Segmentation of cell nuclei in tissue section analysis.

Area of Science:

• Digital image analysis
• Biomedical imaging
• Cellular morphology

Background:

• Accurate segmentation of cell nuclei in tissue sections is crucial for biological information extraction.
• Existing segmentation methods are often inadequate due to the optical properties of tissue section images.

Purpose of the Study:

• To develop novel gradient-based image segmentation methods for improved cell nucleus identification.
• To address the limitations of current techniques in segmenting cell nuclei within tissue sections.

Main Methods:

• Developed two new gradient methods for segmenting presegmented images.
• The first method divides nuclei images into eight segments with specialized gradient filters.
• The second method transforms images into a polar coordinate system (PCS) for gradient computation and contour finding, using nuclear shape as a priori information.

Main Results:

• The segment method enhances gradient images, enabling contour detection via local maxima.
• The PCS method provides precise segmentation and detailed morphological analysis of cell nuclei.
• Both methods account for image optical properties and the approximately circular shape of cell nuclei.

Conclusions:

• The developed gradient segmentation methods offer significant improvements over existing techniques for cell nucleus analysis in tissue sections.
• The PCS method is particularly effective for precise segmentation and morphological characterization.
• These advancements enhance the biological information obtainable from digital picture analysis of tissue samples.