PeriNet: Automated Multi-Task Deep Learning for Periapical Radiograph Analysis and Lesion Detection

Overview

Dental disease is one of the most underdiagnosed conditions worldwide. Apical periodontitis affects nearly half of adults, yet often remains asymptomatic until irreversible damage occurs. Early detection is critical for preventing bone loss, tooth loss, and costly complications. This project introduces a novel artificial intelligence system trained on 5,000 images, transcending conventional image analysis by embedding anatomical understanding into neural networks. By incorporating spatial priors, knowledge of where structures should exist, we enable detection of subtle diseases that standard models often miss. The system automatically analyzes periapical radiographs, identifies infection, quantifies bone loss, and evaluates periodontal ligament spacing for each tooth. The framework achieved a 0% miss rate for apical lesions and outperformed existing methods using 40% less training data. This work is among the first fully automated frameworks to convert dental radiographs into meaningful assessments, with applications in diagnostics, particularly in underdeveloped regions where dental specialists are limited.

Awards (2)

• Bronze Medal
• Selected for CWSF 2026

Competition history

• CWSF 2026 Disease & Illness

Related projects

Closest projects by meaning, across every fair and year in the corpus.