Atlas-based segmentation of temporal bone surface structures.

Area of Science:

• Medical Imaging
• Anatomical Segmentation
• Surgical Simulation

Background:

• Accurate segmentation of temporal bone anatomy is crucial for surgical training and planning.
• Current manual segmentation methods can be time-consuming and prone to variability.

Purpose of the Study:

• To develop a time-efficient automated segmentation approach for temporal bone surface structures.
• To enable the use of segmented anatomy in surgical simulation software and preoperative training.

Main Methods:

• An atlas-based segmentation approach was employed for key temporal bone structures (tegmen, sigmoid sulcus, auditory canals, posterior canal wall).
• The method utilized deformable registration of Gaussian-smoothed CT images and surface landmark refinement.
• Validation involved comparison with manual segmentation in 20 cadaver temporal bone CT datasets using quantitative metrics.

Main Results:

• Automated segmentation achieved high accuracy, with Mahalanobis distance < 0.232 mm for all structures.
• Average Hausdorff distance was consistently low (< 0.464 mm) for most structures.
• Volume similarity exceeded 0.80 for most structures, indicating robust performance.

Conclusions:

• The developed atlas-based automated segmentation is effective for temporal bone anatomy.
• This technique supports enhanced pre-surgical planning and realistic surgical training simulations.
• The approach offers a time-efficient and accurate alternative to manual segmentation.