Articles - Forum Implantologicum

Article

Feature Article

DE FR IT ES 日本語中文 TR PT

Aug 14, 2024 // 20 min read

4P Medicine and AI in Implant Dentistry

DOI: 10.3290/iti.fi.45736

artificial intelligence prediction personalization prevention participation

Abstract

4P medicine is envisioned as the future of medical practice, emphasizing prediction, personalization, prevention, and participation for precise patient management. Artificial intelligence (AI) technologies are rapidly transforming many medical fields and playing a crucial role in paving the way towards 4P medicine. Implant dentistry is relatively more digitalized compared to many other dental disciplines, using various advanced digital technologies throughout the entire treatment process. Most recently, a growing number of innovative AI-driven tools are being developed and validated in the field of implant dentistry. These AI tools may assist clinicians in achieving more accurate diagnosis, personalized treatment planning, precise guided surgery, and predictable treatment outcomes. This article provides a summary of AI tools related to dental implant treatment across various stages, from diagnosis to treatment and beyond, and discusses their possible contribution to advancing a predictive, preventive, personalized, and participatory approach in implant patient management.

Password

Introduction

For decades, many risk assessment tools have been widely used in determining management strategies for various dento-maxillofacial diseases and conditions. These tools primarily stratify individuals into different risk groups based on shared phenotypic characteristics. For example, risk factors such as smoking, diabetes mellitus, and poor oral hygiene have been used for risk assessment of periodontal and peri-implant diseases, and subsequently uniform preventative measures or treatment options are assigned to all individuals within a specific risk group (Schwendicke et al. 2022). True personalized care should tailor management strategies according to an individual's biological, social, and behavioral characteristics, aligning closely with the concept of “4P medicine" (Hung et al. 2023).

Introduced in 2010, “4P medicine” has been seen as a future form of medical practice that emphasizes the significance of the 4Ps in patient management (Auffray et al. 2010). The 4Ps stand for prediction, personalization, prevention, and participation. Prediction aims to assess an individual's susceptibility to specific diseases or conditions by analyzing genetic, lifestyle and environmental factors. This allows for early identification of potential health problems and the implementation of appropriate preventive measures. Prevention focuses on reducing the risk of diseases and health issues through lifestyle modifications, targeted interventions, and personalized health plans based on an individual's risk factors to maintain overall well-being. Personalization seeks to develop customized diagnostic, treatment and prevention strategies considering an individual's biological, social, and behavioral characteristics, enabling more precise and effective therapies with fewer side effects. Participation encourages patients to actively engage in their healthcare through patient-centered preventive measures and health monitoring (Bartold et al. 2022).

Currently, several innovative cutting-edge technologies, such as artificial intelligence (AI), robotics, virtual/augmented reality (VR/AR) as well as wearable and implantable smart devices, are being integrated into the healthcare industry. Among these disruptive technologies, AI has emerged as a driving force behind the advancement towards 4P dentistry (Hung et al. 2023). Recently, a wide range of AI tools has been developed across various dental disciplines to assist clinicians in diagnosis, treatment planning, outcome prediction, and patient management (Hung et al. 2022). These AI tools were primarily developed based on dental radiographic images using deep learning, particularly convolutional neural networks, as the training image data were readily available.

In implant dentistry, an increasing number of AI-driven tools has been developed to support clinicians at various stages from diagnosis and treatment planning to post-surgery and maintenance (Wu et al. 2024) (Table 1 & Fig. 1).

open_in_full

Table 1: AI-driven tools developed to assist clinicians at various stages of dental implant treatment

Fig. 1: A potential workflow for AI-assisted implant treatment

Table 1 Fig. 1

Table 1: AI-driven tools developed to assist clinicians at various stages of dental implant treatment

Fig. 1: A potential workflow for AI-assisted implant treatment

Fig. 2: An example of fully automated superimposition of CBCT and intraoral scans using the Relu® Creator (Relu, Leuven, Belgium) platform

Fig. 3: An example of fully automated identification of teeth (green contours), crowns/bridge/implant (yellow contours), mandibular canals (purple contours), endodontically-treated canals (orange contours), and maxillary sinuses (blue contours) on a single panoramic image using DeepCare Dental AI (DeepCare, Los Angeles, USA)

Fig. 4: An example of fully automated segmentation of dento-maxillofacial anatomical structures (a), including teeth (b), maxillary complex (c), mandible (d), maxillary sinus (e), mandibular canal (f), and pharynx (g) on CBCT images using Relu® Creator (Relu, Leuven, Belgium) platform

Fig. 5: An example of AI-based automated segmentation of bilateral zygomatic bones of a patient on CBCT images using the deep learning model by Tao et al. (Tao, Yu, et al. 2023). The top (green areas in a) displays the automatically segmented zygomatic bones. The middle (b) demonstrates the surface distance errors between the automated and manual segmentations (prediction vs ground truth). The bottom shows the overlay of edges of the prediction and ground truth

Fig. 6: An example of AI-based automated segmentation of a mucous retention cyst and mucosal thickening in the right maxillary sinuses of two patients on CBCT images using the deep learning model by Hung et al. (Hung, Ai, et al. 2022). The top row shows the original CBCT images. The second row shows the CBCT images with manually segmented sinus air space (green area), the cyst (yellow area), and mucosal thickening (orange area). The third row shows the segmentation of the cyst and mucosal thickening (grey area) by automation. The bottom row shows volume-rendered images illustrating the manually (yellow/orange area) and automatically (white area) segmented cyst and mucosal thickening

Fig. 7: An example of automated segmentation of the bone graft material on the maxillary sinus floor in immediate postoperative (green areas in a) and six-month follow-up (green areas in b) CBCT images, using the deep learning model by Tao et al. (Tao, Xu et al. 2023)

Fig. 8: Examples of inaccurate tooth labeling and segmentation from an AI platform. The tilted, shifted right lower third molar was mislabelled as the right lower second molar. The left upper and lower second premolars in contact were misidentified as the entirety of the upper second premolar, resulting in the lower second premolar being mislabelled as missing. The pulpal chambers and root canals of all teeth were not entirely segmented

Fig. 9: The applications of an AR-based dynamic navigation system developed by Fan et al. (2023) for guiding the placement of regular dental implants (a) and zygomatic implants (b)

Fig. 10: Examples of AI-generated crowns for central (a) and lateral (c) incisors using the convolutional neural network by Feng et al. (2023), along with their corresponding ground truth (b and d)