Comprehensive visual index of carious lesions

This specialized dataset consists of 6,377 high-resolution photographic images designed for training machine learning algorithms in dental caries detection and caries risk prediction. Core components of this dataset include the comprehensive assimilation of all three dental stages: primary (deciduous/ baby teeth), permanent (adult teeth), and mixed dentition (a combination of both). This diverse coverage captures the unique anatomical differences across various age groups. The dataset has Multi-View Standardization i.e. every subject is documented using 8 distinct anatomical views. This multi-angle approach ensures the machine learning models learn to identify carious lesions across multiple surfaces (such as occlusal, buccal, and front regions) while minimizing blind spots caused by reflections or positioning. Due to this coverage this dataset is apt for Machine Learning Application for Diagnostic and Predictive Testing. The primary utility of this dataset is to develop automated computer-aided design (CAD) systems. It serves as ground-truth data for classification, segmentation, and detection models aiming to identify early, moderate, and advanced dental caries. This will have a clinical impact on dentistry as by training models on this comprehensive image base, researchers can build non-invasive, objective diagnostic tools. These tools can support clinicians in accurate screening, tele-dentistry, and tracking caries progression over time.