Grant: Idea to Innovation

Patients presenting with scoliosis or disk degenerative disease (DDD) can face significant lower back pain and limited mobility. Patients with severe cases of scoliosis or DDD will undergo spinal surgery involving the attachment of metallic rods to the spine with pedicular screws. Pre-operative or intra-operative planning of pedicle screw placement is a critical step, particularly for patients with pathological bone anatomy. However, planning is a complex and time-consuming task due to challenges such as abnormal vertebral angles and the need to prevent screw breaches in the bone. Automated planning holds the potential to significantly enhance surgical efficiency in the operating room (OR) and lower the risk of complications. We recently developed a tool automating pedicle screw trajectory and size planning using a combination of geometric models of vertebrae, generative AI, and optimization techniques. Using the patient's scan as input, the model predicts an initial plan which is then optimized plan based on weighted factors like screw length, width, and angulation. The model achieves a 92.9% accuracy in predicting screw trajectories compared to surgeon's plans on two separate datasets. Despite this, further development is required before commercialization and integration in the OR. In this proposal, we present one-year plan to improve the tool by integrating within automatic vertebra segmentation, adding compatibility with multi-axial screws and recent surgical strategies, as well as achieve real-time processing. These enhancements are necessary for seamless integration into surgical workflows. The current market has no existing solution offering fully unsupervised planning tools that account for surgeons' preferences. With the support of EOS Imaging, this proposal aims to fill this gap by automating the entire process, from segmentation to screw placement, incorporating local and global anatomical constraints. The tool promises to deliver safer, more accurate, and time-efficient planning, thus enhancing the surgical workflow and patient outcomes.