2D Motion Tracking for Vascular Wall in Ultrasound Imaging

The analysis of vascular wall motion is crucial for identifying cardiovascular disorders. In this study, we have utilized Siamese neural networks for tracking 2D vascular wall motion in ultrasound imaging. Our Siamese network model was trained and tested using ultrasound data represented in I/Q format, which emulated the movement of a carotid artery. The simulated motions were composed of three cases - axial direction only, lateral direction only, and random motion in both axial and lateral directions. To evaluate the performance of the model, the mean squared errors (MSE) for both axial and lateral movements were calculated. The proposed Siamese model demonstrated successful tracking of vascular motion in both axial and lateral directions.