Optimising Insertion Axis Determination in intraoral scans - ON-1012
Project type: ResearchDesired discipline(s): Engineering - computer / electrical, Engineering, Dentistry, Life Sciences, Computer science, Mathematical Sciences
Company: iCAD Dental
Project Length: 6 months to 1 year
Preferred start date: 11/01/2024
Language requirement: English
Location(s): ON, Canada
No. of positions: 2
Desired education level: Master'sPhDRecent graduate
Open to applicants registered at an institution outside of Canada: No
About the company:
iCAD Dental is focused on transforming digital dentistry through advanced AI-driven solutions. We are currently developing automated 3D and AI-powered tools that streamline the processing of dental arches and enable the detection of key dental issues, such as improper tooth preparation, underbites, and overbites. By leveraging AI to analyze data from intraoral scanners, our goal is to provide dentists with the capability to perform real-time, chairside diagnostics. The development of innovative technologies led by iCAD Dental, aims to set new standards for patient care in dentistry.
Describe the project.:
Our project focuses on optimizing the automated process of finding the alignment axis for each tooth within a dental arch scan. The alignment axis is a crucial imaginary line that runs through the center of the tooth, from the root to the crown. It divides the tooth symmetrically when viewed from any side and points through the center of the occlusal face on top. Given that the alignment axis varies significantly among individuals, accurately determining it is a challenging research topic with critical applications in automating the design of orthodontic treatment and design of crown, dentures and aligners. Although our team has developed initial solutions, further research and optimization are needed to enhance both accuracy and runtime performance.
Given the nature of the research topic, we are open to acknowledging the candidate's contributions to the project as a potential part of their thesis. Furthermore, we are open to recognizing the candidate as a contributor to our forthcoming patents.
There are 2 primary tasks that must be solved to find the alignment axis:
Task 1: Segmenting the Gingiva from the Tooth Using Intraoral 3D Scans
The intraoral 3D scan files are a 3D mesh that describes the visible surface of the dental arch in the mouth. The task is to find some way to segment the mesh such that the triangles belonging to the tooth’s surface are separated from the triangles belonging to the gingival surface.
Additional data, such as vertex colour, vertex normals, and convexity, are available for consideration in the model. Other feature extraction techniques should also be explored to identify the most effective solution.
Task 2: Detecting the Occlusal Surface of a Tooth Using 2D Images
Using specially structured 2D mappings of the tooth, develop a 2D segmentation model. This model should accurately segment the occlusal surface (the top surface) of each tooth in both the maxillary and mandibular arches. The segmentation should cover all four categories of teeth: molars, premolars, canines, and incisors.
The classification of the tooth is known before making the prediction. It is up to the candidate to decide whether a single model that covers all tooth classifications is sufficient, or if multiple models, each tailored to a specific tooth classification, are necessary.
Required expertise/skills:
• Theoretical understanding of the structure of ply files
• Understanding of 3D computer Vision models
• Good experience with Python, Pytorch or Tensorflow
• Ability to process 3D data, in particular perform feature extraction on 3D data
• Comfortable in working with 3D data
