Advanced Robotic ultrasound Imaging for ultraprecise Spine surgery

Low back pain (LBP) is one of the most prevalent symptoms of degenerative spine disease, affecting millions of people around the world. It significantly reduces patient quality of life and places a substantial burden on healthcare systems. While spinal surgery offers effective pain relief in severe cases, minimally invasive pedicle screw placement (MIPSP) has gained increasing attention due to its reduced invasiveness and faster recovery times.  Advancements in medical navigation and robotics are being explored to move towards MIPSP techniques. However, current computer-assisted techniques like fluoroscopy expose patients and surgeons to cumulative radiation. Therefore, it is important to explore non-radiative intraoperative navigation solutions. This doctoral work explores a radiation-free approach to MIPSP, addressing the limitations of current image-guided techniques. The research focuses on developing and validating a robot-assisted ultrasound (US) system for 3D spine reconstruction and surgical intervention. To enhance the accuracy of 3D reconstruction, a US image calibration approach was developed and validated using several dedicated calibration phantoms. This calibration approach was compatible with robotic arms and various tracking systems, ensuring broad applicability. Subsequently, a robot-assisted US navigation system was developed. A hybrid control was developed to maintain constant scanning force and improve US image quality. Subsequently, the preoperative surgical plan was registered to the surgical site by using the 3D US reconstruction. However, it is challenging to register preoperative CT and intraoperative US data due to missing anatomical features on the US reconstruction. To address this, a deep learning-based registration approach was developed. This approach employed shape completion algorithms to reconstruct occluded bone structures from partial US data, facilitating robust and efficient registration. Finally, this doctoral work investigated the feasibility of combining US navigation with a robotic arm for accurate screw placement. The system autonomously positioned the drill along a pre-planned trajectory and executed the drilling procedure. Experimental validations were conducted on synthetic phantoms and ex-vivo human spines. The results reported an 87.5% success rate according to the Gertzbein-Robbins grade in clinical practice. This proposed system provides a radiation-free navigation approach to MIPSP. The developed technology has the potential to improve patient outcomes, reduce healthcare costs, and increase operating room efficiency.