Rahman Rubaba Binte, Tanim Sharia Arfin, Alfaz Nazia, Shrestha Tahmid Enam, Miah Md Saef Ullah, Mridha M F
American International University Bangladesh Kuratoli 408/1, Dhaka, Bangladesh.
Data Brief. 2024 Sep 21;57:110970. doi: 10.1016/j.dib.2024.110970. eCollection 2024 Dec.
This article presents a dental dataset for the improvement of research on deep learning-based detection and classification of dental diseases. The dataset is consisted of 232 panoramic dental radiographs, categorized into six major classes: healthy teeth, caries, impacted teeth, infections, fractured teeth, and broken-down crowns/roots (BDC/BDR). The images were collected from three renowned private clinics in Dhaka, Bangladesh, with the help of an experienced dental practitioner who ensured the confidentiality of patients and high-quality data acquisition using a 64-megapixel Android phone camera. To enhance the value of the dataset for machine and deep learning applications, we applied Contrast-Limited Adaptive Histogram Equalization (CLAHE) for image enhancement and augmented the data. The images were annotated using the CVAT tool and reviewed by dental experts. This benchmark dataset is publicly available and provides a valuable resource for researchers in artificial intelligence, computer science, and dental informatics to promote interdisciplinary collaboration and the development of advanced algorithms for dental disease detection.
本文介绍了一个牙科数据集，用于改进基于深度学习的牙科疾病检测和分类研究。该数据集由232张全景牙科X光片组成，分为六大类：健康牙齿、龋齿、阻生牙、感染、牙齿折断和牙冠/牙根破损（BDC/BDR）。这些图像是在孟加拉国达卡的三家知名私人诊所收集的，在一位经验丰富的牙科医生的帮助下，他确保了患者的隐私，并使用6400万像素的安卓手机摄像头获取了高质量的数据。为了提高数据集在机器学习和深度学习应用中的价值，我们应用了对比度受限自适应直方图均衡化（CLAHE）进行图像增强并扩充了数据。这些图像使用CVAT工具进行注释，并由牙科专家进行审核。这个基准数据集是公开可用的，为人工智能、计算机科学和牙科信息学领域的研究人员提供了宝贵的资源，以促进跨学科合作以及开发用于牙科疾病检测的先进算法。
