Nucleus Segmentation for Cervical Cancer Screening with Pap Smears

Cervical cancer, the second most common cancer globally, is highly curable if detected early. However, rural areas face high mortality rates due to poor resources and limited screening programs. Automated diagnosis can address these gaps by distinguishing abnormal Pap smear cells based on nuclear shape. This study evaluates segmentation methods on the AGMC-TU Pap-Smear dataset, achieving a classification accuracy of 92.83% with SVM Linear and improving to 97.65% using optimized features and the FCM method. Accurate nucleus segmentation is crucial for reliable abnormal cell prediction, enhancing cervical cancer screening efficacy.