Table 8.

Advanced AI applications in restorative dentistry.

Authors	Summarized Abstract	Methods Used	Results	Conclusions
Zheng [65]	-Diagnosis of deep caries and pulpitis using convolutional neural networks.	-Assessed each CNN for accuracy, precision, sensitivity, specificity, and AUC. -Employed Grad-CAM to identify critical image features influencing CNN decisions.	-Integrated clinical parameters to enhance CNN performance. -Utilized Grad-CAM to identify important image features for CNNs.	-Multimodal CNN enhanced performance when integrated with clinical parameters.
Fontenele [64]	-Study evaluated AI tool performance for tooth segmentation on CBCT images.	-AI convolutional neural networks assessed segmentation performance for control and experimental groups.	-Dental fillings significantly influenced segmentation performance, showing high accuracy metrics. -AI-driven tool provided 3D tooth models from CBCT images efficiently.	-Dental fillings significantly influenced segmentation performance, but AI tool showed accuracy.
Schwendicke [63]	-Deep learning used for caries lesion detection in near-infrared light images.	-Resnet18 and Resnext50 were trained with data augmentation and 10-fold cross-validation, applying a one-cycle learning rate policy.	-Resnext50 model had a mean AUC of 0.74. -Model was sensitive to areas affected by caries lesions.	-Moderately deep CNN showed satisfying ability to detect caries lesions. -CNNs may assist in NILT-based caries detection in various dental settings.
Abdalla-Aslan [62]	-AI system detected and classified dental restorations on panoramic radiographs.	-Features related to shape and gray-level distribution were extracted and used to classify restorations into 11 categories via a trained algorithm.	-Algorithm detected 94.6% of restorations, with 93.6% classification accuracy.	-Machine learning algorithm excelled in detecting and classifying dental restorations.