Wisdom_DS

This collection includes 500 high-resolution panoramic dental radiographs along with their associated pixel-level segmentation masks and clinically approved classification labels. It aims to facilitate research in dental image analysis, computer vision, and medical imaging, particularly concentrating on automated segmentation and detailed diagnosis of third molar teeth.
The dataset is organized into three main components:
Segmentation/ – Contains paired data for semantic segmentation tasks:
Images/ – 500 full-HD panoramic dental radiographs (2440 × 1280 px) in JPEG format.
Image_Masks/ – Corresponding multiclass semantic segmentation masks in PNG format, aligned with each radiograph and following a consistent naming convention.
Classification Label/ – Includes a CSV file (e.g., Objective_Classification_Label.csv) providing detailed objective-wise clinical annotations for each radiograph. The labels cover multiple diagnostic criteria for 3rd molar assessment, including:
Archer’s classification
Occlusal surface detection
Winter’s classification
Level of eruption measurement
Pell & Gregory’s classification
Each label is encoded using predefined data label indices, enabling structured and reproducible classification tasks.
Scripts/ – Provides utility Python scripts for visualization, inspection, and modification of segmentation masks. A README file is included with instructions for using these tools.
By offering aligned radiographs, pixel-level segmentation masks, and structured diagnostic labels, this dataset supports a wide range of applications, including third molar localization, pathology assessment, surgical planning, and benchmarking of deep learning models for automated dental diagnosis. It provides a valuable resource for advancing AI-assisted clinical decision-making in dentistry.