Classification of DCE-MRI Data for Breast Cancer Diagnosis Combining Contrast Agent Dynamics and Texture Features

Classification of breast tumors via dynamic contrast-enhanced magnetic resonance imaging is an important task for tumor diagnosis. In this paper, we present an approach for automatic tumor segmentation, feature generation and classification. We apply fuzzy c-means on co-occurrence texture features to generate discriminative features for classification. High-frequency information is removed via discrete wavelet transform and computation is simplified via principal component analysis before extraction. We evaluate our approach using different classification algorithms. Our experimental results show the performances of different classifiers with respect to sensitivity and specificity.