Component-Based Modeling and Processing of Medical Ultrasound Signals

Component-based modeling and processing can be found in many signal and image processing applications. In this paper, we apply this approach to cardiac medical ultrasound imaging. We show how the raw data generated in the ultrasound scanning process can be modeled as consisting of two distinct components. By decomposing these signals into the two components, we are able to apply separate processing pipelines tailored to each component. The potential benefits of this approach to medical ultrasound imaging are demonstrated in two central issues of the commonly used processing: 1) the very large amounts of data generated by the scanning process, and 2) artifacts in the images and videos generated by standard processing, commonly referred to as side lobe artifacts. We manage to compress the data by a factor of over 20 as well as remove the side lobe artifacts. These benefits are demonstrated on both simulated data and real cardiac scanning data.