Influence of the Depth of the Convolutional Neural Networks on an Artificial Intelligence Model for Diagnosis of Orthognathic Surgery

Abstract

The aim of this study was to investigate the relationship between image patterns in cephalometric radiographs and the diagnosis of orthognathic surgery and propose a method to improve the accuracy of predictive models according to the depth of the neural networks. The study included 640 and 320 patients requiring non-surgical and surgical orthodontic treatments, respectively. The data of 150 patients were exclusively classified as a test set. The data of the remaining 810 patients were split into five groups and a five-fold cross-validation was performed. The convolutional neural network models used were ResNet-18, 34, 50, and 101. The number in the model name represents the difference in the depth of the blocks that constitute the model. The accuracy, sensitivity, and specificity of each model were estimated and compared. The average success rate in the test set for the ResNet-18, 34, 50, and 101 was 93.80%, 93.60%, 91.13%, and 91.33%, respectively. In screening, ResNet-18 had the best performance with an area under the curve of 0.979, followed by ResNets-34, 50, and 101 at 0.974, 0.945, and 0.944, respectively. This study suggests the required characteristics of the structure of an artificial intelligence model for decision-making based on medical images.

Automatic feature type selection in digital photogrammetry of piping

“…ResNet18, ResNet34, ResNet50, and ResNet101 indicate that the efficiency and accuracy would not be improved by blindly increasing network depth with a limited quantity of data, as was also reported by Kim et al. (2021).…”

“…The data set for training and testing the compared networks was the same as the data set utilized in the ablation study, and the compared networks were fine-tuned based on the default settings. indicate that the efficiency and accuracy would not be improved by blindly increasing network depth with a limited quantity of data, as was also reported by Kim et al (2021). The FTSNet method produces the highest F1-score and prediction speed compared with state-of-art mod-…”

Deep convolutional neural network-based skeletal classification of cephalometric image compared with automated-tracing software

“…The diverse images might be associated with higher performance of the current DCNN model compared with that of the other DCNN models from the earlier studies as well as the AI software 6 , 8 . Proper neural network depth might be another factor leading to better performance, as observed in this study 4 .…”

“…In terms of orthodontic analysis and diagnosis, research is being increasingly conducted on DCNN systems based on dental x-ray images. Moreover, several software methodologies based on their own specific AI algorithms are already being effectively used 3 , 4 . There are two issues in deep learning studies using the cephalogram.…”

Effectiveness Of Cone-Beam Computed Tomography (CBCT)- Generated Cephalograms Using Artificial Intelligence (AI) Cephalometric Analysis

“…Cohen and Linney et al 13 in 1984, various studies to improve the automatic measurement point recognition accuracy have been reported, and most of the measurement points showed a high correlation with the measurement results of the examiner. The development of AI has signi cantly in uenced image analysis, particularly medical image analysis 14 . Several algorithms have been developed to automatically recognize these anatomical indicators using various AI models, and dentistry is no exception.…”