BREAST CANCER DETECTION TECHNIQUE BASED ON MULTI-SUBSPACE RANDOMIZATION AND COLLABORATION FEATURE SELECTION

Breast cancer is one of the common cancer among women and early detection of breast cancer helps in better treatment. Many researches have been conducted to detect breast cancer with high efficiency, but still it is challenging in providing classifier for efficient performance of the breast cancer detection. In this research, the Unsupervised Feature Selection with Multi-Subspace Randomization and Collaboration (MSRC) is proposed to improve the detection performance. The proposed MSRC method has the ability to explore the various subspaces features of the image. In the pre-processing process, the normalization and Adaptive Histogram equalization are applied to enhance the contrast of the image. Region growing and Otsu threshold segmentation are applied to select the neighborhood pixels in the image. The Kernal Fuzzy C-Means (KFCM) is used to select the features from the images. The advantage of using Region growing, Otsu threshold and KFCM method in segmentation is that provide clear edge segmentation based on intensity value. The Dual-Tree Complex Wavelet Transform (DTCWT), Weber Local Descriptor and GreyLevel Co-occurrence Matrix (GLCM) methods are used to extract features from the MRI breast cancer images. The combination of DTCWT, Weber Local Descriptor and GLCM has the advantage of extract the features based on histogram, gradient and orientation. The Support Vector Machine (SVM) classifier is used to detect the breast cancer in the image. The SVM provides the clear margin based on selected and extracted features and more efficient in high dimensional space. The experimental analysis shows that the proposed MSRC method