Scholarlink

<p>Despite the considerable progress in dentistry, the automated detection and diagnosis of oral pathologies remains a major challenge. The great variability of the lesions, the subtleties of some of them, and the anatomical variations greatly complicate this task, even for high-performance AI systems. This article presents a three-step pipeline exploiting deep learning techniques to meet these challenges and enable automated detection of dental diseases from panoramic X-rays. The proposed pipeline uses different YOLO models (v5, v6, v7, v8, and v9) with variable sizes (nano, small, medium, large, and x-large) in order to segment the dental quadrants, accurately detect each individual tooth, and finally detect the main pathologies, such as caries or impacted teeth. The results obtained demonstrate the high potential of these YOLO models to overcome the limitations of previous approaches, thus paving the way for more reliable and efficient automated dental analysis solutions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

YOLO-based panoramic dental X-ray image analysis

  • 期刊论文

  • 发表日期：

    2025年9月23日

查看全文
  • Khalid Nassiri,
  • Moulay A. Akhloufi

摘要

Despite the considerable progress in dentistry, the automated detection and diagnosis of oral pathologies remains a major challenge. The great variability of the lesions, the subtleties of some of them, and the anatomical variations greatly complicate this task, even for high-performance AI systems. This article presents a three-step pipeline exploiting deep learning techniques to meet these challenges and enable automated detection of dental diseases from panoramic X-rays. The proposed pipeline uses different YOLO models (v5, v6, v7, v8, and v9) with variable sizes (nano, small, medium, large, and x-large) in order to segment the dental quadrants, accurately detect each individual tooth, and finally detect the main pathologies, such as caries or impacted teeth. The results obtained demonstrate the high potential of these YOLO models to overcome the limitations of previous approaches, thus paving the way for more reliable and efficient automated dental analysis solutions.