RESUMO
PURPOSE: To develop and evaluate the accuracy of a computer-assisted system based on artificial intelligence for detecting and identifying dental implant brands using digital periapical radiographs. MATERIALS AND METHODS: A total of 1,800 digital periapical radiographs of dental implants from three distinct manufacturers (f1 = 600, f2 = 600, and f3 = 600) were split into training dataset (n = 1,440 [80%]) and testing dataset (n = 360 [20%]) groups. The images were evaluated by software developed by means of convolutional neural networks (CNN), with the aim of identifying the manufacturer of the dental implants contained in them. Accuracy, sensitivity, specificity, positive and negative predictive values, and the receiver operating characteristic (ROC) curve were calculated for detection and diagnostic performance of the CNN algorithm. RESULTS: At the final epoch (25), system accuracy values of 99.78% were obtained for group training data, 99.36% for group testing data, and 85.29% for validation data. The latter value corresponded to the actual accuracy of carrying out the system learning process. CONCLUSION: This study demonstrated the effectiveness of CNN for identifying dental implant manufacturers, which was proven to be a precise method of great clinical significance.