Segmentation of whole cells and cell nuclei from 3-D optical microscope images using dynamic programming

Area of Science:

• Cell biology
• Biomedical imaging
• Computational biology

Background:

• Cell-to-cell communication is crucial for tissue development, function, and diseases like cancer.
• Quantifying molecules in intact tissues requires accurate segmentation of cells or nuclei in 3-D images.
• Current methods lack efficiency and accuracy for segmenting objects in complex 3-D tissue structures.

Purpose of the Study:

• To develop a reliable and accurate semi-automatic algorithmic method for segmenting cells or nuclei from 3-D tissue images.
• To overcome limitations of existing methods in segmenting objects within intact tissue specimens.
• To facilitate quantitative molecular analysis in biological research.

Main Methods:

• A semi-automatic algorithm utilizing dynamic programming (DP) for 2-D object delineation.
• A combined DP and combinatorial searching approach to identify object surfaces in adjacent planes.
• Interactive correction of segmentation errors for enhanced accuracy.
• Robustness tested against challenges like intermittent labeling, diffuse surfaces, and spurious signals.

Main Results:

• The algorithm reliably and accurately segments fluorescence-labeled cells and nuclei from 3-D tissue images.
• Segmentation accuracy is maintained despite variations in labeling and image quality.
• The method successfully segmented all cells, including irregularly shaped ones, across diverse biological tissue samples.
• Demonstrated high performance on various biological tissue types.

Conclusions:

• The developed semi-automatic method provides a significant advancement for segmenting cells and nuclei in 3-D tissue imaging.
• This technique enhances the ability to perform quantitative molecular analyses in intact tissues.
• The algorithm's reliability and accuracy support its application in studying tissue development, function, and disease processes.