Periodontal Disease Prediction System

Overview

The purpose of this study was to develop a periodontal disease prediction software and a patient-based gingival recession simulator for clinical practice aiming at improving oral hygiene motivation of patients with periodontal problems.

Periodontal Disease Prediction (PDP) software has three components: a) Data Loading Window (DLW) b) Three-Dimensional Mouth Model (3DM) and c) Periodontal Attachment Loss Indicator (PLI). Demographic and clinical examinations of 1057 volunteers were recorded to DLW. An unsupervised machine learning K means clustering analysis was used to categorize the data obtained from the study population and identified the periodontal risk groups. An intraoral scanner was utilized to capture direct optical intraoral data of a patient and transferred to the 3DM. The intraoral model went under two algorithm steps for obtaining a recessed model. First, gingival curves separating gingiva and tooth were extracted using a Dijkstra's algorithm. Limit curves determining boundaries of recessed regions in the intraoral model were then obtained using gingival curves. The gingival recession was then mimicked by losing gingiva and disappearing tooth roots at proper locations in the intraoral model and Additionally, the final four different 3 dimensional recessed model (maxilla, mandibula, anterior teeth and posterior teeth) belonging to the same patient were scored based on the similarity to the real gingival recession by 25 periodontology specialist via online survey.

Conditions

Interventions

Eligibility

Locations (1)

Outcomes