Int J Performability Eng ›› 2021, Vol. 17 ›› Issue (11): 918-925.doi: 10.23940/ijpe.21.11.p2.918925

Previous Articles     Next Articles

An Investigation on Cervical Cancer with Image Processing and Hybrid Classification

Kavitha Ravindran*, Srinivasan Rajkumar, and Kavitha Muthuvel   

  1. Department of CSE, Vel Tech Rangarajan Dr.Sagunthala R&D Institute of Science and Technology, Chennai, 600062, India
  • Submitted on ; Revised on ; Accepted on
  • Contact: *E-mail address:

Abstract: Cervical cancer may be avoided by having frequent tests to detect and cure precancers. The Pap analysis evaluates the cells of the cervix for any unusual or precancerous alterations. Therefore, manual Pap smear testing in the microscope is arbitrary, with criteria that is difficult to repeat. In the cervical cancer diagnostic method, image processing of Pap smears is critical. Our cervical cancer diagnostic system has four main components. Nuclei were recognized through a shape-based adaptive approach and redundant cytoplasm was segregated utilizing a marker-control watershed technique in cell division. Three essential characteristics were recovered from the areas of fragmented cytoplasm and nuclei during the characteristics extraction process. As a component extraction technique, the RF algorithm was applied. A bagging clustering algorithm was used in the classification step, which integrated the findings of five different classifiers: support vector machine, bagged trees, linear discriminant, boosted trees, and k-nearest neighbor. The efficacy of our suggested method was demonstrated using the Herlev datasets and SIPaKMeD. According to the observational data, the two-class accuracy rate of 95.12 percent and five-class accuracy rate of 96.12 percent was considerably better in 2 and 5 class tasks.

Key words: classifier, random forest, cancer diagnostic method, image processing, SIPaKMeD, Herlev