KR20130112130A - Registration accuracy improvement method using stereo endoscope - Google Patents

Abstract

PURPOSE: A method for improving matching accuracy by using a stereo-endoscope is provided to improve sailing speed through interaction with a virtual reality (VR) device. CONSTITUTION: A stereo-endoscope image is combined with optical information. A reconstructed 3D image is formed by combining the stereo-endoscope image with the optical information. A computed tomography (CT) image is passed through a virtual endoscope. The CT image is passed through a depth map. Matching accuracy is improved by passing the CT image through the depth map.

Description

How to Improve Matching Accuracy Using a Stereo Endoscope {REGISTRATION ACCURACY IMPROVEMENT METHOD USING STEREO ENDOSCOPE}

The present disclosure relates to a method of improving matching accuracy using a stereo endoscope as a whole.

Herein, the background art relating to the present disclosure is provided, and these are not necessarily meant to be known arts.

When using a conventional mechanical endoscope, there are problems such as pain due to insertion, expensive cost, and difficulty in observing a small diameter part. Therefore, in order to replace or supplement the conventional mechanical endoscope, internal navigation is possible through three-dimensional reconstruction of an actual image, and a virtual endoscope device that interacts with a VR device has emerged. However, the conventional virtual endoscope has a problem in speed during navigation, and also has a problem that the image quality is impaired by surface rendering.

This will be described later in the Specification for Implementation of the Invention.

SUMMARY OF THE INVENTION Herein, a general summary of the present disclosure is provided, which should not be construed as limiting the scope of the present disclosure. of its features).

According to one aspect of the present disclosure, in a method of improving matching accuracy using a stereo endoscope, a stereo-endoscope image and a stereo-endoscope image are combined with optical information, and Depth Creating a reconstructed 3D image via a map; And, through the CT image (MPR), virtual endoscope, Depth-Map, improving the matching accuracy; there is provided a matching accuracy improvement method using a stereo endoscope.

This will be described later in the Specification for Implementation of the Invention.

1 is a view showing an embodiment of a virtual endoscope diagnostic method using a three-dimensional image processing method according to the present invention,
2 is a view showing another embodiment of a virtual endoscope diagnostic method using a three-dimensional image processing method according to the present invention,
3 is a diagram illustrating an example of spatial division;
4 illustrates an embodiment of a virtual endoscope navigation;
5 is a view showing an embodiment of a collision detection method applied in the present invention,
6 is a diagram illustrating an example of a method of generating a 3D model using a virtual endoscope according to the present disclosure;
7 and 8 are views illustrating an example of a matching accuracy improvement method using a virtual endoscope according to the present disclosure;
9 is a view showing an example of a matching accuracy improvement method using a stereo endoscope according to the present disclosure.

The present disclosure will now be described in detail with reference to the accompanying drawing (s).

1 illustrates an embodiment of a virtual endoscope diagnosis method using a 3D image processing method according to the present invention. Acquisition of the cross-sectional images output from the device for outputting the cross-sectional image, such as MRI, CT, ultrasound (S110). Before the 3D volume data is generated through interpolation and segmentation of the obtained cross-sectional image, a preprocessing process such as interpolation of necessary images is performed (S120). Here, interpolation is a process of obtaining a difference before and after the current sequence image between successive images in order to increase the resolution of the image, and interpolating the difference image. Segmentation is a process of setting a threshold corresponding to an intensity of an image of a region of interest and filtering only a value that is equal to or larger than a threshold to generate a 3D volume. In the present invention, it is possible to convert the 3D volume format so that the 3D volume data S130 can be converted into the VRML format 20 for observation and image navigation through a web interface: 10. The 3D volume data is spatially partitioned (S140). This process is called binary space partitioning (BPS). This process is necessary because the 3D volume data is too large for real time interaction (more than 200,000 polygons exist in Trachea), and the BPS process involves spatially correlating spatially small three-dimensional polygonal mesh data. This is the process of creating a hierarchical tree structure. The binary space partitioning process is described in detail as follows. The basic three-dimensional volume of trachea consists of about 200,000 triangular meshes. Each vertex is called a vertex in three-dimensional space. Since the spatial coordinates of these vertices are known, they can be grouped into a hierarchical sub-tree. At this time, the structure of tree should be well structured so that each part can be interworked to produce one unified motion. It is possible to navigate in the interior in two or three dimensions after the spatial partitioning (S150) at this time only the portion shown in the current view point for real-time navigation is rendered (S160). The output device module 40 is an HMD or a monitor, and the input device 30 used by the user for image navigation includes a 2D mouse or a space ball or a 3D mouse used in virtual reality.

Figure 2 shows another embodiment of a virtual endoscope diagnostic method using a three-dimensional image processing method according to the present invention. Acquisition of the cross-sectional images output from the device for outputting the cross-sectional image, such as MRI, CT, ultrasound (S210). The obtained cross-sectional image is subjected to a preprocessing process such as interpolation of the necessary image before generating 3D volume data through an interpolation process and a segmentation process (S220). Here, interpolation is a process of obtaining a difference before and after the current sequence image between successive images in order to increase the resolution of the image, and interpolating the difference image. Segmentation is a process of setting a threshold corresponding to an intensity of an image of a region of interest and filtering only a value that is equal to or larger than a threshold to generate a 3D volume. In the present invention, it is possible to convert the 3D volume format so that the 3D volume data can be converted into the VRML format 20 for observation and image navigation through a web interface (10). The 3D volume data is generated through volume rendering of the image output through the preprocessing process (S230). 3D volume rendering refers to a method of generating 3D volume data by MIP (maximum intensity projection) or general raycasting. However, in order to visualize the volume data by the 3D volume rendering, spatial coordinate data for interaction is required, thereby performing 3D surface rendering (S240). A route to be navigated is generated using the 3D volume data by the 3D surface rendering (S250). For the navigation path generation projection, all the center diameters of the tubes to be navigated can be mathematically calculated from the polygon volume data obtained from 3D surface rendering. The method of sailing along the center of the navigation route by applying mathematically calculated navigation coordinates to the three-dimensional volume data was applied in the present invention. Navigation in the interior is possible in two or three dimensions (S260). At this time, only the portion shown in the current view point is rendered for real-time navigation (S270). The output device module 40 is an HMD or a monitor, and the input device 30 used by the user for image navigation includes a 2D mouse or a space ball or a 3D mouse used in virtual reality.

3 shows an example of spatial division.

4 illustrates an embodiment of a virtual endoscope navigation, which is an example of a human bronchial application. In 2D / 3D navigation, a collision detection method is applied. This method prevents the viewing point from escaping the organs during virtual endoscopic navigation.

5 shows an embodiment of a collision detection method applied in the present invention. It is a technique of detecting the inner surface boundary of the trachea by repeatedly casting a ray from a virtual camera of a virtual bronchoscope. If detected, it is recognized as a wall and the observation point cannot be passed in the direction of the detected wall. By such a collision detection method, it is possible to diagnose like an actual endoscope without escaping to the outside of the organ at the time of image navigation of the hollow organ of the human body.

FIG. 6 is a diagram illustrating an example of a method for generating a 3D model using a virtual endoscope according to the present disclosure. FIG. 6 illustrates Pituiary Gland, Sella Turcica, and Sphenoid Sinus. A method of performing endoscoping and forming a 3D model from them is presented.

7 and 8 are diagrams showing an example of a matching accuracy improvement method using a virtual endoscope according to the present disclosure. After registration, an endoscope image is captured, a gray scale image edge detection is performed, and an edge point is performed. The point matching by the ICP Algorithm is performed through the process of extracting the image, reconstructing the CT image, performing the gray scale image 2D projection, performing the gray scale image edge defeature, and extracting the edge point.

9 is a diagram illustrating an example of a method of improving matching accuracy using a stereo endoscope according to the present disclosure, and combining optical information between a stereo-endoscope image and a stereo-endoscope image, and performing a depth-map to reconceive the image. Create a trucked 3D image. On the other hand, matching accuracy is improved through a CT image (MPR), a virtual endoscope, and a depth map.

S270: Render

Claims (1)

In the matching accuracy improvement method using a stereo endoscope,
Combining optical information into a stereo-endoscope image and a stereo-endoscope image, and passing a Depth-Map to create a reconstructed 3D image; And,
Improving matching accuracy through a CT image (MPR), a virtual endoscope, and a Depth-Map.

Images