An Intelligent Cephalometric Method for Lateral Films Based on Deep Neural Network

Abstract

The invention discloses an intelligent cephalometric measurement method based on a deep neural network for lateral slices, and relates to the technical field of processing lateral slice image data. To solve the problem of missing levels, the present invention includes the following steps: data preparation: importing the data of the lateral slice image and calibrating the data of the lateral slice image, and dividing and preprocessing the data of the lateral slice image; constructing an end-to-end model : including building an encoder module, building a relational reasoning module, and building a decoder module; end-to-end model training: optimize the position, space, and loss functions of measurement indicators on the lateral slice image to obtain a trained model; An end-to-end model for landmark detection and cephalometric analysis. The present invention can learn more spatial information, and fully consider the priority of detecting mark points when constructing the hidden layer of the relational reasoning network module.

Description

technical field [0001] The present invention relates to the technical field of lateral image data processing, and more specifically relates to the technical field of a lateral image intelligent cephalometric measurement method based on a deep neural network. Background technique [0002] Cephalometric analysis is a basic operation to evaluate tooth development, treatment effect and facial aesthetics. Cephalometric analysis is a very important step in stomatology, especially one of the main tasks of orthodontists. The key to intelligent cephalometric measurement The steps include image acquisition, intelligent fixed point, scale setting, measurement generation, and report generation. The intelligence is mainly reflected in the intelligent fixed point module; specifically, the task of the orthodontist is to determine the position of the anatomical landmarks of the craniomaxillofacial region, and then use the Cephalometric technology is used to design treatment plans. The curre...

AI Technical Summary

Problems solved by technology

[0010] The purpose of the present invention is: in order to solve the problems of lack of spatial information and lack of priority of marker points in the cephalometric method in the prior art, the present invention provides a lateral slice intelligent cephalometric method based on a deep neural network, through The relationship reasoning module is used to fuse the potential relationship features between pairs of landmark points, and in the model training stage, the regression learning of the landmark point feature map and the regression learning of the direct measurement value are combined to enable the model to learn more spatial information; build relationship reasoning The hidden layer of the network module fully considers the priority of detecting landmarks, so that cephalometric analysis can be performed more quickly, accurately and intelligently

Images