A new annotation method for 3D cephalometric landmark in CBCT

In this study, we present a novel multi-landmark annotation system for cephalometric landmark annotation in 3D cone-beam CT (CBCT) images. Unlike previous annotation methods where only the axial view is visible for annotation, the new system provides users simultaneously three orthogonal views, i.e., axial, sagittal and coronal views. We compared the new system with the widely used Maeda´S method in terms of both annotation efficiency and accuracy. On a dataset containing 30 CBCT volumes, the proposed method demonstrated a clear advantage over the comparison.