Spine feature point automatic recognizing method based on average curvature flow

Abstract

A spine feature point automatic recognizing method based on average curvature flow includes the steps of firstly, conducting three-dimensional reconstructing on a CT image of a human spine to obtain a three-dimensional spine model of to-be-marked feature points; secondly, conducting meshing on the spine model obtained in the first step; thirdly, manually selecting a certain point on the model, and calculating the average curvature value of all vertexes in the spherical space, with infinitesimal radius R, around the manually-selected point; finally, selecting n points, with the maximum average curvature, around the manually-selected point, and obtaining inner products between the n points with the maximum average curvature and the selected point respectively to obtain the point with the minimum included angle. The marking precision of the feature points can be improved, and therefore data measurement precision is improved, and the method has the advantages of being high in precision and easy to implement.

Description

technical field [0001] The invention belongs to the technical field of medical image processing, and in particular relates to an automatic identification method of spine feature points based on average curvature flow. Background technique [0002] The spine is an important supporting bone for the human body. In recent years, with the changes in people's living and working styles, the incidence of spinal diseases is increasing. Due to the complexity of the vertebral structure and the characteristics of the spine surgery itself, the current spine surgery is very difficult and dangerous. The precise positioning and registration of the spine model is still an unsolved problem in diagnosis and treatment. [0003] In the localization and registration of the spine model, the main task is the localization of feature points. Using a purely manual method to label feature points, the labeling of feature points is inaccurate, resulting in inaccurate measured data and inaccurate model ...

AI Technical Summary

Problems solved by technology

Using a purely manual method to label feature points, the feature point labeling is inaccurate, resulting in low accuracy of measured data and inaccurate model matching

The method of using Gaussian curvature as a reference value for selecting feature points can improve the accuracy of labeling to a certain extent, but the effect of this method is limited for errors caused by manual selection of feature points

Images