Comparison of Low-Dose, Standard-Dose, and High-Definition CBCT Protocols for STL Reconstruction Accuracy in Digital...

Accurate three-dimensional reconstruction of bone defects from cone-beam computed tomography (CBCT) datasets is a critical step in digital workflows for custom scaffold fabrication in guided bone regeneration (GBR). Although low-dose (LD) acquisition protocols are recommended to comply with ALARA (as low as reasonably achievable)/ALADA (as low as diagnostically acceptable) principles, concerns persist regarding their potential impact on STL geometric fidelity. Evidence comparing surface trueness across different CBCT acquisition protocols remains limited. This study aimed to evaluate and compare the dimensional accuracy of STL models derived from CBCT datasets acquired using six imaging protocols, including three LD and three standard-dose (SD) configurations, each tested at 0.150, 0.200, and 0.400 mm voxel resolutions.

Standardized cortical bone defects were prepared on fresh porcine scapulae. A structured-light scanner was used to acquire high-resolution reference meshes. The specimen underwent six CBCT scans using predefined LD and SD settings. DICOM (digital imaging and communications in medicine) datasets were segmented with a standardized global threshold method and exported as STL files. Meshes were aligned to the reference model using iterative closest point (ICP) registration, and point-to-mesh deviation analysis was performed to assess surface trueness. Statistical comparisons were conducted using one-way ANOVA with Tukey's post hoc test (p < 0.05).

All protocols demonstrated high overall accuracy, with more than 95% of surface points deviating less than ± 0.4 mm from the reference. The LD 0.400 mm protocol showed the highest conformity, with over 60% of surface points within ± 0.1 mm. High-definition (HD) protocols (0.150 mm) exhibited greater local deviations despite finer nominal resolution. Differences were more pronounced in cortico-cancellous transition areas.

LD CBCT acquisition at 0.400 mm voxel size provides STL reconstructions with geometric fidelity comparable to or superior to higher-resolution settings, supporting its clinical use in scaffold-GBR while minimizing radiation exposure.

This study demonstrates that LD CBCT protocols can provide STL reconstructions with sufficient geometric accuracy for clinical use in scaffold-GBR. These findings support safer imaging practices without compromising digital workflow reliability, aiding clinicians in balancing radiation exposure with the precision required for personalized surgical planning.