Effect of the CBCT data segmentation threshold on registration accuracy with surface scanning

Area of Science:

• Dental Imaging and Diagnostics
• Computer-Aided Implantology

Background:

• Accurate spatial alignment between CBCT and IOS data is critical for dental applications.
• Segmentation thresholds in CBCT processing can alter 3D bone model geometry, affecting registration accuracy.

Purpose of the Study:

• To evaluate the impact of varying CBCT segmentation thresholds on the accuracy of IOS registration.
• To determine if segmentation threshold choices influence the precision of 3D model alignment for dental procedures.

Main Methods:

• CBCT and IOS data from 19 patients were analyzed.
• Standard tessellation language (STL) models were created using seven CBCT segmentation thresholds (250-1750 HU).
• IOS data were aligned to CBCT-derived bone models, with deviations measured across X, Y, and Z axes.

Main Results:

• Segmentation thresholds significantly affected registration accuracy across all axes (P<.001).
• Deviations ranged from -0.08 to 0.33 mm (X-axis), 0.01 to 0.03 mm (Y-axis), and -0.13 to -0.66 mm (Z-axis).
• Higher thresholds correlated with increased Z-axis (apicocoronal) deviation.

Conclusions:

• CBCT segmentation thresholds substantially influence IOS to STL bone model registration accuracy, especially in the apicocoronal direction.
• Potential inaccuracies can compromise surgical and prosthetic planning, including surgical guide fabrication and implant positioning.
• Clinicians must consider segmentation threshold selection as a critical factor for accurate dental treatment planning.