JP2874445B2 - Region extraction method and device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for processing multidimensional data, and more particularly to a method and apparatus for processing individual organs from three-dimensional digital data imaged by MRI, X-ray CT or the like.
The present invention relates to a region extraction method that is indispensable for extracting an individual organ, which is indispensable for performing dimensional display, and a device therefor.

[0002]

2. Description of the Related Art Known examples related to the present invention include the following documents (1) and (2) and the prior application (3).

(1) Azriel Rosenfeld, “Digital Picture Processi” (translated by Makoto Nagao), “Digital Picture Processing”
n ") p. 335 (2) JP-A-03-140140 (3) 322966 As a method of extracting a specific area from certain data, the following two methods are known roughly.

[0004] A method of classifying according to properties (for example, density) of individual pixels of a region: A method of extracting using spatial continuity of a region For example, when extracting a bone region from an X-ray CT image, Since the pixel value (CT value) of the bone region is higher than the other regions, the above-described method can be used. On the other hand, in an MRI image, since it is difficult to specify individual regions from pixel values, the above-described method is used.

[0005] As a specific method, there are well-known a method called "edge tracing" by tracking a region boundary and a method called "region growing" by expanding a connected region. The edge trace is suitable for high-speed three-dimensional display using a surface model because the extraction result is obtained as a boundary line or a boundary surface. On the other hand, the boundaries are likely to be interrupted due to noise or the like, so that correction operation processing by the operator is indispensable for actual image data.

[0006] On the other hand, region growing is a method of expanding a connection region from the inside of the region to the periphery. The above-mentioned document (1) describes a basic method of region growing. In region growing, noise does not hinder region expansion unlike edge tracing, and application to three-dimensional data is easy. Therefore, for an MRI three-dimensional image, an extraction method based on region growing is basically considered appropriate.

In the region growing, the shape and size of the extraction area change depending on the setting of the connectivity determination condition (extension condition). Therefore, in order to perform extraction with high reliability, it is necessary to optimize expansion conditions for each image to be processed and each extraction target. The above document (2) discloses a method for improving the reliability of the extraction process by using an extended condition that combines a local condition and a global condition.

[0008] In the region growing, at the start, after a growth start point and a criterion are given, originally,
Requires no manpower. This is the most ideal in terms of automation of extraction. However, in this method, once the extension point enters another region, the extension is started in the entered region, so that there is a problem from the viewpoint of reliability in that the method is left to the machine. In fact, even if region growing is applied as it is to image data having noise or uneven density, good results can hardly be expected.

The above-mentioned prior application (3) describes an operator assisting method for region growing, which is considered to be effective for improving extraction reliability. In this method, the process of extracting region growing from three-dimensional voxel data is constantly monitored by a three-dimensional image, and when an extended point enters an unintended area, the area extension is immediately interrupted. The removal is performed using a correction function.

[0010]

The biggest problem in the connected area expanding method represented by the region growing is that a target extraction target area (hereinafter, referred to as a “target area”) and an area existing outside the target extraction target area. (Hereinafter, "external area"
Is connected at one point, the extension point enters the external area, and the extraction of the area starts. As a method for preventing the extension point from entering the external area, there are several proposals as follows.

(1) Tightening the connection criterion (extension condition) If the extension condition is tight, the probability that the extraction target area extends to the outside area is reduced accordingly. However, as the expansion condition becomes tighter, it becomes more difficult to extract the entire target area. As a solution to this problem, there is considered a method in which: a plurality of determination criteria are provided, and the determination of the extension point is diversified in multiple directions; and the area of the extracted data is uniformly enlarged toward the periphery. (Refer to document (2).) (2) Add correction to data during expansion of connected area In the final result of expansion, usually, the protruding area has grown greatly, and this should be deleted by correction work. Is almost impossible. So, to reduce the amount of work to delete, when the extension point enters the external area,
That is, this portion is deleted when the protruding area is still small. If the external area and the target area are separated in this way, only the target area is extracted in the subsequent area expansion. (See prior application (3).) The present invention has been made to further improve the above-described technology, and an object thereof is to add an artificial operation to the extraction process to ensure the reliability of extraction. It is an object of the present invention to provide a multidimensional data processing method and apparatus which minimizes

[0012]

The above objects of the present invention are attained in a multidimensional data processing method for extracting a region from multidimensional digital data, wherein a region other than the target is extracted as a result of the region extraction. A multi-dimensional data processing method characterized by designating one point belonging to the region on the three-dimensional display image and executing a deletion process of the region other than the target, and a connected region expansion process. In a multidimensional data processing apparatus for extracting a region from three-dimensional voxel data and displaying the extracted region three-dimensionally, means for storing extraction history data indicating at which point each point in the extraction region was extracted. A three-dimensional display unit that converts the extracted region at an arbitrary time point into a three-dimensional display image with reference to the extraction history data and outputs the extracted region, and sets one point in the extracted region as a start point of the next extended process. Means for instructing, means for extracting the inside of the area to which the indicated point belongs using a connected area expanding method for the extraction history data, means for displaying the extracted area on a three-dimensional image, A means for deleting the extracted area from the three-dimensional voxel data or registering the area again as a non-display area, and for extracting the target area by expanding the connected area again for the three-dimensional voxel data from which the area has been deleted. This is achieved by a multidimensional data processing device characterized by having means.

[0013]

As described above, in the extraction method based on the connected area expansion processing, if the expansion point once protrudes into an unintended area, the result is completely wrong.
In particular, in an X-ray CT or MRI image, the above-mentioned protrusion can easily occur due to the influence of image distortion, noise, and the like caused by the apparatus. Therefore, in order to ensure the reliability of extraction,
The operator needs to perform some kind of support operation for the above-described extended processing. Naturally, the amount of work of the support operation greatly affects the clinical utility of the extraction process. A main object of the present invention is to enable efficient and highly reliable extraction with minimum effort. Hereinafter, the means provided for this purpose and the operation thereof will be described.

The overall flow of the extraction process in the multidimensional data processing method according to the present invention is shown below. The main content of the present invention relates to the processing for removing the protruding area.

[0015] The operator sets an expansion start point in the destination area.

From the starting point, the connected area extension processing inside the destination area is started.

The extraction process is monitored on a three-dimensional display image, and the extraction process is stopped when a region protruding to the outside is clearly found.

The operator sets an extension start point in the protruding area.

From the start point, a connected area extension process is started, and a protruding area is extracted.

The protruding region is clearly indicated on the three-dimensional display image.

When the operator looks at the three-dimensional display image and finds that the extracted area is not a protruding area,
Delete the area.

Steps (1) to (4) are repeated until the extraction of the entire target area is completed.

By the way, in order for the operator to add some processing to the intermediate result of the extraction, means for monitoring the extraction process, and the extraction processing is stopped when a protruding area appears on the image being monitored. We need a means to make it happen. In particular, when extraction processing is performed on three-dimensional voxel data, a function of outputting an extraction region in real time as a three-dimensional display image is required, as disclosed in the above-mentioned document (3). A detailed description of these means and a realizing method is omitted here. In the present invention, the following functions are newly provided to perform the above-described “deletion of the protruding area” processing.

(1) A region extraction function by expanding a connected region using the time when each point is extracted, that is, the extraction history.

(2) A function of three-dimensionally displaying an area extracted at an arbitrary time based on the above-mentioned extraction history data.

(3) A three-dimensional coordinate designating function for designating a point belonging to the protruding area using the three-dimensional display image.

(4) A region extracting function for extracting a protruding region in a lower direction at the time of extraction by using the above-mentioned extraction history data by a connected region expanding method.

(5) An area extraction function for extracting an area at the same extraction time in the protruding area by using the above-mentioned extraction history data by the connected area expansion method.

(6) A function of checking the connectivity of a newly expanded area and extracting a connected area at each stage of expansion in the connected area expansion.

(7) A function of judging whether or not the extension point is a connection between the extraction target area and the protruding area by examining the number of extension points in each extraction stage in the connection area extension.

(8) A function of extracting the entire protruding area by expanding the area within the protruding area, starting from the connection section.

(9) A function of displaying the protruding area in a manner superimposed on a three-dimensional display image.

(10) A function of deleting the protruding area from the original data or giving each pixel of the protruding area an invisible attribute.

(11) The area extraction by the area extraction function (1) is monitored by the three-dimensional display function (2). When the protruding area appears, the area extraction is stopped, and the three-dimensional coordinate designation function (3). A function of designating the protruding area to be deleted, extracting the protruding area by the functions (4) to (9), and deleting the area by the function (10). (12) The connection specified by the functions (4) to (7)
After the removal or extension prohibition area, the extraction stage is returned to the point where the connection is generated, and the extraction processing by the area extraction function (1) is started again.

The operation (1) to (11) of each of the above functions (1) to (11) will be described below with reference to FIGS.

FIG. 1 shows an example of original image data to be subjected to extraction processing. Although the present invention actually deals with multi-valued three-dimensional voxel data, the description will be made using binary two-dimensional data to simplify the description.

In FIG. 1, the shaded portion has the value "1" and the outside is "0". The area 101 is a target area to be extracted. However, the region 103 which does not belong to the target region
Is also connected to the region 101 at the connection unit 102,
When the connected area extension method is used, the area 103 is also extracted as a protruding area.

(1) In the area extraction by the conventional connected area expansion method, the result is expressed as binary data indicating whether each pixel belongs to the extraction area or as a probability of belonging to the extraction area. Was. In the area extension processing in the target area, which is performed first in the present invention, the number of the extracted points generated by counting from the beginning is output. This number is stored as attribute data for each pixel. Hereinafter, this attribute data is referred to as “extraction history data”.

With respect to the data of FIG. 1, one point 2 in the target area
FIG. 2 shows an extraction result (extraction history data) obtained by performing connected region expansion with 01 as a starting point. The number of each pixel indicates the time when each pixel was extracted.

From the extraction history data, it is possible to extract the extraction result at any time. It is also possible to reverse the extension process. In particular, the latter feature enables automatic extraction of the protruding region. This will be described in detail in (4) and thereafter.

(2) In the case where the outside of the target area occurs, it is necessary to terminate the extraction at an appropriate time. However, it is difficult to automatically determine the time of interruption, and this determination is made by the operator. This means (2)
Provides means for observing the progress of area expansion necessary for the operator to make this determination.

Since it is impossible to directly observe a three-dimensional area on a two-dimensional display, a function of converting the extracted area into a three-dimensional display image is required. Further, in order to observe the state of the region growth, the extracted region that grows sequentially must be sequentially 3
It must be displayed as a three-dimensional image. Further, in the display from one direction, since the presence or absence of the protrusion on the back side cannot be confirmed, it is necessary to provide a function of changing the viewpoint as needed and displaying a plurality of directions simultaneously.

(3) In the process of deleting the protruding area, first, means for designating the area to be deleted is required.
In this instruction, as described in (4) below, it is not always necessary to set the indicated point at the center of the area or a point having a representative pixel value, and it is sufficient to give the coordinates of any one point in the protruding area. It is. By using this, it is possible to indicate the above point directly on the three-dimensional display image. This is extremely effective in reducing the operational burden.

FIG. 3 is an explanatory diagram of a method of setting a starting point on a three-dimensional image. In the figure, reference numeral 302 denotes a three-dimensional image data space, and reference numerals 304 and 305 each denote a target region and a protruding region in the space. 3D image 30
One point 306 in the protruding area designated at 1 is the above 3
It is designated as one point 307 on the surface of the protruding region in the dimensional space.

(4) For the region expansion to go back to the extraction point, reference is made to the extraction source point and the extraction history data of points adjacent to the expansion source point. When the area is small, in this area expansion realized by setting the adjacent point as a new expansion point, the expansion always proceeds in the order of the protruding area → connection part → target area. Therefore, when the extension point is at the connection between the two regions, there is no extension point at any other location.

(5) Whether the extension point has reached the connection portion is determined based on the increase or decrease in the number of extension points, as described in (7) below. When a recursive extension process is used, in which each of the extension points generated from the extension source further generates an extension source, the extension of the region is promoted, so that the increase or decrease in the number of extension points becomes more apparent. Also, when entering the target area from the connection, the number of extension points increases at a stretch,
It becomes extremely easy to specify the connection part.

The recursive extension process includes an extension point and
With reference to the extraction history data of the point adjacent thereto, when the extraction step of the adjacent point is larger than the extraction point of the extension point by one,
This is realized by adding the adjacent point to the point of the extension source.

One stage of area expansion by means (4) and (5) will be described with reference to FIGS.

FIG. 4 shows the first stage of the extension processing. The starting point is shown in FIG. P0 is extracted twelfth in the area expansion processing of (1) (hereinafter, this is described as "extraction stage 12"). (Refer to FIG. 2) First, the first extension source P0 creates extension points c1 and c2 at the adjacent extraction stage 11 points. c1 is the adjacent extraction stage 12
The expansion sources P1 and P2 and c2 make P3. The new extension source P2 then creates an extension point c3, and c3 again creates a new extension source. Thus, the extension points generated in the first stage are finally the points c1 to c8 in FIG.

(6) When a plurality of protruding regions generated from two or more connection portions are overlapped, it is impossible to specify these connection portions at the same time. for that reason,
If the extended area is divided into two or more areas,
Extends the area to the connection area for one area. The method and effect of this function will be supplementarily described at the end of this section using another original image data (FIG. 10).

(7) The connection part is specified from the transition of the number of extension points in the area extension. In this area expansion,
Since the extension progresses in the order of the protruding area → connection part → target area, the number of extension points increases first in the protruding area, then converges once in the connection part, and finally increases again in the target area. . Table 1 shows the number of extension points generated for each step with respect to the data of FIG. Since the number of extension points has converged when the extraction stages are 6 and 5, it is known that the extension points have passed through the connection portion at this time.

[0052]

[Table 1]

Depending on the state of the connection, the number of extension points is not always one as in the above example. However, as a result of the recursive extension, the number of extension points has a characteristic of increasing at a stretch after passing through the connection part. Therefore, it is easy to specify the connection part from the transition of the number of extension points (time-series data). . (See FIG. 5) (8) The shaded area in FIG. 6 is an extraction area obtained by the above-described area expansion. As can be seen from this, this extraction area includes only the range from the designated point input in (3) to the connection part. Therefore, the entire protruding region is extracted by using another region expansion starting from the connection part.
This is realized by an area extension process in which, when the extraction stage of a point adjacent to the extension source is larger than the extraction stage of the point of the extension source by one, the adjacent point is set as a new extension point. FIG. 7 shows the result of this region extraction processing.

(9) If the designated position of the protruding area is incorrect or the specification of the connection part fails, the extraction result in (8) above will be different from the actual protruding area. Therefore, before actually deleting this region, the region extracted in the above (8) is explicitly specified on the three-dimensional display image, and a function for asking the operator for confirmation is provided. The operator observes the displayed area, and if there is no problem as the protruding area, performs the following (10) deletion processing, otherwise cancels the result, and executes the protruding area extraction processing from (3) again. Do.

(10) After confirming the above (9), the protruding area is deleted. However, when data is actually deleted, it is impossible to return to the original state. Therefore, here, a method of attaching an attribute as an invisible area to a pixel instead of deleting it is used.

(11) When there are a plurality of protruding areas, the extraction processing of the target area of the above (1) and (2) to (1)
The process of extracting / deleting the protruding region in step 0) is repeated, and the protruding regions are removed one by one. Thus, a final extraction result (FIG. 8) is obtained.

(12) Instead of deleting the entire protruding area by the processing of (8) to (10) above, after deleting only the connection point, the area within the target area is re-started from the time before the connection point was generated. An extension method is also conceivable. FIG. 9 shows an extraction region at the time of extraction immediately before a connection point appears.
Since the connection point (901) has already been removed, even if the area is expanded again from this point, no protrusion occurs, and FIG.
The same extraction result can be obtained.

Finally, regarding (6), FIGS.
Supplement using

When the area expansion of (1) is performed on actual data, a plurality of protrusions often occur at the same time. FIG. 10 shows extraction history data when two protruding portions occur. FIGS. 102 and 103 are connection portions.

When both areas are separated from each other, (6)
Each connection point can be specified without performing the processing of (1). However, as shown in FIG. 10, when both areas grow into one area, the extraction points of both connection points are different, so that the number of extension points does not converge, and it is not possible to specify the connection points. It becomes possible.

Therefore, when the area is separated in the protruding area, the connection point search is performed by focusing on only one of the areas. For the remaining area, after one of the processes is completed, a connection point search is performed again.

FIG. 11 shows how the area shown in FIG. 10 is expanded. In the figure, reference numeral 803 denotes an expansion start point, and a hatched area is an expansion area. In the extraction stage 13, the region 802 (black portion) formed by the extension points is separated into two. Here, a lower region near the extension start point 803 is selected, and the extension process is continued for this region, thereby extracting one connection point 801.

The area extracted from the connection point by the extension process (7) is shown by the hatched portion in FIG. After deleting this area, a similar connection point search is performed for the remaining upper area,
By performing the area deletion process, the target area is extracted.

[0064]

Embodiments of the present invention will be described below in detail with reference to the drawings.

First, a first embodiment of the present invention will be described with reference to FIGS.

The present embodiment describes a technique relating to a process for extracting a target region from in-vivo three-dimensional voxel data obtained by three-dimensional imaging by MRI.

FIG. 13 shows an example of the configuration of a system to which the present invention is applied. This apparatus uses three-dimensional voxel data obtained by a three-dimensional imaging method by MRI, and performs a process of extracting a specific organ from the three-dimensional voxel data. FIG.
3, a straight line with an arrow indicates a data transmission path, and an arrow indicates the direction of data flowing between the devices.

The three-dimensional voxel data obtained by MRI is taken into the present system from the I / O device 101 and stored in the three-dimensional image memory 102. The extraction processor 103 refers to the data in the three-dimensional image memory 102 and extracts a target area to be extracted by the connected area expansion method. The parent table 111 and the child table 112 connected to the extraction processor 103 are used as work areas for area extraction processing. The extraction result is stored in the extraction history data memory 104 as information indicating the number of the extension in which each voxel of the three-dimensional voxel data is extracted. Therefore, the extraction history data memory 10
4 has the same size as the three-dimensional image memory 102, and each pixel has one-to-one correspondence. Extraction information at an arbitrary time during the extraction is obtained by passing information in the extraction history data memory 104 through a filter 105. The three-dimensional display processor 106 converts the three-dimensional display image data of the extraction area at an arbitrary extraction time into a VRAM based on the data of the three-dimensional image memory 102 and the output data from the filter 105.
107 is created. This is CR as a 3D display image
It is displayed on T108.

The input device 109 sets connection determination conditions,
It is used for setting the extraction start point, interrupting the extension processing, and inputting the instruction of the protruding area to be deleted. An input device 110 for the three-dimensional display processor 106 is used for inputting rendering parameters used for three-dimensional display calculation, viewpoint positions, and the like.

FIG. 14 shows an outline of the area extraction processing according to the present invention.

The outline of the extraction processing operation of the present invention will be described below with reference to the system configuration diagram of FIG. 10 and the flowchart of FIG.

<Step 201> First, the coordinates of one or more points in the region to be extracted are input by the input device 109, and the region to be extracted is specified. The coordinates input in this step are the starting points of the connected area extension processing performed in step 203 and subsequent steps.

<Step 202> A criterion for determining connectivity (hereinafter, referred to as “connection criterion”) in the connection area expansion processing by the input device
"Extended conditions"). In the region extraction processing of the present invention, the extraction of the target region and the extraction and
By repeating deletion, a final extraction result is obtained.
Therefore, the expansion condition does not necessarily need to be optimized. However, by setting the conditions more appropriately, it is possible to reduce the time and effort for this repetition.
It is also effective to use a combination of a plurality of conditions, as disclosed in Document 2 shown above, as extended conditions.

<Step 203> Store the extraction result.
The extraction history data memory 104 is initialized.

<Step 204> Extraction processor 103
Performs an extraction process based on the connected area expansion method from the expansion start point input in step 201 using the parameters input in step 202 as expansion conditions. Connected area expansion 1
The step is performed until all the expansion source points at that time (hereinafter, referred to as “parent”) create an expansion point (hereinafter, referred to as “child”) satisfying the expansion condition in the adjacent area. 1
After completion of the connected area expansion method at the stage, the extraction processor 103 determines on the three-dimensional extraction data memory 104 as an extraction result how many expansion stages the child has been counted from the start of the extraction. Write to the coordinates corresponding to the "child" of.

<Step 205> The rendering processor 106 always refers to the extraction history data 104 updated in step 204, creates a three-dimensional display image of the extraction area in the current extraction stage on the VRAM 107, and displays it on the CRT 108. . The viewpoint position and display parameters of the three-dimensional image can be changed as needed by inputting them to the input device 110.

<Step 206> If no new "child" is generated at a certain stage of the area expansion, the connected area expansion processing automatically stops at this point. The extraction history data 104 at this point is the final area extraction result. On the other hand, if the entire target area has not been extracted yet, that is, if “children” have been generated in step 204, these “children” are set as new extension points (parents), and the extension process is repeated.

<Step 207> The operator proceeds to step 2
The three-dimensional display image of the extraction region created in 05 is evaluated. If the three-dimensional display image clearly protrudes into the external region, the extraction of the target region is temporarily stopped, and the process proceeds to the protruding region deletion routine after step 208. If there is no problem, the extraction of the target area from step 204 is continued.

<Step 208> One point in the protruding area to be deleted is designated. In order to specify one point in the area for the three-dimensional voxel data, it is general to select a certain cross section and specify it on the cross section. However, here, in the connection area expansion for the protruding area, which is performed in the next step, the protruding area is specified on the three-dimensional display image by utilizing the fact that the starting point can be set at an arbitrary point in the area.

<Step 209> A connecting portion between the protruding region and the target region is extracted by the second connecting region expanding method. A detailed description of this step will be described later in step 30.
1 to 308 and steps 401 to 412.

<Step 210> The entire protruding area is extracted by the third area expanding method starting from the connection point extracted in step 209.

<Step 211> The protruding area extracted in step 210 is superimposed and displayed on the three-dimensional display image of the entire extraction area created in step 205. At this time, the two are easily distinguished by using a technique such as changing the colors of the two.

<Step 212> The operator proceeds to step 2
From the three-dimensional display image created in step 11,
The protruding region extracted up to 0 is evaluated. If there is no problem with deletion as a protruding area, the deletion process in the next step is performed. If not, the process returns to step 208 to re-extract the protruding area.

<Step 213> The region extracted as the protruding region is deleted. However, as described in the section of the operation, if the data itself is deleted, it becomes impossible to redo the processing and to extract another area existing in this area. Therefore, here, an attribute as an invisible area is simply added to pixels in the target area.

<Step 214> The operator proceeds to step 21
The three-dimensional display image after the deletion processing of No. 3 is evaluated again, and the presence or absence of another protruding area is checked. If there is a protruding area, the protruding area is extracted from step 208. If there is no protruding area, or if the area is still small and cannot be confirmed as the protruding area, the process returns to step 204 to resume the extraction processing of the target area. I do.

Next, step 20 in the flowchart of FIG.
9 will be described with reference to the configuration diagram of FIG. 10 and the flowchart of FIG. 15. The details of the backward area expansion are shown in FIG.
This will be described with reference to the flowchart of FIG.

Note that “parent” described below indicates an expansion source point in area expansion, and “child” indicates an expansion point generated from “parent”. Further, the “parent table” stores the coordinates of all the “parents” in the one-stage expansion, and the coordinates of the “child” generated by the expansion from the “parent” are sequentially stored in the “child table”. .

<Step 301> The parent table 111 and the child table 112 are initialized.

<Step 302> One point in the area to be deleted specified in step 208 is set as the first “parent” and registered in the parent table 111.

<Step 303> For all the "parents" registered in the parent table, the adjacent points are checked, and those satisfying the extension condition are registered as "child" in the child table. Details of this extension processing will be described in steps 401 to 412.

<Step 304> The connectivity of the region created by the “child” generated in step 303 is checked, and if the region is divided into a plurality, one of them is taken out and the pixels existing in the region are replaced with the real “ Re-register as "child".

<Step 305> The number of “children” generated in step 304 is checked, and the connection portion is determined based on the number or its transition. As this method, for example, a method of comparing the average values of several points before and after each point in time, or searching for a point in time when the variance greatly changes can be considered.

<Step 306> When the connection can be specified, the area expansion is terminated, and the flow proceeds to the connection output processing of step 308. If the extension area has not yet reached the connection portion, the area extension from step 307 onward is continued.

<Step 307> The generated “child” is defined as a “parent”. That is, the newly expanded area is set as an expansion source in the next expansion. This can be achieved by copying the contents of the child table to the parent table. Thereafter, the process returns to step 303 to perform the next-stage extension.

<Step 308> The coordinates of the "child" at the time when the connection is specified are output as the coordinates of the connection. After the connection point is output, the protruding area extraction processing of step 210 and subsequent steps is performed.

Next, details of the extension processing in step 303 will be described with reference to the flowchart of FIG.

<Step 401>"Parent" from parent table
Take out one.

<Step 402> One pixel adjacent to "parent" is extracted. Here, six pixels adjacent in the front, rear, left, right, up, and down are taken as adjacent pixels.

<Step 403> Referring to the "child" table, it is determined whether or not the extracted adjacent pixel is already registered in the child table.

<Step 404> Referring to the extraction history data, the extraction stage of “parent” is compared with the extraction stage of the extracted adjacent pixel.

<Step 405> Steps 403 and 40
As a result of step 4, the extracted neighboring pixel is not registered in the child table yet, and in the first area expansion performed in step 204, the neighboring pixel is extracted one step before “parent”. If the pixel is a child pixel, the adjacent pixel is registered in the child table as a “child”.

<Step 406> Pixels adjacent to the “child” registered in the child table in step 405 are extracted. Similarly to step 402, six pixels adjacent in the front, rear, left, right, up, and down are taken as adjacent pixels.

<Step 407> Referring to the "parent" table, it is checked whether or not the extracted adjacent pixel is already registered in the parent table.

<Step 408> Referring to the extraction history data, the "child" extraction stage is compared with the extracted adjacent pixel extraction stage.

<Step 409> Steps 407 and 40
As a result of 8, if the extracted neighboring pixel has not been registered in the parent table yet and the neighboring pixel is a pixel extracted one stage after “child” in the first area expansion, The adjacent pixel is additionally registered as a “parent” in the parent table.

<Step 410> The above steps 406 to 4 are performed until all the adjacent pixels of the “child” have been checked.
09 (additional registration processing of “parent”) is repeated.

<Step 411> The above steps 402 to 4 are performed until all the adjacent pixels of “parent” have been checked.
10 (the registration process of “child”) is repeated.

<Step 412> Until the “parents” of all parent tables have been checked, the above steps 401 to 401 are completed.
412 is repeated.

According to the above example, 3 obtained by MRI
It becomes possible to extract the target area from the dimensional voxel data while minimizing human operations. More specifically, in the area extraction method by expanding the connected area suitable for 3D voxel data having a huge amount, it is possible to automate most of the extraction area correction work that had to rely on humans in the past Medical 3
An organ extraction process for dimensional voxel data can be put to practical use.

Next, a second embodiment of the present invention will be described with reference to the flowchart of FIG.

In the present embodiment, similarly to the first embodiment, a target area is extracted. However, the processing after extracting the connection part is significantly different between the two. Steps 501 to 509 in FIG. 17 are the same as those in the first embodiment, and a description thereof will be omitted.

<Steps 501 to 509> Omission (refer to steps 201 to 207 in the previous embodiment) <Step 510> The connection portion extracted in step 509 is removed from the data. Instead of removing the attribute, an attribute as an extension prohibition point may be provided.

<Step 511> The extraction time point is returned to the time point at which the connection unit is output. As a result, all the protruding areas are regarded as unextracted areas. At the same time, a part of the target area is also set as an unextracted area. FIG. 9 shows an extraction region in which the extraction time point has been returned until immediately before the connection point appears (extraction time point 5) in the data of FIG. The crosses indicate connection points extracted in step 509, and have an extended prohibition attribute.

After performing the processing of steps 510 and 511, the extension of the connected area in the extraction area from step 204 is restarted.

The connection to the protruding point has already been performed in step 21
Since it has been removed at 0, no extension beyond this occurs, and only the target area is extracted.

In the present embodiment, there are problems that it takes time and effort to extract the target area once extracted again, and there is no means for confirming the protruding area. However, only the connection part is actually deleted. Because it is limited, careless deletion of the target area does not occur.

Next, a third embodiment of the present invention will be described below with reference to the system configuration diagram of FIG. 10 and the flowchart of FIG.

In the present embodiment, the extraction processing of the target area is automatically performed by setting the extraction prohibition point or the extraction prohibition area in the three-dimensional voxel data prior to the extraction processing. In the present embodiment, a brain region is assumed as a target region, and a skin region is assumed as an extraction prohibited region.

<Step 601> The extraction history data memory is initialized.

<Step 602> Display a cross-section including the region which does not need to be extracted, designate one or more points included in the region which should not be extracted on the cross-sectional image, and store the point in the extraction history data memory 104. As an extraction prohibition point.

<Step 603> If there is another area that should not be extracted, the process returns to step 602 to continue specifying the extraction prohibition point. Here, when the unnecessary area is present over the target area (for example, when the target area is a brain area and the unnecessary area is skin), the three-dimensional voxel data is scanned from the peripheral part toward the central part. Thus, the above-mentioned extraction prohibition point can be automatically set.

<Step 604> A cross section including the target area is displayed, a point in the target area existing in the cross section is designated, and the point is stored in the parent table 111 as an extraction start point.

<Step 605> The target area is extracted by expanding the connected area. For this extraction, the method described in step 204 in the above embodiment is used.

<Step 606> When the expansion cannot be performed any more, the extraction process is stopped, and the extraction result is output.

<Step 607> The extension point is set at step 60.
When reaching the extraction prohibition point set in step 2, the target area extraction process is stopped.

<Step 608> Unnecessary areas are extracted and deleted from the extraction prohibition points reached in step 607 by using the area expansion described in the above embodiment, which reverses the extraction time point. After the deletion, the process returns to step 605, and the target area is extracted again by expanding the connected area.

[0127]

As described above in detail, according to the present invention, it is possible to realize a multidimensional data processing method and apparatus for minimizing an artificial operation for the extraction process, which is added to secure the reliability of the extraction. This is a remarkable effect.

[Brief description of the drawings]

FIG. 1 is a diagram showing an example of original image data used for describing an operation.

FIG. 2 is a diagram showing extraction history data for the original image data of FIG.

FIG. 3 is an explanatory diagram of a protruding area designating method on a three-dimensional display image.

FIG. 4 is an explanatory diagram of an extension process for the extraction history data of FIG. 2;

FIG. 5 is a diagram showing the transition of the number of extension points generated in the extension process.

FIG. 6 is a diagram showing an extraction area up to a connection point.

FIG. 7 is an explanatory diagram of a method of extracting a deleted region by expanding a region from a connection point.

FIG. 8 is a diagram showing a final extraction result.

FIG. 9 is a diagram showing an extraction area in an expansion stage where a connection point has occurred.

FIG. 10 is a diagram showing another original image data used for explaining the operation.

FIG. 11 is an explanatory diagram of a connection point search process for the original image data of FIG. 10;

FIG. 12 is an explanatory diagram of a protruding region extraction process for the original image data of FIG. 10;

FIG. 13 is a diagram showing an example of a configuration of a system to which the present invention is applied.

FIG. 14 is a flowchart showing the operation of the first embodiment of the present invention.

FIG. 15 is a flowchart of a connection point extraction process according to the present invention.

FIG. 16 is a flowchart of a backward area expansion process according to the present invention.

FIG. 17 is a flowchart showing the operation of the second embodiment of the present invention.

FIG. 18 is a flowchart showing the operation of the third embodiment of the present invention.

[Explanation of symbols]

101: I / O device, 102: three-dimensional image memory, 10
3. Extraction processor 104 Extraction history data memory
105: filter, 106: three-dimensional display processor, 1
07 VRAM, 108 CRT, 109 Input device for extraction processor (103), 110 Input device for 3D display processor (106), 111 Parent table, 112 Child table.

──────────────────────────────────────────────────の Continued on the front page (72) Inventor Tetsuo Yokoyama 1099 Ozenji Temple, Aso-ku, Kawasaki City, Kanagawa Prefecture Inside Hitachi, Ltd.System Development Laboratory Co., Ltd. (72) Ryuichi Suzuki 1099 Ozenji Temple, Aso-ku, Kawasaki City, Kanagawa Prefecture Co., Ltd. (56) References JP-A-1-201236 (JP, A) JP-A-3-140140 (JP, A) JP-A-4-195479 (JP, A) (58) Field (Int.Cl. ⁶ , DB name) G06T 1/00 A61B 5/00 A61B 6/00

Claims (9)

(57) [Claims] A means for extracting a desired region in a target from three-dimensional voxel data of the target by a connected region expanding process;
Means for outputting extraction history data indicating at which point each point in the extraction region was extracted, and referring to the extraction history data and three-dimensional voxel data to create a three-dimensional image of the extraction region at an arbitrary time point Means for extracting a region outside the desired region in the course of the connected region expanding process, and indicating one point in the region outside the desired region and obtaining coordinates of the specified point. When,
Starting from the designated point, an area comprising means for specifying a connecting portion between the protruding region and the extraction target region by a connected region expanding process for the extraction history data, and means for deleting the protruding region based on the connecting portion Extraction device. 2. An area extracting apparatus according to claim 1, wherein said deleting means includes means for executing a desired area extracting process again by a connected area expanding process on the three-dimensional voxel data from which the protruding area has been deleted. . 3. The method according to claim 1, wherein the deleting means starts from the connecting part after the connecting part is specified, and extracts the entire area to which the designated point belongs by a connected area expanding process for the extraction history data; Execute the entire deletion process,
2. The area extracting apparatus according to claim 1, further comprising means for adding an attribute as a non-extracted area to the entire area. 4. The pointing means instructs a point in an extraction area on a three-dimensional display image displayed on a projection plane, and sets the three-dimensional coordinates of the designated point to the projection plane passing through the designated point. 2. The region extracting apparatus according to claim 1, further comprising means for obtaining coordinates of a point at which the normal line intersects with the extracted region in a three-dimensional space for the first time. 5. The method according to claim 1, wherein the specifying means executes a connection area expansion process for specifying a connection portion between a designated area starting from the expansion start point and an area protruding from the designated point, thereby obtaining the expansion source point and the expansion source point. The extraction time point of a new extension point to be connected is obtained by referring to the extraction history data output by the connection area extension processing. When the extraction time point of the connection point is one point earlier than the extraction time point of the expansion source point, 2. The region extracting apparatus according to claim 1, further comprising means for expanding a region by setting a connection point as a new expansion point. 6. The new expansion by executing a connection area expansion process for specifying a connection portion between a desired area and an area outside of a designated point from the designated point to start the expansion. After the point is obtained, the extraction point of the extension point and the connection point further connected to the extension point is obtained with reference to the extraction history data output by the connection region extension processing. 6. The region extracting apparatus according to claim 5, further comprising means for adding the connection point to a point of the expansion source when the point is one point after the point in time. 7. The process of specifying a connection area extending from an instruction point for starting the expansion to a connection part with an area outside of a desired area by executing a connection area expansion process, thereby performing each expansion step. 2. The area extracting apparatus according to claim 1, further comprising a step of checking the number of the extension points and, when the number of the extension points becomes lower than a predetermined value, setting an area where the extension point is present at that time as the connection unit. . 8. The method according to claim 1, wherein the specifying means executes a connection area expansion process for specifying a connection area between a desired area and an area outside the designated point from which the expansion is to be started. Check the number of points, compare with the number of extension points before and after several points, and when the number of extension points takes the minimum value, or when the number of extension points changes significantly after a certain point in time, the point in time 2. The area extracting apparatus according to claim 1, further comprising means for setting an area where the extension point exists as the connection part. 9. A desired region in the object is extracted from the target multidimensional digital data by the connected region expanding process, the extracted region is displayed on a display surface, and the extracted region extends outside the desired region by the connected region expanding process. The extracted region is extracted, and from the one point in the protruding region, the connection area expansion processing is performed to reverse the time point extracted in the process of the connected region expansion to connect the desired region to the protruding region. A region extracting method for specifying a portion and deleting the protruding region based on the connection portion.