Doctor of Philosophy
Terry, Peters M.
Elvis, Chen C. S.
Ultrasound-guided needle insertions at the site of the internal jugular vein (IJV) are routinely performed to access the central venous system. Ultrasound-guided insertions maintain high rates of carotid artery puncture, as clinicians rely on 2D information to perform a 3D procedure. The limitations of 2D ultrasound-guidance motivated the research question: “Do 3D ultrasound-based environments improve IJV needle insertion accuracy”. We addressed this by developing advanced surgical navigation systems based on tracked surgical tools and ultrasound with various visualizations. The point-to-line ultrasound calibration enables the use of tracked ultrasound. We automated the fiducial localization required for this calibration method such that fiducials can be automatically localized within 0.25 mm of the manual equivalent. The point-to-line calibration obtained with both manual and automatic localizations produced average normalized distance errors less than 1.5 mm from point targets. Another calibration method was developed that registers an optical tracking system and the VIVE Pro head-mounted display (HMD) tracking system with sub-millimetre and sub-degree accuracy compared to ground truth values. This co-calibration enabled the development of an HMD needle navigation system, in which the calibrated ultrasound image and tracked models of the needle, needle trajectory, and probe were visualized in the HMD. In a phantom experiment, 31 clinicians had a 96 % success rate using the HMD system compared to 70 % for the ultrasound-only approach (p= 0.018). We developed a machine-learning-based vascular reconstruction pipeline that automatically returns accurate 3D reconstructions of the carotid artery and IJV given sequential tracked ultrasound images. This reconstruction pipeline was used to develop a surgical navigation system, where tracked models of the needle, needle trajectory, and the 3D z-buffered vasculature from a phantom were visualized in a common coordinate system on a screen. This system improved the insertion accuracy and resulted in 100 % success rates compared to 70 % under ultrasound-guidance (p=0.041) across 20 clinicians during the phantom experiment. Overall, accurate calibrations and machine learning algorithms enable the development of advanced 3D ultrasound systems for needle navigation, both in an immersive first-person perspective and on a screen, illustrating that 3D US environments outperformed 2D ultrasound-guidance used clinically.
Summary for Lay Audience
Central line insertions are used to access the central venous system require inserting a needle into a vein, such as the internal jugular vein (IJV) on the neck. Ultrasound-(US)-guidance is used to navigate the needle into the vein using 2D US images, but still results in high complication rates, such as carotid artery (CA) puncture, as clinicians rely on 2D information to guide a 3D procedure. The use of 3D US environments for surgical navigation is an active research area and this thesis focuses on the development of 3D US-based systems for needle navigation into the IJV. A surgical navigation system was developed that employed a spatial tracking system and provides tracked US images aligned with tracked models of the needle, needle trajectory, and US probe. Tracked US requires US probe calibration that enables the US image to be positioned and scaled to the field-of-view of the US beam, such that the interactions between the image and surgical tools can be visualized for the user. This system can be rendered on a monitor or within a head-mounted display (HMD) by means of a co-calibration apparatus that places the surgical navigation system described above in a first-person perspective in the HMD. The components required to develop this system were evaluated and found to be highly accurate. The surgical navigation system displayed on the monitor and in the HMD were compared to the US-guidance technique, and the HMD system significantly improved the needle insertion accuracy in a phantom study with 31 experienced clinicians compared to the US-guidance approach. As clinicians have a preference for monitor-based systems, a second surgical navigation system was developed, based on a method to automatically obtain 3D models of the CA and IJV given a tracked US recording that was developed in this work. In a phantom study with 20 experienced clinical practitioners a system that employed models of the 3D vessel models and surgical tools, delivered significantly improved the needle insertion accuracy, compared to the system that employed US-only guidance. Overall, 3D US environments outperformed US-guidance for needle navigation.
Groves, Leah, "Mixed-reality visualization environments to facilitate ultrasound-guided vascular access" (2021). Electronic Thesis and Dissertation Repository. 8151.