ARTIFICIAL INTELLIGENCE FOR SEGMENTATION OF MIDFACIAL STRUCTURES ON CBCT IMAGES

The first and the most essential step in the digital workflow of the majority of digital dental workflows is known as segmentation, which involves the generation of 3D models of the dentomaxillofacial structures. Manual segmentation is not a feasible task in a daily clinical practice since it is a time-consuming task and requires high operator experience. The current commonly applied methodologies for segmentation are either thresholding - or template-based and semi-automatic in nature. These techniques are prone to certain limitations, such as missing thin bony structure, observer variability and need for manual refinement. 
Furthermore, the currently available dentomaxillofacial segmentation software programs have been optimized based on computed tomography (CT) data, which cannot be applied to cone-beam computed tomography (CBCT) scans due to the presence of uncalibrated absolute Hounsfield units (HU), beam hardening artifacts, noise, and low-contrast resolution. However, CBCT imaging has been widely employed in the field of oral and maxillofacial radiology for the visualization of orofacial structures, presurgical planning and follow-up assessment considering its low cost, relatively lower dose, and increased accessibility.
The midfacial complex holds a unique position in the workflows of orthognathic and reconstructive surgery, and orthodontics, for ensuring an accurate diagnosis, patient-specific treatment planning and follow-up assessment. It is one of the most difficult anatomical regions to segment with conventional approaches. The maxillary sinus is the largest of the four paranasal sinuses, lies in the body of the maxilla and surrounded by the bones of midfacial complex. An accurate 3D segmentation of the sinus is crucial for multiple diagnostic and treatment applications, where evaluation of sinus changes, remodelling at follow-up, volumetric analysis or creation of 3D virtual models is required. Furthermore, the most relevant surgical procedures requiring sinus
assessment include implant placement, sinus augmentation and orthognathic surgery.
Considering the limitations of the conventional segmentation methods, recent application of deep convolutional neural networks (CNNs) has outperformed the previously available algorithms for modelling of the dentomaxillofacial region. These CNNs have been successfully applied with promising results for the CBCT-based automated segmentation of the teeth, pharyngeal airway space, and mandible. However, a lack of evidence exists related to the CNN-based automated segmentation of the midfacial structures (bone and air components).