The patent based on my Master’s thesis work at the Johns Hopkins University has been granted.

This patent introduces a deep-learning-driven system that dynamically adjusts the source trajectory of a C-arm Cone-Beam CT (CBCT) scanner for applications in surgical procedure verification. Traditional circular trajectories often yield poor reconstructions—especially when metal implants like spinal screws cause severe artifacts. Our model predicts, in real time, which next view will deliver the best image quality, steering the device into a more optimal, non-circular orbit suited to each patient and task.