A Fuzzified Euclidean Framework for Accurate Cervical Cancer Cell Segmentation and Nuclei Detection

The accurate identification and segmentation of cervical nuclei in cell images are crucial for the early diagnosis of cervical cancer. However, challenges such as overlapping cells, uneven staining, and poor contrast often hinder this process. While traditional techniques demonstrate efficacy in segmenting clearly defined individual nuclei, they often fail to yield satisfactory results for adherent nuclei. To overcome the limitations, this paper introduces fuzzification-defuzzification process followed by Euclidean distance transformation (FEDT). Further, exponential dynamic growth analysis is done to improve the identification and segmentation of cervical nuclei in cell images. The model also incorporates exponential dynamic growth analysis to improve image quality and separation. Gaussian noise is added to enhance robustness, and EDT post-noise addition provides a framework for processing tasks. The method achieves a test accuracy of 93.75%, high resolution segmented nuclei, and precision and recall scores of 92.6% and 93%, enhancing cervical cancer early detection and diagnosis.