Artificiële Intelligentie in de Prechirurgische Planning van Implantaatplaatsing

This PhD thesis investigates the application of artificial intelligence (AI) in presurgical dental implant planning, with a focus on both technical and clinical validation. The work is structured in three parts, beginning with a scoping review, followed by the creation of a virtual implant patient using AI-driven tools, and concluding with the validation of AI-assisted virtual implant placement. Part I provides a comprehensive overview of the current landscape of AI applications in implant dentistry regarding the preoperative phase. It maps existing research, highlighting the strengths and limitations of different AI approaches and identifying key challenges that need to be addressed to support clinical integration. This foundation guided the subsequent experimental work. Part II of the thesis focuses on the digital creation of the virtual implant patient. It includes the validation of deep learning models for segmenting dental implants on CBCT scans. Another key contribution is the implementation of an AI-based system for automated multimodal image registration, aligning CBCT and intraoral scans to support precise planning. Additionally, an innovative approach was introduced for segmenting prosthetic crowns in cases with high imaging artifacts, improving the robustness and accuracy of the registration process in challenging scenarios. Part III transitions from image preparation to virtual implant placement. It introduces an AI system capable of fully automated implant location and dimension selection and compares its performance to that of human experts. The findings demonstrate that the AI tool can achieve a level of accuracy comparable to experienced clinicians, while also delivering greater consistency and time efficiency. A preclinical validation study confirmed the clinical acceptability of the AI-generated plans, and a further clinical validation study was designed to compare AI-assisted planning with routine workflows conducted by specialists. The integration of AI into presurgical implant planning presents several significant advantages. It enhances time efficiency by streamlining the planning process, thereby allowing clinicians to dedicate more attention to direct patient care. Additionally, the use of AI promotes greater consistency in treatment planning by minimizing the variability often seen with human operators, contributing to more standardized and reliable outcomes. Crucially, AI demonstrates the potential to emulate expert-level decision-making in implant planning, which represents a major advancement in dental technology. However, it is important to emphasize that AI should serve as an assistive tool to support, not replace, clinical judgment. Final treatment decisions must always be made by qualified clinicians who oversee the planning process. Furthermore, this thesis provides validation of specific AI models applied to distinct stages and clinical scenarios, including dentulous patients for IOS-CBCT registration and virtual implant placement in cases involving missing mandibular premolars and molars. It is imperative to delineate key avenues for future research, encompassing the comparative evaluation of multiple AI systems, the incorporation of a broader spectrum of clinical scenarios (i.e. multiple missing teeth, compromised bone conditions, and implant placement within the esthetic zone) and the inclusion of diverse patient cohorts. Additionally, prospective multicenter clinical trials are warranted to rigorously assess and confirm the generalizability, accuracy, and clinical reliability of AI-driven technologies in implant dentistry.