Nucleus Segmentation Using Gaussian Mixture based Shape Models.

Area of Science:

• Cellular biology
• Biomedical imaging
• Computational pathology

Background:

• Accurate identification and segmentation of cell nuclei are crucial for quantitative analysis in microscopy.
• Existing methods often struggle with detecting faint nuclei or accurately separating touching or overlapping nuclei.

Purpose of the Study:

• To develop and validate a robust method for nucleus detection and segmentation in microscopy images.
• To improve the accuracy of nucleus region identification and the separation of connected nuclei.

Main Methods:

• Nucleus candidates identified using extrema in Laplacian-of-Gaussian space.
• Ellipse fitting and clustering used to eliminate weak candidates.
• Region of interest defined by combined local and global thresholding.
• Iterative merging and splitting based on nucleus shape modeling and boundary roughness measurement.

Main Results:

• The proposed method demonstrated superior performance in detecting nucleus regions.
• Significantly improved accuracy in splitting boundaries of connected nuclei compared to other techniques.
• Achieved higher scores across eight evaluation metrics in experimental comparisons.

Conclusions:

• The developed method offers a significant advancement in automated nucleus segmentation for biological image analysis.
• This technique provides a more reliable tool for researchers in fields requiring precise cell counting and morphological analysis.