Semi-Automatic Features Extraction Of Cervical Cells

• Main Author: Mohanadas, Veerayen
• Format: Monograph
• Language: English
• Published: Universiti Sains Malaysia 2005
• Subjects: T Technology; TK Electrical Engineering. Electronics. Nuclear Engineering
• Online Access: http://eprints.usm.my/57982/1/Semi-Automatic%20Features%20Extraction%20Of%20Cervical%20Cells_Veerayen%20Mohanadas.pdf; http://eprints.usm.my/57982/

This project is entitled ‘Semi-automatic Features Extraction of Cervical Cells’. The project is aimed to create a user friendly software which can be able to analyze Pap smear images via image processing. Cytological screening using the Pap smear test is the most effective strategy for the detection of precancerous state and consequent control of cervical cancer. Cytological samples that are taken from Pap smear test will undergo further analysis to detect the degree of abnormality of the cervical cells. The results of the abnormality of the samples can be inaccurate since some types of the medical images are blurring and highly affected by unwanted noise. Those bottlenecks in the medical images are believed that can be reduced via implementations of an adaptive fuzzy c-means (AFCM) and moving k-means (MKM) clustering techniques. These clustering techniques were used to segment the Pap smear images and later the features of the cells were extracted using region growing based feature extraction (RGBFE) technique. The performance of AFCM and MKM were analyzed based on the segmentation results of 6 Pap smear images. In overall, MKM was produced much better images than AFCM. Although the results have revealed that AFCM was suffering from centre redundancy and poor final centres in most of the cases, but it has also shown an advantage over MKM where AFCM was not sensitive to initial centres.