This systematic review and meta-analysis aimed to investigate the efficacy of fluorescence-based methods, visual inspections, and photographic visual examinations in initial caries detection. A literature search was undertaken in the PubMed and Cochrane databases. Preferred Reporting Items for Systematic Review and Meta-Analyses (PRISMA) guidelines were followed, and eligible articles published from 1 January 2009 to 30 October 2019 were included if they met the following criteria: they (1) assessed the accuracy of methods of detecting initial tooth caries lesions on occlusal, proximal, or smooth surfaces in both primary and permanent teeth (in clinical); (2) used a reference standard; (3) reported data regarding the sample size, prevalence of initial tooth caries, and accuracy of the methods. Data collection and extraction, quality assessment, and data analysis were conducted according to Cochrane standards Quality Assessment of Diagnostic Accuracy Studies-2. Statistical analyses were performed using Review Manager 5.3 and STATA 14.0. A total of 12 eligible articles were included in the meta-analysis. The results showed that the sensitivity and specificity of fluorescence-based methods were 80% and 80%, respectively; visual inspection was measured at 80% and 75%, respectively; photographic visual examination was measured at 67% and 79%, respectively. We found that the visual method and the fluorescence method were reliable for laboratory use to detect early-stage caries with equivalent accuracy.

In dental diagnosis, recognizing tooth complications quickly from radiology (e.g., X-rays) takes highly experienced medical professionals. By using object detection models and algorithms, this work is much easier and needs less experienced medical practitioners to clear their doubts while diagnosing a medical case. In this paper, we propose a dental defect recognition model by the integration of Adaptive Convolution Neural Network and Bag of Visual Word (BoVW). In this model, BoVW is used to save the features extracted from images. After that, a designed Convolutional Neural Network (CNN) model is used to make quality prediction. To evaluate the proposed model, we collected a dataset of radiography images of 447 patients in Hanoi Medical Hospital, Vietnam, with third molar complications. The results of the model suggest accuracy of 84% ± 4%. This accuracy is comparable to that of experienced dentists and radiologists.