RESUMO
INTRODUCTION: This study proposed an automatic diagnosis method based on deep learning for adenoid hypertrophy detection on cone-beam computed tomography. METHODS: The hierarchical masks self-attention U-net (HMSAU-Net) for segmentation of the upper airway and the 3-dimensional (3D)-ResNet for diagnosing adenoid hypertrophy were constructed on the basis of 87 cone-beam computed tomography samples. A self-attention encoder module was added to the SAU-Net to optimize upper airway segmentation precision. The hierarchical masks were introduced to ensure that the HMSAU-Net captured sufficient local semantic information. RESULTS: We used Dice to evaluate the performance of HMSAU-Net and used diagnostic method indicators to test the performance of 3D-ResNet. The average Dice value of our proposed model was 0.960, which was superior to the 3DU-Net and SAU-Net models. In the diagnostic models, 3D-ResNet10 had an excellent ability to diagnose adenoid hypertrophy automatically with a mean accuracy of 0.912, a mean sensitivity of 0.976, a mean specificity of 0.867, a mean positive predictive value of 0.837, a mean negative predictive value of 0.981, and a F1 score of 0.901. CONCLUSIONS: The value of this diagnostic system lies in that it provides a new method for the rapid and accurate early clinical diagnosis of adenoid hypertrophy in children, allows us to look at the upper airway obstruction in three-dimensional space and relieves the work pressure of imaging doctors.
Assuntos
Tonsila Faríngea , Aprendizado Profundo , Criança , Humanos , Tonsila Faríngea/diagnóstico por imagem , Tomografia Computadorizada de Feixe Cônico/métodos , Nariz , Hipertrofia/diagnóstico por imagem , Processamento de Imagem Assistida por Computador/métodosRESUMO
INTRODUCTION: The purpose of this study was to compare predicted anterior teeth intrusion measurements with the actual clinical intrusion measurements using cone-beam computed tomography. Understanding the precision of the software in anticipating changes may help practitioners predict the need for overcorrection. METHODS: Twenty-two patients, with a mean age of 23.74 years, who underwent Invisalign (Align Technology, Santa Clara, Calif) clear aligners treatment for both arches only after having completed treatment with an initial series of aligners were included in this study. The pretreatment and posttreatment cone-beam computed tomography scans after the initial series were acquired by a single orthodontist practitioner. ClinCheck measurements were recorded with Align Technology. The long axis of the anterior tooth intrusion movement was measured in 142 teeth. A comparison between the predicted and actual measurements of anterior intrusion of the teeth was performed, and the intraclass correlation coefficients showed an almost perfect agreement in the linear measurements. RESULTS: A statistically notable difference between the predicted and actual measurements of anterior intrusion. The predicted intrusion movement of the maxillary canines (P = 0.001), maxillary lateral incisors (P <0.0001), and maxillary central incisors (P <0.0001) significantly differed from the actual values. Similarly, the intrusion movement in the mandibular teeth seemed to be inaccurate, with significant differences in the mandibular canines (P <0.0001) and mandibular lateral and central incisors (P <0.0001). CONCLUSIONS: The mean precision of true anterior intrusion with Invisalign clear aligners was 51.19%, and the mean amount of correction was 48.81%. The use of other supplementary methods of anterior teeth intrusion may be helpful to reduce the rate of midcourse corrections and refinements.
Assuntos
Desenho de Aparelho Ortodôntico , Aparelhos Ortodônticos Removíveis , Adulto , Tomografia Computadorizada de Feixe Cônico , Humanos , Incisivo/diagnóstico por imagem , Técnicas de Movimentação Dentária , Adulto JovemRESUMO
Photorealistic stylization of 3D scenes aims to generate photorealistic images from arbitrary novel views according to a given style image, while ensuring consistency when rendering video from different viewpoints. Some existing stylization methods using neural radiance fields can effectively predict stylized scenes by combining the features of the style image with multi-view images to train 3D scenes. However, these methods generate novel view images that contain undesirable artifacts. In addition, they cannot achieve universal photorealistic stylization for a 3D scene. Therefore, a stylization image needs to retrain a 3D scene representation network based on a neural radiation field. We propose a novel photorealistic 3D scene stylization transfer framework to address these issues. It can realize photorealistic 3D scene style transfer with a 2D style image for novel view video rendering. We first pre-trained a 2D photorealistic style transfer network, which can satisfy the photorealistic style transfer between any content image and style image. Then, we use voxel features to optimize a 3D scene and obtain the geometric representation of the scene. Finally, we jointly optimize a hypernetwork to realize the photorealistic style transfer of arbitrary style images. In the transfer stage, we use a pre-trained 2D photorealistic network to constrain the photorealistic style of different views and different style images in the 3D scene. The experimental results show that our method not only realizes the 3D photorealistic style transfer of arbitrary style images, but also outperforms the existing methods in terms of visual quality and consistency. Project page:https://semchan.github.io/UPST_NeRF/.