IDENTIFIKASI KEBERADAAN KANKER PADA CITRA MAMMOGRAFI MENGGUNAKAN METODE WAVELET HAAR
ABSTRACT: Breast cancer is the
most common kind of cancer suffered by women. Mammography has been a common
method for early detection of breast cancer. Recently mammograms are examined
manually, so it demands good knowledge, intuition, and experience in this
particular field. In many cases the breast normal tissue can hide malignant so
that it can’t b seen on the mammogram. With image processing tissue into
mammogram image can be effored to know location. Much methode are used in
digital image processing. Methode is used in this final task is texture
analysis. Based on that methode, this simulation program is made for
identification tissue into mammogram image using wavelet Haar methode. Data
about mammogram image any 42 image are get from Telogorejo Hospital Semarang.
This program simulation is started with reading image processing and then continued
to ROI (region of interest) process, in image from ROI used image enhancement
quality with median filter to strech the contrast, after that texture analysis
is used to get coefficient from that image. The classification is started with
the decomposition process to obtain the wavelet coefficients which then counted
the energy and entropy values of each images and then incorporated to database.
The next process is comparing the energy and entropy between images which will
be classified with the images on the database. The final step is to find
Euclidean distance to show that the tested images is one of the class on the
database. From the 42 sample observed, the testing result image after ROI and
enhancement show that it has recognition rate 86% and testing result without
image enhancement show recognition rate at 50%. The observed with using image
enhancement ang wavelet Haar from 14 normally image, 13 image can identified,
from 20 masses image, 15 can identified, from 8 microclasification image, 7 can
identified. The observed without image enhancement and wavelet Haar is using 10
image analyzed by doctor. from 2 normally image, 1 image can identified, from 6
masses image, 2 can identified, from 2 microclasification image, 2 can be
identified as microclassification.
Penulis: DANE KURNIA PUTRA,
IMAM SANTOSO, AJUB AJULIAN ZAHRA
Kode Jurnal: jptlisetrodd090099
