JP3461201B2 - Image processing apparatus and image processing method - Google Patents

Description

Detailed Description of the Invention

[0001]

The present invention relates to a pixel value (density value) data of a region of interest of an image (for example, the contour of a region of interest such as the left ventricle of the heart when performing functional analysis of the left ventricle). The present invention relates to an image processing device and an image processing method for rendering based on the above.

[0002]

2. Description of the Related Art A contrast agent is injected into the body of a subject for imaging and
When using a fluoroscopic X-ray image for medical treatment, image processing for extracting a region of interest in the image is frequently performed. As an example of the image processing, there is a left ventricle contour extraction method for extracting a contour of the contrasted left ventricle from a left ventricle contrast image, for example.

The left ventricle contour extraction method is performed as follows. The operator sets a threshold value and determines whether the pixel value (density value) of each pixel is larger (or smaller) than the threshold value. A region having a pixel value whose pixel value is larger (or smaller) than the threshold value is defined as the left ventricle, and the boundary of this region is extracted as the contour of the left ventricle.

However, the above-mentioned conventional method has the following problems. First, in setting the threshold value, the threshold value should be set so as to include all desired regions in an image in which the variation or gradient of the density of the background (that is, a portion other than the desired region) is large with respect to the concentration of the contrast agent. When is set, an area other than the desired threshold value is also detected. Conversely, if you set the threshold value so that it does not include any areas other than the desired area, you may not be able to detect part of the desired area.
As a result, the contour of the desired region cannot be extracted.

Secondly, for example, when radiographing a left ventricular contrast image and fluoroscopy, it is not always possible to perform X-ray irradiation with the best positioning, and depending on the positioning, ribs or the like may be formed around the left ventricle in the image. The structure may be reflected. There are various difficulties in extracting the contour of the left ventricle by applying threshold processing to an image including such a structure.

Since the pixel value of the structure portion is also small as in the left ventricle, when the pixel value data of the entire image is discriminated by one threshold value set to discriminate the left ventricle, the Misdetection frequently occurs such that even a region is misidentified as the region of the left ventricle, and the extraction accuracy is significantly reduced.

[0007] In order to avoid the above-mentioned decrease in extraction accuracy, it is assumed that an appropriate threshold value is selected for each part in the image so that a structure is not erroneously detected. However, the method of selecting the threshold value for each part significantly increases the labor of the operation, increases the operational burden on the operator, and reduces the efficiency of the contour extraction processing.

Thirdly, in the end diastole image, blood that is not mixed with a contrast medium flows in from the left atrium, so that a region where the contrast medium becomes thin is likely to occur near the left atrium. In this portion, it is difficult to determine a threshold value equal to or higher than the concentration of the left ventricle below the concentration of structures such as ribs outside the ventricle, and this thin portion of the contrast medium may be mistaken as outside the ventricle.

[0009]

As described above, the conventional method has the following problems. First, depending on the setting of the threshold value, an area other than the desired threshold value may be detected or the contour of the desired area may not be extracted.

Second, depending on the positioning, a structure such as a rib may appear around a desired region in the image, for example, around the left ventricle. In this case, if it is attempted to discriminate the pixel value data of the entire image with one threshold value, erroneous detection frequently occurs, such that the region of the structure is erroneously recognized as the region of the left ventricle. Markedly reduced. If it is attempted to avoid the reduction of the extraction accuracy, the labor of the operation remarkably increases, the operation load on the operator increases, and the efficiency of the contour extraction processing decreases.

Thirdly, in the end diastole image, blood in which the contrast medium is not mixed flows from the left atrium, so that a region where the contrast medium becomes thin is likely to occur near the left atrium. In this portion, it is difficult to determine a threshold value equal to or higher than the concentration of the left ventricle below the concentration of structures such as ribs outside the ventricle, and this thin portion of the contrast medium may be mistaken as outside the ventricle.

The present invention has been made in view of the above, and an object thereof is to provide an improved image processing apparatus and image processing method. Specifically, (1) The contour of the desired region can be extracted even if the density variation or inclination of the background (that is, the portion other than the desired region) is large with respect to the density of the desired region.

(2) Even in an image in which a structure other than the desired object is reflected, the contour of the object or the region of interest of the object is not affected by the pixel value data of the structure. It should be possible to extract with higher accuracy and operating efficiency.

(3) When there are many variations in the density of the desired region (for example, the density of the contrast agent) (particularly when the density is low)
However, the desired area can be accurately recognized. With the goal.

[0015]

The present invention has taken the following means in order to solve the above problems. First of the present invention
The image processing apparatus of, an image collecting means for collecting an image including a desired part of the subject, an image storing means for temporarily storing the collected image, and a contour for extracting a region of interest of a desired portion of the subject. Extracting means, wherein the contour extracting means divides the image stored in the image storage means into a plurality of areas, and each of the plurality of areas.
First profile to create the first profile data
Based on the first profile data
The background area outside the region of interest and the inside of the region of interest
The inner area on the side to find the density of the background area and the inner area.
The density setting means for obtaining the density of the area and the density of the background area
And the density of the internal area, the first profile data
Threshold value determining means for determining a threshold value for each data
The threshold value determined for each first profile data
Contour points that are the boundaries between the outside and the inside of the region of interest using
And a contour setting means for determining .

In the above arrangement, the contour extracting means sets the long axis in the first direction of the image.
Further comprising a first profile creating means, in a second direction perpendicular to the long axis on the image, draw a plurality of perpendicular, for the plurality of vertical, respectively to create a plurality of first profile data , The threshold value determining means,
The threshold value is determined by a weighted average of the density of the background area and the density of the internal area.

The contour extracting means further includes means for searching a background area on the first profile data, and the first profile creating means draws the plurality of perpendicular lines at substantially equal intervals, and the background area. And a search start point setting means for setting the representative position as a contour search start point to a predetermined position outside a predetermined distance from the contour point obtained by the immediately preceding first profile data.

The first image processing method of the present invention comprises: a first step of collecting an image including a desired portion of a subject; a second step of temporarily storing the collected image; ; and a third step of extracting a portion of the region of interest, the steps of the third step is to divide the image stored in the image storing means into a plurality of regions, prior to
The first profile data for each of the
First profile creating means for creating the first profile
A background region outside the region of interest based on file data
A background and an interior region inside the region of interest to search for the background
Determining the density of the area and the density of the internal area,
Threshold from the density of the background area and the density of the internal area
And determining the first profile data
Outside the region of interest using the threshold determined as
And a contour point which is a boundary between the inside and the inside . In addition, the third step further includes the step of setting a major axis in the first direction of the image.
E , the step of creating the first profile data
Further includes a step of drawing a plurality of perpendiculars to the image in a second direction perpendicular to the long axis, and creating a plurality of first profile data for each of the perpendiculars.
Wherein the step of determining the threshold value, characterized in that it further comprises the step of determining the threshold by the weighted average of the density and concentration of the internal region of the background area.

According to a second image processing apparatus of the present invention, in the first image processing apparatus, the first profile creating means in the first profile data is directed from the outer predetermined position toward the major axis direction. Position detecting means for detecting a maximum density position between the outer predetermined position and the long axis, and background density replacing means for replacing the density at the maximum density position detected by the position detecting means with a background density. It is characterized by including. Further, the contour extracting means generates second profile data indicating density distribution at a plurality of pixel positions on the long axis, and a position near a boundary between the first profile data and the region of interest. Region of interest edge determining means for automatically determining one end of the region of interest on the long axis based on the above. In addition, the contour setting means includes a first
At least the contours of the first profile data that are adjacent to the distance data creating means for finding the vertical distance to the long axis based on the contour points determined for each profile data and the vertical distance created by the distance data creating means. The method further comprises contour smoothing means for replacing with an average value of the vertical distance created based on the points.

Further, the region of interest of the subject is the left ventricle of the heart, and the long axis is a line segment connecting the position near the apex of the left ventricle and the midpoint of the aortic valve. The means includes means for creating first profile data starting from a position in the vicinity of the apex, and the contour setting means has the first profiles in the first direction from the apex position toward the aortic valve. It is characterized in that it includes means for detecting contour points of data. Then, the region-of-interest edge determining means differentiates the second profile data twice, determines the maximum value and the minimum value after the second differentiation, and positions both the maximum value and the minimum value. And calculating the weighted average of the second profile data in, and specifying the position on the long axis corresponding to the obtained result as the position of the apex.

The background density replacing means determines whether or not the maximum density position is between the contour point set in the preceding first profile data and the predetermined outside position; Means for replacing the background density with the density at the maximum density position based on the determination result of the determination means, and in addition, the background density replacement means includes the first profile data before the maximum density position. Determination means for determining whether or not it is between the contour point set in step 3 and the outside predetermined position, and changing the start point for changing the contour search start point to the maximum density position based on the determination result of the determination means. And means.

According to a second image processing method of the present invention, in the above-mentioned first image processing method, the first profile creating step is performed in the first profile data from the predetermined outer position in the long axis direction. Toward, a step of detecting a maximum density position between the outer predetermined position and the long axis, and a step of replacing the density of the maximum density position detected by the position detection step with a background density Is characterized by.

Further, in the third step, a step of creating second profile data showing a density distribution at a plurality of pixel positions on the long axis, and in a position near a boundary between the first profile data and the region of interest are provided. The step of automatically determining one end of the region of interest on the long axis based on the contour point determined for each of the first profile data. And a step of replacing the vertical distance created by the distance data creating step with an average value of at least the vertical distance created based on the contour points of the adjacent first profile data. Is further included.

Further, in the first and second image processing apparatuses, the density setting means sets the background density of the first profile data currently read and the already read first density.
The first profile creating means includes means for creating a number of first profile data proportional to the length of the long axis. Is characterized by.

It is characterized in that the search start point setting means includes means for setting a point outside the distance proportional to the length of the major axis as a search start point. At least the distance setting means for calculating the vertical distance to the long axis based on the contour points determined for each first profile data and the vertical distance generated by the distance data generating means. The method further comprises a contour smoothing means for replacing the average value with the vertical distance created based on the contour points of the adjacent first profile data.

The region of interest of the subject is the left ventricle of the heart, the long axis is a line segment connecting the position near the apex of the left ventricle and the midpoint of the aortic valve, and the first data creating means is ,
The contour setting means includes means for creating first profile data starting from a position near the apex, and the contour setting means is a contour of each of the first profile data in a first direction from the apex position toward the aortic valve. It is characterized by including means for detecting points. Then, the contour setting means detects the contour points along the first direction, and then, along the second direction from the aortic valve toward the apex of the heart, starting from the designated position on the long axis. It is characterized by including means for re-detecting contour points up to a position near the apex.

Further, the threshold value determining means is the first
The profile setting means includes means for setting first and second threshold values on the left and right sides of the major axis of the profile data, respectively, and the contour setting means has first and second threshold values for the left and right sides of the major axis, respectively. And a means for detecting the contour point based on the above. Further, the first profile creating means includes means for determining the minimum value of the first profile data as the internal density, and the position detecting means detects the maximum value of the first profile data as the maximum density position. It is characterized by including means.

In the third image processing apparatus of the present invention, the contour extracting means subtracts the first image read from the image storage means from the second image one frame before the first image. And a means for performing absolute value processing on the image obtained by the subtraction, and means for multiplying the absolute value processed image by a predetermined coefficient and subtracting from the first image.

The third image processing method of the present invention is the third image processing method.
The step of subtracting the first image read in the second step from the second image one frame before the first image; and the step of performing absolute value processing on the image obtained by the subtraction. ,
And multiplying the absolute-value-processed image by a predetermined coefficient, and subtracting the image from the first image.

Further, in the first to third image processing apparatuses of the present invention, the image collecting means includes means for collecting a contrast image of a radiation image in which the desired portion is filled with a contrast agent, and , And a means for collecting an end-diastolic left ventricular contrast X-ray image.

[0031]

As a result of taking the above-mentioned means, the following effects occur. According to the first image processing apparatus and the image processing method of the present invention, the image is divided into a plurality of areas on parallel straight lines, and a threshold value suitable for each area is determined for each area.
Since the contour is decided by this threshold,
Even if the density variation or inclination of the background (portion other than the desired region) is large with respect to the density of the contrast-enhanced region, which is the desired region, the contour can be extracted to obtain the desired contour.

Since the contour extracting means searches for the background area on the first profile data, the desired contour of the left ventricle can be extracted even in an image having a background inclination, for example, a left ventricular contrast image.

A first profile creating means draws the plurality of perpendicular lines at substantially equal intervals, and sets a representative position of the background area as a contour search start point, and a predetermined distance outside a predetermined distance from the contour point obtained by the immediately preceding first profile data. Since the position is set to the position, the desired contour of the left ventricle can be extracted even in a contrast image in which the left ventricle is overlaid on the nonuniform background tissue, for example.

As described above, according to the first image processing apparatus and the method of the present invention, the variation or inclination of the density of the background (that is, the part other than the desired region) causes the density of the contrasted part, which is the desired region, to increase. Even if it is large, the contour of the desired region can be extracted.

According to the second image processing apparatus and method of the present invention, when a left ventricular contrast image is given as grayscale image data as a diagnosis target, for example, the long axis connecting the midpoint of the aortic valve to the vicinity of the apex of the heart. Is set, and the first profile data indicating the profile of the pixel value (density value) of the pixels forming the long axis is created. Further, second profile data indicating a profile on a vertical axis perpendicular to the long axis is created at each pixel position on the long axis. The background concentration of the left ventricle is obtained from the second profile data on each of the vertical axes.

Specifically, if the contour point on the vertical axis adjacent to the preceding line by one line has already been determined, a position outside the contour point by a predetermined distance is designated, and a position outside the designated outer position is designated. The background density is read from the second profile data. However, when a structure such as a rib is included in the image data, the density value around the left ventricle may decrease. In such a case, the second profile data does not change gently between the designated outer position and the long axis, and has a peak value larger than the background density at the designated outer position (however, the density of the left ventricular contrast image). If the value is lower than the density value of the surrounding area). Therefore, such a peak value is detected as the maximum fluctuation position. When this peak value is detected, the pixel value at the position of the peak value is replaced as the background density.

On the other hand, a representative internal concentration value of the left ventricle (for example, the minimum value of the second profile data) is obtained on each of the vertical axes. A threshold value for contour extraction is calculated for each vertical axis on the left and right of the long axis by weighted averaging the background density and the internal density value. Therefore, the threshold value and the second profile data are compared in the range from the contour search start point, which is provided at a position outside the contour point one line before on the vertical axis by a predetermined distance, to the major axis. As a result, the position of the pixel value below the threshold value is detected as the left ventricle contour point on each vertical axis.

As a result, even if a structure such as a rib appears in the periphery of the left ventricular contrast image and the pixel value of the structure is lower than that of the normal peripheral portion, regardless of the decrease. , The background density that can almost avoid the influence of the structure is automatically determined, and the accurate threshold value can be set. This improves the accuracy of contour extraction. Even if the influence of the pixel value variation due to the structure is large and the magnitude relationship between the set threshold value and the background density is reversed, the contour search start point is changed to a position closer to the long axis, and false detection is avoided. Therefore, according to the second image processing apparatus and the method of the present invention, even when a structure such as a rib appears in the periphery of the left ventricle to be diagnosed and the background density is reduced by the structure. , The outer density value, that is, the background density, can be accurately selected without being affected by the decrease in the density value, which allows the threshold value to be determined with the same accuracy as when the structure is not visible, and thus the contour can be more accurate. In addition, it is possible to extract stably with good operation efficiency.

According to the third image processing apparatus and the method thereof of the present invention, the preprocess for enhancing the change in the contrast agent concentration from the previous frame is performed even in the portion where the contrast agent concentration is low, and the enhanced left heart is emphasized. A contour point is determined by obtaining a threshold value for the contrast agent concentration signal in the room. Therefore, a desired contour can be automatically extracted even in a thin portion of the contrast medium in the left ventricle.

Further, since the desired contour can be automatically extracted by a few operations, the correction is unnecessary or reduced, and the burden on the operator is reduced. In addition, since it is not necessary to increase the concentration of the contrast medium to be injected and the amount of the contrast medium to an unnecessarily large amount, the side effects of the contrast medium are reduced and the burden on the subject is reduced.

[0041]

Embodiments of the present invention will be described with reference to the drawings. The image applied to the present invention is not limited to an X-ray contrast image, and may be any image as long as it has different pixel values between the inner and outer pixels of the diagnosis target. For example, a nuclear medicine diagnostic apparatus, MRI (magnetic resonance imaging) ) Device or X-ray CT image may be used. Furthermore, the present invention can be applied to a DF (digital fluorography) device. Hereinafter, for convenience of explanation, an X-ray diagnostic apparatus will be described as an example.

Further, in the following embodiments, the case where the diagnostic region related to X-ray diagnosis is the heart of the subject and a left ventricular contrast image is taken while injecting a contrast medium will be exemplified. FIG. 1 shows an X including an image processing apparatus according to the first embodiment of the present invention.
It is a figure which shows schematic structure of a line diagnostic device.

The X-ray diagnostic apparatus of the present invention comprises an X-ray tube 1, a detection unit 2, a data processing unit 3, an image display unit 4, and an image display unit 4.
It has an input unit 5 and an analysis unit 6. The X-ray tube 1 irradiates the subject P with X-rays. A high voltage generator (not shown) for supplying a high voltage is connected to the X-ray tube 1, and the operation of the high voltage generator is controlled by an X-ray controller (not shown). As a result, the X-ray tube 1 irradiates the subject P with pulsed X-rays.

The detector 2 detects the transmitted X-rays that have passed through the subject P and converts them into image signals. The detection unit 2 includes an image intensifier 2a, an optical system 2b, and a TV camera 2c. The image intensifier 2a converts the X-rays transmitted through the subject P into visible light, and the optical system 2b
Guides this visible light to the TV camera 2c. TV camera 2
The image sensor c is, for example, a CCD device as an image sensor and converts visible light corresponding to transmitted X-rays into an image signal. TV camera 2c
Is connected to a data processing unit 3 for processing an image signal.

The data processing unit 3 includes the A / D converter 3
a, an image memory 3b, and a contour extraction unit 3c.
The A / D converter 3a converts the analog image signal output from the TV camera 2c into digital image data. The image memory 3b temporarily stores digital amount of image data for each frame. The contour extraction unit 3c reads the image data stored in the image memory 3b and extracts the contour of the desired portion. The contour extracting unit 3c is equipped with a computer and can execute processing for contour extraction, which will be described later.

The image display unit 4 includes a display device such as a CRT, takes in the storage data of the image memory 3b at a constant timing and at a frame-by-frame basis, and displays an X-ray image of the diagnostic region of the subject almost in real time.

The input section 5 includes, for example, a keyboard, a mouse,
A trackball or the like is mounted, and information such as coordinate input set by the operator is output to the contour extraction unit 3c and the image display unit 4, and a command necessary for functional analysis is output to the analysis unit 6. The input unit 5 may be integrated with the image display unit 4.

The analysis unit 6 receives the extracted data from the contour extraction unit 3c, and performs a functional analysis of the region related to the extraction, for example, the left ventricle of the heart. As a left ventricular functional analysis, for example, the volume of the ventricle is calculated from the contour data, and the cardiac ejection fraction is calculated from the end-diastolic volume and the end-systolic volume. There is a heart wall motion analysis, etc. for obtaining the movement amount of the heart wall from the contour shape. The analysis result is displayed on the image display unit 4 as needed.

The operation of the first embodiment constructed as above will be described with reference to FIGS. 2 to 6C. FIG. 2 is a flowchart showing the operation of the contour extracting unit according to the first embodiment of the present invention.

Image data is fetched from the image memory 3b (step S1). This image data targets an image taken with a contrast agent injected into the left ventricle. A smoothing filter (low-pass filter) is applied, if necessary, at the time of capturing the image data or after capturing the image data to reduce roughness due to X-ray noise. In parallel with this image data acquisition, a left ventricular contrast image is displayed on the image display unit 4.

The major axis LA is set (step S2). The major axis LA is a line segment connecting the midpoint (black circle) and the apex (white circle) of the aortic valve as shown in FIG. The long axis is shown in Figure 3.
The rough area where the left ventricle is present is grasped on the image of (a), and is set by designating the position of x in FIG. 3 (b), which is used to limit the area to be subjected to the contour extraction processing. It The long axis LA is obtained by moving the cursor on the screen of the image display unit 4 using a mouse, a trackball or the like (not shown), and displaying the midpoint of the aortic valve and the apex of the aortic valve shown in FIG.
It is set by designating three points, or both end points of the aortic valve and the apex of the heart. The long axis is set, for example, by connecting the midpoint of the aortic valve and the two points of the apex. With this long axis LA, the region in which the left ventricle is present can be roughly pointed out, and the target of contour extraction processing can be limited.

A long axis perpendicular profile is created (step S3). As shown in FIG. 4, the long-axis perpendicular line profile has a large number of perpendicular lines VS drawn at substantially equal intervals with respect to the long-axis LA.
It is created by obtaining the pixel value on each perpendicular VS. This means that the target image is divided into a number of partial areas corresponding to the long-axis perpendicular profile.

After the long axis perpendicular profile is created, the contour OL is determined by sequentially determining the contour point OLP of the left ventricle on the long axis perpendicular profile from the apex to the midpoint of the aorta as shown in FIG. Extract. The procedure is shown below.

A background region is searched (determined) on the long-axis perpendicular profile (step S4). In this case, FIG. 6 (a)
As shown in, the point from the center (intersection with the long axis) of the long-axis perpendicular profile toward the periphery is found to be a provisional threshold value or more, and the area outside the predetermined distance from this point is set as the background area.
As the temporary threshold value, for example, the pixel value of the apex which is one end point of the long axis is used.

The density (pixel value) of the background area is read (step S5). When the background area includes a plurality of pixels, the plurality of pixel values are averaged and the pixel value is set as the density of the background area.

The internal density is read (step S6). The position of the minimum pixel value between the background regions on both sides on the long-axis perpendicular profile is defined as the internal region, and the pixel value of the internal region is defined as the internal density.

A threshold value for detecting the contour point OLP is determined (step S7). The threshold value is a weighted average of the internal density and the density of the background area, as shown in FIG.
Desired. The weighting coefficient related to the weighted average calculation may be a fixed value in advance or may be arbitrarily changeable by the operator.

The threshold value determined in step S7 is used as shown in FIG.
As shown in (c), the contour point OLP of the profile is determined. Pixel values are sequentially examined from the previously determined background area to the center of the profile, and the pixel that is equal to or less than the threshold value is detected as a contour point first (step S8). Determine on both sides of the profile.

Steps S4 to S8 are executed for all long-axis perpendicular profiles to detect contour points (step S9). When the contour points for all the long-axis perpendicular profiles have been extracted, the contour points of adjacent profiles are connected, and the connected contour data is output as the left ventricle contour (step S10).

According to the first embodiment of the apparatus of the present invention, the X-ray image is divided into parallel sub-regions, and a threshold value suitable for the sub-region is determined for each sub-region. Since the contour is determined based on the value, the contour is extracted in order to obtain the desired contour even if the variation in the density of the background (the portion other than the desired area) or the inclination is large with respect to the density of the contrasted portion that is the desired area. it can.

According to the first embodiment described above, the following effects can be obtained in addition to the above. (1) Since the contour of a desired area can be automatically extracted with a small number of operations, the burden on the operator can be reduced.

(2) Since the image display mechanism used when the operator sets the threshold value while watching the threshold value and the area to be extracted is unnecessary, the structure of the apparatus can be simplified. (3) Since the time required to adjust the threshold is not needed,
The time required for contour extraction is shortened.

(4) Since the concentration of the contrast agent may be low,
The amount of contrast agent used is reduced and the burden on the patient is reduced. A second embodiment of the present invention will be described with reference to FIGS. Since the device configuration of the second embodiment is the same as that of the first embodiment, its illustration is omitted. 7, the same parts as those in FIG. 2 are designated by the same reference numerals, and detailed description thereof will be omitted.

FIG. 7 is a flow chart showing the operation of the contour extracting section according to the second embodiment of the present invention. In the second embodiment, similarly to the first embodiment, the image is captured, the long axis is set, and the long axis perpendicular profile is created (steps S1 to S1).
Step S3).

The background area in the first (ie, first line) profile in which the contour points of adjacent profiles have not been determined is input (step S11). The operator sets a background area using the input unit 5, or automatically sets an area having a predetermined positional relationship from the separately set long axis as the background area. That is, as shown in FIG. 8, a point separated from the long axis LA by the same distance as a contour point determined by an adjacent profile is set as a temporary contour point of the profile, and an area outside a predetermined distance from the temporary contour point is set as a background. The background area is determined by setting the area. Alternatively, the pixel value of a region outside a predetermined distance from the contour point on the adjacent profile may be measured and used as the background density of that profile.

Similar to the first embodiment, as shown in FIG. 9, the contour point is determined by the threshold value determined for each profile,
From the apex of the heart to the midpoint of the aorta, the contour points of the left ventricle are sequentially determined on the perpendicular profile, and the contour points are connected to extract the contour of the left ventricle as shown in FIG. (Steps S5 to S10).

According to the second embodiment, the desired contour of the left ventricle can be extracted even in the contrast image in which the contrasted left ventricle overlaps the nonuniform background tissue. Furthermore,
In the first and second embodiments, X-rays are given as an example of radiation, but other radiations can be used.
In the above-described embodiment, the case where the region of interest filled with the contrast agent is extracted from the contrast pixel has been described, but the same can be performed for a contrast image other than the contrast image.

The third embodiment of the present invention will be described with reference to FIGS. The configuration of the device of the third embodiment is also the same as that of the first embodiment, so the illustration and description thereof are omitted. FIG. 11 is a flowchart showing the operation of the contour extracting unit according to the third embodiment of the present invention. The same parts as those in the first embodiment are designated by the same reference numerals, and detailed description thereof will be omitted.

Image data is fetched from the image memory 3b (step S1). Set the long axis LA (step S
2). In this case, the apex of the long axis LA is a temporary apex position.

A density value profile for a set position (pixel) on the long axis LA (hereinafter referred to as "long axis profile P").
L ”) is created (step S13). The data of the long-axis profile PL is created by reading the density value of each pixel existing on the long-axis LA. When obtaining this concentration value, the long axis LA is extended in the direction of the apex, and the concentration value profile data for this extension is also obtained.

The apex search is performed (step S1).
4). Based on the concentration value of the long-axis profile PL, the point of the maximum first derivative existing in the vicinity of the point B near the designated apex is obtained. A position of a predetermined distance obtained from the length of the long axis LA is designated from the point of the maximum of the first derivative toward the outside of the apex, and the density value at the designated position is read together with the minimum value of the long axis profile PL. . Furthermore,
An average value of the density value and the minimum density value at the read designated position is obtained, and points below this average value are sequentially searched from the apex side of the data of the long-axis profile PL. The point AC that is less than or equal to the average value thus obtained is the final determined position of the apex.

A profile of the perpendicular PE of the long axis LA (hereinafter referred to as "long axis perpendicular profile PP") is created (step S3). The contour extracting unit 3c sets the initial position of the background area P _out of the long-axis perpendicular profile PP which is the first contour point search target given from the operator via the input unit 5 (step S4). The situation is shown in FIG. 12, and the background area P _out in FIG. 12 is the contour point search for the second and subsequent lines, and shows the case where the contour points of the adjacent long axis perpendicular profile PP have been determined last time. As the initial position, an outer position (a side on the perpendicular NL away from the long axis LA) that is separated from the long axis LA by a predetermined distance may be automatically designated.

The initial background area P set in step S4
_{The out} density value (that is, the background density) is read (step S5). The density value of the internal point P _in (that is, the internal density) is obtained (step S6). As shown in FIG. 12, the minimum value on the long-axis perpendicular profile PP is obtained as the internal density D _in . When obtaining the internal density D _in , the long-axis perpendicular NL inside (close to the long-axis LA) of the background region P _out so as not to be affected by the peripheral portion of the image that is not related to the left ventricle VS.
The upper position is the search target.

A threshold value for detecting the contour point OLP is determined (step S7). The left and right threshold values TH of the long-axis perpendicular NL set in step S7 are used to separately determine the contour points BD on the long-axis perpendicular NL (step S).
8). As shown in FIG. 13, the representative position of the set background area P _out is set as the contour search start point P _st, and this start point P _{st is set.}
To the center side of the long-axis perpendicular profile PP, the pixel value is searched in order, and the searched pixel value is compared with the threshold value TH. Then, the pixel position of the pixel value that first becomes equal to or less than the threshold value TH is determined as the contour point BD. The search for the contour point BD is performed by deriving the long axis perpendicular profile P on both the left and right sides of the long axis LA.
Executed individually in P.

The representative position is set by selecting any one point in the background area P _out . In particular, the background area P
It is desirable to set the pixel located closest to the long axis in _out (the point closest to the long axis) or the point located one pixel inside the background area P _out as the representative position.

It is determined whether or not the long-axis perpendicular profile PP for searching the contour point BD still remains, that is, whether or not the pixels on the long-axis LA remain (step S).
9). In the case of YES (remaining) in this determination, step S15 to step S17 and step S6 to
The process of step S8 is sequentially executed.

In obtaining the contour point BD _n of the n-th line, a temporary contour point TB _n is set (step S1).
5). This provisional contour point TB _n has the same value as the contour point BD _n−1 determined on the already determined adjacent profile PP. That is, the contour point BD _{n-1 on the} preceding line and the provisional contour point TB _n are points separated by the same distance from the long axis LA (see FIGS. 12 and 14). Further, a position outside the provisional contour point TB by a predetermined distance L _d (on the side away from the long axis) is set as a background area P _out on the long axis perpendicular to the n-th line. The predetermined distance L _d for swinging the background area P _out to the outside is the long axis L
It is adjusted to a value proportional to the length of A. The reason for setting the predetermined distance L _d in this way is to set the same position relatively regardless of the conditions at the time of shooting such as the image enlargement ratio and the size of the object. The predetermined distance L _d may be, for example, a value proportional to the image matrix size and the image enlargement ratio.

The background density D _out is obtained by reading the pixel density value of the specified background area P _out , and
If necessary, the background density D _out is replaced and the contour search start point P _st is changed (step S16). The reading of the background density D _out will be described in detail with reference to FIGS. 15 and 16. FIG. 15 shows the subprogram executed for this purpose in step S16.

In step S15, the pixel density value D _out of the background area P _out designated at the position of the predetermined distance L _d is read (step S16 ₁ ). The density value D _out may be calculated by applying an average value of a plurality of pixels.

The density value D _x and the pixel position P _x of the pixel one inside (closer to the major axis) than the background area P _out (that is, the pixel at the contour search start point P _{st in} the first process) are read (step S16 ₂ ). Between the readings D _out and D _{x obtained in} steps S16 ₁ and S16 ₂ , D
_It is determined whether _x > D _out (step S16 ₃ ). If NO in this judgment, that is, if D _x ≦ D _out , the background area P is searched even if the search is performed on the long-axis vertical line NL toward the long-axis side.
_It is recognized that there is no pixel having a density value larger than the density value of _out (background density D _out ), and the process proceeds to step S16 ₆ described later.

YES in step S16 ₃ (D _x > D
_out ), the background density D of the designated background area P _out is due to the fact that structures such as ribs appear in the image.
_It is recognized that there is a pixel having a density value higher than _out, and the processes of steps S16 ₄ and S16 ₅ are sequentially performed. In step S16 ₄ , the density value D _x related to the determination of D _x > D _out
Replace the background density D _out up to then. Step S
At 16 ₅ , the position P _{x of} the pixel exhibiting the density value D _x determined to be larger is replaced with the variable P _x0 .

It is determined whether or not there is a pixel to be searched for on the long-axis vertical line NL (step S16 ₆ ). When the search target pixel does not reach the long axis LA of (YES), and repeats the above process returns to step S16 _2. As a result, when the density value becomes larger as the search is performed in the long axis LA direction, the background density D _out is updated to the larger value and the value of the variable P _x0 representing the pixel position is also updated. Therefore, when the search up to the long axis LA is finished, the background density D _out is replaced with the maximum density value D _max on the long axis perpendicular NL (one side), and the variable P _x0 contains the pixel position of the maximum density value D _max. Remembered

[0083] Step S16 ₆ in NO, that when it is determined that finished search on the long axis perpendicular line NL, the pixel position of the maximum density value D _max (the value of the variable P _x0) is the temporary contour point T
It is determined whether or not it is located between B _n and the background area P _out (step S16 ₇ ). If NO in this determination (for example, when the pixel position of the maximum density value D _max is as shown in FIG. 16 (b)), it is returned to the meting program as it is, but Y
(When for example, the pixel position of the maximum density value D _max in FIG. 16 (c)) when the ES is Rita after performing the processes of steps S16 ₈ - main routine. In step S16 _8, adjacent the position of the pixel position of the maximum density value D _max (the value of the variable P _x0) by setting as the contour search start point P _st, starting point P _st is the background area P _out so far ( Old) to maximum density value D
_It is automatically replaced with the _max pixel position (new).

As described above, in step S16, the background density D
There are various advantages by considering the maximum density value D _max when determining _out . When no structure or the like is reflected in the periphery of the left ventricle VS, the long-axis perpendicular profile PP is normally calculated from the background density D _out to the long-axis L as shown in FIG.
It decreases as it goes to A. In this case, even if the processes of steps S16 ₁ to S16 ₈ described above are performed, there is no pixel having a density value higher than this inside the background area P _out (closer to the major axis), so the density of the background area P _out The background density D _{out as} it is (step S16 ₁ ) and the pixel next to the background area P _out becomes the contour search start point P _st as it is.

However, when the image data showing the structure is processed, the density value of the pixel representing the structure is as low as that of the left ventricle VS injected with the contrast agent. The values are shown in FIG. 16 (b) and FIG.
As shown in (c), a peak is often shown between the background area P _out and the long axis LA.

When this peak, that is, the position of the maximum density value D _max is between the provisional contour point TB _n and the long axis LA as shown in FIG. 16 (b), the structure is far from the left ventricle VS. Normally, the background density D _out is still at a relatively high level (a level higher than the threshold TH set in step S7) such as when the maximum density value D is reached.
_The profile curve connecting _max and the background density D _out is also gentle. At this time, the background density D _out is simply the maximum value D.
_The background density D _{out, which} is replaced by _max and is less affected by the reflection of the structure, is determined.

On the other hand, when the position of the maximum density value D _max is between the provisional contour point TB _n and the background area P _out as shown in FIG. 16C, the structure is in the left ventricle VS. The image is very close, and the effect of the pixel value reduction due to the structure is large. Therefore, by performing the processing of FIG. 15 described above, the background density D _out is replaced with the maximum density value D _max , and further, the contour search start point P _st is newly moved to the position of the maximum density value D _max .

When the background density D _out is determined as described above, the contour extracting section 3c performs the processing in step S1 of FIG.
Move to 4. In step S14, step S16
The background density D _out determined in step 1 is averaged with the background density determined by the long-axis perpendicular profile PP for the number of adjacent predetermined lines whose background density has already been determined, and the average value is finally determined as the background density D _out this time. Is set. As a result, the background density is stable and highly accurate. The predetermined number of lines is the number of lines proportional to the length of the long axis LA.

After execution of step S17, step S
Return to 6. As a result, the internal density D related to step S6
_in the reading, the determination of the threshold value TH in the step S7,
And the detection of the contour point BD according to step S8 is repeated. As a result, as shown in FIG. 14, the contour points BD are sequentially determined for each long-axis perpendicular profile PP (and for each left and right of the long-axis LA).

However, if it is judged NO in step S9 described above, that is, if the detection of the contour points for all the long-axis perpendicular profiles PP on the long axis LA is completed, the processes of steps S10 to S19 are performed.

The contour extracting unit 3c connects the contour points BD-BD on each long-axis perpendicular line LN to form the contour B as shown in FIG.
DL data is formed (step S10). This contour data is held as a value of the vertical distance from the long axis LA to each contour point BD for each point (each pixel) on the long axis LA. For example, referring to the i-th pixel (black circle in the figure) on the long axis LA in FIG. 17, the position at which the vertical line at the i-th pixel position intersects the contours BDL on both sides of the long axis (the contour point BD ₁ is a white circle). , BD ₂ are stored as data of lengths (distances) f ₁ (i) and f ₂ (i) from the long axis LA up to white triangles.

For each point on the long axis LA, the contour BDL is smoothed by obtaining the average value of the contour data at each point and a predetermined peripheral point (step S1).
8). The range of the predetermined point used for this smoothing is set to a value proportional to the length of the long axis LA, for example. Since the contour data is held as the length from the long axis LA as described above, even a two-dimensional figure called a contour can be accurately smoothed by a one-dimensional calculation process, and a smooth contour can be obtained.

When the smoothing is finished, the contour BDL data is sent to the image display unit 4 and is displayed on the X-ray image, for example, while being output to the analysis unit 6 (step S1).
9). The analysis unit 6 performs a functional analysis of the left ventricle on the basis of the contour data drawn by the contour extractor 3c as described above.

Since the third embodiment is processed as described above, the structure is not shown in the X-ray image, and the long-axis perpendicular profile PP has a smooth shape as shown in FIG. 16 (a). In this case, the threshold value TH is determined based on the density value of the background area P _{out and} the density value of the internal point P _in , and the pixel position of the density value below this threshold value TH is the contour point BD.
Are sequentially detected (see FIG. 13).

On the other hand, in the case where the structure is reflected in the X-ray image and the long-axis perpendicular profile PP has a shape having peaks as shown in FIGS. 16 (b) and 16 (c), The background density D _out and the contour search start point P _st are appropriately determined while avoiding the influence of the structure.

In explaining the advantages of the third embodiment of the present invention, the decrease of the threshold value TH due to the influence of the structure is shown in FIG.
It will be described based on 8. As shown in FIG. 18A, when the image includes the left ventricle VS and the rib RB as a structure, the long axis perpendicular profile PP along the line BB ′ in the figure is, for example, the same figure. It becomes like (b). Since the density value of the rib RB portion is low, if the background area P _out reaches the rib RB portion, the background density D as described above.
_The threshold TH related to the weighted average of _out (the density value of the background area P _out ) and the internal density D _in becomes lower than the normal value when the structure is not shown. As a result, the contour point BD determined by the unreasonably lowered threshold value TH
Does not represent the true contour of the left ventricle VS, and causes a detection error by the distance R in FIG.

In the third embodiment of the present invention, this detection error can be eliminated as follows. In particular,
The situation in which the structure appears in a position far away from the left ventricle VS or the structure is small, and the influence of the pixel value decrease due to the structure is small, makes the pixel position of the maximum density value D _out a parameter. To be determined. That is, in FIG.
As in (b), when the pixel position of the maximum density value D _out is far from the background area P _out , the effect of the pixel value reduction due to the structure is small. At this time, the background density D _out is replaced with the maximum density value D _max in order to substitute the correction of the pixel value decrease, and the threshold value TH determined by the weighted average of the background density D _out and the internal density D _in decreases. Is reliably prevented. This allows
It is possible to prevent erroneous detection of the contour points described with reference to FIG. 18, and to draw a highly accurate contour line.

As shown in FIG. 16C, the structure is close to the left ventricle VS or the structure is large.
When the pixel position of the maximum density value D _out is close to the background area P _out , the influence of the pixel decrease due to the structure is considerably large. At this time, the background density D _out is replaced with the maximum density value D _max to prevent the background density D _out from decreasing and the contour search start point P _st is moved to the position of the maximum density value D _max . As a result, a more accurate threshold value TH that is hardly affected by the structure is set, and the contour point BD can be detected with high accuracy. As illustrated in FIG. 16C, if the threshold TH set appropriately is higher than the background density D _out (old), it is first detected in step S8 that it is smaller than the threshold TH. Since this point becomes the contour point, the contour search start point P _st = contour point BD is erroneously detected, but in the present embodiment, the contour search start point P _st is shifted to the position of the maximum density value D _max . The false detection is also prevented. That is, in the range closer to the long axis than the position of the maximum density value D _max ,
The pixel position that first becomes less than or equal to the threshold value TH is the contour point BD
Is accurately determined as

Furthermore, according to the third embodiment, the contour of the desired area can be automatically drawn with a smaller amount of operation.
It is possible to significantly reduce the operational burden on the operator.
Positioning restrictions during image capturing are alleviated, the positioning time for image capturing is shortened, and inspection efficiency is improved.

Further, since the positioning restriction is relaxed, it is not necessary for the patient to take an unreasonable posture at the time of photographing, and the burden on the patient is remarkably reduced.

A fourth embodiment of the present invention will be described with reference to FIGS. 19 and 20. The fourth embodiment relates to improvement of the apex search in the third embodiment. The same or equivalent components as those of the third embodiment are designated by the same reference numerals, and the description thereof will be omitted or simplified.

The contour extracting section 3c performs the processing shown in FIG. 19 in step S14 of FIG. 11 in the third embodiment. Long axis profile PL created in step S13
The data shown in FIG. 20A is smoothed by averaging the neighboring pixel values as shown in FIG. 20B, and a smoothed long-axis profile PLS is created (step S14).
₁ ).

The primary differential value of the smoothed data is calculated as shown in FIG. 7C (step S14 ₂ ).
As a result, a peak exhibiting a maximum appears near the provisional apex position B set by the operator.

The peak obtained in step S14 ₂ is specified as the position of the maximum value (step S14 ₃ ). Again first derivative profile data, differentiating (Step S14 _4). In this second derivative, a minimum point and a maximum point appear as shown in FIG.

[0105] Step S14 ₃ in identifying the minimum point and maximum point of the secondary differential on either side of the maximum value of the first derivative determined (Step S14 _5). Step S14 ₅ in reading the density value of the smoothing length axis profile PLS in both positions of the minima and maxima of the specified second derivative (step S
14 ₆ ).

A weighted average value of the two read density values is obtained (step S14 ₇ ). FIG. 20 (e) shows the position where the weighted average value is obtained between the minimum and maximum positions of the second derivative.
Automatically set as the position of the apex, as shown in (Step S14 _8).

The process after the automatic apex setting described above is the third process.
Since it is the same as the embodiment, its illustration and description are omitted. .
As described above, according to the fourth embodiment, since the position of the apex can be automatically set, the operator only needs to specify the position in the vicinity of the apex for searching the apex. Therefore, compared to the case of manually specifying the apex, it is possible to specify the position with higher accuracy, while reducing the operational burden on the operator and significantly improving the operation efficiency.

A fifth embodiment of the present invention will be described with reference to FIGS. 21 and 22. The fifth embodiment relates to a further improvement of the contour tracking method. The contour extracting unit 3c sequentially performs the process shown in FIG. 21 after the process of step S9 of FIG. 11 in the fifth embodiment.

In step S9 of FIG. 21, it is judged whether or not the target profile PP for contour search is still present on the long axis LA, as in the fifth embodiment, and the processing is completed (NO is confirmed.) 22 (a), the first contour search is completed from the apex toward the valve direction, and the data of the contour line BDL ₁ related to this is prepared. After the second search, the re-search relating to the processing in step S9a and subsequent steps in FIG. 21 is performed in the reverse direction (that is, from the valve position toward the apex position) from a predetermined position.

The predetermined re-search start position Pr is read (step S9a). The start position Pr is, for example, a half position of the long axis LA is suitable, but is not necessarily limited to this and may be an end position corresponding to the valve position of the long axis LA, for example.

When the re-search start position Pr is determined, as shown in steps S9b to S9h, the contour point B is moved in the opposite direction toward the apex, as in the process of the third embodiment.
D is detected (see the arrow in FIG. 22B).

When it is determined that the backward search up to the apex is completed (step S9h), the process proceeds to step S9i, and the contour line BDL ₁ detected in the first search is changed to the contour line detected in the second search. Data of the contour line BDL corrected by the BDL ₂ (for example, the contour line BDL ₂ is adopted as a true value) is created as shown in FIG. 22C.

As a result, in the first search, since the contour of the heart is round, especially near the apex, the distance between the contour point and the major axis increases, so that an error occurs in contour detection near the apex. Although it is easy, the contour near the apex can be appropriately corrected by the re-search in the reverse direction according to the fifth embodiment. That is, for example, the contour portion (the virtual line portion in the figure) detected in the first search in FIG. 22C can be excluded from the regular contour data. This makes it possible to more accurately delineate the contour line of the left ventricle. At this time, by determining the start position of the re-search at an appropriate position close to the apex, it is possible to re-search only the necessary part and suppress the increase of the rendering time to the necessary minimum.

In the left ventricle contrast images of the first to fifth examples, it has been explained that the density value of the left ventricle part is smaller than that of its peripheral part, but it is not always necessary to have such a density value relationship. For example, if the density value of the pixel in the left ventricle part is larger than that of its peripheral part, the density value of the left ventricle part is subtracted from a certain value in advance so that the left ventricle part has a smaller density value than its peripheral part. You just have to process it.
When the present invention is applied while the density value of the pixel in the left ventricle part is larger than that in the peripheral part, in the description of each of the above-described embodiments, "the density value is small" is "the density value is large".
In addition, "the minimum value may be read as the" maximum value "by reversing the magnitude relationship of the density values.

A detailed procedure of creating contour data and a procedure of smoothing contours in the first to fifth embodiments will be described with reference to FIGS. 23 to 25. FIG. 23 is a flowchart showing the procedure for creating contour data. Figure 24
FIG. 7 is a flowchart showing a contour smoothing procedure. FIG. 25 is a diagram for explaining the creation and smoothing of the contour data.

First, the creation of contour data will be described. The variable i is set to 0 (step S10a). The position of the i-th pixel on the long axis LA is read (step S10).
b). Next, the position of the intersection of the vertical line from the i-th pixel and the contour is read (step S10c). Step S10
In b and step S10c, the pixel position is given in coordinates. For example, in FIG. 25A, the long axis LA
The position LA (i) of the i-th pixel above is (x _L , y
_{At L} ), the position of the intersection BD ₁ (i) is represented by (x _B , y _B ).

Find the distance between the major axis and the intersection (step S
10d). In this case, the distance f ₁ of the i-th pixel of the long axis LA LA (i) the intersection BD ₁ (i) (i) is
It is calculated by the following formula.

F ₁ (i) = | x _B −x _L | / cos θ or f ₁ (i) = {(x _L −x _B ) ² + (y _L −y _B ) ² } ^{1/2 In the} above equation, , Θ is the angle formed by the long axis LA and the y axis.

From the above, the calculated value calculated for each contour point is temporarily stored in a memory (not shown) (step S1).
0e). If the variable i does not reach i _max (the number of final data), the value of the variable i is incremented (step S
10g), and steps S10b to S10e are repeated. If not, it is considered that the processing has been completed for all data, and the processing is completed.

By the above processing, the distance for each contour point as shown in FIG. 25A is calculated. Next, the smoothing processing of the contour points obtained by FIG. 23 will be described. Variable i
Is set to 1 (step S18a).

The (i-1) th to (i + 1) th data are sequentially read (steps S18b to S18).
d). An average value f _s1 of the read (i−1) th to (i + 1) th data is obtained (step S18e). In this case, the average value f _s1 of the data is obtained by the following formula.

F _s1 (i) = {f ₁ (i-1) + f ₁
(I) + f ₁ (i + 1)} / 3 The calculation result for each contour point obtained by the above equation is temporarily stored in a memory (not shown) (step S18).
f).

The value of the variable i is incremented (step S18g). When the variable i does not reach i _max (the number of final data), steps S18b to S18g
Is repeated, and if not, it is considered that the processing has been completed for all data, and the processing is completed.

As described above, the desired portion is smoothed and the smooth contour can be extracted. FIG. 26 is a diagram showing a schematic configuration of an X-ray diagnostic apparatus including an image processing apparatus according to the sixth embodiment of the present invention. The same parts as those in FIG. 1 are designated by the same reference numerals, and detailed description thereof will be omitted. Sixth Embodiment devices are that the contour extracting section 3c has a profile creation portion 3c ₁ and edge determining section 3c ₂ differs from that of Figure 1.

The profile creating section 3c ₁ reads an image of a desired frame from a moving image composed of a plurality of frames of images recorded in the image memory 3b, performs the processing described later, and then designates it by the edge determining section 3c _2. A density profile at the specified position is created and delivered to the edge determination unit 3c ₂ .

The edge determining section 3c ₂ is equipped with a computer and is capable of executing the processing for contour extraction described later. FIG. 27 shows a schematic configuration of the profile creation section 3c ₁ .

The profile creating section 3c ₁ includes a first image memory 31, a second image memory 32, and a first logarithmic converter 3
3, a second logarithmic converter 34, a subtractor 35, an absolute value processing unit 36, a coefficient setting unit 37, an adder 38, and a profile curve extraction unit 39.

The profile creating section 3c ₁ stores the end diastolic image (hereinafter referred to as “first ED image (end diast image)” from the image memory 3b.
olic image) ”) and select the first image memory 3
Take in 1. The image is selected by the operator while looking at the image displayed on the image display unit. Then, the profile creation unit 3c ₁ selects the E one frame before the selected image.
The D image (hereinafter referred to as “second ED image”) is loaded into the second image memory 32. First and second image memories 31 and 32
It is desirable to apply a smoothing (low-pass) filter to the first and second ED images taken in to reduce roughness due to X-ray noise.

Subsequently, the first ED image and the second ED image are logarithmically converted (LOG converted) by the first and second logarithmic converters 33 and 34, respectively, and the logarithmically converted first ED image is converted into the logarithmically converted first ED image. The subtractor 35 obtains the difference value for each pixel to obtain a subtraction image. This subtraction image has a negative value in the area where the contrast medium near the mitral valve is thin, a positive value in the area where the ventricles have spread, and a zero value in the area outside the ventricle where the ribs / diaphragm, etc. are visible. It becomes an image. In other words, an image is obtained in which only changes in the contrast agent are extracted.

The absolute value processing unit 36 performs absolute value processing on the obtained subtraction image. That is, by the absolute value processing, the pixels with a positive pixel value in the region where the ventricle near the edge of the ventricle has spread are still positive, while the pixels with a negative pixel value in the region near the mitral valve where the contrast agent has become thin have a sign. Invert to a positive value.

The coefficient setting unit 37 sets a coefficient for adjusting the pixel value of the subtraction image subjected to the absolute value processing, multiplies the obtained subtraction image by the set coefficient, and outputs it to the adder 38. .

The adder 38 adds the subtraction image obtained by multiplying the first ED image by a predetermined coefficient. In the X-ray contrast image, the signal of the contrast agent appears in the direction in which the pixel value becomes smaller. Therefore, by subtracting the absolute value processed image in which the change component of the contrast agent has a positive value from the first ED image, the contrast agent signal is imaged. An image enhanced with the changing component of the agent is obtained.
The same result can be obtained by adding the first ED image and the subtraction image with a negative coefficient in the coefficient setting unit 37.

Subsequently, the profile curve extraction unit 39
Reads the density value of the pixel on the specified line (i.e. the major axis) from the edge determining unit 3c _2, sent to the edge determining section 3c ₂ as a density profile.

The operation of the edge determining section 3c ₂ is exactly the same as the operation after step S13 in FIG. The operation of the profile creating section 3c ₁ configured as described above will be described with reference to FIG. FIG. 28
28B to FIG. 28F respectively show the AA ′ profile of FIG. 28A obtained in the profile creating unit 3c ₁ .

Assume that the initially obtained image is as shown in FIG. 28 (a). In this case, the shaded portion is equal to or larger than the threshold value, so that the image is not regarded as a part of the left ventricle and becomes a missing image.

The profile creating section 3c ₁ takes in an image from the image memory 3b. At this time, the first image memory is
First E having profile as shown in FIG. 28 (b)
The D image and the second image memory respectively store the second ED image having the profile as shown in FIG.

The first ED image and the second ED image are logarithmically converted by the first logarithmic converter 33 and the second logarithmic converter 34, respectively, and then subtracted by the subtractor 35 to obtain a subtraction image. The AA ′ profile at this time is shown in FIG. By performing the absolute value processing on the image as shown in FIG. 28D, the image shown in FIG. 28E is obtained.

Finally, the adder 38 performs image enhancement processing as shown in FIG. 28 (e) and performs addition with the first ED image shown in FIG. 28 (b), and FIG. 28 (g). An image with no defects is obtained as shown in.

Therefore, according to this embodiment, an accurate contour can be extracted even if there is a variation in the concentration of the contrast agent. FIG. 29 shows the profile creation unit 3c ₁
The 1st modification of is shown. The same parts as those in FIG. 27 are designated by the same reference numerals, and detailed description thereof will be omitted.

In the first modification, the profile curve extraction unit 39 is provided not after the adder 38 but the first profile curve extraction unit 39a is provided at the subsequent stage of the first memory, and the second profile curve extraction unit 39b is provided as the subtractor 35. It is provided in the latter stage.

After the first ED image and the second ED image are captured and the subtraction image 36 is created, the profile curve at the position designated by the edge determining section 3c ₂ is extracted from the first ED image and the subtraction image. The absolute value processing unit 36 performs absolute value processing on the profile curve extracted from the subtraction image. Coefficient setting unit 37
Multiplies the absolute-value processed profile by a predetermined coefficient. The adder 38 adds the profile curve extracted from the first ED image and the output from the coefficient setting unit 37.
The profile curve obtained by this addition is sent to the edge determining unit 3c ₂ .

Also in the first modified example, the same profile curve as in the case where the profile curve is created after the enhanced image is created can be obtained. In the first modified example, the enhanced image is created as compared with the embodiment of FIG. Since the area is unnecessary, the configuration is simplified.

FIG. 30 shows the second section of the profile creating section 3c ₁ .
A modified example is shown. The same parts as those in FIG. 27 are designated by the same reference numerals, and detailed description thereof will be omitted. The second modification is different from the configuration of FIG. 27 in that the first and second logarithmic converters 33 and 34 are provided.
It is the point that omitted.

When the left ventricular contrast image taken by the moving image is imaged so as to overlap with a site having a large X-ray absorption (for example, a rib), the change of the contrast medium in the overlapping part such as the rib cannot be emphasized. As in the embodiment of FIG. 27, it is desirable to create an emphasized image using an image that has been logarithmically converted and then subtracted. However, when the left ventricular contrast image does not overlap with a site having a large X-ray absorption, an image in which the change line segment of the contrast agent is emphasized can be obtained by the process of the second modified example. Compared with the embodiment of FIG. 27, the logarithmic conversion processing part is not required, so the configuration is simplified.

FIG. 31 shows the third section of the profile creating section 3c ₁ .
A modified example is shown. The same parts as those in FIG. 29 are designated by the same reference numerals, and detailed description thereof will be omitted. The third modification is different from the configuration of FIG. 29 in that first and second logarithmic converters 33 and 34 are provided.
It is the point that omitted.

According to the third modification, the same profile curve as when the profile curve is created after the enhanced image is created can be obtained. In the third modification, as compared with the embodiment of FIG. 30, the area for creating the emphasized image is unnecessary and the configuration is simplified.

The description of the above embodiment is made in the first to the sixth.
Although the embodiments have been described independently, they can be applied in an appropriate combination. Basically, a more improved image processing apparatus and method thereof are provided by appropriately combining the first to second embodiments of the present invention with the second to sixth embodiments. Besides, the present invention is not limited to the above-described embodiments, and it goes without saying that various modifications can be made without departing from the scope of the present invention.

[0148]

According to the present invention, the following effects can be obtained. A first aspect of the present invention corresponding to the first and second embodiments described above.
According to the image processing apparatus and the image processing method of, the image is divided into a plurality of regions on parallel straight lines, a threshold value suitable for each region is determined for each region, and the contour is determined by the threshold value. Since the determination is made, the contour can be extracted to obtain the desired contour even if the variation or the gradient of the density of the background (the portion other than the desired area) is large with respect to the density of the contrasted portion that is the desired area.

Since the contour extracting means searches the background area on the first profile data, a desired contour of the left ventricle can be extracted even in an image having a background gradient, for example, a left ventricle contrast image.

The first profile creating means draws the plurality of perpendicular lines at substantially equal intervals, and sets a representative position of the background area as a contour search start point to a predetermined distance outside the contour point obtained by the immediately preceding first profile data. Since the position is set to the position, the desired contour of the left ventricle can be extracted even in a contrast image in which the left ventricle is overlaid on the nonuniform background tissue, for example.

As described above, according to the first image processing apparatus and the method thereof of the present invention, the variation or inclination of the density of the background (that is, the part other than the desired region) causes the density of the contrasted part which is the desired region. Even if it is large, the contour of the desired region can be extracted.

According to the second image processing apparatus and method of the present invention corresponding to the above third to fifth embodiments, when a left ventricular contrast image is given as grayscale image data as a diagnosis target, for example, A long axis connecting the midpoint of the aortic valve and the vicinity of the apex is set, and first profile data indicating a profile of pixel values (density values) of pixels forming the long axis is created. Further, second profile data indicating a profile on a vertical axis perpendicular to the long axis is created at each pixel position on the long axis. The background concentration of the left ventricle is obtained from the second profile data on each of the vertical axes.

Specifically, assuming that the contour point on the vertical axis adjacent to the preceding line by one line has already been determined, a position outside the contour point by a predetermined distance is designated, and a position outside the designated outer position is designated. The background density is read from the second profile data. However, when a structure such as a rib is included in the image data, the density value around the left ventricle may decrease. In such a case, the second profile data does not change gently between the designated outer position and the long axis, and has a peak value larger than the background density at the designated outer position (however, the density of the left ventricular contrast image). If the value is lower than the density value of the surrounding area). Therefore, such a peak value is detected as the maximum fluctuation position. When this peak value is detected, the pixel value at the position of the peak value is replaced as the background density.

On the other hand, a representative internal concentration value of the left ventricle (for example, the minimum value of the second profile data) is obtained on each of the vertical axes. A threshold value for contour extraction is calculated for each vertical axis on the left and right of the long axis by weighted averaging the background density and the internal density value. Therefore, the threshold value and the second profile data are compared in the range from the contour search start point, which is provided at a position outside the contour point one line before on the vertical axis by a predetermined distance, to the major axis. As a result, the position of the pixel value below the threshold value is detected as the left ventricle contour point on each vertical axis.

As a result, even if a structure such as ribs appears in the periphery of the left ventricular contrast image, and the pixel value of the structure may be lower than the normal peripheral portion, regardless of the decrease. , The background density that can almost avoid the influence of the structure is automatically determined, and the accurate threshold value can be set. This improves the accuracy of contour extraction. Even if the influence of the pixel value variation due to the structure is large and the magnitude relationship between the set threshold value and the background density is reversed, the contour search start point is changed to a position closer to the long axis, and false detection is avoided. Therefore, according to the second image processing apparatus and the method of the present invention, even when a structure such as a rib appears in the periphery of the left ventricle to be diagnosed and the background density is reduced by the structure. , The outer density value, that is, the background density, can be accurately selected without being affected by the decrease in the density value, which allows the threshold value to be determined with the same accuracy as when the structure is not visible, and thus the contour can be more accurate. In addition, it is possible to extract stably with good operation efficiency.

A third aspect of the present invention corresponding to the sixth embodiment described above.
According to the image processing apparatus and the method thereof, the preprocessing is performed to emphasize the change in the contrast agent concentration from the previous frame even in the portion where the contrast agent concentration is low, and the contrast agent concentration signal in the left ventricle is emphasized. , The threshold is determined to determine the contour point. Therefore, a desired contour can be automatically extracted even in a thin portion of the contrast medium in the left ventricle.

Furthermore, since the desired area can be automatically extracted with a small number of contours, correction is unnecessary or reduced, and the burden on the operator is reduced. In addition, since it is not necessary to increase the concentration of the contrast medium to be injected and the amount of the contrast medium to an unnecessarily large amount, the side effects of the contrast medium are reduced and the burden on the subject is reduced.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of an X-ray diagnostic apparatus including an image processing apparatus according to a first embodiment of the present invention.

FIG. 2 is a flowchart showing the operation of the contour extracting unit according to the first embodiment of the present invention.

FIG. 3 is a diagram for explaining a method of setting a long axis in the first embodiment.

FIG. 4 is a diagram for explaining a method of dividing a region and creating a long-axis perpendicular profile in the first embodiment.

FIG. 5 is a diagram for explaining a contour extracting method according to the first embodiment.

FIG. 6 is a diagram for explaining a background area determination method, a threshold value determination method, and a contour point determination method according to the first embodiment.

FIG. 7 is a flowchart showing the operation of the contour extracting unit according to the second embodiment of the present invention.

FIG. 8 is a diagram for explaining a method of determining a background area in the second embodiment.

FIG. 9 is a diagram for explaining how to obtain contour points in the second embodiment.

FIG. 10 is a diagram for explaining a method of extracting a left ventricle contour according to the second embodiment.

FIG. 11 is a flowchart showing the operation of the contour extracting unit according to the third embodiment of the present invention.

FIG. 12 is an explanatory diagram showing a relationship between a long axis perpendicular profile and a threshold value.

FIG. 13 is an explanatory diagram showing a relationship between a long axis perpendicular profile, a threshold value, a search direction, and contour points.

FIG. 14 is an explanatory diagram showing tracking of contour points for the left ventricle.

FIG. 15 is a detailed flowchart showing a background density determination process in FIG.

FIG. 16 is an explanatory diagram showing a relationship between a long axis perpendicular profile and a maximum density value.

FIG. 17 is an explanatory diagram illustrating the contour points of the left ventricle by the distance from the long axis.

FIG. 18 is a diagram for explaining a conventional inconvenience, (a) is an image diagram of the left ventricle showing the ribs, and (b) is B in (a).
The long-axis perpendicular profile figure which shows the mode of the density value along a -B 'line.

FIG. 19 is a flowchart showing the procedure of an apex search according to the fourth embodiment.

FIG. 20 is an explanatory diagram showing the procedure of an apex search according to the processing of FIG.

FIG. 21 is a partial flowchart showing the procedure of contour re-search according to the fifth embodiment.

22 is an explanatory diagram of contour re-search according to the process of FIG.

FIG. 23 is a flowchart showing a procedure for creating contour data.

FIG. 24 is a flowchart showing a contour smoothing procedure.

FIG. 25 is a diagram for explaining creation and smoothing of contour data.

FIG. 26 is a diagram showing a schematic configuration of an X-ray diagnostic apparatus including an image processing apparatus according to a sixth embodiment of the present invention.

FIG. 27 is a diagram showing a schematic configuration of a profile creation section 3c ₁ .

FIG. 28 is a diagram for explaining the operation of the profile creation unit 3c ₁ .

FIG. 29 is a diagram showing a first modification of the profile creation section 3c ₁ .

FIG. 30 is a diagram showing a second modification of the profile creation section 3c ₁ .

FIG. 31 is a diagram showing a third modification of the profile creation section 3c ₁ .

[Explanation of symbols]

1 ... X-ray tube, 2 ... Detection part, 2a ... Image intensifier, 2b ... Optical system, 2c ... TV camera, 3 ... Data processing unit, 3a ... A / D converter, 3b ... Image memory, 3c ... Contour Extraction unit, 3c ₁ ... Profile creation unit,
3c ₂ ... Edge determination unit, 4 ... Image display unit, 5 ... Input unit, 6
... analysis unit, P ... subject.

─────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A 61-253044 (JP, A) JP-A 62-104263 (JP, A) JP-A 1-238637 (JP, A) JP-A 2- 220638 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) A61B 6/00-6/14

Claims (24)

(57) [Claims] 1. An image collecting unit for collecting an image including a desired region of a subject, an image storing unit for temporarily storing the collected image, and a contour extraction for extracting a region of interest of a desired portion of the subject. Means for dividing the image stored in the image storage means into a plurality of areas, and a first profile for creating first profile data for each of the plurality of areas. Creating means, and a density setting for finding the density of the background area and the density of the internal area by searching a background area outside the area of interest and an internal area inside the area of interest based on the first profile data. Means, threshold value determining means for determining a threshold value for each of the first profile data from the density of the background area and the density of the internal area, and the first profile data. An image processing apparatus comprising: a contour setting unit that determines a contour point that is a boundary between the outside and the inside of the region of interest using the threshold determined for each data. 2. The contour extracting means further includes a long axis setting means for setting a long axis in a first direction of the image, wherein the first profile creating means is perpendicular to the long axis with respect to the image. A plurality of perpendiculars are drawn in the second direction, a plurality of first profile data are created for each of the plurality of perpendiculars, and the threshold value determining unit weights the density of the background area and the density of the internal area. The image processing apparatus according to claim 1, wherein the threshold value is determined on average. 3. The image processing apparatus according to claim 2, wherein the contour extracting means further includes means for searching a background area on the first profile data. 4. The first profile creating means draws the plurality of perpendicular lines at substantially equal intervals, and a contour point obtained from immediately preceding first profile data with a representative position of the background area as a contour search start point. 3. The image processing apparatus according to claim 2, further comprising a search start point setting unit that sets a predetermined position outside a predetermined distance. 5. The first profile creating means is configured such that, in the first profile data, a maximum density position between the outer predetermined position and the long axis in the longitudinal direction from the outer predetermined position. 5. The image according to claim 2 or 4, further comprising: a position detecting unit that detects the image density; and a background density replacing unit that replaces the density at the maximum density position detected by the position detecting unit with a background density. Processing equipment. 6. The second data creating means for creating the second profile data indicating the density distribution at a plurality of pixel positions on the long axis by the contour extracting means, and the boundary between the first profile data and the region of interest. 6. The image processing apparatus according to claim 2, further comprising a region-of-interest edge determining unit that automatically determines one end of the region of interest on the long axis based on a position near the region. 7. The contour setting means creates distance data creating means for finding a vertical distance to the major axis based on the contour points determined for each first profile data, and the distance data creating means creates the vertical data. A contour sum that replaces the vertical distance with an average value of at least the vertical distance created based on the contour points of the adjacent first profile data.
7. The image processing apparatus according to claim 6, further comprising a closing means. 8. The region of interest of the subject is the left ventricle of the heart, and the long axis is a line segment connecting a position near the apex of the left ventricle and the midpoint of the aortic valve. Item 6. The image processing device according to item 6. 9. The first data creating means includes means for creating first profile data starting from a position near the apex of the heart, and the contour setting means extends from the apex of the heart toward the aortic valve. 9. The image processing apparatus according to claim 8, further comprising means for detecting a contour point of each of the first profile data in the first direction toward which the image data is directed. 10. The region-of-interest edge determining means is the second device.
Differentiation means for differentiating the profile data twice, means for specifying the maximum value and the minimum value after the second differentiation, and a weighted average of the second profile data at both the maximum value and the minimum value are calculated and obtained. The image processing apparatus according to claim 9, further comprising means for specifying a position on the long axis corresponding to the obtained result as a position of the apex. 11. The determination means for determining whether or not the background density replacing means determines whether the maximum density position is between a contour point set in the preceding first profile data and the predetermined outer position. 6. The image processing apparatus according to claim 5, further comprising means for replacing the background density with the density at the maximum density position based on the determination result of the determination means. 12. The determination means for determining whether or not the background density replacing means is for determining whether the maximum density position is between the contour point set in the immediately preceding first profile data and the outer predetermined position. The image processing apparatus according to claim 5, further comprising: a start point changing unit that changes the contour search start point to the maximum density position based on the determination result of the determination unit. 13. The density setting means includes means for setting an addition average of the background density of the first profile data currently read and the background density of the already read first profile data as a new background density. The image processing device according to claim 2 or claim 5, 14. The image processing according to claim 2, wherein the first profile creating means includes means for creating a number of pieces of first profile data proportional to the length of the long axis. apparatus. 15. The search start point setting means includes means for setting a point outside a distance proportional to the length of the major axis as a search start point. Image processing device. 16. The contour setting means generates distance data creating means for finding a vertical distance to the major axis based on the contour points determined for each first profile data, and the distance data creating means creates the vertical data. 3. A contour smoothing means for replacing the vertical distance with an arithmetic mean value of at least the vertical distance created based on the contour points of the adjacent first profile data, further comprising:
The image processing apparatus according to claim 4. 17. The region of interest of the subject is the left ventricle of the heart, and the long axis is a line segment connecting a position near the apex of the left ventricle and the midpoint of the aortic valve. The image processing apparatus according to claim 2, claim 4, claim 5, or claim 16. 18. The first data creating means includes means for creating first profile data starting from a position near the apex of the heart, and the contour setting means in the direction of the aortic valve from the position of the apex of the heart. 18. The image processing apparatus according to claim 17, further comprising means for detecting a contour point of each of the first profile data in the first direction toward which the image data is directed. 19. The contour setting means detects a contour point along the first direction and then follows a second direction from the aortic valve toward the apex of the heart with the designated position on the long axis as a starting point. The image processing apparatus according to claim 18, further comprising means for re-detecting a contour point up to a position near the apex. 20. The threshold value determining means includes means for setting first and second threshold values respectively on the left and right sides of the long axis of the first profile data, and the contour setting means includes: 7. A means for detecting the contour points on the left and right sides of the long axis based on the first and second threshold values, respectively, any one of claim 2, claim 4, claim 5, and claim 6. The image processing device according to claim 1. 21. The first profile creating means includes means for determining a minimum value of the first profile data as the internal density, and the position detecting means determines a maximum value of the first profile data as the maximum density position. 17. The image processing apparatus according to claim 5, further comprising means for detecting as. 22. A means for the contour extracting means to perform a subtraction between the first image read from the image storage means and a second image one frame before the first image.
The method further comprising: means for performing absolute value processing on the image obtained by the subtraction; and means for multiplying the image subjected to absolute value processing by a predetermined coefficient and subtracting from the first image. The image processing apparatus according to claim 2, claim 4 to claim 7, claim 10, or claim 12. 23. The image collecting means includes means for collecting a contrast image of a radiographic image in which the desired region is filled with a contrast agent, according to any one of claims 1 to 5 or 12. The image processing apparatus according to claim 22. 24. The image processing apparatus according to claim 23, wherein the image acquisition unit includes a unit for acquiring an end-diastolic left ventricular contrast X-ray image.