Localisation of the canal in CBCT images is complicated by anatomical variations in the course and shape of the canal according to individual and ethnicity. To avoid compression or other surgical complications, a safety margin of 2 mm above the mandibular canal is recommended in implantology. Precise knowledge of canal position is also important for various other oral and maxillofacial surgical procedures, such as jaw surgery or removal of third molars.

Researchers from Aalto University in Espoo, Planmeca and the Finnish Center for Artificial Intelligence (FCAI) developed a deep learning system and trained it with 3D images rendered with CBCT. The database consisted of images from five different CBCT scanners from four vendors and patient cohorts of two ethnicities—869 Finnish patients (79%) and 234 Thai patients (21%).

The performance of the deep learning system was clinically evaluated by comparing its results with those of four experienced dental and maxillofacial radiologists. The model accurately segmented the mandibular canal and performed better overall than the radiologists. In addition, it showed promising generalisability with regard to new CBCT scanners and ethnic groups.

“When a huge amount of data is fed to the neural network and the location of the mandibular canal is marked in it, it learns to optimise its own internal parameters. The neural network resulting from this learning quickly finds the mandibular canal from the individual 3D data input,” said co-author Vesa Varjonen, vice president of research and technology at dental equipment manufacturer Planmeca, which is based in Helsinki, in a press release.

“In clinical assessments, experts went through the results produced by the model and discovered that in 96% of the cases they were fully usable in clinical terms. We are highly confident that the model works well,” commented co-author Jaakko Sahlsten, a doctoral researcher at Aalto University.

“The collaboration arose from the needs of experts practising clinical work and from seeking ways to help their everyday work. A lot of time can be saved by using artificial intelligence in patient treatment planning,” said Varjonen.

“Tampere University Hospital provided us with extensive and versatile clinical materials produced with several 3D-imaging devices. The data was divided at random and part of it used for training the neural networks and part of it isolated for testing and validating the designed method,” said Sahlsten.

Planmeca to integrate the model in its imaging portfolio

For Planmeca, a Finnish family business and one of the world’s leading equipment manufacturers in health technology, the collaboration with FCAI and Tampere University Hospital means significant new business potential.

“Digitality and AI used in imaging equipment are important for us. We will integrate the neural network model developed in this research into our imaging software. This will improve the usability and performance of our equipment,” said Varjonen.

Model for orthognathic surgery

In addition, the collaborative research project developed a neural network model for orthognathic surgery. “The model helps to identify landmarks in the skull area for correcting malocclusion and planning jaw alignment surgery,” said Varjonen.

“I see artificial intelligence as a very powerful tool that physicians and other experts can use when making their first assessments or to get alternative opinions. The challenge with deep learning models is that we cannot give definite grounds as to why the model reaches a specific outcome. Further research is needed to increase the explainability and transparency of the models,” concluded Sahlsten.

The study, titled “Comparison of deep learning segmentation and multigrader-annotated mandibular canals of multicenter CBCT scans”, was published on 3 November 2022 in Scientific Reports.

Tags: