Artificial intelligence‑assisted analysis of prevalence patterns and phenotypic distribution of dental pulp stones using cone‑beam computed tomography imaging

Abstract

Background: Artificial intelligence (AI) is increasingly being integrated into dental radiology to enhance diagnostic accuracy and support epidemiological research. AI‑driven techniques, such as K‑Means clustering and principal component analysis (PCA) offer novel approaches for analyzing cone‑beam computed tomography (CBCT) data.

Aims: This study applied AI‑based clustering techniques to data derived from CBCT images, in a sample of Saudi population, to identify pulp stone prevalence patterns, and evaluate the ability of these methods to reveal hidden patterns and stratify distributional phenotypes.

Materials and Methods: Retrospective analysis of 300 CBCT scans (150 males, 150 females) from archives of university dental hospital in Saudi Arabia was conducted. Pulp stones were manually identified, and the CBCT‑derived data were analyzed using Chi‑square tests and AI‑based clustering (K‑Means, PCA) to detect prevalence patterns and stratify patient phenotypes.

Results: Pulp stones were identified in 15.3% of patients involving 6.1% of teeth. Significant associations were observed with age (P = 0.027), tooth type (P = 0.036), and dental arch (P = 0.022), but not gender (P = 0.136). AI‑based clustering identified three clusters, severe, moderate, and mild, based on prevalence patterns, not captured by conventional analysis.

Conclusions: AI‑based clustering of CBCT‑derived data effectively revealed hidden patterns and stratified distributional phenotypes of pulp stones, enhancing epidemiological understanding and supporting personalized, and data‑driven decision‑making in dental radiology.