Prediction of Tooth Wear
Conditions
Brief summary
Tooth wear, resulting from gradual loss of dental hard tissue due to mechanical and chemical factors, impacts tooth structure, texture, and function. It affects quality of life, with varying prevalence (26.9% to 90.0%), and is traditionally detected visually during check-ups, often at advanced stages. Monitoring alterations in tooth shape via intraoral scanners aids early detection, but restoration remains challenging. Prevention through early detection is vital, as patients may not fully comprehend tooth structure loss until visible. Recently, statistical shape analysis (SSA) used to learn the tooth anatomy and define a reference shape (biogeneric tooth) using. However, assuring landmark consistency is challenging mostly due to biases of the operator. Recently, a robust method called MEG-IsoQuad offered automated, isotopological remeshing. Combining this with SSA holds promise for diagnostic and simulation purposes. This study aims to assess the reliability of a remeshing-SSA approach for altered and intact premolar analysis and compare machine learning algorithms for simulating the shape of the initially intact tooth or future altered one. The clinical perspective of the current work offers possibilities to: * Prevent future tooth wear by detecting it at an early stage; and communicate better to the patient by presenting him/her potential future altered teeth * Simulate the adapted reconstruction for the altered tooth by simulating the initially intact one
Interventions
OTHERTooth Shape assessed using an intraoral scanner one after avulsion
* Tooth Shape assessed using an intraoral scanner one after avulsion and stored as StereoLithography (STL) file * Age, gender, reason of avulsion, type of tooth taken from the database and stored in a google sheet
Sponsors
Hospices Civils de Lyon
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to 80 Years
Healthy volunteers
No
Inclusion criteria
* teeth avulsed presenting a tooth wear index between 0 and 3 * mature incisor, canine, premolar or molars (1st and 2nd only)
Exclusion criteria
* teeth avulsed presenting a tooth wear index over 3 (or presenting an oral rehabilitation representative of a similar wear) * immature teeth or teeth without root edification * wisdom teeth
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Prediction of the tooth wear index based on a dataset of dental shapes:a retrospective study | only once | Four machine learning (ML) algorithms: a linear discriminant analysis (LDA) a support vector machine (SVM), a random forest (RM) and a gradient boosting machine (GBM) will be used to predict the tooth type and the alteration of the anatomy. The data set will be split into a 60/40 train and holdout test data set and models will be three-fold cross validated. Model performances will be evaluated in confusion matrices leading to define precision, recall, F1 score and accuracy. |
Countries
Belgium, France, India, Israel, United States
Contacts
Primary ContactRaphael Richet, Clinical Assistant
Backup ContactMaxime Ducret, Professor
Outcome results
None listed