Detection and segmentation of cancer regions in ervical images using fuzzy logic and ANFIS classification method

Abstract

Cervical cancer is one of the most severe death cause cancers in newlinedeveloping countries. The mortality rates of the cervical cancer are high in newlinedeveloping countries due to their unawareness about such cancer. This type of newlinecancer can be cured if it is detected at an earlier stage, by detecting and newlineremoving the cancer affected regions in cervical regions. The severity of cervical newlinecancer can be categorized into Stage I to IV based on the cell affection with its newlinesurrounding. In case of stage I, the cells in cervix region are affected and it is newlinealso called as mild stage. In case of stage II, the cells outside the cervix regions newlineare affected and it is also called as moderate stage. In case of stage III, cancer newlinecells are spread in the pelvis and vagina region. In case of stage IV, the cancer newlinecells are spread into bladder or rectum region, which bleeds the blood. Stages III newlineand IV are called as severe stage. Death is occurred in stage IV. In this thesis, newlinethe cancer can be detected at stage I and II, so that the patient can be cured at an newlineearlier stage. newlineThis research work proposes an efficient Fuzzy logic and Adaptive newlineNeuro Fuzzy Inference System (ANFIS) classification method based cancer newlineregion detection and segmentation in cervical images. The thick and thin edges newlineare detected using fuzzy logic and these detected edges are fused pixel level newlineimage fusion technique. Then, Gabor transform is applied on the fused cervical newlineimage. The texture features are extracted from the Gabor transformed image and newlinethese features are classified using ANFIS classification approach. Further, the newlinemorphological operations are used to segment the cancer regions in classified newlineabnormal cervical image. newline