PhD Thesis Defense: Ross Warner

Optional ZOOM LINK
Meeting ID: 928 8144 3597
Passcode: 589514

"Methods for Radiation-Sparing Navigation in Endovascular and Spine Surgery"

Abstract

Modern surgical navigation systems can enhance accuracy, reduce complications, and limit reliance on intraoperative ionizing radiation. Yet, their adoption remains uneven across surgical domains. This thesis develops radiation-sparing navigation methods in two areas where current guidance is costly, intermittent, or fundamentally limited: carotid endovascular surgery and lumbar spine surgery.

The first component focuses on the prerequisite step for any tracked-ultrasound navigation system: spatial ultrasound calibration. This work investigates calibration methods that improve robustness and usability, culminating in an image-based optimization technique that yields accurate spatial calibration. This approach provides a practical and adaptable foundation for using tracked ultrasound as a quantitative 3D imaging modality during surgery.

Building on this foundation, the thesis introduces a radiation-reducing navigation framework for carotid endovascular procedures. Spatially tracked ultrasound is used to reconstruct vessel anatomy and localize endovascular instruments without reliance on fluoroscopy. Phantom experiments demonstrate high registration and instrument-localization accuracy, and a complementary clinical data study evaluates methods to integrate preoperative information, such as vessel geometry and treatment planning, into the intraoperative environment. This establishes the technical underpinnings for a navigation strategy that can extend beyond the carotid artery to broader peripheral vascular applications.

Finally, the thesis investigates dynamic, non-radiative registration methods for open spine surgery, where navigation accuracy can degrade as the spine shifts during the procedure. Building on the concept of stereovision-based surface capture, this work explores a stereovision-to-stereovision registration paradigm that enables more frequent or on-demand updates to spinal alignment without repeated intraoperative CT scans. Deep learning–based techniques for automatic bone-surface extraction are evaluated to provide the consistent inputs required for reliable registration throughout the case.

Together, these contributions advance the development of radiation-sparing surgical navigation by improving ultrasound calibration, enabling fluoroscopy-free endovascular guidance, and establishing a pathway toward more dynamic, CT-free registration in spine surgery.

Thesis Committee

• Keith Paulsen (co-chair)
• Xiaoyao Fan (co-chair)
• Sohail Mirza, MD, MPH
• Richard Powell, MD
• Elvis Chen (Western University)

Contact

For more information, contact Thayer Registrar at thayer.registrar@dartmouth.edu.