Breast Cancer Detection using Local Binary Patterns

Abstract

Breast cancer is a leading cause of cancer type for death among women in most of popular countries, breast cancer detection is important and challenging role in worldwide to save women's life. Due to inexperience to detect cancer, the doctors and radio logistic can miss the abnormality, which leads to death. Mammography is the most used method for breast cancer detection used by the radiologists. In this experiment, the MIAS (Mammogram Image Analysis Society) database is used and the MIAS database consists of normal and abnormal type of 322 mammograms. The pre-processing is most important step to capture quality mammogram image for next study and processing in mammogram analysis. Texture analysis plays important role to identify normal and abnormal types. Texture feature extraction can be done by local binary patterns (LBP) operator and by using LBP we can consider only sign parameters, it may loss the some texture information. The local binary pattern is a rotation invariant approach for the texture analysis. In this experiment famous completed LBP (CLBP) method used for extracting texture features. Completed LBP considering the sign, magnitude and centre gray level values. By using the joint or hybrid distributions combine CLBP_Sign, CLBP_Magnitude and CLBP_Center gray level values.LBP is one type of Completed LBP for texture analysis, advantage of CLBP is rotation invariant. Finally extracted texture features can be trained and classified by using the SVM classifier for identifying the normal and abnormal cancer type.

A Survey of Breast Cancer Detection and Classification Based on Texture Feature

“…Dependent on threshold parameter selection and size of the averaging filter Oliveira et al [5] DDSM database Texture characteristics Accuracy achieved -85% S. Naresh et al [7] MIAS database LBP features Results of CLBP are better than LBP Oliver et al [8] ---Eigen faces approach Result is A Z = 0.92. Peirrira et al [9] MIAS database LBP features ROC performance > 0.8…”

“…However, there is a dependency on manual selection of threshold parameter and size of the averaging filter. S. Naresh et al [7] proposes an approach for breast cancer detection, which uses completed Local Binary Pattern (CLBP) for texture feature extraction with support vector machine (SVM) for classification. The experiments are performed on MIAS database.…”

“…For this purpose, classifier like Support Vector Machine [4], [5], [7], [9], [10], Decision tree [8], KNN [8], thresholding [6] were use…”

“…Feature extraction can be described as extracting significant fine information from given input while rejecting all other data. LBP texture feature [4], [7], [9], [10], texture [2], [5], intensity value [11], eigenfaces approach [8], shape [6] feature were extracted from digital mammograms.…”

Classification of Breast cancer Using Local Binary Pattern and Gabor Filter

“…It is believed that the texture plays an important role in the visual system for recognition and interpretation of data. Local binary pattern [4], [7], [9], [10] and Gabor filter is used in proposed system to extract the texture features from the processed image.…”