Apr 22 2020
Scientists from the Finnish Center for Artificial Intelligence (FCAI) have designed a novel automatized method to localize mandibular canals.
Comparison of the model segmentation and the ground truth, from the secondary test data annotations, for a CBCT scan. For further explanation, see the research article in Nature Scientific Reports. Image Credit: FCAI.
To plan a dental implant operation as well as the implant position and size, dentists must know the accurate location of the mandibular canal—a canal situated on both sides of the lower jaw and comprising the alveolar nerve.
The lower jaw is an anatomically complicated structure, and specialists in the medical field utilize X-ray and computer tomography (CT) models to detect and diagnose such structures.
Normally, the position of mandibular canals is determined by radiologists and dentists manually from the CT or X-ray scans, which makes the task time-consuming and difficult. Hence, an automatized method to do this could render the work of the dentists and positioning of dental implants simpler.
To address this issue, scientists at the Finnish Center for Artificial Intelligence (FCAI), Tampere University Hospital, Planmeca, and the Alan Turing Institute have designed the latest model that shows the exact location of mandibular canals in an automated and precise way.
The model is based on training and making use of deep neural networks. The team used a dataset comprising 3D cone-beam CT (CBCT) scans to train the model.
Additionally, the model is based on a completely convolutional architecture, rendering fast and data-efficient to the maximum. Depending on the study outcomes, a deep learning model of this kind can localize the mandibular canals more precisely. The model transcends the statistical shape models, which have been the best-automatized technique to localize the mandibular canals until now.
In regular cases–when the patient does not present any unique conditions, like osteoporosis–the model acts as accurately as a human expert does. A majority of the patients who visit a dentist come under this category.
In more complex cases, one may need to adjust the estimate, so we are not yet talking about a fully stand-alone system.
Joel Jaskari, Study First Author and Doctoral Candidate, Aalto University, Finnish Center for Artificial Intelligence
The use of Artificial Intelligence has another evident benefit—the fact that the machine carries out the job equally fast and precisely all the time.
The aim of this research work is not, however, to replace radiologists but to make their job faster and more efficient so that they will have time to focus on the most complex cases.
Kimmo Kaski, Professor, Aalto University, Finnish Center for Artificial Intelligence
Planmeca, a Finnish company that develops, manufactures, and markets dental equipment, 2D and 3D software and imaging equipment, is in collaboration with FCAI. At present, the company is combining the current model with its dedicated software, for use with Planmeca 3D tomography equipment.
The findings of the study were published recently in the renowned publication series Nature Scientific Reports.