Classification of cervical cancer cells using the K-nearest neighbor (KNN) method based on geometric feature extraction

Lentera Afrida Kusumawardani, Riries Rulaningtyas, Winarno Winarno

Research output: Contribution to journalConference articlepeer-review


Cervical cancer is a malignant tumor disease in the cervix of women. Cervical cancer is also the cause of death from cancer with the second-highest number of cases after breast cancer. A Pap smear test will be done to find out the next diagnosis when a woman is suspected of having cervical cancer. Early detection of cancer needs to be done to get the right treatment and diagnosis. Since manual diagnosis is error-prone and time-consuming, an automated system that uses a computerized method of image processing is found to be more accurate. The existence of computerized detection is expected to make it easier to use and produce a higher level of accuracy. This study uses an image processing computer obtained from cervical cytology images from the Herlev database at the Danish Hospital to classify cervical cells using a geometric feature extraction process as an input parameter into the classification testing process. The object used for feature extraction is the nucleus from the cervical cell image. The classification method used is k-Nearest Neighbor. Accuracy values, precision values, and good recall values were obtained from k-Nearest Neighbor with a value of k=1, with the results being 94.29%, 95.24%, and 94.29%, respectively.

Original languageEnglish
Article number030003
JournalAIP Conference Proceedings
Issue number1
Publication statusPublished - 16 Aug 2023
Event11th International Conference on Theoretical and Applied Physics: The Spirit of Research and Collaboration Facing the COVID-19 Pandemic, ICTAP 2021 - Virtual, Online, Indonesia
Duration: 27 Oct 202128 Oct 2021


  • Classification of Cervical Cancer Cells
  • Confusion Matrix
  • Extraction of Geometric Features
  • k-Nearest Neighbor


Dive into the research topics of 'Classification of cervical cancer cells using the K-nearest neighbor (KNN) method based on geometric feature extraction'. Together they form a unique fingerprint.

Cite this