.
• Extensive tissue reorganization was observed in the tumor microenvironment compared to normal tissue.
• A distinct, patterned cell-EV neighborhood was identified in cancerous tissue, highlighting TME heterogeneity.

Conclusions:

• The developed computational framework enables quantitative, label-free analysis of the TME.
• This approach facilitates a deeper understanding of the complexity and heterogeneity of cancer tissue.
• The pipeline holds potential for advancing biomedical research in areas requiring single-cell, single-EV, and cell-to-EV analysis.