An Efficient Deep Learning Model for Breast Lesion Segmentation and Classification of Ultrasound Images

Breast cancer, a major life threatening disease, has been affecting more women nowadays. To effectively reduce the mortality rate, diagnostic accuracy has to be greatly improved. Various techniques, including deep and machine learning, are employed in open literature. For effective breast ultrasound image classification, a model incorporating a shearlet transform based image preprocessor, a ResNet-18 based image segmentator, and a convolution neural network based classifier is proposed in this work. This work makes use of a model based on deep learning for precise segmentation and classification of breast cancer. The proposed model exhibits a high accuracy rate in ultrasound image based breast cancer classification.