KLASIFIKASI CITRA MAMOGRAFI MENGGUNAKAN JARINGAN SYARAF TIRUAN

Abstract

There are abnormalities in breast tissue which can be detected by mammogram images analysis. One of those abnormalities is microcalcification. Microcalcifications are small calcium deposits in the breast tissue that can be seen only on a mammogram and can be an indicator of breast cancer. The main objective of this research is to automatically recognize the pattern of two types of breast tissues, i.e. normal tissue and breast tissue which contain microcalcifications in digital mammograms using Matlab software tools.  In this research, pattern recognition is carried out using Artificial Neural Network (ANN), i.e. LVQ (Learning Vector Quantization). The pattern recognition is formulated as a supervised-learning problem and classification was based on six-feature input given to the ANN. The system recognizes the pattern in three steps. Firstly, a tophat transformation is applied on the images, and then features of the images are extracted based on images pixel values. Finally, image classification is carried out in recognizing the pattern. The research uses 26 digital mammograms, consist of 16 normal mammograms and 10 mammograms which contain microcalcifications. The results show that the LVQ best performance in recognizing the pattern is 97%, using learning rate and decrement of learning rate equal to 0.1.