AUTOMATIC DETECTION AND FEATURE SELECTION TO CLASSIFY THE BREAST LESIONS ON ULTRASOUND IMAGE USING MORPHOLOGICAL FEATURES
Journal of Critical Reviews · 2020
Breast lesion identification as early is a dynamic role to reduce the death rate. Mammography is a technique for breast cancer finding, which are harmful, and can be embarrassing for the younger women. In this paper focusing on automatic detection and feature selection to classify the breast lesions from ultrasound images by using Morphological features. Here filtering technique is used for the reduction of noise in the image before automatic detection. An active contour procedure is used for lesion segmentation. From ROI of Tetrolet filtered and Input Image morphological features are extracted. These features are reduced by feature selection algorithm then given to the SVM Classifier. From the results observed that the SVM classifier with Tetrolet morphological features outperforms than without filter morphological features with 85.45 % accuracy, 85.21% sensitivity and specificity of 88.89%.