Mark point labeling method and device for skeleton structure image based on neural network

Abstract

The invention discloses a mark point labeling method and device for a skeleton structure image based on a neural network. The method comprises the following steps: inputting a skeleton structure imageinto a mark point initial prediction model based on a neural network to output initial prediction coordinate values of various mark points; according to the initial prediction coordinate values of the various mark points, respectively determining the local areas of the various mark points; for each type of mark points, inputting the images of the local areas of the type of mark points into a single mark point labeling model which corresponds to the type of mark points and is based on a neural network; and for each single mark point labeling model, performing coordinate reverse return on the coordinate values output by the single mark point labeling model to obtain the coordinate values of the mark points in the skeleton structure image for labeling. According to the invention, each mark point of the skeletal structure image can be automatically marked based on the neural network, so that the diagnosis efficiency and accuracy of doctors are greatly improved.

Description

technical field [0001] The invention relates to the technical field of medical imaging, in particular to a neural network-based marker point labeling method and device for skeletal structure images. Background technique [0002] X-ray bone measurement analysis is one of the important research issues in the medical field. It refers to the positioning and shooting of bones through X-rays, the positioning and labeling of the landmarks on the captured bone structure images, and then drawing certain landmarks of key points of the bones. The angle of the line is measured. For example, X-rays are used for cephalometric photography to obtain anteroposterior and lateral images of the skull, and positioning and labeling of landmarks on the anteroposterior and lateral images of the skull, and then draw a certain line angle for each marker point on the jaw and craniofacial to measure and analyze the teeth and jaws. , craniofacial soft and hard tissue structure, so that the inspection a...

AI Technical Summary

Problems solved by technology

However, the process of manually labeling and analyzing bone structure images is very time-consuming and subjective, and it is difficult to quickly obtain accurate measurement data only through manual labeling by doctors.

Images