UCLA

Ultrasound imaging systems provide a low-cost, real-time, noninvasive and safe way to examine soft tissues inside the human body. Yet the ability of medical practitioners to understand and mentally register two-dimensional (2D) ultrasound image slices within the three-dimensional (3D) anatomy is a difficult task, so training is needed. Current ultrasound training methods are expensive, inefficient, and pose a major obstacle to the wide deployment of ultrasound imaging systems in routine clinical practice. In this thesis, we present a new approach to ultrasound training, where complex and expensive phantoms are replaced by a 3D virtual patient model, which represents the anatomy of any desired body part or organ and is simulated on a standard laptop computer. The advantage of our system is not only its cost effectiveness, but also its ability to emulate different disease states or conditions in different virtual patients and to visualize the underlying body structures of interest through different examination procedures with a virtual ultrasound probe.

Conventional 3D models of the human body are purely geometric and do not model soft-tissue mechanics and deformation, which is an important factor in the practice of clinical ultrasound imaging. To address this limitation, we introduce real-time interactive soft tissue simulation in our 3D patient model. For this purpose, we adapt and evaluate two well-known deformable model simulation methods: mass-spring-damper systems (MSDS) and the finite element method (FEM) with a quasistatic solution of isotropic linear elastic materials with Cauchy strain. We apply these methods to the simulation of ultrasound in soft tissues. The soft tissue model in its undeformed state is determined by static real-patient data captured by applying a linear ultrasound probe on the neck and on the left upper arm. A visual tracking system is used to control the virtual probe. We achieve real-time interactive simulation rates by carefully adapting the code to run efficiently on multicore personal computers.

Our real-time, interactive simulation of a 3D virtual patient with deformable soft tissues enables cheaper, more efficient, and more effective ultrasound training. Our ultrasound training system promises to facilitate the broader use of ultrasound in healthcare and reduce the number of medical procedure complications.