In this paper, we investigated the behavior of 22 co-occurrence statistics combined to six gray-scale quantization levels to classify breast lesions on ultrasound (BUS) images. The database of 436 BUS images used in this investigation was formed by 217 carcinoma and 219 benign lesions images. The region delimited by a minimum bounding rectangle around the lesion was employed to calculate the gray-level co-occurrence matrix (GLCM). Next, 22 co-occurrence statistics were computed regarding six quantization levels (8, 16, 32, 64, 128, and 256), four orientations (0° , 45° , 90° , and 135°), and ten distances (1, 2,...,10 pixels). Also, to reduce feature space dimensionality, texture descriptors of the same distance were averaged over all orientations, which is a common practice in the literature. Thereafter, the feature space was ranked using mutual information technique with minimal-redundancy-maximal-relevance (mRMR) criterion. Fisher linear discriminant analysis (FLDA) was applied to assess the discrimination power of texture features, by adding the first m-ranked features to the classification procedure iteratively until all of them were considered. The area under ROC curve (AUC) was used as figure of merit to measure the performance of the classifier. It was observed that averaging texture descriptors of a same distance impacts negatively the classification performance, since the best AUC of 0.81 was achieved with 32 gray levels and 109 features. On the other hand, regarding the single texture features (i.e., without averaging procedure), the quantization level does not impact the discrimination power, since AUC = 0.87 was obtained for the six quantization levels. Moreover, the number of features was reduced (between 17 and 24 features). The texture descriptors that contributed notably to distinguish breast lesions were contrast and correlation computed from GLCMs with orientation of 90° and distance more than five pixels.

This work aims at investigating texture parameters in distinguishing malign and benign breast tumors on ultrasound images. A rectangular region of interest (ROI) containing the tumor and its neighboring was defined for each image. Five parameters were extracted from the complexity curve (CC) of the ROI. Another five parameters were calculated from the grey-level co-occurrence matrix (GLCM) also for the ROI. The same was carried out for internal tumor region, hence, totaling 20 parameters. The linear discriminant analysis was applied to sets of up to five parameters and then the performances were assessed. The most relevant individual parameters were the contrast (con) (from the GLCM over the ROI) and the maximum value (mvi) from the CC just for the tumor internal region). When they were taken together, a correct classification slightly over 80% of the breast tumors was achieved. The highest performance (accuracy=84.2%, sensitivity=87.0%, and specificity=78.8%) was obtained with mvi, con, the standard deviation of the pixel pairs and the entropy, both for GLCM, and the internal region contrast also from GLCM. Parameters extracted from the internal region generally performed better and were more significant than those from the ROI. Moreover, parameters calculated only from CC or GLCM resulted in no statistically significant performance difference. These findings suggest that the texture parameters can be useful to help radiologist in distinguishing between benign or malign breast tumors on ultrasound images.

The segmentation method proposed was capable of delineating the lesion contours with high accuracy in comparison to both the radiologists' delineations and the true delineations of simulated images. Moreover, this method was also found to be robust to human-dependent parameters variations.

No abstract

No abstract