A bi-stage technique for segmenting cervical smear images using possibilistic fuzzy c-means and mathematical morphology

Khaled A. Abuhasel, Chastine Fatichah, Abdullah M. Iliyasu*

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review

3 Citations (Scopus)

Abstract

Accurate detection of cervical cells in microscopic smear images is an integral part of image-based efforts for cervical cancer diagnosis. The study presents a bi-stage technique to segment smeared cervical (CS) images. In the first stage, the Possibilistic Fuzzy C-Means (PFCM) algorithm, an appendage to the standard FCM algorithm that was developed mainly to address some weaknesses of the standard Fuzzy C-Means (FCM) algorithm in terms of its poor sensitivity to noise, is employed in order to segment the cervical cell into clusters. Following this, in the second stage, the cervical cell components, i.e., nucleus and cytoplasm, are detected and delineated using the mathematical morphological operations. Using a dataset of cervical smear images, the proposed technique achieves average values of 0.98, 0.98, 0.97, and 0.93 for sensitivity, specificity, accuracy, and Zijdenbos Similarity Index (ZSI) respectively, for the segmented nucleus. Similarly, for the segmented cytoplasm values of 0.89, 0.96, 0.91, and 0.95, respectively are obtained for the same parameters. These experimental results suggest that the PFCM algorithm obtains higher average ZSI values than the standard FCM algorithm for both the segmented nucleus and cytoplasm indicates the potential application of the proposed study in cervical cancer diagnosis.

Original languageEnglish
Pages (from-to)1663-1669
Number of pages7
JournalJournal of Medical Imaging and Health Informatics
Volume6
Issue number7
DOIs
Publication statusPublished - Nov 2016

Keywords

  • Cervical Smear Images
  • Clustering Algorithm
  • Disease Diagnosis
  • FCM
  • Mathematical Morphology
  • Medical Image Processing
  • PFCM

Fingerprint

Dive into the research topics of 'A bi-stage technique for segmenting cervical smear images using possibilistic fuzzy c-means and mathematical morphology'. Together they form a unique fingerprint.

Cite this