JPH07265300A - Diagnosis supporting device - Google Patents

Abstract

PURPOSE:To reduce the number of images read by a doctor by cross-section- converting the images in the interested region extracted from multiple axial tomographic images to prepare multiple columnar tomographic images, and using these tomographic images for diagnosis. CONSTITUTION:The projection data outputted from an X-ray detection section 4 are temporarily stored in a raw data memory section 5 in a helical scan type CT device. The read projection data are fed to a reconstitution section 6, reconstituting calculations including convolution calculation and back projection calculation are made, and axial tomographic images are reconstituted and stored in an image memory section 7. A region extraction section 8 extracts a specific interested region, e.g. lung region, from the axial tomographic images, a cross section conversion section 9 cross-section-converts the axial tomographic images in the extracted interested region, and columnar tomographic images are prepared and stored in the image memory section 7. A display section 10 displays the axial tomograhic images or columnar tomographic images stored in the image memory section 7 according to the switching instruction.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a diagnosis support device for supporting diagnosis using a tomographic image.

[0002]

2. Description of the Related Art Recently, the mortality rate of lung cancer tends to increase year by year even in Japan. Mass screening is considered to be effective for early detection of lung cancer. Conventionally, mass screening for lung cancer has been performed on the basis of simple X-ray images, but simple X-ray images have limited ability to detect early cancer. On the other hand, a computer tomography apparatus (hereinafter abbreviated as CT apparatus)
It is known that the CT image taken by the method is easier to detect lung cancer than the plain X-ray image. Therefore, it is considered to perform a mass screening for lung cancer using a CT device.

However, there is a problem that must be solved even in mass screening using CT images. One of them is the examination time. Many people undergo group medical examinations as compared to individual medical examinations at medical institutions. Furthermore, the number of examinees tends to increase more and more as Japan shifts to an aging society in the future. Therefore, it is necessary to examine a large number of examinees in a short time. However, since it takes more time to capture an image with a CT apparatus than a simple X-ray imaging, it is practically impossible to use the CT apparatus for mass examination. However, recently, a CT device called a so-called helical scan system has been developed, and it has become possible to capture a CT image in a relatively short time. If such a CT device capable of high-speed imaging is applied to mass screening for lung cancer, the problem of screening time can be solved.

However, the mass examination by CT images has a problem not only in the examination time but also in that the number of CT images read by the doctor increases. A simple X-ray image can be used to read the entire lung field with only one image, but a CT image is an axial tomographic image orthogonal to the body axis, and therefore a large number of C
Unless the T image is used, the entire lung field cannot be read. For example, the length of the lung field in the body axis direction is 30 cm, C
If the slice pitch of the T image is set to 10 mm, 30 CT images are needed to read the entire lung field, and the doctor needs to read 30 times as many images as the plain X-ray image. Therefore, there is a problem that interpretation takes time.

Given that the number of people who undergo group medical examinations is on the increase, the time required for doctors to read images will increase in the future. Therefore, it is desired to support the interpretation (diagnosis) of doctors.

The ultimate goal of diagnostic support is automatic diagnosis. C
If the pattern recognition of the T image is performed and only the CT image of the examinee suspected to have lung cancer is presented to the doctor, the number of CT images to be interpreted by the doctor is considerably reduced. However, pattern recognition of natural images such as CT images is very difficult. Therefore, the misdiagnosis (false negative or oversight) of abnormal cases occurring in the process of pattern recognition is much more than that of the case of image interpretation by a doctor, and the practical application of automatic diagnosis has not yet become a reality.

[0007] Further, a display method such as displaying a large number of CT images continuously (cine display) or displaying them at the same time has been devised. Shortening cannot be achieved.

Therefore, it is required to reduce the number of CT images to be interpreted as the first step of diagnosis support.
In the CT device of the helical scan system, it is possible to reduce the number of CT images to be photographed by increasing the feed pitch of the top plate per one rotation of the X-ray tube. This is not preferable because the image quality deteriorates.

The request for shortening the interpretation time mentioned above is not limited to the case of the group medical examination but also exists in the case of general individual diagnosis. Further, the diagnosis target is not limited to lung cancer, and the support device is useful for diagnosis of any part.

[0010]

As described above, conventionally, there has been a demand for the development of a diagnostic support device that shortens the time required for doctors to interpret images in medical examinations, especially in mass examinations. Therefore, an object of the present invention is to provide a diagnosis support device that supports a doctor's diagnosis so that interpretation can be performed in a short time.

[0011]

According to the present invention, in a diagnosis support apparatus utilizing a plurality of axial tomographic images, a means for extracting a region of interest from at least one of the plurality of axial tomographic images, and a plurality of axial tomographic images. A plurality of coronal tomographic images are used for diagnosis by means of cross-section conversion of the image in the region of interest extracted by the extracting means of the images to create a plurality of coronal tomographic images.

[0012]

According to the present invention, the axial tomographic image is cross-sectionally converted only in the region of interest to obtain the coronal tomographic image, and the diagnosis is performed by the coronal tomographic image, so that the doctor interprets the image as compared with the case of interpreting the axial tomographic image. The number of images to be displayed can be reduced.

[0013]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the diagnosis support device according to the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of this embodiment. The present embodiment is a system for displaying an image for diagnosis support based on a large number of CT images taken by using a CT device of helical scan type.

The CT apparatus includes a bed 3 on which the subject 2 is placed, an X-ray generator 1 for irradiating the subject 2 with a fan-shaped X-ray beam, and detections arranged in an arc shape. X-ray detector 4 for detecting the X-ray beam transmitted through the subject 2
Have and. The X-ray detector 4 collects projection data indicating the X-ray transmittance. Here, the projection data at an angle of 360 ° or more can be continuously collected so that the helical scan can be executed. That is, although not shown, the gantry rotating unit that holds the X-ray generation unit 1 and the X-ray detection unit 4 is attached to the gantry fixing unit via a slip ring, and the X-ray generation unit 1 is attached to the subject. It is possible to rotate continuously around. Sleeper 3
Moves continuously in the body axis direction during continuous rotation of the X-ray generation unit 1 and the X-ray detection unit 4. As a result, the helical scan is performed.

The projection data output from the X-ray detection unit 4 is temporarily stored in the raw data storage unit 5. Raw data storage 5
The projection data read from the image data is supplied to the reconstruction unit 6, and reconstruction operations including convolution calculation and back projection calculation are performed on the projection data to obtain an image (axial (axial) tomographic image). Is reconstructed. The reconstructed image is stored in the image storage unit 7. A region extraction unit 8, a cross section conversion unit 9, and a display unit 10 are connected to the image storage unit 7.
The region extraction unit 8 extracts a specific region of interest (ROI), for example, a lung field region, from the axial tomographic image. Section conversion unit 9
Creates a coronal tomographic image by subjecting the axial tomographic image in the region of interest extracted by the region extracting unit 8 to cross-section conversion processing. The coronal tomographic image is also stored in the image storage unit 7. The display unit 10 displays the axial tomographic image or the coronal tomographic image stored in the image storage unit 7. Although not shown, the image storage unit 7 or the display unit 10 is also connected to a console for giving an instruction to switch the image (axial image / coronal image) displayed on the display unit 10.

The operation in the case of conducting the group medical examination by using the embodiment thus constructed will be described in detail with reference to the flowchart of FIG. The operation described with reference to FIG. 2 is performed by imaging a lung field region of a subject with a helical scan CT apparatus, and creating a coronal tomographic image that a doctor actually reads from a large number of reconstructed axial tomographic images (cross section). It is the process of converting and displaying.

In step S1, the pixel size p (mm) and slice pitch d of the axial tomographic image to be reconstructed are set.
(Mm) is determined. Set the image matrix size to 512
× 512, the size of the shooting area is 400 mm × 400 mm
Then, in general, the pixel size p is as follows.

P = 400/512 = 0.78125 mm (1) However, in this embodiment, since the axial tomographic image is cross-sectionally converted into a coronal tomographic image and used for image interpretation, the pixel size p of the axial tomographic image and the coronal tomographic image are compared. It is preferable to determine the pixel sizes to be the same. Therefore, the pixel size p of the axial tomographic image is determined as follows.

In the helical scan method, the moving distance in the body axis direction of the subject per rotation of the X-ray generator 1 and the X-ray detector 4 is D, and the number of X-ray exposures per rotation (the number of views). ) Is M, it is known that an axial tomographic image can be reconstructed at the following pitch d. This d is the pixel size of the coronal tomographic image.

D = (D / M) × k (2) Here, k is a constant (integer). D = 15 mm, M =
When 1000 and k = 120, the pixel size d of the coronal tomographic image becomes 0.8 mm, and d is the pixel size p (= of the general axial tomographic image defined by the above formula (1).
The value is close to 0.78125). Therefore, here, the pixel size p of the axial tomographic image is determined by the following equation.

P = d = 0.8 mm (3) In step S 2, the part of the subject including the lung field is scanned, and the projection data is stored in the raw data storage unit 5.

In step S3, as indicated by a thick solid line in FIG. 3, the pitch is an integral multiple of the pitch d of the above expression (3) (for example, 50 times 40 mm) and the pixel size p of the above expression (3) is set. So that the axial tomographic image Ij (j = 1
~ M) is reconstructed. The pitch of the axial tomographic image to be reconstructed may not be an integral multiple of d, but may be a value obtained by dividing the total length of the scan by an appropriate integer. For example, when the scan length (length of the lung field in the body axis direction) is 30 cm, 11 axial tomographic images are reconstructed at a pitch of 30 mm. In step S4, this axial tomographic image Ij (j =
The lung field region is extracted from 1 to m). Various techniques for extracting a specific region from a CT image are known. Here, since the accuracy is not required so much and there is a considerable difference in CT value between the lung field and other parts, a method of extracting using a threshold value is used as shown below.

(1) Determine the threshold value TH. Since most of the lung field and the outside of the body are air, it may be set to about TH = −500. (2) The CT image is binarized with the threshold value TH. That is, if the CT value of each pixel is less than −500, the pixel value is set to 0, and if the CT value is −500 or more, the pixel value is set to 1, and a binary image is created.

(3) It should be noted that the pixels having a pixel value of 1 in the binarized image include objects other than the subject, such as a bed top, and thus need to be excluded. Since the shape, area, CT value, etc. of objects other than the subject are known in advance, C
A threshold value is set according to the T value, and the contours of objects other than the subject such as the bed top are obtained by the same procedure as above. Then, by changing the pixel value of the pixel inside the obtained contour to 0, it is possible to remove objects other than the subject such as the bed top from the binarized image. As a result, as shown in FIG.
A binarized image is obtained. The shaded area represents a set of pixels having a pixel value of 1.

(4) The pixel value of the binary image is searched to find the pixel having the minimum Y coordinate and the minimum X coordinate among the pixels having the pixel value of 1. This pixel is one of the pixels at the end of the image of pixels having a pixel value of 1.
The pixel value is 0 because the edge of the image is usually air. The pixel value of the skin is 1. Therefore, the pixel found here can be regarded as the skin, that is, one point on the body contour.

(5) A body is extracted from one point of the body contour by a known contour extraction method (for example, “Medical Image Processing”, Yuichi Imazato, et al., Shokodo Publishing, page 171). Extract the contour (closed curve). Furthermore, if necessary, points that are considered to be affected by noise are removed from the body contour, or the body contour is smoothed.

(6) Similar to the extraction of the body contour, the contour of the lung field is next extracted. The contour of the lung field is inside the body contour,
In addition, it is obtained as a contour line that surrounds the region where the pixel value is 0. Since there are two lung field regions, the right lung and the left lung, it is assumed that two lung field contours are extracted.

(7) Determine the horizontal lines L1 and L2 contacting the top and bottom of the lung field contour. As is clear from FIG. 4, the horizontal line L1 is the smallest Y coordinate Y among the pixels on the contour of the lung field.
A horizontal line passing through the pixel of 1 and L2 is a horizontal line passing through the pixel of the largest Y coordinate Y2.

(8) The procedure described above relates to one axial tomographic image, and all axial tomographic images Ij
The above processing is performed for (j = 1 to m) to determine the horizontal lines L1j and L2j for each tomographic image.

(9) The minimum value Y1m of the Y coordinate Y1j of the horizontal line L1j and the maximum value Y2m of the Y coordinate Y2j of the horizontal line L2j of all the images are obtained. (10) The range of Y coordinates of Y1m to Y2m thus obtained is the range of the lung field region in the axial tomographic image. However, since the extraction accuracy of the lung field contour may not be so good in some cases, the following Ymim-Ymax range is determined as the range of the lung field region in order to provide a margin.

Ymin = Y1m−α × (Y2m−Y1m) (4) Ymax = Y2m + α × (Y2m−Y1m) (5) where α is a constant for giving a margin, and α =
Set to about 0.1.

In step S5, as shown by the thin solid line in FIG. 3, a large number of axial tomographic images Ii (i = 1 to n) with the pitch d and the pixel size p shown in the equation (3).
Reconfigure. The reconstructed image is stored in the image storage unit 7. Note that the reconstruction range of the axial tomographic image in step S5 may be the entire area of one screen, or in order to shorten the reconstruction time, the range Ymin to be extracted in step S4 as indicated by the diagonal lines in FIG. You may make it reconstruct only the range of Ymax. Furthermore, for continuous slice reconstruction, images can be reconstructed more efficiently and in a shorter time by using the continuous reconstruction method disclosed in Japanese Patent Application No. 4-168919 by the same applicant. In step S6, the range Ymin-
An axial tomographic image of Ymax is supplied from the image storage unit 7 to the cross-section conversion unit 9 and is subjected to cross-section conversion processing, and the lung field region Ymi
A coronal tomographic image in the range of n to Ymax is created. The pitch of the coronal tomographic image can be set to an integral multiple of the pixel size of the axial tomographic image, but it is preferable to make it equal to the pixel size of the axial tomographic image. The coronal tomographic image is also stored in the image storage unit 7. The cross-section conversion is a technique widely used in CT apparatuses and is well known, so a detailed description thereof will be omitted. Since the pitch of the axial tomographic image captured by the normal scan method, which is not the helical scan method, is relatively long, the normal cross-section conversion (creation of a coronal tomographic image) requires interpolation, but this embodiment adopts the helical scan method. However, since the pixel size d of the coronal tomographic image and the slice pitch of the axial tomographic image are made equal, interpolation is not necessary. Therefore, a relatively accurate coronal tomographic image can be created.

In step S7, the display unit 10 sequentially displays the coronal tomographic images stored in the image storage unit 7,
The doctor interprets this. The images may be displayed one by one, a plurality of images may be divided into one screen, or may be displayed simultaneously on a plurality of screens.

The doctor may observe only the coronal tomographic image and interpret the image, or if he or she suspects that there is a shadow on the coronal tomographic image, it is displayed by giving a switching instruction from a console not shown. Instead of the existing coronal tomographic image, a plurality of axial tomographic images used to create the coronal tomographic image may be sequentially displayed, and the image may be read using the axial tomographic image.

As described above, according to this embodiment, since a coronal tomographic image of only the region of interest is created from a plurality of axial tomographic images, the number of CT images to be read is reduced,
Interpretation time can be shortened. Therefore, the CT device can be applied to mass screening for lung cancer and the like. Further, since the pitch and the pixel size of the axial tomographic image are made equal by the CT device of the helical scan system, interpolation processing is not required, so that the cross section can be converted from the axial tomographic image to the coronal tomographic image with high accuracy. Further, since the display can be switched from the coronal tomographic image to the axial tomographic image as needed, the deterioration of the diagnostic ability due to the cross-sectional conversion can be reduced and the diagnostic ability can be enhanced.

The present invention is not limited to the above-mentioned embodiments, but can be implemented with various modifications. For example, although the above description has been made with respect to diagnosis support at the time of mass screening for lung cancer, the present invention is not limited to this, and can be applied to general individual diagnosis, and the target of diagnosis is not limited to lung cancer. It can also be applied to site diagnosis. Further, although the region of interest is extracted by the threshold processing automatically, the operator may manually set Ymin to Ymax using a line cursor. Instead of strictly extracting the lung field region, the body contour range may be simply set as the lung field region extraction range Ymin to Ymax. This is easier to process, but the range Ymin to Y
On the other hand, since max becomes longer, the number of coronal tomographic images to be created also increases. Further, although the radiogram interpretation image was captured using a helical scan type CT device, a conventional scan type CT device may be used, or another modality,
For example, an image may be taken with a magnetic resonance imaging device or the like.

[0037]

According to the present invention, a cross-sectional conversion of an axial tomographic image only in a region of interest is performed to obtain a coronal tomographic image, and a diagnosis is performed using the coronal tomographic image. It is possible to provide a diagnosis support device capable of reducing the number of images to be interpreted by the.

[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 flow chart for explaining the operation of creating and displaying a coronal tomographic image in order to explain the operation of the embodiment of the present invention.

FIG. 3 is a schematic view of a human body seen from the side and a schematic view showing slice positions of a CT image.

FIG. 4 is a diagram showing a binarized image including a lung field region and a region extraction range.

FIG. 5 is a diagram showing a positional relationship between an axial tomographic image and a coronal tomographic image.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... X-ray generation part, 2 ... Subject, 3 ... Bed, 4 ... X-ray detection part, 5 ... Raw data storage part, 6 ... Reconstruction part, 7 ... Image storage part, 8 ... Area extraction part, 9 ... Section conversion unit, 10 ... Display unit.

Claims (3)

[Claims] 1. A diagnostic support apparatus utilizing a plurality of axial tomographic images, comprising: a unit for extracting a region of interest from at least one of the plurality of axial tomographic images; and a unit for extracting the plurality of axial tomographic images. A diagnostic support apparatus comprising: a unit configured to convert a cross-section of the extracted image in the region of interest to create a plurality of coronal tomographic images, wherein the plurality of coronal tomographic images are used for diagnosis. 2. A first display unit for displaying the plurality of coronal tomographic images, and a first display unit for displaying a plurality of axial tomographic images used to create a specific coronal tomographic image displayed by the display unit. The diagnosis support apparatus according to claim 1, further comprising two display means and a means for switching between the first and second display means. 3. The helical tomographic computer tomography apparatus captures the plurality of axial tomographic images, and the pixel size of the axial tomographic images is equal to the pitch of the axial tomographic images. Item 2. The diagnostic support device according to item 2.