Aspects of artificial intelligence (AI) in dentistry

Abstract

Artificial intelligence (AI) is the science and engineering of machines that act intelligently, and it has been a hot topic over the past decade. A wide range of applications has shown to be improved by embedding AI into various industries. Some of the dental diagnostic software and tools using AI have been rolled out into market. However, dentistry is a teamed work that also involves dental technology, research and safety. For example, Computer-Aided Design and Computer-Aided Manufacturing (CAD/CAM) is gradually (about 30 years) replacing the traditional workflow into digital workflow. Although the current dental CAD/CAM systems have been advocated to be more advanced and claimed to have a higher accuracy and better efficiency compared with the traditional hand-crafted method, the ground truth is still human-dependant. On the other hand, in terms of dental materials research, bacterial adhesion to dental materials has been of great interest for decades. The colony-forming unit (CFU) is one of the most widely used methods to quantify the bacterial adhesion; however, this indirect measuring method is limited to detecting microorganisms that develop colonies on agar plates, and it is a time and labour consuming process. Thus, simple and versatile approaches to analyse the bacterial adhesion are vital. Further, in the COVID-19 pandemic era, a safe clinical environment would be essential for any dental practice. Nonetheless, the aerosol generating procedures, that are crucial in dentistry, are one of the inherent risks that might transmit coronavirus. A proper understanding of how the aerosols and droplets being dispersed in the clinic is indeed necessary. So, how can AI tap into these aspects of dentistry?

Four studies were conducted in this PhD project. In the first two studies, a novel AI algorithm with 3D-DCGAN for designing dental crowns was proposed and evaluated. Dental crowns designed by AI algorithm on premolars and molars were compared respectively to natural teeth, CEREC biogeneric design and technician design with the parameters of cusp angle, 3D volume discrepancy, occlusal contact point number and area, and in silico fatigue load. The results revealed that the AI algorithm can design a dental crown-mimicking natural tooth morphology, such that the performance of load outweigh than other designs. The latter two studies examined the application of AI image segmentation. In the third study, the AI tool was used to quantitatively measure the initial bacterial adhesion on scanning electron microscope images. To evaluate the efficiency of different dental suction systems in the COVID-19 pandemic, in the fourth study this AI tool was used to measure the number and area of aerosols/droplets produced by a high-speed dental handpiece powered by an electrical surgical motor. The AI tool was shown to be accurate and efficient in measuring and detecting for these purposes, able to find a new relationship, and can be an alternative method in evaluation of initial bacterial adhesion and dental aerosol/droplet measurement.

In conclusion, AI could be a useful and practical approach in solving dental problems such as crown design, bacterial adhesion measurement, and aerosol/droplet detection.