Table 9.

Overview of recent methodologies and reported performance in dental radiograph analysis

Research Study	Dataset Size	Imaging Modality	Clinical Focus	Computational Approach	Reported Performance
Jae-Hong Lee (2023) [18]	11,980 images	Panoramic	Dental implant system identification	Deep CNN with professional validation	95.4% accuracy
Vasdev et al. (2023) [19]	16,000 images	Mixed dental images	Multi-class dental disease detection	Pipeline: AlexNet, ResNet-18, ResNet-34	AlexNet: 85.2% accuracy
Muhammad Adnan Hasnain (2023) [20]	N/A	Not specified	Radiographic dental pathology classification	Transfer learning: ResNet-101, Xception, DenseNet-201, EfficientNet-B0	EfficientNet-B0: 98.91% accuracy
Chisako Muramatsu (2023) [21]	100 images	Panoramic	Tooth detection and classification for dental charting	Object detection network with 4-fold CV	93.2% classification performance
Kailai Zhang (2023) [22]	1,000 images	Radiographic images	Individual tooth detection and classification	Hierarchical label tree with cascade network	95.8% detection accuracy
W. Park (2023) [23]	150,733 images	Dental images	Implant system classification	Modified ResNet-50 with adaptations	82% accuracy
L. Toledo Reyes (2023) [24]	639 images	Clinical images	Caries progression prediction	Ensemble: Decision Trees, RF, XGBoost	AUC > 0.70
F. Schwendicke (2022) [25]	3,293,252 samples	Radiographic samples	AI-assisted caries detection cost-effectiveness	ML-based detection algorithms	80% diagnostic accuracy
Present Investigation	1,512 images (3,576 balanced samples)	Panoramic	Multi-condition classification: Fillings, Cavities, Implants, Impacted Teeth	Custom CNN, Hybrid CNN-ML, Pre-trained (VGG16, Xception, ResNet50)	Custom CNN: 74.29%, CNN + RF: 85.40%, VGG16: 82.23%