Pengekstrakan Ciri Secara Manual Sel Serviks (Manual Features Extraction Of Cervical Cells)

• Main Author: Jamal, Noor Zaihah
• Format: Monograph
• Language: English
• Published: Universiti Sains Malaysia 2005
• Subjects: T Technology; TK Electrical Engineering. Electronics. Nuclear Engineering
• Online Access: http://eprints.usm.my/58759/1/Pengekstrakan%20Ciri%20Secara%20Manual%20Sel%20Serviks%20%28Manual%20Features%20Extraction%20Of%20Cervical%20Cells%29_Noor%20Zaihah%20Jamal.pdf; http://eprints.usm.my/58759/

Digital image processing is a wide field and covers various fields including the medical diagnosis field. The objective of this project is to design a new software using Borland C++ Software, to extract features from cervix cancer cell which is the second most killer for women. This software is build to overcome the method which is used normally. Extraction method through conventional often encounters problem in differentiating between object and background because certain images are blur or consist a lot of impurities. Measurement through conventional method is just an assumption because here, ability of human vision is limited and the process takes a long time. The characterize that is being extract are size and grey level of nucleus and cytoplasm. Extraction characteristics can be carried out by implementing two methods which are segmentation and extraction. Segmentation is done using threshold technique whereas extraction is done based on region growing technique. Segmentation is apply to to increase the difference between nucleus, cytoplasm and its background before region growing is done against the segmented image. 60 images which were acquired from Hospital of University Science Malaysia were extracted using this software. From the correlation test which was been done between the data extracted using software and the data extracted through conventional, showed that designed software was able to extract the cell characteristic efficiently and confidently. Data that was extracted will be used by neural network to categorize the risk of cell. Hopefully with this software, diagnosis process for cervix cancer would be easier and eventually more life’s can be saved in the future.