Automatic cytoplasm and nuclei segmentation for color cervical smear image using an efficient gap-search MRF

Area of Science:

• Medical image analysis
• Computational pathology
• Biomedical imaging

Background:

• Automated cervical cell analysis requires accurate segmentation of cellular components.
• Existing methods may lack efficiency or precision in segmenting nucleus, cytoplasm, and background.

Purpose of the Study:

• To develop a novel superpixel-based Markov random field (MRF) segmentation framework for cervical smear images.
• To accurately segment nucleus, cytoplasm, and background regions within cell images.

Main Methods:

• Utilized a superpixel-based Markov random field (MRF) segmentation framework.
• Employed an automatic label-map mechanism for region determination.
• Developed a gap-search algorithm to enhance model efficiency.

Main Results:

• The proposed framework achieved high accuracy in segmenting cervical cell components on real-world and public datasets.
• The gap-search algorithm significantly improved the speed of segmentation compared to traditional pixel-based and superpixel-based methods.

Conclusions:

• The novel superpixel-based MRF framework offers an accurate and efficient solution for cervical cell image segmentation.
• The developed gap-search algorithm enhances computational speed, making it suitable for automated analysis systems.