Artificial intelligence helps with digital smile design

Search Dental Tribune

Study evaluates whether AI can craft a universally aesthetically pleasing restoration

A study has found that, while the use of artificial intelligence in smile design requires further understanding of the relationship between facial landmarks and of human aesthetic perception, it can already be used satisfactorily for symmetrical faces. (Image: Login/Shutterstock)

ISTANBUL, Turkey: Digital smile design (DSD) programs have become instrumental in dental treatment, enabling dentists and patients to plan treatments in harmony with the entire face. A recent innovation in DSD is the integration of artificial intelligence (AI). A study in Turkey has compared aesthetic preferences among dental professionals, dentistry students and laypeople regarding smile designs created manually and using AI, considering factors such as sex, professional experience and specialty. The findings offer insights into the potential of AI in DSD.

For the study, four cases representing major smile design groupings were selected. They were chosen based on relationships between the trichion, glabella, subnasale and menton, these being crucial for smile designs, employing the concept of “facial flow”, which refers to the direction of facial structures. Case 1 showed a facial flow towards the right side, Case 2 showed a facial flow towards the left side, the nose and chin pointed in different directions in Case 3, and Case 4 was a symmetrical face.

Two smile designs were created for each case using Smile Designer App: one via AI mode and another manually. The app utilises the Microsoft Face API, a robust AI tool with facial recognition capabilities. The API identifies 68 facial landmarks essential for determining the patient’s facial type and appropriate tooth sizes, ensuring a precise and personalised treatment plan.

To gather perceptions on these designs, an online survey was conducted. The 807 participants were classified into three occupational groups—dentists, dentistry students and other professionals (laypeople)—and were asked about their professional knowledge, expertise in smile design and usage of a smile design program, in the case of dentists. They were then asked to choose whether they found the AI-created or the manually created design more attractive for each case.

The socio-demographic breakdown showed that the majority of the dentists had 0–4 years of experience and were general practitioners. Almost half the laypeople and over half the dentistry students were familiar with aesthetic smile design. Age, education and clinical experience did not influence aesthetic preferences.

For Cases 1–3, both dentists who used smile design programs and those who did not favoured the manually created designs. However, for Case 4, dentists who used a smile design program preferred the manually created design, whereas those who did not preferred the AI-generated design. For Case 3, orthodontists notably favoured the AI-generated design. The authors suggested that this might have been due to their familiarity with AI values or experience with treatments based on the landmarks used.

Aesthetic preferences varied significantly between all three occupational groups for Cases 1–3, but notably not for Case 4. The authors suggested that dentists’ aesthetic perception may be different from that of laypeople in complex cases. There were significant differences between dentists and both dental students and laypeople for certain cases. The survey revealed an overall perceptual gap between dentists and laypeople, however, for symmetrical faces, AI-generated designs were acceptable to both dentists and laypeople, suggesting a potential time-saving tool for clinicians in such cases.

The challenge of incorporating AI lies in recognising that faces and smiles are not always symmetrical. Studies have found a link between facial symmetry and perceived beauty, but crucial landmarks used in smile designs on asymmetrical faces still need research for using AI. AI relies on mathematical models to create symmetrical smiles. However, because facial flow takes into account human perception, it allows for more natural smile designs for both symmetrical and asymmetrical faces. While AI-driven DSD works well for symmetrical faces, manual techniques may be better for asymmetrical cases. Given AI’s infancy in healthcare, there is potential for misinterpretations owing to algorithmic constraints. The authors thus recommended ongoing research to optimise AI in DSD and understand aesthetic perception.

The study, titled “Evaluating the facial esthetic outcomes of digital smile designs generated by artificial intelligence and dental professionals”, was published online on 6 August 2023 in Applied Sciences.

To post a reply please login or register