Differential Profile Diagram for Breast Tumor Classification Using Vibro-Acoustic Tactile System

We introduced an approach for breast tumor characterization with a Vibro-acoustic Tactile System (VTS), which is based on an ultrasound transducer for vibration and a tactile sensor for imaging. This system captures both tactile and vibro-tactile images of breast tumor phantoms, which are then integrated into two representative diagrams: the Tactile Profile Diagram and the Vibro-tactile Profile Diagram, respectively. These diagrams provide valuable touch-based information about the breast tissue and tumor characteristics. We quantify the differences induced by ultrasound vibrations by comparing pixel values between the Tactile Profile Diagram and the Vibro-tactile Profile Diagram, resulting in the development of Differential Profile Diagrams. Finally, we employ convolutional neural network (CNN) algorithms to classify these Differential Profile Diagrams into benign and malignant tumors with 86.4% classification accuracy.