Structure Tensor Based Analysis of Cells and Nuclei Organization in Tissues

Area of Science:

• Biomedical imaging
• Quantitative biology
• Image analysis

Background:

• Analyzing spatial organization of cells and nuclei in 2D/3D biomedical images is crucial for understanding morphogenesis.
• Microscopy images often suffer from low quality and large sizes, posing challenges for automated analysis.
• Existing segmentation methods rely on object-based modeling, which may not capture macroscopic tissue properties.

Purpose of the Study:

• To develop an automated method for analyzing the spatial organization of cells or nuclei in 2D and 3D biomedical images.
• To quantify nuclear orientation and the ratio of main axes lengths using a macroscopic image analysis approach.
• To apply the method to analyze nuclei orientation and anisotropy in multicellular tumor spheroids.

Main Methods:

• Utilized the structure tensor, a descriptor for analyzing texture orientation, to analyze spatial organization.
• Applied the structure tensor at a macroscopic scale, treating the sample as a continuous medium.
• Quantified privileged orientation and the ratio between the lengths of main nuclear axes.

Main Results:

• The structure tensor method successfully quantifies nuclear orientation and anisotropy in 2D and 3D images.
• Quantitative evaluation on synthetic and real images validated the method's robustness.
• Analysis of multicellular tumor spheroids revealed cells elongated parallel to the spheroid boundary.

Conclusions:

• The structure tensor provides a robust tool for quantifying geometrical information from large biomedical images.
• This method offers insights into tissue organization by analyzing nuclear orientation and shape anisotropy.
• Available software (MATLAB toolbox, Icy plugin) facilitates the application of this technique in biological research.