Optimizing CBCT acquisition for detecting non-displaced dentoalveolar fractures: effects of field of view and...

To evaluate the diagnostic accuracy of cone-beam computed tomography for detecting non-displaced dentoalveolar process fractures using an ex vivo sheep head model, and to compare acquisition protocols differing in field of view (FOV) and manufacturer-defined resolution mode.

Thirty sheep heads underwent baseline cone-beam computed tomography to confirm the absence of pre-existing fractures, followed by controlled mechanical loading. Fracture status (present/absent) was established by direct surgical inspection after imaging. Each specimen was scanned using four predefined protocols combining two FOV settings (8 × 16 cm and 18 × 16 cm) and two resolution modes (standard and high), with voxel size varying across protocols. Four blinded experts assessed fracture presence (binary decision) and confidence (five-point scale); repeat readings after one month were used to estimate intraobserver agreement. Diagnostic discrimination was assessed using the area under the receiver operating characteristic curve derived from confidence scores, while sensitivity and specificity were calculated from binary decisions.

All protocols showed high discrimination (area under the curve, 0.817-0.917), with the best performance for the 18 × 16 cm high-resolution protocol (sensitivity 83.3%; specificity 100.0%). Within the same FOV, high-resolution imaging improved discrimination compared with standard-resolution imaging; however, no meaningful difference was observed between the two high-resolution protocols. Interobserver agreement was substantial (Fleiss kappa 0.67-0.84) and intraobserver agreement was substantial to almost perfect (Cohen kappa 0.605-1.000).

Spatial resolution was the dominant driver of diagnostic performance and reader confidence for non-displaced dentoalveolar fractures, whereas enlarging the FOV provided limited additional benefit once high-resolution imaging was used.