JPH05130989A - Processor for ct image - Google Patents

Abstract

PURPOSE:To exactly extract a concerned area such as a lung area, etc., by providing means to obtain a mask pattern divided into the areas of closed areas between means to obtain two-dimensional binary images and means to execute an MIP processing. CONSTITUTION:The closed areas of the mask pattern is labeled by an area processor 24 so as to calculate the areas of the respective areas, and the mask pattern is prepared and stored in a mask image memory 25. Images reconstituted for each two-dimensional CT image are successively inputted to an original image memory 22 of a concerned area extracting device 21, an original image stored in the original image memory 22 is binarized with a threshold value set by a threshold value processor 23, and a picture element equal to or larger than the threshold value is defined as '0' and the picture element smaller than the threshold value is defined as '1' and they are outputted. A mask processor 26 ORs this mask pattern and the original image and only the picture element whose mask pattern is '1' is outputted. This masked original image is stored in an image memory 27. Thus, the CT value information of a three-dimensional concerned area is stored in the image memory 27 of a projection transformation processor 22.

Description

Detailed Description of the Invention

[0001]

The present invention relates to a CT image processing apparatus for efficiently observing a plurality of tomographic images in a CT apparatus for obtaining a tomographic image of a subject.

[0002]

2. Description of the Related Art In the case of current image diagnostic equipment such as an X-ray CT apparatus, image observation is performed by CRT or film. CR
The observation method on T is to display each slice image sequentially,
Reduce or display some slice images at the same time. Also, as a special method, it takes a huge amount of calculation time to
The three-dimensional image is reconstructed so that it can be observed from any direction by changing the viewpoint with a trackball or mouse. When observing with a film, hang it on Schaucasten and observe.

By the way, in photographing, for example, when observing a range of 200 mm, generally 20 to 40 images having a slice thickness of 5 to 10 mm are projected at intervals of 5 to 10 mm. The doctor will read all 20 to 40 images and make a diagnosis. When 10 similar patients are photographed on one day, the number of images to be interpreted on one day is as high as 200 to 400. Furthermore, if a mass screening system using an X-ray CT apparatus is implemented in the future, the number of subjects will be on the order of several hundreds and the time required for screening will be enormous. In particular, the effectiveness of X-ray CT has attracted attention for lung cancer, but this problem is important in realizing a lung cancer mass screening system using X-ray CT.

The possibility of three-dimensional image diagnosis has been investigated as a solution to this problem. However, a conventional three-dimensional image requires a lot of calculation time because it has a plurality of processes such as a rendering process for converting three-dimensional information into two-dimensional image information (which may be called a 2.5-dimensional image for this process).
Also, the display symmetry is limited to bones, skins, etc. where the CT value is clear and can be extracted by simple threshold processing, and since the extracted display symmetry is a binary image, the CT value is not reflected in the three-dimensional image, and the shape is Can only diagnose. For this reason, it has not been practically used routinely as much as it is used for simulation of surgical operations, and it is also inappropriate in terms of throughput.

On the other hand, in MRI, MIP (Maximum Inte
nsity Projection) has been attempted. M
IP is one of the methods for projecting three-dimensional image information onto a two-dimensional plane.
Then, as shown in FIG. 5, the method of projecting the maximum pixel value on the projection line can obtain a three-dimensional reconstructed image in a short time, and the image information of a plurality of images is compressed into one image and observed at one time. It will be possible. That is, when a three-dimensional image is composed of a plurality (n) of slice images in FIG. 5, a viewpoint to a pixel position of interest is set, and among the n slice images of a plurality of n existing at this viewpoint position. Pixels P ₁ , P ₂ , ..., P _n are taken out. Of the pixels P ₁ , P ₂ , ..., P _n , the pixel having the maximum density (gradation) is selected. These pixel selections are performed over the entire region of interest, and one image thus obtained becomes a projected two-dimensional image, and when this is displayed on the display surface, the two-dimensional image can be observed.

[0006]

In the case of MRI, the above MIP method is a convenient method for obtaining a two-dimensional image in which only blood vessels are extracted from a three-dimensional image including blood vessels. This is because the pixel density of blood vessels is high compared to the background. This MI
It has been attempted to apply P to a chest image of an X-ray CT apparatus. In the case of X-ray CT, the pixel value is a value corresponding to an absorption coefficient called CT value, the lung field region such as the bronchus is about -800 to -400, the lung cancer lesion is about -500 to 0,
Other soft tissues such as liver and heart are about -100 to +1
The value of 00 is shown. Therefore, when the MIP process is performed, the value of the soft tissue is larger, so that the bronchus, which is the site where the lung cancer or the like exists, becomes difficult to be projected. Therefore, masking is performed using a threshold value to exclude soft tissue having a high absorption value, and only the region of interest such as the bronchus is extracted. For example, the processing device includes the region-of-interest extraction device 1 and the projection conversion processing device 2 as shown in FIG. The region-of-interest extraction device 1 includes an original image memory 3, a threshold value processing device 4, a mask image memory 6, a mask processing device 5, and the like. The projection conversion processing device 2 includes an image memory 8, a projection conversion device 7, a processed image memory 9, and the like. The threshold value processing device 4 binarizes the original image A stored in the original image memory 3 with a set threshold value to obtain a binarized image B, which is stored in the mask image memory 6. The binary image B has a pattern in which the pixels corresponding to the region of interest (in this example, lung regions such as lesions and bronchus) are 1, and other soft tissues and the like are 0. The logical sum of the mask pattern B and the original image A is calculated, and only the pixels for which the mask pattern is 1 are output. The mask processing image C as the mask processing result is input to the image memory 8 of the projection conversion processing apparatus 2 and projected and converted by the projection conversion apparatus 7, and a desired three-dimensional image is obtained in the processed image memory 9. However, since the distribution range of the CT value of the soft tissue and that of the cancer tissue overlap, the lesion of the carcinoma is also deleted, or the soft tissue cannot be completely removed. Further, in the case of a partial volume (Partial Volume) in which the soft tissue is partially included in the slice, the value is lower than the original CT value. According to experiments, when the liver or heart is measured as a partial volume, it may have a value of about −500, which makes it difficult to separate it from the bronchi. That is, the pixel corresponding to the cancerous tissue on the mask pattern becomes 0, or a part of the soft tissue region that obstructs the observation becomes 1.

It is an object of the present invention to provide a CT image processing apparatus capable of accurately extracting a region of interest such as a lung field region.

[0008]

The processing apparatus of the present invention comprises means for performing threshold processing on each two-dimensional original image to obtain each two-dimensional binary image, and each two-dimensional binary image. For each, a means for obtaining an area for each closed region having a pixel value of 1, and for each two-dimensional binarized pixel, generate a mask pattern divided into pixels to be masked and pixels not to be masked according to the size of the area. For each two-dimensional original image, the original image pixel at the pixel position to be masked according to this mask pattern has a pixel value of 0, and the original image pixel at the pixel position not to be masked is left as it is, and The mask-processed two-dimensional original image is subjected to MIP processing to obtain a two-dimensional projected image (claim 1).

Further, in the MIP processing in the processing apparatus of the present invention, the maximum pixel value among a plurality of pixels on the projection line is calculated,
The thus obtained maximum pixel value for each projection line is set as the pixel value of the two-dimensional projection image on that projection line (claim 2).

Further, in the means for obtaining the area in the processing apparatus of the present invention, the two-dimensional binarized image is contracted to label the closed area having the pixel value 1 with respect to the contracted image. After the labeling, the binarized image is expanded, and the area obtained for each of the labeled closed regions is calculated for the image obtained by this expansion (claim 3).

[0011]

According to the present invention, since the means for obtaining the mask pattern divided by the area of the closed region is provided between the means for obtaining the two-dimensional binarized image and the means for performing the MIP processing, the extra soft portion is provided. It is possible to avoid leaving tissue on the projected image. (Claims 1 to 3).

Further, according to the present invention, the maximum value is obtained in the MIP process (claim 2), and the contraction process and the expansion process are performed in the process of obtaining the area to correctly segment the closed region. Achieve (claim 3).

[0013]

DETAILED DESCRIPTION FIG. 2 is a block diagram of a general X-ray CT apparatus.
Gantry-1 consisting of X-ray tube, X-ray detector and measuring circuit
0, a patient table 11 for carrying a patient, a high voltage generator 12 for supplying a high voltage to an X-ray tube, and an image diagnostic apparatus 1.
It consists of three. FIG. 3 shows an embodiment of the image diagnostic apparatus 13. The image diagnostic device 13 includes a magnetic disk 14, a main memory 15, an image reconstruction processing device 16, and a high-speed computing unit 1.
7, a simple three-dimensional processing device 18, a display unit 19, and a high-speed internal bus 20 connecting them. The simple three-dimensional processing device 18 includes a region of interest extraction processing device 21 and a projection conversion processing device 22.

With the above configuration, the magnetic disk 14 stores various data such as the original image A and is transferred to the main memory 15 via the bus 20 during the extraction processing. In addition, the measured projection data is temporarily stored in the main memory 15, and the image reconstruction processing device 16 processes the measured projection data to obtain a tomographic image on the slice plane. It becomes the original image A and the magnetic disk 14
Will be stored in. The high-speed computing unit 17 is a general CP
In U, processing (logarithmic processing, trigonometric function processing, other image processing, etc.) that slows down the processing speed is provided for high-speed processing. The simple three-dimensional processing device 18 is for the above-mentioned MIP processing, and is a characteristic part of the present invention, and its detailed configuration is shown in FIG. The display unit 19 is used for displaying various images.

The patient's chest 400 having the above configuration
Consider the case of shooting a range of mm. The scan is a spiral scan in which the table 11 is measured while moving and the wide range can be taken at high speed. Spiral scan (JP-A-62-87137, JP-A-62-139)
It is needless to say that in order to carry out No. 630), it is necessary to supply electric power from the stationary system to the rotating system using a slip ring or the like, and to deliver and receive signals. In the spiral scan, if the projection data at the slice position is obtained by interpolation, a slice image at an arbitrary position can be obtained. If the slice images are obtained at 5 mm intervals, the image reconstruction processing device 16
Projection data at a specified position is obtained by interpolation processing, and the well-known filtered back projection method (Filtered Back Projec
The image reconstructed by the motion method) is sequentially transferred to the magnetic disk 14 and the display unit 19. Finally, about 80 images are obtained and stored in the magnetic disk 14. As described above, in the normal use method, the tomographic images are sequentially displayed on the display unit 19 as the images are reconstructed even during photographing.

If the simplified 3D mode, which is a feature of this embodiment, is selected before photographing, an MIP-processed image is displayed on the screen. Next, the simple 3D processing will be described. The simplified 3D processing device 18 is composed of a region of interest extraction device 21 and a projection conversion processing device 22, and its detailed configuration is shown in FIG. The region-of-interest extraction device 21 of the present embodiment is the original image memory 2
2, a threshold processing device 23, an area processing device 24, a mask image memory 25, and a mask processing device 26. The projection conversion processing device 22 includes the image memory 27 and the projection conversion device 2.
8 and a processed image memory 29. These configurations are different in that an area processing device 24 is added, and other configurations are the same as those in FIG. However, the symbols are different. The ROI extraction device 21 performs the following internal operation for each two-dimensional CT image (two-dimensional original image). First, the reconstructed images are sequentially stored in the original image memory 22 of the ROI extracting device 21.
Entered in. The threshold value processing device 23 binarizes the original image stored in the original image memory 22 with the set threshold value -100, and outputs the pixels above the threshold value as 0 and the pixels below the threshold value as 1. Since the mask processing device 26 takes the logical sum of this mask pattern (binary image) and the original image and outputs only the pixels whose mask pattern is 1, the soft tissue must be 0 in the mask pattern. .. However, as described above, the soft tissue having a value of −800 to −100 remains as one pixel in the mask pattern. Therefore, the area processing device 24 in the present embodiment
Then, the closed area (collection of connected pixel values) of the mask pattern is labeled, and the area of each closed area is obtained.

In this embodiment, in order to prevent different tissue regions from becoming the same closed region, a shrinking process (also referred to as a thinning process) F ₁ is performed in advance before labeling. After the labeling, the original size is restored by the expansion process (also called thickening process) F ₂ by the amount reduced by the contraction process F ₁ . The pixel value at this stage is divided by the label (number) of the closed region to which the pixel belongs. For example, the bronchus is L
1. The lung cancer part is labeled as L2 ... By counting the pixels having the same label number, the area of each corresponding closed region is obtained in the unit of the number of pixels. (Area treatment F _3).
According to the experiment, the soft tissue region has a larger area than that of the bronchus or the affected area, and for example, 100 pixels or more are the soft tissue, and thus the soft tissue region can be separated by the threshold processing by the area value. That is, in the area processing F ₃ , the mask value is created by setting the pixel value of the closed area larger than the threshold area to 0 and the pixel value of the small closed area to 1, and the mask pattern is stored in the mask image memory 25. As a result, the mask pattern becomes almost completely maskable with the binarized pixel value at the soft tissue corresponding position being 0, and a mask pattern image convenient for projection conversion processing can be obtained.

Next, in the mask processing device 26, the mask processing is performed between the mask pattern image and the original two-dimensional original image (that was in the original image memory 22).
In this mask processing, the pixel position of the two-dimensional original image and the pixel position of the mask pattern image correspond to each other. Therefore, at these corresponding pixel positions, the pixel value on the mask pattern is 0. If there is, the pixel of the corresponding original image is forcibly set to 0, and if the pixel value on the mask pattern is 1, the pixel value of the corresponding original image is left as it is. Thus, by this masking process, the pixels of the original image corresponding to the pixel position indicating 0 on the mask pattern are masked, and the original image is convenient for the projection conversion process. This is stored in the image memory 27.

When these region-of-interest extraction processes are applied to all two-dimensional original images (slice images), the projection conversion processing device 2
The second image memory 27 stores CT value information of the three-dimensional region of interest. The projection conversion device 28 executes MIP processing,
The projection line is selected perpendicular to the slice to simplify the process.
That is, n mask processed images I1 (i, j) to In
If (i, j), the simplified 3D image P (i, j) output to the processed image memory 29 is

## EQU1 ## P (i, j) = MAX (I1 (i, j), I2
(I, j), ..., In (i, j)). Where M
AX means a value obtained by extracting the maximum pixel value for each pixel position in n images.

In the above embodiment, one threshold value is set and implemented. However, in the case of the X-ray CT apparatus, the background air indicates about -1000 and remains in the mask processed image.
It also remains as noise on the simple 3D image. Therefore, in the second embodiment, two threshold values are set, and the threshold value processing device 23 outputs the binarized image in which the pixel of T1 ≦ CT value ≦ T2 is set to 1. Subsequent processing does not change. T1,
T2 is a value such as −800 and 0. In addition, although the present embodiment has been described as an example of obtaining the maximum value, it may be applied to the extraction of the minimum value or the statistical processing amount such as the average value of M pixels depending on the extraction purpose. In addition, X-ray C such as MRI images
It can be applied to CT images other than T images.

[0021]

According to the present invention, only the region of interest can be accurately extracted in the MIP processing from the original image in the X-ray CT image or the like.

[Brief description of drawings]

FIG. 1 is a diagram showing an embodiment of a simple three-dimensional processing apparatus of the present invention.

FIG. 2 is an overall configuration diagram of an X-ray CT apparatus according to the present invention.

FIG. 3 is a diagram showing an embodiment of the image diagnostic apparatus of the present invention.

FIG. 4 is a configuration diagram of a conventional MRI MIP processing system.

FIG. 5 is an explanatory diagram of MIP processing by MRI.

[Explanation of symbols]

 21 Region of Interest Extractor 22 Projection Conversion Processor 24 Area Processor

Claims (3)

[Claims] 1. A CT image processing apparatus for projecting and outputting three-dimensional original image information composed of a plurality of reconstructed two-dimensional tomographic original images (CT images) of a subject on a two-dimensional plane.
Means for performing threshold processing on each of the two-dimensional original images to obtain each of the two-dimensional binary images; and for each of the two-dimensional binary images,
A means for obtaining an area for each closed region having a pixel value of 1, and a means for generating, for each two-dimensional binarized image, a mask pattern divided into pixels to be masked and pixels not to be masked according to the size of the area. Then, for each two-dimensional original image, the original image pixel at the pixel position to be masked according to the mask pattern has a pixel value of 0, and the original image pixel at the pixel position not to be masked is left as it is, and the mask processing. A CT image processing device comprising means for performing MIP processing on each of the two-dimensional original images thus obtained to obtain a two-dimensional projection image. 2. The MIP processing obtains the maximum pixel value among a plurality of pixels on a projection line, and the maximum pixel value for each projection line thus obtained is the pixel value of a two-dimensional projection image on that projection line. The CT image processing apparatus according to claim 1, wherein 3. In the means for obtaining the area, a two-dimensional binarized image is contracted to label a closed region having a pixel value of 1 on the contracted image, and the labeled image is labeled. 3. The CT image processing apparatus according to claim 1, wherein the binarized image is expanded afterward, and the area obtained for each of the labeled closed regions is calculated for the image obtained by this expansion.