Deep Learning-based Implant Placement Planning and Application

Abstract

Placement of the dental implant is the most optimal way to restore partially or completely edentulous patients, and its use is increasing steadily. For precise implant placement, it is essential to establish an implant placement plan according to the patient, such as the patient’s dental condition and the position of nerves and bones. However, due to a lack of automated technologies, clinicians had to plan dental implant placement manually. As a result, we propose fully automated implant placement planning generation and its application in this study. First, Chapter 1 introduces the purpose of our study and related works. In Chapter 2, for the first time, we proposed deep learning-based method for reconstruction of panoramic image from Cone Beam Computed Tomography (CBCT). A preprocessing method using segmentation networks to obtain segmented masks is introduced in Chapter 3. This method extracts the information required for implant placement from a dental panoramic image with low contrast. In Chapter 4, a method for generating the implant placement planning in the panoramic images is proposed. By designing an auxiliary module and objective function utilizing characteristics of dental panoramic images, the performance of the implant placement planning is improved. Finally, the design of an implant placement planning application from CBCT is covered in Chapter 5, which is used as a diagnostic tool with panoramic images in actual clinical practice. Since deep learning operation in CBCT requires a lots of computing costs, automated method with low computational cost is needed while taking the advantages of CBCT. Therefore, we designed cost-effective fully automated implant placement planning application in CBCT by integrating the proposed methodologies from the Chapter 2 to the Chapter 4. As a result, we expect that this research will support clinicians in making quicker diagnoses and provide patients with more understanding about their treatments.