Pengenalan Pola Jaringan Normal dan Jaringan Bermikrokalsifikasi pada Citra Mamografi Digital Menggunakan Support Vector Machines (SVM)

TRANSMISI Jilid 10, Nomor 4, Desember 2008 : 197-202, ISSN : 1411-0814


Abstract :

Features can be extracted from images based on pixel values. A classifier uses these image features to classify the image by the mean of pattern recognition. In this research, six features are extracted from mammogram image, i.e. normal mammogram and mammogram with microcalsification, based on pixel values. Because of the presence of microcalsification in mammograms can be an indicator of breast cancer, a pattern recognition of these two type of mammograms is important. In this research, Support Vector Machines (SVM) is used to recognize the pattern and classify the images based on six-feature input. The proposed method is developed and evaluated using 98 samples. All samples are preprocessed using linear scalling function and tophat transformation. The result show that the SVM best perfomance in recognition the microcalcification is 100%, using linear kernel and regularization parameter C = 100. 

Keywords: normal, microcalsification, features, linear scalling function, tophat, SVM

Full text : Download PDF  (732)

Back ...