Artificial Intelligence Evaluation of Fillings

Eskisehir Osmangazi University4,323 enrolled

Overview

The goal of this Non-Interventional Clinical Research is to detect the prevalence and distribution of filling and overhanging filling without the need for additional bitewing radiographs using panoramic images, based on a deep CNN (Convolutional Neural Network) architecture trained through supervised learning. In this study, retrospectively obtained radiographs were used in the development of artificial intelligence models for relevant situations. These datasets were obtained from the images of the patients who applied to ESOGU (Eskişehir Osmangazi University) Dentistry Faculty, Dentomaxillofacial Radiology clinic for various dental purposes. Eskisehir Osmangazi University Non-Interventional Clinical Research Ethics Board (decision date and decision number: 04.10.2022/22) approved the study protocol. The principles of the Helsinki Declaration were followed in the study.

Study Type

OBSERVATIONAL

Enrollment

4,323

Conditions

Dentomaxillofacial Radiology

Interventions

Panoramic RadiographyDIAGNOSTIC_TEST

this retrospective study includes analysis of radiographs previously taken from patients for various purposes

Eligibility

Sex: ALL

Medical Language ↔ Plain English

Inclusion Criteria: * Images of individuals in the permanent dentition period * Artifact-free images in the examination region * Individuals with a history of restorative dental treatment Exclusion Criteria: * Images of individuals in mixed dentition * Radiographic images obtained by incorrect positioning of the patient or containing artifacts

Outcomes

Primary Outcomes

The success of artificial intelligence models for filling and overhanging filling

It is obtained by calculating the sensitivity, precision, and F1 scores values for filling and overhanging filling.

Time frame: 1 year