JP2000217818A - Diagnostic imaging device - Google Patents

Abstract

(57) [Summary] [PROBLEMS] To perform accurate contour extraction and region extraction from a heart image. SOLUTION: A ventricle range detecting means 3 for detecting a range of a ventricle from a cardiac tomographic image, a valve position detecting means 4 for detecting a position of a valve portion, and a contour shape is corrected by a valve position, thereby forming a contour of the ventricle. And a region extracting unit for extracting a region of interest including a ventricle. This can prevent erroneous extraction of contours due to intracardiac structures such as papillary muscles and chordae, and enable accurate contour extraction. In addition, an appropriate ventricle range can be extracted for each time phase by a simple operation, and an appropriate region of interest can be set. As described above, the accuracy and efficiency of the image diagnosis are improved.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to an image processing apparatus for extracting a region of interest or an outline based on image data of a subject imaged by an ultrasonic diagnostic apparatus or MRI.

[0002]

2. Description of the Related Art Binarization of an image and edge detection are often used to extract the contour of a region of interest or an organ from a biological image. As a method of extracting a contour by edge detection, there is a method of performing edge detection radially from a reference position on an image as shown in FIG. 4 and extracting a portion having a high edge strength as a contour. FIG. 10 schematically shows an example in which the contour of the ventricle is extracted from the tomographic image of the heart by this method. There are parts such as papillary muscles and chordae in the ventricle, and it is not possible to determine whether the detected edge is a myocardium edge or a chordae edge using the edge information alone, and an incorrect contour is extracted. There was a problem.

As a method of extracting a region by binarizing an image, there is a method of performing binarization by setting a region of interest.
FIG. 15 shows an example in which the region of interest for indicating the position of the ventricle is manually set. However, since the position of the region of interest is fixed, an organ that moves like the heart is unnecessary when contracted. There is a problem of including an unnecessary area.

Japanese Patent Application Laid-Open No. 9-122123 discloses a method of detecting a position having a peak value at a pixel in a predetermined area as a pixel serving as a boundary between the left ventricle and the left atrium of the heart. It is necessary to appropriately determine a predetermined area including the area, and the peak value is not always shown at the boundary, and it has been difficult to detect the boundary stably.

[0005]

As described above, in the conventional contour extraction, there has been a problem that the contour is erroneously extracted due to the structure of an organ which cannot be determined only by detecting the edge of the image. In addition, a fixed region of interest cannot always indicate an appropriate region for a moving organ, and it is necessary to manually set an appropriate region of interest for a large number of images such as moving images. There were many operations and difficulties. In the method of using the peak value of the pixel as a method of determining the boundary position between the ventricle and the atrium, the peak value may not always indicate the boundary position, and it is necessary to determine an appropriate search area in order to obtain a good result. there were.

Accordingly, an object of the present invention is to detect a partial position of an organ from an image, thereby preventing erroneous extraction of a contour caused by a structure of an organ that cannot be determined only by detecting an edge of the image, and achieving high accuracy. This enables contour extraction. Another object of the present invention is to provide an image diagnostic apparatus capable of easily setting an appropriate region of interest for a moving organ. It is another object of the present invention to provide an image diagnostic apparatus that can efficiently detect a valve position by limiting a search range of a valve portion after detecting a ventricular range.

[0007]

SUMMARY OF THE INVENTION The present invention provides a ventricle range detecting means for detecting a range of a ventricle from a cardiac tomographic image, a valve position detecting means for detecting a position of a valve portion from a cardiac tomographic image, and a valve position detecting means. And a region extracting unit for extracting a region of interest including the ventricle.

According to the present invention, it is possible to stably extract a specific part of an organ, to prevent erroneous extraction of a contour due to a structure of an organ that cannot be discriminated only from edge detection of an image, and to obtain a contour with high accuracy. Extraction becomes possible. Regarding the region of interest setting, it is possible to easily set an appropriate region of interest in accordance with the movement of the organ. Further, the valve position can be efficiently detected.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram showing a structure of an image diagnostic apparatus to which a contour extraction method according to one embodiment of the present invention is applied, and FIG. 2 is a flowchart showing a flow of operation of the contour extraction method according to one embodiment of the present invention. is there. The present embodiment is an example of use of the contour extraction method, and is an image diagnostic apparatus that inputs a moving image of a heart, extracts an inner contour of a ventricle, measures and displays an area of a ventricular cross section.

In FIG. 1, reference numeral 1 denotes an image input unit for inputting image data, 2 denotes an image memory for storing image data, 3 denotes a ventricle range detecting unit for detecting a range of a ventricle in an image, Reference numeral 4 denotes a valve position detecting unit for detecting a position of a heart valve in an image, 5 denotes a contour extracting unit for extracting a ventricular contour, and 6 calculates a measurement value useful for diagnosis from the extracted contour information. And a display unit 7 for displaying the extracted contour information, the measured information, and the image.

First, moving image data is input from the image input unit 1 and stored in the image memory 2. The operator inputs the approximate position of the heart ventricle in the image using a pointing device (not shown) or the like (S1). If the position where the heart is to be imaged is substantially determined, the input of the approximate position in S1 may be omitted, and a preset coordinate value may be used as the approximate position. Next, the ventricle range is detected by the ventricle range detector 3 based on the approximate position of the ventricle (S2).

A case where a contour model that converges on a portion having a high luminance value in an image is used as an example of the ventricle range detection processing will be described with reference to a detailed flow shown in FIG. This contour model is a contour represented by a plurality of control points. For example, in the case of a heart image captured by an ultrasonic diagnostic apparatus, the myocardial portion has a high luminance value and the blood portion has a low luminance value.

Therefore, in this method, a myocardial portion surrounding the ventricle, that is, a range of the ventricle, is detected using a contour model converging on a portion having a high luminance value. For an image in which the myocardial portion is imaged with a lower luminance value than the blood portion, a contour model that converges to a portion with a lower luminance value in the image may be used. The contents of the processing of the ventricular range detection will be described. First, the search range of the ventricle range is determined based on the approximate position of the ventricle (S
10).

For the first frame of the moving image processing, a search range may be set for each control point as shown in FIG. 4, for example, using the approximate position. Further, for the second and subsequent frames of the moving image processing, the search range may be set as shown in FIG. 4 similarly to the first frame, or the search range may be set as shown in FIG. May be set.

Next, a pixel having the maximum luminance is searched in the determined search range, and a contour point candidate is generated by moving a control point to the pixel position (S11). Next, the contour shape is smoothed (S12). The smoothing of the contour shape is performed in order to correct a portion that is partially erroneously extracted due to noise or the like and to search again. Next, if the change amount of the contour shape is equal to or larger than the threshold value, the processing of S10, S11, and S12 is repeated. In the second and subsequent processes of S10 to S12, by setting the search range smaller than the first time, it is possible to generate a contour having the maximum luminance near the contour smoothed in the first process. When the change amount of the contour shape becomes equal to or less than the threshold value, the process is terminated (S13), and the ventricle range contour is determined.

After the ventricle range is determined, the search range of the valve portion is determined (S3). Since the shape of the valve portion in the tomographic image of the heart has a characteristic image pattern as compared with the other portions of the heart, it is effective to detect the valve portion. Generally, an image taken from the apex side of the transthoracic wall by the ultrasound diagnostic apparatus is the apex on the upper side of the image and the valve part on the lower side of the image. Thus, the search range for the right and left valve portions may be set. In an image taken by the ultrasound diagnostic apparatus in the transesophageal tract, the upper part of the image is generally the valve part and the lower part of the image is the apex, so that, for example, the search area may be the right and left valve parts as shown in FIG. Thus, the search range of the valve portion may be determined from the ventricle range or the like according to the posture of the heart. By determining the search range based on the ventricular range, the search range can be narrowed compared to searching for the valve portion directly from the entire image, so that the processing can be speeded up and the detection rate can be improved.

Next, a valve portion is detected from the search range determined in S3 (S4). In the detection of the valve portion, the position is detected by collation with a dictionary image created from an image of the valve portion or the like (FIG. 8). The matching process may be performed by template matching, or may use a subspace method or a composite similarity method. Through the above processing, the position of the valve in the image is detected. Next, a contour initial shape for extracting the inner contour of the ventricle is calculated (S5).

The initial shape of the contour is set, for example, by correcting a part of the elliptical shape to a valve position as shown in FIG. 9 using the ventricle range set in S2 and the valve position detected in S4. I do. By detecting the valve portion position and using it for contour extraction in this manner, a correct contour can be extracted without being caught by an erroneous structure in the ventricle, as compared with a case where the contour is searched only from the approximate ventricle position. .

Next, an image edge is detected near the initial contour corrected at the valve position, and an inner contour of the ventricle is extracted (S).
6). By performing the processing from S2 to S6 on all frames, the inner contour of the ventricle is extracted from the moving image (S
7). For the second and subsequent frames, it is efficient to generate the initial shape using the contour position of the previous frame.

Further, as a modification, when the distance between the valve position in the previous frame and the detected valve position in the current frame is greater than or equal to a threshold value when calculating the contour initial shape in S5, the contour initial shape based on the valve position is determined. When the correction is not performed, and when the value is smaller than the threshold value, it is possible to correct the initial shape of the contour based on the valve position. By controlling the application of the correction in this way, even when an erroneous valve position is detected, the contour can be stably extracted. With the above processing, the extraction of the contour ends.

Next, the sectional area of the ventricle is calculated by the measuring unit 6 from the extracted contour information (S8), and the display unit 7 displays, for example, FIG.
2 is displayed (S9). The detected position and movement amount of the valve portion may be displayed in synchronization with an electrocardiogram or the like.

In the present embodiment, the valve position can be efficiently detected by detecting the ventricle range, and erroneous extraction of the ventricular contour as shown in FIG. 10 is prevented by performing contour correction using the valve position. 11, it is possible to extract the contour with high accuracy, and it is possible to improve the accuracy of the measurement result at the subsequent stage.

A second embodiment relating to the detection of a ventricular range according to the present invention will be described. The present embodiment is an example in which the region extraction method of the present invention is applied. An image diagnosis in which a region of interest is automatically generated by detecting a ventricular range, and a pixel value in the region of interest is binarized to calculate an area of the ventricular region. It is an example of an apparatus. FIG.
Is a block diagram of the present embodiment, and FIG. 14 is a flowchart showing a flow of processing of the present embodiment.

In FIG. 13, 1 is an image input unit for inputting image data, 2 is an image memory for storing image data, 3 is a ventricle range detecting unit for detecting a ventricular range in the image, 11 Is a binarization processing unit that binarizes pixel values in the region of interest using the detected ventricle range as the region of interest;
Reference numeral 12 denotes an area measurement unit that calculates the area in the ventricle from the binarized image data, and 13 denotes a display unit that displays extracted contour information, measured information, and an image.

First, moving image data is input from the image input unit 1 and stored in the image memory 2. As in the first embodiment, the operator inputs the approximate position of the cardiac ventricle in the image using a pointing device (not shown) or the like (S2).
1). If the position at which the heart is to be imaged is substantially determined, the input of the approximate position in S21 may be omitted, and a preset coordinate value may be used as the approximate position.

Next, the ventricle range is detected by the ventricle range detector 3 based on the approximate position of the ventricle (S22). The method described in the first embodiment can be used as a method for detecting the ventricle range. Further, the ventricle range may be determined by collation with a dictionary created from an image of the ventricle part or the like. The matching method can use template matching, a subspace method, or a composite similarity method as in the valve position detection in the first embodiment. The range of the detected ventricle is used as a region of interest in the following processing.

Next, using the ventricle range detected by the ventricle range detector as a region of interest, the pixel values in the region of interest are binarized (S23). In the case of an ultrasonic image, generally, the luminance value of the myocardial part is high and the luminance value of the blood part is low. Therefore, the area of the blood part in the ventricle range is calculated from the number of low pixel values among the binarized pixel values. (S24). Next, the measurement result and the image are displayed on the display unit 7 (S25).

As described above, by detecting the ventricle range for each frame and setting it as the region of interest, an extra portion such as the atrium is not included as compared with the related art using a fixed region of interest. It becomes possible. Also, the setting of the region of interest is unnecessary, and only by specifying a simple approximate position or adjusting the display position of the organ to be almost constant,
Since no position input is required, operability is greatly improved.

In the above embodiment, a mode separated from the imaging apparatus has been described, but it may be incorporated in an ultrasonic diagnostic apparatus or the like. Also, add an image input function to the calculator,
The processing of the present invention can be realized by software.

[0030]

According to the present invention, it becomes possible to accurately extract the contour of an organ and to easily set an appropriate region of interest.
The efficiency of diagnosis and measurement is dramatically improved. Therefore, the effect is great.

[Brief description of the drawings]

FIG. 1 is a block diagram of a first embodiment according to a contour extraction method of the present invention.

FIG. 2 is a flowchart of a first embodiment according to a contour extraction method of the present invention.

FIG. 3 is a flowchart of a ventricular range detection unit according to the first embodiment.

FIG. 4 is a diagram showing an example of a search range of an edge or a ventricle range.

FIG. 5 is a diagram showing an example of a search range of a ventricle range.

FIG. 6 is a diagram showing an example of a search range of a valve portion.

FIG. 7 is a diagram showing an example of a search range of a valve portion.

FIG. 8 is a diagram showing a detected valve portion.

FIG. 9 is a diagram illustrating a setting example of an outline initial shape.

FIG. 10 is a diagram illustrating an example of a contour that is erroneously extracted;

FIG. 11 is a diagram illustrating an example of a contour correctly extracted using a valve position.

FIG. 12 is a diagram illustrating a display example of a measurement result.

FIG. 13 is a block diagram of a second embodiment according to the region extracting method of the present invention.

FIG. 14 is a flowchart of a second embodiment according to the region extracting method of the present invention.

FIG. 15 is an explanatory diagram of an example in which a fixed region of interest includes an erroneous region.

[Explanation of symbols]

 Reference Signs List 1 image input unit 2 image memory 3 ventricular range detecting unit 4 valve position detecting unit 5 contour extracting unit 6 measuring unit 7 display unit

Claims (7)

[Claims] 1. A valve position detecting step of detecting a position of a valve portion from a heart image, a contour extracting step of correcting a contour shape based on the detected valve position to extract a ventricular contour, or a region of interest including a ventricle. An image diagnostic apparatus comprising: an area extracting step of extracting an image. 2. A ventricular range detecting step of detecting a ventricular range from a heart image, a contour extracting step of extracting a ventricular contour based on the detected ventricular range, or a region extracting step of extracting a region of interest including the ventricle. An image diagnostic apparatus characterized by the above-mentioned. 3. The valve position detecting step according to claim 1,
A ventricle range detection step of detecting a ventricle range from a heart image, a valve search range determination step of determining a search range of a valve portion based on the ventricle range, and a valve search step of detecting a position of the valve portion from the search range. The diagnostic imaging apparatus according to claim 1, wherein the diagnostic imaging apparatus is configured. 4. The ventricle range detecting step according to claim 2, wherein the ventricle range is detected by collating with a dictionary created from a partial ventricle image. Diagnostic imaging device. 5. The method according to claim 2, wherein the step of detecting the ventricle range uses a predetermined coordinate or a manually input coordinate as a reference position and a point where the luminance becomes an extreme value in a search range based on the reference position. 4. The image diagnostic apparatus according to claim 2, further comprising a luminance search step of searching for a contour curve and generating a contour curve. 6. The valve search step according to claim 3, wherein the valve portion is detected by performing a search for the valve portion by collating with a dictionary created from the valve portion image. The image diagnostic apparatus according to claim 4, wherein 7. The contour extracting step according to claim 1, wherein, based on the positional relationship between the valve position and the candidate contour, a correction determining step for determining whether or not to perform a contour shape correction based on the valve position is determined. 3. A contour correcting step for performing a correction only when there is a correction.
The image diagnostic apparatus according to the above.