Ultrasound probe localization with respect to the patient's body is essential for freehand three-dimensional (3D) ultrasound and image-guided intervention. However, current methods for probe localization, such as optical or electromagnetic tracking, typically involve bulky and expensive equipment, and the resulting volume reconstruction could suffer from patient motion. In this thesis, a miniature-mobile and cost-effective system is described for 6-degrees-of-freedom (6-DOF) probe localization that is robust to rigid patient motion. In this system, each time an ultrasound image is acquired, skin features in the corresponding scan region are recorded synchronously by a lightweight camera rigidly mounted to the probe. Through visual simultaneous localization and mapping (visual SLAM), a skin map is built based on contrast-enhanced skin features and the probe pose corresponding to each ultrasound image is estimated. This estimate is then optimized in a Bayesian probabilistic framework that incorporates visual SLAM, a prior motion model, and ultrasound images. Motion errors from freehand scans on human bodies were quantified through comparison with independent measurement from a tri-camera tracking device.
Two applications of this system are demonstrated. One application is freehand 3D ultrasound, in which conventional 2D ultrasound images are accurately registered in space for 3D volume reconstruction. Through freehand scans on three body parts: lower leg, abdomen, and neck, it is shown that the synthesized re-slices generated by the system are structurally consistent with real ultrasound images. Probe re-localization with respect to an existing skin map by skin feature identification is also demonstrated and quantitatively evaluated. The second application is computer-guided ultrasound probe re-alignment for longitudinal studies, where an intuitive user interface was developed in conjunction with the proposed localization system to help sonographers return the probe to a target position and orientation. The overall outcome demonstrates the system's practical value and potential to be a highly cost-effective and mobile alternative for 6-DOF ultrasound probe localization.
Thesis Supervisors: Dr. Brian W. Anthony and Prof. Charles G. Sodini