JPH07236634A - Diagnosis supporting device - Google Patents

Abstract

PURPOSE:To provide a diagnosis supporting device for precisely determining a tumor by using three-dimensional spheroidicity. CONSTITUTION:A three-dimensional image formed by cubic box cells, the length, width and height of which are equal to each other to a tested body is reconfigured by a reconfiguring device 5 by using a CT device to be taken as three- dimensional image information. A region suspected to be a tumor is extracted from the three-dimensional image information by tomographic images. Subsequently, the volume of the extracted region is obtained, and the center of gravity of the extracted region is obtained. Subsequently, the radius of a sphere having the same volume as the volume of the region is obtained, and the spheroidicity defined by the ratio of the volume of the extracted region included in the sphere to the volume of the sphere having the above radius with the center of gravity as the center is obtained. In the characteristic space of the obtained radius and spheroidicity, when the radius and spheroidicity of the extracted region are within a fixed range, the extracted region is regarded as a tumor.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a diagnostic support device for detecting the presence or absence of a tumor based on three-dimensional image information.

[0002]

2. Description of the Related Art Various diagnostic support devices for detecting the presence or absence of a tumor have been proposed. One of them is a method of determining the circularity of a specific area using two-dimensional image information and determining whether or not the area is a tumor. The procedure is as described below.

First, a specific region (shadow) for which it is judged whether or not it is a tumor is extracted from the two-dimensional image information. Next, the area S of the extracted region is calculated, and the center of gravity of the extracted region is calculated. Next, the radius r of a circle having the same area as the extracted region
(Hereinafter, referred to as the equivalent radius of the extraction area) is calculated by the following equation.

R = (S / π) ^1/2 (1) Subsequently, the circularity of the extracted region is obtained. The circularity is defined by the following equation, based on the ratio of the area of the extraction region included in the circle having the equivalent radius r obtained above with the center of gravity obtained above as the center.

Circularity = area of extraction region included in circle / S (2) As a result, in the feature space determined by the equivalent radius r and the circularity, the equivalent radius r and the circularity of the extraction region are included within a certain range. If it is, the extracted area is considered to be a tumor.

As described above, not only one image but also an area in each two-dimensional image is extracted based on a large number of three-dimensionally distributed two-dimensional image information, and the equivalent radius and circularity of the extracted area are extracted. There was also a device for finding each image and detecting it for each image.

[0007]

In the above-mentioned conventional example, since it is based on the two-dimensional information of the equivalent radius and the circularity, it is difficult to correctly judge the three-dimensionally distributed tumor. In a conventional computer tomography apparatus (hereinafter simply referred to as a CT apparatus) or the like, the slice interval is large compared to the pixel size of a pixel, so that even when making judgments on a large number of images, they are processed as three-dimensional images. There was a limit.

The present invention has been made in view of the above circumstances, and provides a diagnosis support apparatus for accurately determining whether or not a suspicious lesion part exists in a certain region of a subject, if the region is a lesion part. The purpose is to do.

[0009]

The diagnosis support apparatus of the present invention obtains a tomographic image in which the size of the two-dimensional pixels of the tomographic image and the interval between the tomographic images are substantially equal to each other, and the vertical, horizontal and height are substantially equal to each other. It is provided with a means for obtaining three-dimensional image information of the subject constituted by three-dimensional pixels, a means for extracting a specific three-dimensional area, and a means for obtaining at least one characteristic amount of this area, and according to the characteristic amount. It is characterized in that the diagnosis support information of the area is obtained.

[0010]

The three-dimensional region to be diagnosed is extracted from the three-dimensional image information, and the sphericity (three-dimensional feature) related to this extracted region is used as a feature amount, that is, a parameter for determining the attribute such as a tumor. As a result, the diagnostic ability is improved.

[0011]

EXAMPLES Before describing the examples of the present invention, the outline of the present invention will be described first. The present invention utilizes the property that a tumor or the like becomes round (becomes a sphere) when it deteriorates or progresses. Therefore, a three-dimensional specific area is extracted from the three-dimensional image information, the sphericity of the area is obtained, and the sphericity is used as one of the feature amounts to determine whether the area is a tumor. The procedure is described below.

First, three-dimensional image information composed of a large number of tomographic images is created using, for example, a CT device. At this time, the interval in the slice direction of each tomographic image is made substantially equal to the two-dimensional pixel (pixel) size of the image. That is, the three-dimensional image information is composed of cubic three-dimensional pixels (voxels) having the same height, width, and height. For this purpose, it is preferable to use a CT device of helical scan type. Next, a region suspected of being a tumor (hereinafter abbreviated as a region of interest) is extracted from the three-dimensional image information. The extraction is performed for each tomographic image, but the extraction result represents a three-dimensional area. As an extraction method, for example, there is a method of dividing a pixel value into a region and a region outside with a threshold value. Next, the volume V of the extracted area is obtained, and the center of gravity of the extracted area is obtained. Then, the radius r of the sphere having the same volume as this volume (hereinafter referred to as the equivalent radius of the extraction region) is obtained from the volume of this region.

R = (3V / 4π) ^1/3 (3) The ratio of the volume of the extraction region included in the sphere to the volume of the sphere having the equivalent radius r centered on the center of gravity obtained above is defined as the sphericity. , The sphericity is calculated by the following formula.

Sphericity = 3 × (volume of extraction region included in sphere) / 4πr ³ (4) The sphericity is not limited to this definition, and various definitions are possible. In the feature space of the equivalent radius r and the sphericity calculated above, when the equivalent radius r and the sphericity of the extraction region are within a certain range, the extraction region is considered to be a tumor. The region of interest to be diagnosed is not limited to the tumor.

Embodiments of the present invention will be described below with reference to the drawings. As an example, a helical scan CT
An example of obtaining image information of the chest by the device will be described.

FIG. 1 is a block diagram showing the configuration of a diagnosis support device according to an embodiment of the present invention. The X-ray generation unit 1 that irradiates the subject 2 placed on the bed 3 with a fan-shaped X-ray beam and the detector array arranged in an arc shape detect the X-rays transmitted through the subject 2. It has an X-ray detection unit 4.
The X-ray detection unit 4 collects projection data indicating the X-ray transmittance. Regarding the data acquisition, the projection data at an angle of 360 ° or more can be continuously collected. That is, although not shown, the X-ray generator 1 and X
The rotating unit that holds the line detection unit 4 is attached to the fixed unit via a slip ring. The bed 3 continuously moves in the body axis direction during the continuous rotation of the X-ray generation unit 1 and the X-ray detection unit 4, that is, a helical scan is thereby performed. The reconstruction device 5 is a device that reconstructs an image by performing convolution and back projection processing on the collected projection data. The console 6 is C
Gives various commands to PU7. The CPU 7 controls the reconfiguring device 5 and the display device 8 via the system bus 10. The display device 8 has a function of displaying the reconstructed image. The image storage unit 9 stores the reconstructed image, and the system bus 10 includes the reconstructing device 5 and the CPU.
A bus for connecting the display device 7, the display device 8, and the image storage device 9 to each other and transmitting signals.

The operation of the embodiment thus configured is shown in FIG.
Will be described in detail with reference to. Note that the operation described in FIG. 2 shows a process of obtaining the equivalent radius of the ROI, which is a parameter for determining whether the ROI for one clinical example is a tumor, and the sphericity. It is a thing.

First, in step S1, the pixel size,
That is, the pixel size Δp (mm) and Δd of the image
(Mm) is determined. Helical scan CT
In the apparatus, the X-ray tube in the X-ray generation unit 1 moves in a helical or spiral locus as shown in FIG. 3, assuming that the subject 2 is stationary. Here, the Z axis is the body axis direction of the subject. Therefore, the projection data is also collected in a helical shape. Here, the relationship between the projection data and the reconstructed image will be described with reference to FIG. When the projection position of one projection data is represented by one point on the Z-axis, the helical scan continuously moves as shown in FIG. 3, so the projection data can be represented as one straight line. The first image is reconstructed using the projection data A of 720 degrees shown in FIG. At this time, the pixel size Δp of the reconstructed image is set to be equal to the slice interval Δd of the image. Here, Δd is determined as follows.

As shown in FIG. 5A, for example, the image matrix is 512 × 512, the pixel size is Δp, and the human chest region is 400 mm × 400 mm. Then, in order to include the entire chest in the image, Δ
It is understood that p may be set to a value calculated by the following equation.

Δp = 400/512 = 0.78125 (5) The moving distance D of the projection data in the Z-axis direction during one rotation of the X-ray tube is set to 10 mm, and the number of projection data collected during one rotation is set. Assuming 1000, Δd is as follows.

Δd = (10/1000) × k (6) Here, k is an integer. Since k is an integer,
If k = 80, Δd = 0.8 and Δd =
The value is close to 0.78125.

Therefore, here, Δp = Δd = 0.
8 The initial value of the i-th image to be reconstructed is set to 1 in step S2, and scanning is started in step S3. When the projection data A necessary for reconstructing the first image is collected during the helical scan, the projection data A is convolved and backprojected to obtain Δp determined in step S4, and the first image is obtained. Reconstruct. In step S5, i is incremented (increased by 1), and in step S6 it is determined whether i> n. Here, n is the number of images reconstructed for the subject 2.
If i> n is not satisfied, the process returns to step S4 and the above operation is repeated. In step S4 for the second time, as shown in FIG.
The second image is reconstructed using the projection data B of 0 degree. Similarly, this is repeated until the nth image is reconstructed. Note that these continuous images can be reconstructed more efficiently if they are reconstructed using the continuous reconstructing method disclosed in Japanese Patent Application No. 4-168919 by the same applicant.

The reconstructed first to nth images are Δd
= Δp, as shown in FIG. 6, the voxels have a uniform length, width, and height, and the first image 11 and the second image 1
Three-dimensional image information of the subject 2 is created from the second and third images 13. Further, in FIG. 6, the center of the voxel is represented as a sampling point. When the reconstruction is completed up to the n-th image, that is, when i> n is determined in step S6, i = 1 is set in step S7, and the interest is calculated from the i-th image in step S8, that is, the first image here. The area is extracted. A method of extracting a region of interest will be described below.

The i-th chest CT image is schematically shown in FIG. 5 (a). In FIG. 5A, the region of interest to be extracted is indicated by the diagonal line in the figure. FIG. 5B is an enlarged view thereof. An example of the profile of FIG. 5 (b) is shown in FIG. 5 (c). The tumor generally has a monomodal chevron shape as shown in FIG. 5 (c). In this case, the region of interest can be extracted by extracting the pixels having the threshold value TH or more. Tumors are usually complex and may not be extracted by the threshold alone. In this case, a more complicated method needs to be used. For example, the method described in “Medical Image Processing”, Yuichi Imazato, et al., Shokoido Publishing, pp. 166-171 can be used.

After the region of interest is extracted from the i-th image in step S8, i is incremented (increased by 1) in step S9, and it is determined in step S10 whether i> n. If i> n is not satisfied, the process returns to step S8 and the same operation is repeated until the nth image. When the extraction of the region of interest from the n-th image is completed, that is, when i> n, the volume of the region of interest is obtained in the next step S11.

When step 11 is executed, the shaded area in FIG. 5B is the region of interest. The number of voxels contained in this area is S _i . Letting V be the volume of the entire region of interest, V is obtained by the following equation (7).

[0027] In step 12, the center of gravity of the region of interest is obtained. The center coordinates of voxels in the region of interest are (x ₁ , y ₁ , z ₁ ), (x ₂ ,
y ₂ , z ₂ ), ..., (x _j , y _j , z _j ), ...
., (X _V , y _V , z _V ), the center of gravity (X ₀ , Y
₀ , Z ₀ ) is calculated by the following equation.

[0028] After obtaining the center of gravity, in step 13, the equivalent radius r of the region of interest is obtained. The equivalent radius r is obtained as follows as the radius of a sphere having a volume equal to the volume of the region of interest.

R = (3V / 4π) ^1/3 (9) Here, let Q be a sphere having an equivalent radius r centered on the center of gravity (X ₀ , Y ₀ , Z ₀ ). In step 14, the volume VR of the entire region of interest contained in the sphere Q is obtained. Specifically, if the number of voxels in the region of interest included in the sphere Q is found for each image and this is SR _i , the volume VR of the entire region of interest included in Q is found by the following equation.

[0030] In step 15, the sphericity of the region of interest is calculated. The sphericity is defined by the ratio of the volume of the sphere Q to the volume VR of the entire region of interest contained in the sphere Q, and can be obtained as follows.

Sphericity = 3VR / 4πr ³ (11) Two parameters for determining the attribute (whether tumor or not) of the extracted region of interest for one clinical example by performing the operations from step 1 to step 15. This means that a certain equivalent radius r and sphericity have been obtained.

The above-mentioned steps 1 to 15 are applied to the clinical images of many examples, and the equivalent radius r and the sphericity are obtained for each example. As a result,
FIG. 7 shows, for example, whether the region of interest of each clinical case was a tumor or not in the feature space defined by the equivalent radius r and the sphericity. Figure 7
In, the horizontal axis is the equivalent radius and the vertical axis is the sphericity. ◯ indicates clinical cases with tumors, and × indicates clinical cases without tumors. In the feature space of FIG. 7, a range separating the tumor and the tumor is determined as a closed curve L, for example. Once the closed curve L is obtained, the tumor can be automatically identified from r and the sphericity thereafter. That is, the equivalent radius r and the sphericity of the region of interest are calculated in the image in which the tumor is desired to be determined,
(a) If it is inside the closed curve L, it is determined to be a tumor, and if it is outside the closed curve L, it is determined to be not a tumor.

As described above, according to the present embodiment, for example, in a helical scan type CT device, the slice interval is made equal to the pixel size of pixels, that is, three-dimensional pixels called voxels having uniform vertical, horizontal and height. The three-dimensional image information consisting of is reconstructed, and for the region of interest extracted from this three-dimensional image information, the equivalent radius of this region of interest,
By obtaining the sphericity and using the equivalent radius and sphericity as parameters for diagnosing the region of interest, it becomes possible to three-dimensionally diagnose a three-dimensional diagnosis target such as a tumor. As compared with the two-dimensional diagnosis, a much more accurate diagnosis can be realized for the diagnosis target.

The present invention is not limited to the above-mentioned embodiments, but can be implemented with various modifications. For example, with respect to the method of extracting the region of interest in the above-described embodiment, the region of interest is extracted from the entire image. However, with this method, there may be a case where a region located away from the target region is extracted. Since tumors are generally small, it can be technically difficult to extract them from the entire image. Therefore, as a modified example of the method of extracting a region of interest, for example, as shown in FIG.
The operator sets a frame surrounding the region to be extracted, that is, a rectangular ROI, and extracts the region of interest only inside the cube extending in the Z-axis direction. That is, the area is extracted by the threshold method only inside the ROI. In this way, since the tumor is extracted within a narrow range, the tumor can be extracted easily and accurately, and the calculation time can be shortened. Further, as another modified example, similarly to the modified example described above, the lung field portion may be extracted from the entire image in advance by another means, and the region of interest may be extracted in the lung field portion. If only inside the lung field, the tumor can be detected by a relatively simple method such as a threshold method.

Further, in the above description, the diagnostic support information for the region of interest was obtained from the two parameters of the equivalent radius and the sphericity, but the equivalent radius does not necessarily have to be used and may be determined only by the sphericity. . The target of diagnosis is
It is not limited to the chest of the subject (subject).

Further, in the description of this embodiment, the three-dimensional image information is obtained by using the CT device of the helical scan type, but the present invention is not limited to this, and the three-dimensional image information may be obtained by a magnetic resonance imaging device or the like. good.

[0037]

As described above, according to the present invention, a suspicious region which is considered to be a lesion part in a subject is judged by using a sphericity which is a three-dimensional feature quantity.
It is possible to provide a diagnostic support device capable of improving the ability to detect a tumor or the like.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an embodiment of a diagnosis support device according to the present invention.

FIG. 2 is a diagram for explaining the operation of one embodiment of the present invention.
The flowchart figure explaining operation | movement which calculates | requires equivalent radius r and sphericity about two clinical examples.

FIG. 3 is a diagram for explaining a scan trajectory in a helical scan type CT device.

FIG. 4 is a diagram showing a relationship between projection data and an image generated as a result of the helical scan shown in FIG.

5 is a diagram illustrating an example of a method of extracting a region of interest in the operation of FIG.

FIG. 6 is a diagram showing a three-dimensional image composed of voxels evenly arranged.

FIG. 7 is a diagram showing whether or not a clinical image of a large number of cases is a tumor in the feature space represented by the equivalent radius r and the sphericity according to the present invention.

FIG. 8 is a diagram showing an example of extracting a tumor only within a rectangular ROI, as one modification of the method of extracting a region of interest according to the embodiment of the present invention.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... X-ray generation part, 2 ... Subject, 3 ... Bed, 4 ... X-ray detection part, 5 ... Reconstruction device, 6 ... Operation console, 7 ... CPU, 8 ... Display device, 9 ... Image storage part, 10 ... system bus, 11 ...
First image, 12 ... Second image, 13 ... Third image.

Claims (4)

[Claims] 1. A tomographic image in which the size of a two-dimensional pixel of the tomographic image and the interval between the tomographic images are substantially equal to each other is used to obtain the longitudinal, lateral,
And a means for obtaining three-dimensional image information of the subject composed of three-dimensional pixels having substantially the same height, a means for extracting a specific three-dimensional area, and a means for obtaining at least one characteristic amount of this area. A diagnostic support device, wherein the diagnostic support information of the area is obtained according to the feature amount. 2. The feature quantity is at least one of a radius and a sphericity of a sphere having the same volume as the area, and it is determined whether or not the extracted area is a tumor. The diagnostic support device described in 1. 3. The means for extracting the specific area, wherein a frame surrounding the specific area is set, and the extraction is performed only inside the frame. Diagnosis support device. 4. The diagnosis support apparatus according to claim 1, wherein the means for obtaining the three-dimensional image information of the subject is a helical scan type computer tomography apparatus.