Robotic Ultrasound Makes CBCT Alive
The framework uses robotic ultrasound as a dynamic tool to correct static Cone Beam Computed Tomography (CBCT) images in real time. It employs a lightweight AI network, USCorUNet, trained on optical flow and photometric consistency to estimate tissue motion caused by respiration or surgery. This allows for the creation of deformation-consistent CBCT slices, improving guidance accuracy during robotic procedures.