چكيده به لاتين
Breast Magnetic Resonance Imaging (MRI) acquisition assists the radiologist in detecting breast cancer. MR imaging produces a large number of images. Interpretation and evaluation of these images is time consuming and depends on radiologist’s experience. Thus, Computer-Aided Diagnosis (CAD) systems are introduced to assist the radiologist.
The main purpose of this study is to design a CAD system for detection and classification of lesions in MRIs. After atlas-based breast segmentation, mass and non-mass tumors are detected and segmented. Afterwards, 6 morphologic, 20 kinetic and 11 GLCM (Gray-Level Co-occurance Matrix) features are extracted from the tumors. In this study, a novel feature called Dual-Tree Complex Wavelet Transform (DT-CWT) is extracted along with other mentioned features and then applied to the classification step. Finally, feature vectors are fed to the Support Vector Machine (SVM) classifier with different kernels and combination of Linear Discriminant Analysis (LDA) and k-Nearest Neighbor (k-NN) to classify the lesions into malignant and benign classes. By applying the new feature to the mass lesions, non-mass lesions and combination of them and utilizing SVM classifier, the Az values of 0.71, 0.77 and 0.70, respectively. Also, by applying the hybrid classifier the Az values of 0.70, 0.74 and 0.69 are achieved, respectively.
Keywords: Breast Magnetic Resonance Image (MRI), Atlas-based segmentation, Dual-Tree Complex Wavelet Transform (DT-CWT), Support Vector Machine (SVM) classifier and hybrid classification.
