JPH10283504A - Three-dimensional image generator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To facilitate parameter setting for volume rendering and to simply obtain a three-dimensional image as it is intended. SOLUTION: A volume rendering device 1 converts a voxel value into opacity based on a linear conversion characteristic that is expressed by translucent width W and threshold T which are inputted by an input device 2, also calculates a normal vector, boundary surface strength, a unit normal vector and normalization boundary surface strength that is normalized with the width W from inputted volume data and performs volume rendering processing through normalization boundary surface strength and opacity which are calculated in each voxel.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for performing rendering from three-dimensional volume data obtained by an X-ray CT apparatus or an MRI apparatus (nuclear magnetic resonance imaging apparatus).
The present invention relates to a three-dimensional image generation device that creates a three-dimensional image.

[0002]

2. Description of the Related Art In the medical field, X-ray CT apparatuses and MRI
Images for diagnosing humans have been taken using the device. Although this image is planar, three-dimensionally arranged digital data can be obtained by obtaining a number of images in a direction perpendicular to the plane. Represents

As a technique for generating and displaying a three-dimensional image from such volume data, surface rendering for generating and displaying a surface is conventionally known.
In this method, a voxel value is divided by a threshold value to extract a region, and a boundary of the region is shaded. Further, by performing translucent display according to the material content ratio of voxels from the volume data without performing binarization using a threshold value, the unnaturalness due to aliasing and quantization seen in surface rendering is eliminated. Also known as volume rendering (RA Dreb)
in L. Hanrahan, "Volume Ren
dering ", Computer Graphics
(Proc. SIGGRAPH), Vol. 22, N
o. 3 (Aug. 1988) pp. 65-74).

[0004]

However, in the surface rendering, a portion having a voxel and a portion not having a voxel are divided into 100% and 0% according to a threshold value.
There is a problem that if the threshold value is reduced in order to express details in a light and shade object, a granular noise is generated, and a thin blood vessel or the like is interrupted to form a dotted line.

According to volume rendering,
Although such inconvenience of surface rendering is solved, there are many parameters to be set such as opacity, and it is complicated, and it is unclear what value is given as a parameter and what image is obtained. There is a problem. Therefore, actually, many cuts and tries are required, and experience and intuition are required.

In view of the above, it is an object of the present invention to provide a three-dimensional image generating apparatus in which parameters can be easily set and an intended image can be easily obtained.

[0007]

In order to achieve the above object, in a three-dimensional image generating apparatus according to the present invention, there are provided means for inputting a threshold value and a translucent width; Means for converting a voxel value to opacity by using a straight line represented by a transparency width as a conversion characteristic, a normal vector from volume data, a boundary surface intensity which is an absolute value of the normal vector, a unit normal vector, and Means for obtaining a normalized boundary strength obtained by normalizing the boundary strength with the input translucent width, and volume rendering processing using the normalized boundary strength for each voxel and the opacity. It is characterized in that volume rendering means is provided.

A threshold and a translucent width are input, and a straight line represented by the threshold and the translucency is used as a conversion characteristic from the voxel value to the opacity, and the conversion from the voxel value to the opacity is performed according to the conversion characteristic. . On the other hand, a normal vector, a boundary surface intensity, which is an absolute value of the normal vector, and a unit normal vector are obtained from the volume data, and further, the boundary surface intensity is normalized by the input translucent width, thereby obtaining a normal value. Calculated boundary surface strength is required. Then, a volume rendering process is performed using the normalized boundary surface strength obtained for each voxel and the opacity. In this case, since the gradient of the voxel value (boundary surface strength) is normalized by the translucent width, when the translucent width is changed to change the voxel value range to be translucent, the change in the image is the area to be extracted. And the threshold does not change near the middle of the translucent range. Therefore, an image that correctly reflects the set threshold value is obtained, and appropriate opacity can be set only by setting the threshold value and the translucent width. This means that parameters for volume rendering can be set as easily as surface rendering.

[0009]

Next, embodiments of the present invention will be described in detail with reference to the drawings. As shown in FIG. 1, the three-dimensional image generation device according to the present invention includes a volume rendering device 1, an input device 2, and an image display device 3. The volume rendering device 1
X-ray CT or MR online or offline
This is an image processing device using a computer that performs image processing of volume data received from the I device, and can be configured as dedicated hardware or software. The input device 2 includes a keyboard, a mouse, and the like. The image display device 3 displays an input assist image and a generated image. These can be configured as a device integrated with the X-ray CT device or the MRI device, or as a separate device.

The volume data is a three-dimensional array of voxel values as shown in FIG. The input device 2 determines the threshold value T as shown in FIG.
And the translucent width W. That is, the translucent width W is translucent (opacity in the range of 0% to 100%).
Is a range of voxel values to be displayed, and the threshold value T represents a voxel value having an opacity of 50%. A conversion table from voxel values to opacity as shown in FIG. 2A is formed by the inputted T and W. Ordinarily,
Although the conversion characteristics are freely set as shown in FIG. 2 (b), the conversion becomes complicated and the result becomes unclear, so that the linear conversion characteristics defined by T and W I do.

The volume rendering device 1 converts voxel values to opacity in accordance with the above-described linear conversion characteristics, in addition to normal volume rendering processing. Further, for a voxel having an opacity greater than 0, the normal vector N, the boundary surface intensity | N |, the unit normal vector N / | N |, and the boundary surface intensity | N | / W normalized by W calculate.

Here, the values of the voxels A, B, C, D,... Arranged along the line of sight as shown in FIG.
.., T = 150, W = 200
With the setting of, the opacity on the right side from voxel B is larger than 0. The normal vector N of each voxel is obtained by determining the gradient of the voxel value and the direction of the gradient from neighboring voxel values. The interface strength | N | is the absolute value of the normal vector N, that is, the absolute value of the gradient. The unit normal vector is
N / | N |. In the example of FIG. 3, the boundary surface strength of any voxel is 50 per distance 1 (sampling interval).

By dividing the interface strength | N | by the set translucent width W, the interface strength | N | / W normalized by the translucent width is obtained. In this way, normal volume rendering is performed using the normalized boundary surface strength and opacity obtained for each voxel. At this time, the normalized interface strength is multiplied by the opacity data. Therefore, the normalized interface strength is the probability of voxel existence (material content represented by opacity).
It corresponds to. Shading can be performed by the method of Phong or the like by using the unit normal vector described above (Phong, Bothon).
g, "Illumination for Compu
ter Generated Images ”, CAC
M 18 (6) (June 1975) pp. 311-
317).

In this case, if the threshold value T is fixed and the translucent width W is increased or decreased, the number of voxels integrated along the line of sight increases or decreases, but the normalized boundary surface is inversely proportional to the increase or decrease. Strength decreases or increases. Therefore, the change of the image due to the operation of W is only the ambiguity of the region to be extracted, and the fact that the threshold value T is near the center of the translucent width W does not change (strictly, along the line of sight in the light source model). (The threshold T is not exactly at 50% opacity, since the integration involves light attenuation corresponding to the opacity at each voxel). Therefore, an image that correctly reflects the set threshold value T is obtained, and appropriate opacity can be set only by setting the threshold value T and the translucent width W.

Actually, first, the original image is displayed on the image display device 3.
Is set and the threshold value T is set while observing this. In this case, the threshold value T can be specified with the same feeling as in the case of surface rendering. Next, the contrast of the object to be extracted included in the original image, S
The translucent width W is adjusted according to the / N ratio. If the S / N ratio of the part to be extracted is poor, the setting and adjustment are performed so that W becomes larger than the noise amplitude. Thus, by observing the original image, the appropriate threshold value T and translucent width W can be generally determined, and the setting of the opacity becomes easy. In addition, since the boundary surface intensity is normalized by the translucent width W, the obtained image correctly reflects the set threshold value T even if W is changed, so it is necessary to repeat cut and try. Disappears.

[0016]

As described above, according to the three-dimensional image generating apparatus of the present invention, parameters for volume rendering can be set as easily as surface rendering, and the intended three-dimensional image can be set. Can be easily obtained.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention.

FIG. 2 is a graph showing conversion characteristics from voxel values to opacity.

FIG. 3 is a diagram schematically showing an arrangement of voxels.

[Explanation of symbols]

 1 volume rendering device 2 input device 3 image display device

Claims (1)

[Claims] 1. A means for inputting a threshold value and a translucent width, and means for converting a voxel value to opacity using a straight line represented by the input threshold value and translucent width as a conversion characteristic. From the volume data, the normal vector, the boundary surface strength that is the absolute value of the normal vector, the unit normal vector, and the normalized boundary surface strength obtained by normalizing the boundary surface strength with the input translucent width described above. And a volume rendering means for performing a volume rendering process using the normalized boundary surface strength for each voxel and the opacity.
Dimensional image generation device.