Computed tomography (CT) and cone-beam computed tomography (CBCT) are essential imaging tools for visualization in cranio-maxillofacial (CMF) surgery, providing high-resolution, 3D anatomical data for diagnosis, surgical planning, and follow-up. CT offers broader anatomical coverage and soft tissue contrast, while CBCT provides detailed bone imaging at lower radiation doses. However, both modalities are prone to artifacts – beam hardening, scatter, motion, and metal interference – that reduce image accuracy. Optimization of scanning parameters and protocols is essential to balance diagnostic quality with radiation safety. In parallel, deep learning approaches such as convolutional and generative adversarial networks are being explored for artifact suppression and segmentation enhancement.

Aim. The aim of the study is to review and compare CT and CBCT to identify the most optimal scanning parameters for cranio-maxillofacial imaging, ensuring high diagnostic accuracy while minimizing radiation exposure and artifact impact.

Materials and methods. A systematic search of scientific studies was conducted in the PubMed, Scopus, IEEE Xplore, and Web of Science using keywords: CBCT optimization, CT artifact correction, cranio-maxillofacial imaging, and deep learning in CT / CBCT. Inclusion criteria: studies assessing scanning parameters, image quality, artifact correction techniques in CMF contexts. Clinical, in vitro, and ex vivo studies were included. In total, 85 papers were analyzed.

Results. Optimal parameters – voxel sizes of 0.075–0.125 mm for CBCT and slice thicknesses of 0.50–1.25 mm for CT – improved diagnostic accuracy and segmentation outcomes. CBCT was preferred for bone structures, while CT remained superior for soft tissue and trauma. Traditional correction methods showed Dice gains of 6–15 %. AI-based models demonstrated higher performance, reducing artifacts by up to 70 % and achieving Dice scores up to 0.95. However, clinical adoption remains limited due to regulatory and standardization barriers.

Conclusions. Optimizing scan parameters significantly improves diagnostic performance in CMF imaging. While AI-based artifact correction shows strong potential, integration into clinical workflows requires further validation and regulatory alignment.