Research Article
Segmentation Mammograms based on Level Set for Detection of Breast Cancer as a Second Opinion Radiologist
@INPROCEEDINGS{10.4108/eai.11-12-2019.2290847, author={Endang Supriyati and Tutik Khotimah and Mohammad Iqbal and Tri Listyorini and E Evanita}, title={Segmentation Mammograms based on Level Set for Detection of Breast Cancer as a Second Opinion Radiologist}, proceedings={Proceedings of the Third Workshop on Multidisciplinary and Its Applications, WMA-3 2019, 11-14 December 2019, Medan, Indonesia}, publisher={EAI}, proceedings_a={WMA-3}, year={2020}, month={3}, keywords={segmentation mammography level set classification}, doi={10.4108/eai.11-12-2019.2290847} }- Endang Supriyati
Tutik Khotimah
Mohammad Iqbal
Tri Listyorini
E Evanita
Year: 2020
Segmentation Mammograms based on Level Set for Detection of Breast Cancer as a Second Opinion Radiologist
WMA-3
EAI
DOI: 10.4108/eai.11-12-2019.2290847
Abstract
Detection of signs of cancer using mammograms is a difficult job, this is due to a pathological disorder and noise structures that appear in the image.A mammogram is an X-ray image of the breast that can reveal abnormalities at an early stage. Problems in mammography screening is error a highrate.Early detection of cancer can reduce the death rate. Segmentation is used to separate objects from the background. Segmentation Mass separates the mass of the background and captures the contours of mass. Segmentation is the method of Binary and Gaussian Filtering Selective Regularized Level Set (SBGFRLS). From the test data, the average obtained by using segmentation SBGFRLS RMSE of 8.67, whereas with traditional segmentation level set at 11.4725. For the feature extraction method in this study using the Discrete Wavelet Transformation (DWT), grey Level Co-occurrence Matrix (GLCM), Gabor-Wavelet (GW). The next stage is the classification performed using the method of artificial neural network (ANN) - Lavenberg Marquard (LM). This research resulted in a classification test GW-JST, where the results are given for the better in tests using the training data. In addition, testing is also performed using the data-ANN GLCM testing, where the results of these tests are less stable. This problem is caused because the amount of data that diversification is applied on the size and structure of the mammary pathological.