Automatic generation of pedicle contours in 3D vertebral models

Area of Science:

• Medical Imaging
• Computer-Aided Surgery
• Deep Learning

Background:

• Accurate pedicle location is crucial for spinal morphology analysis and safe orthodontic screw implantation.
• The complex 3D structure of vertebrae presents challenges in precisely identifying pedicles from medical images.

Purpose of the Study:

• To develop an automated method for recognizing vertebral pedicles and generating pedicle contours using deep learning.
• To provide reliable surgical references for simulating pedicle screw implantation and preventing screw deviation.

Main Methods:

• Preprocessing individual vertebral models into point clouds.
• Utilizing a modified PointNet model for pedicle area segmentation.
• Fitting pedicle cross-sections and generating contours from segmentation results.

Main Results:

• The proposed method rapidly generates accurate pedicle contours with minimal manual adjustment.
• The generated contours serve as effective references for simulating screw placement.

Conclusions:

• The deep learning approach significantly improves the efficiency of generating pedicle contours for simulated screw placement.
• This method reduces the complexity of using surgical simulation systems, aiding in pedicle screw implantation and surgical planning.