JPH04236952A - Ultrasonic image analyzing device - Google Patents

Abstract

PURPOSE:To perform an highly effective and precise image analysis possible to detect an image characteristic amount. CONSTITUTION:This image analyzing device is mainly formed of an ultrasonic observing device 1 having an ultrasonic probe conducting radial scanning, a processing unit 3 for conducting image processing, a keyboard 9 and track ball 10 for inputting a designated value related to a concerned area or parameter, and monitors 2, 15. In the image processing by this device, the concerned area or parameter can be changed in conformation with the resolution of the probe, S/N ratio and STC, so that a highly effective and precise analysis can be performed in image analyzing processing of texture analysis.

Description

[Detailed description of the invention]

0001

[Industrial Application Field] The present invention relates to an ultrasonic image analysis device,
Specifically, the present invention relates to an ultrasound image analysis device that performs image analysis based on an image of an observation site obtained by radial scanning with an ultrasound probe.

0002

BACKGROUND OF THE INVENTION In recent years, attempts have been made to apply texture analysis to quantitative diagnosis of images by ultrasound endoscopes using probes that are ultrasound transducers. For example, see the document “Quantitative Diagnosis Using Texture Analysis of Endoscopic Ultrasound Images (Part 1)”.
(Gastroenterol. Endosc)
．． 32:1363-1368, 1990) attempted quantitative diagnosis by importing tomographic images of lesions into a computer using an ultrasound probe at the tip of an endoscope, and performing texture analysis based on the data. This is what was reported.

[0003] The texture analysis applied in the above report is
For example, a 9x9 pixel region of interest is created on the input image plane Ge (640x512 pixels) of the observation site.
Interest) Re is set (see FIG. 8), and a run length matrix is determined from the shading of each pixel P in each region of interest.
The properties of the observed region are determined by applying various parameters obtained from the run-length matrix as feature quantities used for diagnosis.

[0004] The above texture analysis method is described in the document "Basics of Image Recognition [II]" (Ohmsha Publishing, pp. 195-20).
As described in 0), it is well known,
There are methods using concentration co-occurrence matrix, concentration level difference method, concentration level run length method, power spectrum method, etc.

The above method of texture analysis using the concentration co-occurrence matrix basically uses a two-dimensional combined probability density function f(i, j
Id, θ), and (i, jId
, θ) is the distance d in the θ direction from the pixel with the density value i.
This is a probability density function that indicates the possibility that pixels that are separated by a certain distance have a density value j. That is, a matrix representing f(i, jId, θ) for each (d, θ) is a concentration co-occurrence matrix, and i and j indicate the row and column positions, respectively. Usually, the following parameters are used as effective features.

(1) Energy

[
Formula 1]

[0007]

(2) Entropy
)

[Formula 2]

[0009]

(3) Correlation
)

[Formula 3]

[0011]

(4) Local uniformity (local ho
mogeneity)

[Formula 4]

[0013]

(5) Inertia

[Formula 5
]

[0015]

Here, Sθ(i,jId) is Sθ(d)
It is the i-th row and j-column element of the matrix, and NG is the number of density levels of the image. Further, the average density Vx, Vy and the fractions σx, σy are given by the following equation 6.

[0017]

[Formula 6]

[0018]

[0019] Furthermore, the above concentration level run length method is as follows:
For example, this method is effectively used for objects for which run-length coding is effective, such as striped patterns. A density label run is a collection of linearly adjacent pixels with the same density value. is the number of pixels included in the density level run. Then, calculate the number of runs of density value i and length j in the θ direction on the target image, and express it in the form of a matrix for each θ, which is called the density level run matrix R(θ ), then r(i, jIθ
) as its matrix element, R(θ)=[r(i,jIθ)]. Then, as before, the following feature amounts are defined as parameters.

(6) Short run emphasis (short run emphasis)
n emphasis)

[Formula 7]

[0021]

(7) Long run emphasis
(emphasis)

[Formula 8]

[0023]

(8) Concentration level distribution (gray l
distribution)

[Formula 9]

[0025]

(9) Run length distribution (run l
length distribution)

[Formula 10]

[0027]

(10) Run percentage (run
percentage)

[Formula 11]

[0029]

[0030] Here, NG is the number of concentration levels,
NR is the number of run lengths in matrix R(θ). TR is the length in the θ direction, and is the total number of runs regardless of the density value.

The above is the texture analysis method. Then, the value calculated by the above parameters is compared with the "thresholds" α and β set in the processing unit, and if the calculated value is between the "thresholds" α and β, the corresponding interest is The values of the above parameters are employed as the feature amounts of the region. Note that the region of interest is moved up, down, left and right, and the entire screen is processed.

[0032] Regarding the texture analysis shown in the above-mentioned document "Quantitative diagnosis using texture analysis of endoscopic ultrasound images (first report)", Fundamentals of Image Recognition (II), and other documents, etc. , when calculating the parameters of each region of interest R set on the target image data,
The region of interest R had a constant size and shape. Furthermore, there is no particular description of whether the position of the region of interest R is to be scanned in the horizontal and vertical directions on the image plane during the above calculation, or how to do so. Furthermore, the parameters did not particularly change depending on the position of the region of interest.

[0033]

[Problems to be Solved by the Invention] In the ultrasonic image processing apparatus using the above-mentioned probe, for example, one using a mechanical scanning type ultrasonic probe 22 as shown in FIG. is rotated around the center point O, and tomographic image data in each ultrasound propagation direction Z accompanying the rotation is captured. Note that the probe 22 is connected to the operating wire 2.
It is fixed to the tip of 1. In addition, in the electronic radial scanning method, transducers are arranged in an annular shape at the tip, and the operating transducers are electrically switched, allowing observation of the propagation direction Z of each ultrasonic wave as in the mechanical radical method. You can import image data that shows the condition of the body part.

Therefore, as the observation site moves further away from the probe 22 to the periphery, the detection width widens and the resolution decreases. Furthermore, as shown in FIG. 11, the beam diameter B of the probe 22 in the ultrasonic propagation direction Z widens at a portion farther from the position of the probe 22, and from this point of view, the resolution is poorer at the outer periphery. It can be said that it will be. Also, regarding the output |P(f)| of the spatial frequency characteristic of the probe 22, when the detection position is close, the reference frequency fo
However, when the detection position is far away, the attenuation in the high frequency range is large, as shown in the characteristics in Figure 13, and from this point of view, the resolution is low in the peripheral area. can be said to decrease.

However, as described above, in the conventional example, when calculating the above-mentioned parameters, the size and shape of the region of interest R are kept constant, and moreover, the size and shape of the region of interest R are fixed in the horizontal and vertical directions on the image plane. Because it scans, the same processing is applied to image data with different resolutions,
This is a problem in terms of time and accuracy as it wastes calculations. Furthermore, as mentioned above, since the resolution is high near the probe 22, the probe 22 is usually placed near the region to be observed. However, in the conventional example, the same processing is performed for both distance and near, which wastes time, is inefficient, and is not preferable in terms of accuracy.

[0036] Furthermore, when determining the tissue structure of the observed region, parameters such as the calculated values of short run emphasis, long run emphasis, etc. that are between predetermined thresholds are taken in as feature quantities, and each Tissue structure is determined by a combination of parameters. However, in the case of conventional examples,
The calculation threshold value and combination of the parameters, the contribution rate to the discrimination, the calculation formula, etc. are not changed depending on the distance between the probe 22 and the observation site. Therefore, it has been difficult to perform appropriate discrimination processing on the entire range of image data having different resolutions.

[0037] An object of the present invention was to solve the above-mentioned problems, and it is possible to adjust image analysis parameters and/or
By changing the size of the region of interest, efficient
Moreover, it is an object of the present invention to provide an ultrasonic image analysis device that can perform image analysis that allows detection of image feature amounts with high precision.

[0038]

[Means for Solving the Problems] The ultrasonic image analysis device of the present invention provides images based on images obtained by radial scanning by an ultrasonic probe, corresponding to the propagation direction of ultrasonic waves of the probe. It is characterized in that image analysis is performed by changing the analysis parameters and/or the size of the region of interest.

[0039]

[Operation] Image analysis is performed with high precision and in a short time by changing the image analysis parameters and/or the size of the region of interest in accordance with the propagation direction of the ultrasound waves.

[0040]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be explained below based on embodiments shown in the drawings. FIG. 1 shows a block diagram of an ultrasonic image analysis apparatus according to a first embodiment of the present invention. The above analysis device includes an ultrasound observation device 1, a monitor 2 of the observation device, a processing unit 3, a keyboard 9 and a trackball 10 for inputting specified values regarding the region of interest and parameters, and a monitor 15 of the analysis device. It is configured.

The ultrasonic observation device 1 includes, for example, an ultrasonic probe 22 that is inserted into a living body lumen and performs radial scanning as shown in FIG. The processing unit 3 has the operation of each component controlled by a control circuit 7, takes in the output of the observation device, performs texture analysis based on the data, and displays the analysis results on a color monitor 2.
This is what is output to.

That is, the observation output of the observation device is A/D converted by the A/D converter 4, and the output is input to the frame memory 5 and stored as image data. on the other hand,
The feature calculation circuit 6 calculates the control circuit 7 from the image data.
Each pixel data of the area corresponding to the region of interest specified by is extracted, and the value of each parameter, i.e., the feature amount, is extracted based on the arithmetic formula for short run emphasis, long run emphasis, etc., which is similarly selected by the control circuit 7. Calculate.

Subsequently, the determination circuit 8 performs a determination process on the feature amount based on "threshold values" α and β that are separately designated. Based on the results, the properties and tissue structure of the observed site are determined, and the name of the lesion is diagnosed. The result is output to the display control circuit 11. The image analysis results are displayed by superimposing color display data of the discrimination results on the image output of the frame memory 5 in the display control circuit 11, and converting the output into a D/A converter 12. The mixed amplification circuit 14 mixes the signal with the signal from the video synchronization signal generation circuit 13 and displays it on the color monitor 15 for displaying the image analysis results.

FIG. 2 shows the image plane Ga in the data processing state of the image analysis apparatus of the above embodiment. It is assumed that the probe 22 rotates around the center point O on the image plane of this figure. The region of interest in the texture analysis process of this device is on a ring Ca with a substantially constant width Ta in the radial direction, which is the ultrasonic propagation direction Z, centered at the position O, and divided by a substantially constant width in the circumferential direction. A roughly rectangular area is applied. The size of the region of interest (Ra1, Ra2...) is set to, for example, 11 pixels x 11 pixels (see FIG. 3).

In the calculation of parameter values, the starting position of the calculation of the target region of interest position is set to the center of the screen, and the sequential parameter calculations are performed by moving (scanning) the region of interest in the radial direction of the same radius. I will do it. The amount of movement is approximately 1
For pixels. For example, in the case of FIG. 2, it is moved from area Ra1 to area Ra1'. After one round of movement, the radius is increased by one pitch (Ta) and the value of the parameter of the corresponding region of interest is calculated. The parameters remain unchanged among the regions of interest at the same radius, that is, Ra1, Ra2, and Ra3, but the region of interest Ra4 has a different radius.
and Ra5, etc., values are calculated using parameters that take into account the resolution of observation data.

In the above case, the radius is increased by 1 pitch (Ta) after the area has moved around once on the same radius, but this is done by shifting the radial position by 1 pixel to more accurately adjust the parameter value. It is also possible to search for Furthermore, it is not always necessary to calculate parameters in the peripheral area.

[0047] Means for changing the above-mentioned parameters include, for example, changing the "threshold value" for determining the characteristic amount, changing the combination rate of parameters for determining the properties of the observed region, or , the parameter calculation formula itself may be changed.

As described above, in this embodiment,
Calculations are performed starting from the center, which has good resolution, and the outer periphery, which is less important for diagnosis, is processed as quickly as possible to improve the frame rate.In addition, by changing parameters depending on the distance from the center, resolution can be improved. S/N ratio, STC
(Sensitivity Time Control
) can be reduced. FIG. 4 is a diagram showing the processed image plane Gb and the regions of interest Rb1, Rb2, and Rb3 in an ultrasound image analysis apparatus according to the second embodiment of the present invention. In this embodiment, the shape of the region of interest is an area partitioned by a circular ring Cb having a constant radial width Tb and a unit sector angle θb corresponding to the radial scanning of the probe. The center O of the ring Cb is the position of the probe.

FIG. 5 shows an enlarged view of the region of interest, and assumes that the region of interest Rb formed by a plurality of pixels P has a fan shape. It is assumed that scanning for calculating parameter values of a region of interest during texture analysis is performed by first scanning the same ring from the inner circumference, and then scanning toward the outer circumference. Note that the configuration of this device is the same as that of the first embodiment.

In this embodiment, since the region of interest is a fan-shaped unit, the change in resolution with respect to the ultrasonic propagation direction Z of the probe 22 is more easily controlled than in the first embodiment. Compatible. That is, the region of interest is made smaller on the inner circumferential side where the resolution is better, and the region of interest is made larger on the peripheral side where the resolution is worse. Therefore, it is possible to further improve the accuracy of analysis and shorten the time required for observation. Note that the region of interest may be scanned from the outer circumferential side.

FIG. 6 shows a processed image plane Gc and each region of interest Rc in an ultrasound image analysis apparatus showing a third embodiment of the present invention.
1, Rc2, Rc3, and Rc4. In this embodiment, the shape of the region of interest is a circular ring C centered around the rotation center O of the probe 22, as in the first embodiment.
Let it be a rectangle along c. However, the radial width Tc of the ring Cc is narrower toward the inner circumference, and the width of the region of interest in the circumferential direction is similarly narrower toward the inner circumference. Therefore, the area of the region of interest becomes larger toward the periphery.

[0052]Then, on the inner circumferential side of the region of interest where the rectangle is smaller, the amount of movement of the region of interest is reduced to improve the accuracy of analysis. On the other hand, in the peripheral area, the amount of movement for scanning is increased, the detection accuracy is lowered in accordance with the resolution, etc., and it is possible to perform image analysis with less decrease in frame rate.

FIG. 7 shows the processing image plane Gd and each region of interest Rd1 in the analysis apparatus according to a modification of the third embodiment.
It is a figure showing Rd2. In this embodiment, the shape of the region of interest is similar to that of the third embodiment, with the probe 22
Let it be a quadrilateral along a circular ring Cd centered on the rotation center O. However, the radial width Td of the ring Cd is narrower toward the periphery, and the width in the circumferential direction is similarly narrower toward the periphery. Therefore, the area of the region of interest becomes smaller toward the periphery.

[0054]Then, on the larger inner periphery side of the rectangular region of interest, the amount of scanning movement of the region of interest is reduced (for example, from region Rd1 to Rd1') to improve high-precision analysis over a wide range. On the other hand, in the peripheral area, the amount of movement is increased,
This enables efficient analysis by lowering the detection accuracy in accordance with the resolution. In the embodiments other than the first embodiment, the shape of the region of interest is changed in accordance with the ultrasound propagation direction, but parameters may also be changed at the same time.

[0055]

As described above, the ultrasound image analysis apparatus of the present invention changes the parameters for image analysis and/or the size of the region of interest depending on the position of the observation site. According to the present invention, it is possible to perform image analysis corresponding to resolution, S/N ratio, STC, etc. by radial scanning of a probe, and it is possible to perform image analysis with higher precision. Moreover, it is possible to provide an ultrasonic image analysis device that has remarkable effects such as improved frame rate.

[Brief explanation of the drawing]

FIG. 1 is a block configuration diagram of an ultrasound image analysis apparatus showing a first embodiment of the present invention.

FIG. 2 is a diagram showing an image analysis processing screen by the ultrasound image analysis apparatus shown in FIG. 1;

FIG. 3 is a diagram showing a region of interest during image analysis processing by the ultrasound image analysis apparatus of FIG. 1;

FIG. 4 is a diagram showing an image analysis processing screen by an ultrasound image analysis apparatus showing a second embodiment of the present invention.

FIG. 5 is a diagram showing a region of interest during image analysis processing by the ultrasound image analysis apparatus shown in FIG. 4;

FIG. 6 is a diagram showing an image analysis processing screen by an ultrasound image analysis apparatus showing a third embodiment of the present invention.

7 is a diagram showing an image analysis processing screen by an ultrasound image analysis apparatus showing a modification of the third embodiment shown in FIG. 6; FIG.

FIG. 8 is a diagram showing an image analysis processing screen by a conventional ultrasound image analysis device.

9 is a diagram showing a region of interest during image analysis processing by the ultrasound image analysis apparatus of FIG. 8; FIG.

FIG. 10 is a diagram showing a radial scanning state of the ultrasound probe in the conventional ultrasound image analysis apparatus shown in FIG. 8;

FIG. 11 is a diagram showing changes in the beam width of the ultrasound probe in the conventional ultrasound image analysis apparatus shown in FIG. 8;

FIG. 12 is a diagram showing the spatial frequency characteristics of the short-range output of the ultrasound probe in the conventional ultrasound image analysis apparatus shown in FIG. 8;

FIG. 13 is a diagram showing the spatial frequency characteristics of the long-distance output of the ultrasound probe in the conventional ultrasound image analysis apparatus shown in FIG. 8;

[Explanation of symbols]

Claims (1)

[Claims] Claim 1: Based on an image obtained by radial scanning with an ultrasound probe, image analysis parameters and/or the size of a region of interest are determined in accordance with the propagation direction of the ultrasound of the probe. An ultrasonic image analysis device characterized by performing image analysis by changing the image.