JP2005237555A - Evaluation method of heart wall motion, evaluation device and program of heart wall motion - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an evaluation method of heart wall motion capable of automatically evaluating whether or not heart wall motion is normal based on a cardiogram. <P>SOLUTION: The method has a first step of extracting the contour of a heart from a plurality of cardiograms captured continuously in a time series, a second step of determining a predetermined characteristic point from the extracted heart contour, a third step of calculating the momentum of the characteristic point corresponding the characteristic point in the plurality of cardiograms, and a fourth step of determining whether or not the heart wall motion is abnormal using fuzzy inference based on the calculated momentum of the characteristic point. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a heart wall motion evaluation method, a heart wall motion evaluation apparatus, and a program.

Conventionally, in order to evaluate the heart wall motion, the heart wall contour extracted from each heart image obtained continuously in time series is used, and the structural features of the heart with respect to the heart wall contour in each heart image obtained thereby. A predetermined feature point, which is a point having a point, is divided on the basis, and the divided contour is moved from the reference point for each divided region or divided point of the heart wall contours of a plurality of heart images different in time series. And a technique for displaying the obtained movement amount as a numerical value or a graph is disclosed (for example, Patent Document 1).
Japanese Patent Laid-Open No. 10-99334

  However, the method described in Patent Document 1 merely digitizes or graphs the movement change of the heart wall motion, and the evaluation as to whether or not the heart wall motion is normal is ultimately an evaluator. Therefore, the evaluation was not possible without the experience and skills of the evaluator.

  Therefore, an object of the present invention is to provide a heart wall motion evaluation method, a heart wall motion evaluation apparatus, and a program capable of automatically evaluating whether or not heart wall motion is normal based on a heart image. .

As means for solving the above problems, the invention according to claim 1 is a method for evaluating heart wall motion, for example, as shown in FIG.
A first step of extracting a heart contour from a plurality of heart images obtained continuously in time series;
A second step of determining a predetermined feature point from the extracted heart contour;
A third step of calculating the momentum of the feature points by associating the feature points in the plurality of heart images;
A fourth step of determining whether the heart wall motion is abnormal using fuzzy inference based on the calculated momentum of the feature point;
It is characterized by having.

The invention according to claim 2 is a method for evaluating heart wall motion according to claim 1,
The cardiac image is an X-ray CT image.

Invention of Claim 3 is the evaluation method of the heart wall motion of Claim 1 or 2, Comprising:
The second step includes
Determining two break points that are the boundary between the blood vessel portion and the left ventricular portion from the extracted heart contour;
Cutting the segmented blood vessel part and setting the midpoint between the segmentation points as the starting point of the left ventricular axis;
Finding the center of gravity for the left ventricular region, connecting the start point of the left ventricular axis and the center of gravity, and the intersection of the extension line and the left ventricular contour as the end point of the left ventricular axis;
A step of equally dividing a start point of the left ventricular axis and an end point of the left ventricular axis to determine a division point;
Drawing a perpendicular to the left ventricular axis from the determined dividing point, and determining an intersection of the perpendicular and the left ventricular contour as a feature point;
It is characterized by having.

Invention of Claim 4 is the evaluation method of the heart wall motion as described in any one of Claims 1-3,
The third step includes
The momentum of the feature point is calculated based on the movement distance between the start point of the left ventricular axis and each feature point.

Invention of Claim 5 is the evaluation method of the heart wall motion as described in any one of Claims 1-4,
The fourth step includes
Obtaining a contraction rate based on momentum of the feature point in the maximum diastole and maximum systole of the left ventricle, and using the contraction rate as input data for fuzzy inference;
Applying a predetermined membership function set in advance to the input data, and calculating an evaluation value for heart wall motion;
It is characterized by having.

The invention described in claim 6 is a method for evaluating heart wall motion according to claim 5,
The fourth step includes
Before the evaluation value is calculated by the predetermined membership function, the method includes a step of performing weighting according to the importance of the feature point.

The invention according to claim 7 is a heart wall motion evaluation apparatus, for example, as shown in FIG.
Contour extraction means (for example, CPU 31, contour extraction program 330) for extracting a heart contour from a plurality of heart images obtained continuously in time series;
Feature point determination means for determining a predetermined feature point from the heart contour extracted by the contour extraction means (for example, CPU 31, division point determination program 331, start point determination program 332, end point determination program 333, division point determination program 334) , Feature point final decision program 335),
Momentum calculation means (for example, CPU 31, exercise amount calculation program 336) for calculating the momentum of the feature points in association with the feature points in the plurality of cardiac images determined by the feature point determination means;
Based on the momentum of the feature points calculated by the momentum calculation means, determination means (for example, a CPU 31, an input data determination program 337, an evaluation value calculation program 338, which determines whether or not the heart wall motion is abnormal using fuzzy reasoning) Weighting program 339)
It is characterized by having.

The invention according to claim 8 is the heart wall motion evaluation apparatus according to claim 7,
The cardiac image is an X-ray CT image.

The invention according to claim 9 is the heart wall motion evaluation apparatus according to claim 7 or 8,
The feature point determining means includes
A breakpoint determination means (for example, a CPU 31, a breakpoint determination program 331) for determining two breakpoints that are boundaries between the blood vessel portion and the left ventricle portion from the heart contour extracted by the contour extraction means;
A start point determining means (for example, CPU 31, start point determining program 332) that cuts off the blood vessel portion divided by the dividing point determining means and sets the midpoint between the dividing points as the starting point of the left ventricular axis;
A center of gravity is obtained for the left ventricular region, the start point of the left ventricular axis is connected to the center of gravity, and an end point determining means (for example, CPU 31) having the intersection of the extension line and the left ventricular contour as the end point of the left ventricular axis. , End point determination program 333),
Split point determination for equally dividing the start point of the left ventricular axis determined by the start point determining means and the end point of the left ventricular axis determined by the end point determining means to determine a split point Means (for example, CPU 31, division point determination program 334);
Feature point final determination means (for example, CPU 31, final feature point final) that draws a perpendicular line from the division point determined by the division point determination means to the left ventricular axis and determines an intersection of the perpendicular line and the left ventricular contour as a feature point. Decision program 335),
It is characterized by having.

A tenth aspect of the present invention is the heart wall motion evaluation apparatus according to any one of the seventh to ninth aspects,
The momentum calculating means includes
The momentum of the feature point is calculated based on the movement distance between the start point of the left ventricular axis and each feature point.

Invention of Claim 11 is a heart wall motion evaluation apparatus as described in any one of Claims 7-10,
The determination means includes
Input data determination means (for example, CPU 31, input data determination program for obtaining contraction rate based on momentum of the feature point in the maximum diastole and maximum systole of the left ventricle and using the contraction rate as input data for fuzzy inference 337),
An evaluation value calculation means (for example, CPU 31, evaluation value calculation program 338) for calculating an evaluation value for heart wall motion by applying a predetermined membership function set in advance to the input data;
It is characterized by having.

The invention according to claim 12 is the heart wall motion evaluation apparatus according to claim 11,
The determination means includes
Before calculating an evaluation value by the predetermined membership function, weighting means (for example, CPU 31, weighting program 339) for performing weighting according to the importance of the feature point is provided.

The invention according to claim 13 is a program,
On the computer,
A first function of extracting a heart contour from a plurality of heart images obtained continuously in time series;
A second function for determining a predetermined feature point from the extracted heart contour;
A third function for calculating the momentum of the feature points by associating the feature points in the plurality of heart images;
A fourth function for determining whether or not the heart wall motion is abnormal by fuzzy reasoning based on the calculated momentum of the feature point;
It is characterized by realizing.

  According to the first, seventh, and thirteenth aspects of the present invention, a heart contour is extracted from a plurality of heart images obtained continuously in time series, predetermined feature points are determined from the heart contour, and a plurality of hearts are extracted. The amount of motion of the feature point is calculated in association with the feature point in the image, and based on the calculated amount of motion of the feature point, it is determined whether or not the heart wall motion is abnormal using fuzzy reasoning. Therefore, it is possible to automatically evaluate whether or not the heart wall motion is abnormal based on the heart image.

  According to the second and eighth aspects of the invention, the same effect as that of the first aspect of the invention can be obtained. Particularly, since the heart image is an X-ray CT image, the burden on the diagnosis of the subject is reduced. In addition to being reduced, since it has high spatial resolution or contrast resolution, more accurate evaluation can be performed.

  According to the third and ninth aspects of the invention, the same effects as those of the first, second, seventh and eighth aspects of the invention can be obtained, and in particular, the blood vessel portion can be separated from the extracted heart contour. The midpoint between the cut and split points is the start point of the left ventricular axis, connects the start point of the left ventricular axis and the center of gravity, and the intersection of the extension line and the left ventricular contour is the end point of the left ventricular axis. Divide equally between the start and end points of the axis to determine the dividing point, draw a perpendicular to the left ventricular axis from the determined dividing point, and determine the intersection of the perpendicular and the left ventricular contour as the feature point Therefore, the feature points can be determined with high accuracy without being affected by the contraction and expansion of the left ventricle.

  According to the fourth and tenth aspects of the invention, it is obvious that the same effects as those of the first, second, third, seventh, eighth and ninth aspects can be obtained. Since the momentum of the feature point is calculated based on the movement distance between the feature points, the movement of the feature point according to the momentum of the heart wall can be grasped more easily.

  According to the fifth and eleventh aspects of the invention, it is a matter of course that the same effect as that of any one of the first, second, third, fourth, seventh, eighth, ninth and tenth aspects can be obtained. In particular, the contraction rate calculated based on the momentum of the feature point is used as input data, and the inference result of fuzzy inference is calculated based on the input data. Therefore, it is possible to automatically make a judgment approximate to a judgment based on advanced empirical knowledge possessed by a doctor.

  According to the inventions of claims 6 and 12, it is obvious that the same effect as that of the invention of claim 5 or 11 can be obtained. In particular, weighting is performed according to the importance of the feature points. The contraction rate based on the momentum of the feature points is mainly used for the heart wall motion evaluation. Therefore, it is possible to reduce the misidentification rate of normal cases as abnormal cases without lowering the identification rate of abnormal cases, and to further improve the accuracy of judgment.

Hereinafter, the best embodiment according to the present invention will be described in detail.
(About the evaluation method of heart wall motion)
The heart wall motion evaluation method according to the present invention automatically evaluates whether or not the heart wall motion is abnormal based on the heart image. Specifically, for example, a heart image acquisition process, a heart contour extraction process, a feature point determination process, a feature point momentum calculation process, and a fuzzy inference process are performed in this order.

(1) Cardiac image acquisition processing The cardiac image used in the cardiac wall motion evaluation method according to the present invention is a plurality of cardiac images obtained continuously in time series, for example, obtained by left ventricular imaging in cardiac catheterization. The obtained X-ray CT image is used.
As the heart image, for example, a heart image for approximately one heartbeat from the maximum diastole of the left ventricle of the heart to the 12th slice position is used as a processing target.
Note that the left ventricle heart image is used for evaluating heart wall motion because the left ventricle has the thickest myocardium in the heart to pump blood to tissues throughout the body. This is because, for example, an ischemic state or the like is a very large part.

(2) Heart contour extraction processing Next, for example, edge information is acquired from the acquired heart image, and a heart contour is extracted based on the acquired edge information. Here, the acquisition of edge information uses, for example, a Sobel operator and a dynamic net. More specifically, after extracting edge information about the heart contour using the Sobel operator, using the dynamic net composed of a plurality of grid points, the density information is acquired from the heart image, Based on this, the lattice points of the dynamic net are moved to a predetermined position, and the stopped lattice points are connected to obtain a heart contour image.

(3) Feature Point Determination Processing Next, the left ventricular axis and feature points are determined from the extracted heart contour.
1 and 2 are schematic diagrams for explaining a feature point determination process from the extracted heart contour.
An example of the procedure for determining feature points is shown below.
First, as shown in FIG. 1A, two dividing points A that are boundaries between the blood vessel portion and the left ventricle portion are determined from the heart contour. Here, the dividing point A is determined, for example, as the most constricted portion of the boundary between the blood vessel portion and the left ventricle portion.

Next, as shown in FIG. 1 (b), the segmented blood vessel portion is cut out, and the midpoint between the segmentation points A is set as the start point B of the left ventricular axis D.
Next, as shown in FIG. 1C, the center of gravity is obtained for the left ventricular region, the start point B of the left ventricular axis D is connected to the center of gravity, and the intersection of the extension line and the left ventricular contour is defined as the left ventricular axis. Let D be the end point C.
Next, as shown in FIG. 2A, the division point E is determined by equally dividing the start point B and the end point C of the left ventricular axis D.

Next, as shown in FIG. 2B, a new division point F is determined by equally dividing the determined division point E and the start point B and end point C of the left ventricular axis D, As shown in FIG. 2 (c), the newly determined dividing point F and the starting point B and the ending point C of the left ventricular axis D are further divided equally between the dividing point E and the dividing point F. Then, new division points G are sequentially determined.
Next, as shown in FIG. 2 (d), a perpendicular line is drawn with respect to the left ventricular axis D from the determined division points E to G, and an intersection of the perpendicular line and the left ventricular contour is set as a feature point H.
In FIG. 2, the number of division points is set to 7. However, the number of division points is not limited to this, and the setting can be changed as appropriate.

(4) Feature point momentum calculation processing After determining the feature point of the heart contour, the left ventricle is divided into predetermined regions in order to process the prediction of the stenosis of the coronary artery or the evaluation of cardiac function on the image. For example, the left ventricle is divided into three regions, an upper part, a lower part, and a tip part.
Next, the momentum of each feature point is calculated using the left ventricular axis.
Here, the momentum of the feature point is indicated by, for example, a moving distance between the start point of the left ventricular axis and each feature point. Since the start point of the left ventricular axis is on the boundary surface between the aorta and the myocardium, it is least affected by contraction / dilatation, and each feature point is most active depending on the amount of motion of the left ventricular wall. This is because the change in momentum can be easily grasped.
For example, as shown in FIG. 3, a vector A + B obtained by adding a horizontal vector A and a vertical vector B with respect to the left ventricular axis is used as the momentum of the feature point.

FIG. 4 is a diagram showing an example of the momentum of each feature point when 15 feature points are determined. FIG. 4A shows a normal case of the upper left ventricle, and FIG. 4B shows the upper left ventricle. Each abnormal case is shown.
When the heart wall motion is normal, as shown in FIG. 4 (a), it can be seen that the contraction motion of the myocardium is active at each feature point, and that a substantially equal amount of motion is maintained vertically and horizontally.
When the heart wall motion is abnormal, as shown in FIG. 4B, it can be seen that the myocardium is not sufficiently contracting at each feature point.

(5) Fuzzy inference processing Next, fuzzy inference processing is performed to determine whether the heart wall motion is abnormal.
Specifically, first, the contraction rate is obtained based on the momentum of the feature points in the maximum diastole and maximum systole of the left ventricle, and the obtained contraction rate is used as input data for fuzzy inference.
First, the left ventricular image determined by the doctor to be in the maximum diastole is set as the start slice position, and the left ventricular region area is obtained for all images from the image up to the 12th slice position corresponding to one heartbeat, The slice position where the area of the area is the smallest is taken as the maximum systole.

  Next, the difference between the momentum of the feature point of the maximum diastole and the momentum of the feature point of the maximum systole is obtained, and the difference value is divided by the momentum of the feature point of the maximum diastole to obtain the contraction rate (Equation 1) .

Next, a predetermined membership function set in advance is applied to the contraction rate as input data, and an evaluation value for determining whether or not the heart wall motion is abnormal is calculated.
Here, the definition of the membership function will be specifically described.
The membership function is a function defined based on fuzzy rules. For example, two types of fuzzy rules, a normal rule indicating a normal case of heart wall motion and an abnormal rule indicating an abnormal case, are determined. That is, the former indicates that the image processing target area is normal when the shrinkage rate is higher than a predetermined value, whereas the latter indicates that the image processing target is normal when the shrinkage rate is lower than the predetermined value. The area indicates an abnormality. An example of the membership function obtained based on the normal rule and the abnormal rule is shown in Formula 2 and FIG.

  Equation 2 (a) is a membership function based on a normal rule, and Equation 2 (b) is a membership function based on an abnormal rule. FIGS. 5A and 5B are diagrams for explaining the application result of the membership function shown in Equation 2, in which FIG. 5A is an application result for a shrinkage rate of 0.3, and FIG. Application results are shown. 5A and 5B are diagrams showing the membership function shown in Equation 1 (a), and each lower row shows the membership function shown in Equation 1 (b).

As shown in FIG. 5A, when the contraction rate is 0.3, an evaluation value in which the normal rule responds elastically and the abnormal rule responds inelastically is obtained and reflected in the inference result.
On the other hand, as shown in FIG. 5B, when the contraction rate is 0.15, an evaluation value is obtained in which the normal rule is inelastic and the abnormal rule is elastic, and is reflected in the inference result. The

Further, an optimum membership function based on the normal rule and the abnormal rule is extracted. This extracted membership function is set as a predetermined membership function used for fuzzy inference.
6A and 6B are diagrams for explaining extraction of an optimal membership function. FIG. 6A shows a membership function based on a normal rule, and FIG. 6B shows a membership function based on an abnormal rule. FIG. As shown in FIGS. 6 (a) and 6 (b), membership is obtained by obtaining the identification result of normal cases and abnormal cases at that time while changing the slope of the function in increments of 0.05. Extract a function.

  For example, set the output value of fuzzy inference to 0.3 when reacting only to abnormal rules, set to 0.7 when responding only to normal rules, and set the threshold to 0.5, and it should be 0.5 or more. If it is determined that the case is a normal case and if it is less than 0.5, it is an abnormal case, only the membership function with an output value of 0.45 or less in the abnormal case is extracted.

Based on the evaluation value calculated by giving input data to the membership function extracted as described above, the inference result of fuzzy inference is obtained, all the inference results are synthesized, and the final fuzzy inference result is obtained. Based on the result, it is determined whether or not the region to be processed of the heart image is an abnormal heart wall motion.
From the inference results obtained by fuzzy inference, the probability that the image processing target area is estimated to have abnormal heart wall motion (heart wall motion abnormal rate) and the probability that heart wall motion is normal (heart It is also possible to obtain a normal rate of wall motion and present it as a quantitative and numerical index.

It is also possible to improve the identification rate by applying a predetermined weight to the above method.
For example, important feature points with high importance that are likely to cause symptoms of abnormal heart wall motion are extracted, and weighting is performed according to the extracted important feature points. Note that selection of important feature points is determined by, for example, doctor's empirical knowledge or anatomical knowledge, or an analysis result by a genetic algorithm.

Specifically, as shown in FIG. 7, the feature points are grouped into important feature points and feature points other than the important feature points (non-important feature points), and the contraction rate based on the momentum of the non-important feature points. Weighting is performed by changing some identification conditions of the membership function applied to.
First, as shown in FIG. 7, important feature points are extracted. Each feature point is numbered, and the circled circle indicates the important feature point. For example, feature points 3, 4, and 5 are extracted as important feature points in the upper left ventricle, feature points 14, 15, and 16 are extracted in the lower portion, and feature points 8, 9, and 10 are extracted in the distal end portion.
At this time, the extracted important feature points may be distinguished from non-important feature points by differences in color, shape, and the like.

  Next, for example, the membership function shown in Equation 2 is applied to the contraction rate based on the momentum of the important feature point, and the membership function shown in Equation 3 is applied to the contraction rate based on the momentum of the non-important feature point. Apply. Equation 3 (a) is a membership function based on normal rules, and Equation 3 (b) is a membership function based on abnormal rules.

Here, comparing Equations 2 and 3, in Equation 2 (a), when the contraction rate is greater than 0.4, it is determined as a normal case with a very high probability, whereas Equation 3 (a) In, the same determination is made if the shrinkage rate is greater than 0.2. That is, it can be said that the identification condition is relaxed in Equation 3 (a) than in Equation 2 (b). In this way, weighting is performed according to the importance of the feature points.
In addition, Equation 2 (b) and Equation 3 (b) have the same identification conditions, but this is an abnormal case in consideration of the possibility of abnormal heart wall motion even if it is an unimportant feature point. This is in order not to miss the discovery.

  After weighting, the membership function is applied to the input data in the same manner as described above to obtain the inference result of fuzzy inference.

(System configuration)
Next, a system configuration for determining cardiac wall motion abnormality will be described with reference to FIG.
The heart wall motion evaluation apparatus 1 shown in FIG. 8 includes therein an image acquisition unit 2 that acquires a heart image, an evaluation unit 3 that evaluates heart wall motion, an input unit 4, a display unit 5, and the like. Yes.

  The image acquisition unit 2 has, for example, a communication function, and acquires an X-ray CT image of the subject's heart via a communication line with an X-ray CT image capturing device (not shown) of the subject's heart. It has become.

The evaluation unit 3 includes a CPU 31, a RAM 32, a ROM 33, and the like.
The CPU 31 expands a program designated from various programs stored in the ROM 33 in the work area of the RAM 32 and executes various processes according to the program.
The RAM 32 forms a storage area for temporarily storing programs, data, and the like, a work area for executing the programs, and the like in various processes executed by the CPU 31.
The ROM 33 stores various application software and programs necessary for executing the operation of the present invention. Specifically, for example, as shown in FIG. 8, a contour extraction program 330, a breakpoint determination program 331, and a start A point determination program 332, end point determination program 333, division point determination program 334, feature point final determination program 335, momentum calculation program 336, input data determination program 337, evaluation value calculation program 338, weighting program 339, and the like are stored. .

The contour extraction program 330 is a program for extracting a heart contour from a plurality of heart images obtained continuously in time series, and the CPU 31 executes the contour extraction program 330 as a contour extraction unit. Function.
The breakpoint determination program 331 is a program for determining two breakpoints that are boundaries between the blood vessel portion and the left ventricle portion from the extracted heart contour, and the CPU 31 executes the breakpoint determination program 331 by executing the breakpoint determination program 331. Functions as a breakpoint determination means.

The start point determination program 332 is a program for cutting out the divided blood vessel portion and determining the midpoint between the break points as the start point of the left ventricular axis. The CPU 31 executes the start point determination program 332 Serves as a starting point determination means.
The end point determination program 333 obtains the center of gravity of the left ventricular region, connects the start point and the center of gravity of the left ventricular axis, and determines the intersection of the extension line and the left ventricular contour as the end point of the left ventricular axis. The CPU 31 functions as an end point determination unit by executing the end point determination program 333.

The division point determination program 334 is a program for equally dividing the start point and the end point of the left ventricular axis to determine a division point. The CPU 31 executes the division point determination program 334. Functions as a dividing point determining means.
The feature point final determination program 335 is a program for drawing a perpendicular to the left ventricular axis from the determined division point, and determining the intersection of the perpendicular and the left ventricular contour as a feature point. By executing the point final determination program 335, it functions as a feature point final determination means.
As described above, the breakpoint determination program 331, the start point determination program 332, the end point determination program 333, the division point determination program 334, and the feature point final determination program 335 are feature point determination means for determining a predetermined feature point from the heart contour. Act as part of

The momentum calculation program 336 is a program for calculating the momentum of the feature point in association with the feature point in the heart image, and the CPU 31 functions as the momentum calculation means by executing the momentum calculation program 336.
The input data determination program 337 is a program for obtaining a contraction rate based on the momentum of the feature points in the maximum diastole and maximum systole of the left ventricle and determining the contraction rate as input data for fuzzy inference. Functions as input data determination means by executing the input data determination program 337.

The evaluation value calculation program 338 is a program for calculating an evaluation value for heart wall motion by applying a preset membership function to input data, and the CPU 31 executes the evaluation value calculation program 338. Functions as an evaluation value calculation means.
The weighting program 339 is a program for performing weighting according to the importance of the feature points, and the CPU 31 functions as a weighting unit by executing the weighting program 339.
Thus, the input data determination program 337, the evaluation value calculation program 338, and the weighting program 339 function as part of a determination unit that determines whether or not the heart wall motion is abnormal.

The input unit 4 includes, for example, a mouse, a keyboard, or a scanner, and can input data and is used for input processing.
The display unit 5 is composed of, for example, a CRT or LCD, and displays various display data processed by the CPU 31, heart images, and the like.

Next, an example of a heart wall motion evaluation process performed using the heart wall motion evaluation apparatus 1 having the above configuration will be described with reference to a flowchart shown in FIG.
As a premise of this heart wall motion processing operation, a membership function used for fuzzy inference needs to be set in advance.

First, the CPU 31 determines whether or not an evaluation start signal for cardiac wall motion has been output (step S1). If it is determined that an evaluation start signal has been output (step S1; Yes), a plurality of heart images that are sequentially obtained in time series to be processed are acquired (step S2).
Next, a heart contour is extracted from the heart image (step S3), and feature points of the extracted heart contour are determined (step S4). Specifically, the CPU 31 determines a dividing point that is a boundary between the blood vessel portion and the left ventricular portion from the extracted heart contour, cuts off the divided blood vessel portion, and determines a start point and an end point of the left ventricular axis. A division point is determined based on these points, a perpendicular line is drawn from the determined division point with respect to the left ventricular axis, and an intersection point between the perpendicular line and the left ventricular contour is determined as a feature point.
Next, the CPU 31 calculates the amount of motion of the determined feature point (step S5), and determines whether or not the heart wall motion is abnormal using fuzzy reasoning (step S6). Specifically, an inference result is obtained by obtaining a contraction rate based on the momentum of the feature point, applying a membership function to the contraction rate, and calculating an evaluation value. If the CPU 31 determines that the heart wall motion is abnormal (step S6; Yes), the CPU 31 displays a notification of “abnormal” on the display unit 5 (step S7), and ends this processing. To do. On the other hand, if the CPU 31 determines that the heart wall motion is abnormal (step S6; No), the CPU 31 displays a notification indicating “normal” on the display unit 5 (step S8), and ends this process. To do.

According to the heart wall motion evaluation apparatus 1 according to the present invention described above, a heart contour is extracted from a plurality of heart images obtained continuously in time series, a predetermined feature point is determined from the heart contour, The amount of motion of the feature point is calculated by associating the feature points in a plurality of heart images, and based on the calculated amount of motion of the feature point, it is determined whether or not the heart wall motion is abnormal by fuzzy inference. Therefore, it can be automatically evaluated whether or not the heart wall motion is abnormal.
In addition, since the heart image is an X-ray CT image, the burden on the diagnosis of the subject is reduced, and since it has high spatial resolution or contrast resolution, more accurate evaluation can be performed.

  In addition, the blood vessel portion is cut off from the extracted heart contour at the dividing point, the midpoint between the dividing points is set as the starting point of the left ventricular axis, the starting point of the left ventricular axis is connected to the center of gravity, the extension line and the left ventricular contour The left ventricular axis end point is taken as the end point of the left ventricular axis, and the dividing point is determined by equally dividing the start and end points of the left ventricular axis, and a perpendicular is drawn from the determined dividing point to the left ventricular axis. Since the intersection of the perpendicular and the left ventricle outline is determined as a feature point, the feature point can be determined with high accuracy without being affected by the contraction or expansion of the left ventricle.

Further, since the momentum of the feature point is calculated based on the movement distance between the start point of the left ventricular axis and each feature point, the movement of the feature point according to the momentum of the heart wall can be grasped more easily.
In addition, the contraction rate calculated based on the momentum of the feature points is used as input data, and the inference result of fuzzy inference is calculated based on the input data. Therefore, it approximates the judgment based on advanced empirical knowledge possessed by doctors. The decision made can be made automatically.
In addition, since weighting is performed according to the importance of the feature points, a part where an abnormal symptom is likely to appear is preferentially used for the heart wall motion evaluation. Therefore, it is possible to reduce the misidentification rate of a normal case as an abnormal case without reducing the identification rate of the abnormal case, and to further improve the accuracy of the determination result.

The heart wall motion evaluation apparatus according to the present invention is not limited to the above configuration. For example, the feature point extraction is not limited to the method described in the above embodiment, and may be arbitrarily determined.
In addition, although an X-ray CT image is used as a heart image, an ultrasonic image, a magnetic resonance image (MRI), or the like may be used.

  As described above, in the present invention, fuzzy reasoning is used for the evaluation of the heart wall motion, but in addition to this, a neural network can also be used. For example, it is possible to obtain the same evaluation as the evaluation result by the heart wall motion evaluation apparatus 1 according to the present invention by using a hierarchical structure type neural network based on the momentum of the feature points.

  Next, an embodiment of the heart wall motion evaluation method according to the present invention will be described.

In Example 1, the subject's heart was photographed using a digital fluorography apparatus DFP-2000A (manufactured by TOSHIBA). Photographing was performed under the condition of 60 frame / sec from one direction of 30 ° to the subject's body.
Further, as the heart images, a total of 33 types of images of 11 cases (8 normal cases and 3 abnormal cases) in the upper, lower, and distal ends of the left ventricle were used. These heart images are 12 slice images per case.
Abnormal cases include cases in which cardiac wall motion abnormalities extend to the entire upper, lower, and distal regions of the left ventricle, cases that extend only to the upper and distal ends, and cases that extend only to the lower portion. Using.
The feature points of the heart contour used in the evaluation were set to 7 points, and the heart wall motion evaluation was performed without weighting.
For fuzzy reasoning, 14 types of membership functions were extracted and used to calculate evaluation values. An example is shown in Equation 4. Equation 4 (a) represents a membership function based on a normal rule, and Equation 4 (b) represents a membership function based on an abnormal rule.

  Table 1 shows the evaluation results of heart wall motion using fuzzy reasoning. The region name “normal case → normal case” indicates that the case is determined to be a normal case as a result of evaluation based on the heart image of the normal case.

  As shown in Table 1, since all abnormal cases could be identified and abnormal cases were not distinguished from normal cases, it can be seen that the evaluation method according to the present invention is effective. Note that there are about 11 cases in which normal cases are mistakenly recognized as abnormal cases. This is because the membership function identification conditions are set strictly.

In Example 2, using the same heart image as in Example 1, the feature points were set to 7 points, and weighting was added to evaluate the heart wall motion.
In fuzzy reasoning, the membership function shown in Equation 5 was applied to the shrinkage rate based on the momentum of non-important feature points. Equation 5 (a) represents a membership function based on the normal rule, and Equation 5 (b) represents a membership function based on the abnormal rule.

  Table 2 shows the evaluation results of heart wall motion using fuzzy reasoning.

  As shown in Table 2, it can be seen that weighting improves the discrimination rate in the lower left ventricle, and the other discrimination rates are the same as in the first embodiment. Therefore, it can be said that the weighting according to the present invention is an effective technique.

It is a schematic diagram for demonstrating the feature point determination of a heart outline. It is a schematic diagram for demonstrating the feature point determination of a heart outline. It is a schematic diagram for demonstrating calculation of the momentum of a feature point. It is a figure for demonstrating the normal case and abnormal case of heart wall motion. It is a figure for demonstrating the application result of a membership function. It is a figure for demonstrating extraction of the optimal membership function used for evaluation, (a) is a figure which shows the function based on a normal rule, (b) is a function based on an abnormal rule. It is a schematic diagram for demonstrating the weighting in heart wall motion evaluation. It is a block diagram which shows the structure of the heart wall motion evaluation apparatus concerning this invention. It is a flowchart which shows an example of the evaluation process operation | movement using the heart wall motion evaluation apparatus of FIG.

Explanation of symbols

1 Cardiac wall motion evaluation apparatus 31 CPU (contour extraction means, feature point determination means, division point determination means, start point determination means, end point determination means, division point determination means, feature point final determination means, momentum calculation means, determination means , Input data determination means, evaluation value calculation means, weighting means)
330 contour extraction program (contour extraction means)
331 Breakpoint determination program (breakpoint determination means, feature point determination means)
332 Start point determination program (start point determination means, feature point determination means)
333 End point determination program (end point determination means, feature point determination means)
334 Dividing point determining program (dividing point determining means, feature point determining means)
335 feature point final determination program (feature point final determination means, feature point determination means)
336 Exercise amount calculation program (exercise amount calculation means)
337 Input data determination program (input data determination means, determination means)
338 Evaluation value calculation program (evaluation value calculation means, determination means)
339 Weighting program (weighting means, judgment means)

Claims (13)

A first step of extracting a heart contour from a plurality of heart images obtained continuously in time series;
A second step of determining a predetermined feature point from the extracted heart contour;
A third step of calculating the momentum of the feature points by associating the feature points in the plurality of heart images;
A fourth step of determining whether the heart wall motion is abnormal using fuzzy inference based on the calculated momentum of the feature point;
A method for evaluating heart wall motion, comprising:   The method for evaluating heart wall motion according to claim 1, wherein the heart image is an X-ray CT image. The second step includes
Determining two break points that are the boundary between the blood vessel portion and the left ventricular portion from the extracted heart contour;
Cutting the segmented blood vessel part and setting the midpoint between the segmentation points as the starting point of the left ventricular axis;
Finding the center of gravity for the left ventricular region, connecting the start point of the left ventricular axis and the center of gravity, and the intersection of the extension line and the left ventricular contour as the end point of the left ventricular axis;
A step of equally dividing a start point of the left ventricular axis and an end point of the left ventricular axis to determine a division point;
Drawing a perpendicular to the left ventricular axis from the determined dividing point, and determining an intersection of the perpendicular and the left ventricular contour as a feature point;
The method for evaluating heart wall motion according to claim 1, comprising: The third step includes
The heart wall motion evaluation method according to any one of claims 1 to 3, wherein a momentum of the feature point is calculated based on a movement distance between the start point of the left ventricular axis and each feature point. . The fourth step includes
Obtaining a contraction rate based on the momentum of the feature point in the maximum diastole and maximum systole of the left ventricle, and using the contraction rate as input data for fuzzy inference;
Applying a predetermined membership function set in advance to the input data, and calculating an evaluation value for heart wall motion;
The method for evaluating heart wall motion according to any one of claims 1 to 4, further comprising: The fourth step includes
6. The heart wall motion evaluation method according to claim 5, further comprising a step of performing weighting according to importance of the feature points before calculating an evaluation value by the predetermined membership function. Contour extracting means for extracting a heart contour from a plurality of heart images obtained continuously in time series;
Feature point determining means for determining a predetermined feature point from the heart contour extracted by the contour extracting means;
Momentum calculation means for associating the feature points in the plurality of heart images determined by the feature point determination means and calculating the momentum of the feature points;
Determining means for determining whether or not the heart wall motion is abnormal using fuzzy reasoning based on the momentum calculated by the momentum calculating means;
An apparatus for evaluating heart wall motion, comprising:   The heart wall motion evaluation apparatus according to claim 7, wherein the heart image is an X-ray CT image. The feature point determining means includes
A breakpoint determination means for determining two breakpoints that are boundaries between the blood vessel portion and the left ventricle portion from the heart contour extracted by the contour extraction means;
A start point determining means that cuts off the blood vessel portion divided by the dividing point determining means, and the midpoint between the dividing points is the starting point of the left ventricular axis;
Finding the center of gravity with respect to the left ventricular region, connecting the start point and the center of gravity of the left ventricular axis, and an end point determining means having the intersection of the extension line and the left ventricular contour as the end point of the left ventricular axis;
Split point determination for equally dividing the start point of the left ventricular axis determined by the start point determining means and the end point of the left ventricular axis determined by the end point determining means to determine a split point Means,
A feature point final determining means for drawing a perpendicular to the left ventricular axis from the dividing point determined by the dividing point determining means and determining an intersection of the perpendicular and the left ventricular contour as a feature point;
The heart wall motion evaluation apparatus according to claim 7 or 8, further comprising: The momentum calculating means includes
The heart wall motion evaluation apparatus according to any one of claims 7 to 9, wherein a momentum of the feature point is calculated based on a moving distance between the start point of the left ventricular axis and each feature point. The determination means includes
An input data determining means for obtaining a contraction rate based on the momentum of the feature point in the maximum diastole and maximum systole of the left ventricle, and using the contraction rate as input data for fuzzy inference;
Applying a predetermined membership function set in advance to the input data, evaluation value calculation means for calculating an evaluation value for heart wall motion,
The heart wall motion evaluation apparatus according to claim 7, further comprising: The determination means includes
12. The heart wall motion evaluation apparatus according to claim 11, further comprising weighting means for performing weighting according to importance of the feature points before calculating an evaluation value by the predetermined membership function. On the computer,
A first function of extracting a heart contour from a plurality of heart images obtained continuously in time series;
A second function for determining a predetermined feature point from the extracted heart contour;
A third function for calculating the momentum of the feature points by associating the feature points in the plurality of heart images;
A fourth function for determining whether the heart wall motion is abnormal by fuzzy inference based on the calculated momentum of the feature point;
A program characterized by realizing.

Images