Three-D quick division method of target field in surgical operation guide system

Abstract

Firstly, adopting cyclic corrosion algorithm of morphology to separate the target field from adjacent other field and recording cycle-index, then adopting quick travelling method to make the seed point selected by user undergo the process of initial multiplication operation and make the multiplified forward line quickly approach to edge line of target field so as to obtain the initial separation effect, then adopting reconstruction method of the morphology field to further optimize the initial separation effect.

Description

Technical field [0001] The invention relates to a three-dimensional fast segmentation method, in particular to a three-dimensional fast segmentation method of a target field in a surgical navigation system, belonging to the technical field of image processing. Background technique [0002] With the rapid development of science and technology, the continuous emergence of high-tech medical equipment has provided more and more help to the modernization of medical treatment. X-ray tomography (CT), nuclear magnetic resonance (MR), and ultrasound (US) equipment are gradually being widely adopted by humans. They provide high-quality anatomical information for medical diagnosis, surgical navigation, and tissue structure analysis of the human body, greatly improving human cognition of themselves and thereby improving the national medical health level. From another point of view, the huge data of these high-definition multi-dimensional images also poses a challenge to later medical applica...

AI Technical Summary

Problems solved by technology

However, model-based segmentation methods generally have a common problem, that is, the segmentation accuracy is limited

This is caused by the following reasons: (1) Since the interruption parameter of the model propagation (Propagation) depends on the gradient value of the portrait being close to rather than equal to zero, most of the final dividing lines will stop several pixels away from the target dividing line, so The segmentation results are often inaccurate, especially where the arc is large or the gradient value is weak, the error is greater

(2) In order to achieve the effect of smoothing the edge line (or surface), most of these algorithms use a surface tension function, which will inevitably make it difficult for the contour line to enter the narrow target area, thus affecting the segmentation accuracy

(3) The existence of multiple minimum values ​​will also cause deviations in the segmentation results

Although many researchers have proposed some optimization algorithms, the computational complexity is still high, and the requirements for real-time segmentation cannot be achieved.

Images