JP5756812B2 - Ultrasonic moving image processing method, apparatus, and program - Google Patents

Description

  The present invention relates to a technique for obtaining an observation image by irradiating a subject with ultrasonic waves.

  Conventionally, in medical diagnosis using an image, a technique for obtaining an observation image by irradiating a subject with ultrasonic waves has been used. As a technique for clarifying an observation image, there are those described in Patent Documents 1 and 2 below.

  In the following Patent Document 1, in an ultrasonic image processing apparatus used for medical image diagnosis, an elastic modulus distribution of a tissue is estimated based on a change amount of a small area of a diagnostic moving image, and hardness is converted into a color map for display. A technique is disclosed. In this method, when attention is paid to the tissue boundary in order to perform the elastic modulus processing, the sharpness of the image may be deteriorated.

  Patent Document 2 below discloses a technique for creating a scalar distribution image based on a motion vector of a diagnostic moving image to improve the degree of tissue boundary discrimination in order to solve the above-described problems. In addition, a technique for extracting a boundary between a tumor and a normal tissue by applying eigenvalue decomposition to a motion vector field is also disclosed.

JP 2004-135929 A JP 2009-246734 A

  Consider using a medical image to diagnose the degree of progression of a degenerated site such as a lesion (for example, a tumor). In general, characteristics such as the degree of infiltration at the boundary between a tumor part and a healthy part differ depending on a living body part and a case. Therefore, the image processing method suitable for extracting the boundary of the tumor differs depending on whether the objective is to observe the invasiveness of the tumor or to determine the overall shape of the tumor. In the prior art, since these differences are not taken into consideration, it is difficult to use an image processing method suitable for the purpose of diagnosis.

  The present invention has been made to solve the above-described problems, and an object thereof is to provide an ultrasonic moving image processing technique suitable for a body part or a case.

  In the ultrasonic moving image processing method according to the present invention, the luminance value of the pixel of the scalar distribution image is determined using the motion vector distribution image of the observation image. At that time, a process is performed to alleviate the degree to which the luminance change over the boundary portion in the scalar distribution image depends on the direction in the scalar distribution image.

  According to the ultrasonic moving image processing method of the present invention, the luminance value of the boundary portion of the observation image becomes a close value regardless of the direction of the boundary line. As a result, the boundary portion can be stably identified, so that it is possible to provide an ultrasonic moving image processing method suitable for grasping the entire shape of a lesion such as a tumor.

1 is a functional block diagram of an ultrasonic moving image processing apparatus 100 according to Embodiment 1. FIG. It is a figure which shows the operation | movement flow of the process part and the synthetic | combination part. It is a model figure which shows a mode that the shearing stress rightward is applied to the boundary of the upper end part of a uniform structure | tissue structure. It is a figure which shows the difference of the displacement amount of an up-down depth direction. It is a figure which compares the change of a scalar value when changing the value of H from 0 to 1 using the vector field model of 3x3 pixels of FIG. It is a figure which shows the example from which the direction of the motion vector field has shifted | deviated to 2 directions in a 6x6 motion vector field. An example is shown in which the same motion vector field as in FIG. 6 is scalarized by the maximum eigenvalue processing after performing the covariance processing. It is a figure which shows the result of applying the method which concerns on this Embodiment 1 with respect to an actual tumor. It is a figure explaining the block matching implemented when calculating | requiring a motion vector. It is a figure which shows a displacement model when an image displaces 1 pixel. It is a figure which shows the processing flow which determines the search range in a block matching process. It is a figure which shows the result of applying the method which concerns on this Embodiment 2 with respect to an actual tumor. It is a figure which shows the process of converting a motion vector into a complex matrix. It is a figure which shows the power spectrum obtained by carrying out the Fourier analysis of the vector field model of 3x3 pixel demonstrated in FIG. 3 to the depth direction (vertical direction of drawing). It is a figure which shows the relationship between the speckle size which appears on a B mode image, and the resolution | decomposability of an ultrasonic beam.

<Embodiment 1>
In the first embodiment of the present invention, an operation example in which a method for generating an observation image is switched according to the purpose of diagnosis will be described.

  In the first embodiment, when observing with priority given to the shape of the altered portion, the luminance change before and after the boundary between the altered portion and the other portion of the observed image is viewed from any direction of the observed image. However, try to be the same level. Thus, the boundary of the altered portion is made clear from any direction, and the shape of the altered portion is easily identified on the observation image.

  Further, in the first embodiment, when observing with priority given to the infiltration degree of the altered portion, it is noted that the infiltration degree correlates with the magnitude of the tissue movement, and the magnitude of the tissue movement and the image luminance are correlated. An observation image corresponding to is generated.

  FIG. 1 is a functional block diagram of the ultrasonic moving image processing apparatus 100 according to the first embodiment. Hereinafter, each component of the ultrasonic moving image processing apparatus 100 will be described.

  An ultrasonic probe (probe) 1 in which ultrasonic elements are arranged one-dimensionally transmits an ultrasonic beam (ultrasonic pulse) to a living body and receives an echo signal (received signal) reflected from the living body. .

  The control system 4 uses the transmission beamformer 3 to output a transmission signal having a delay time matched to the transmission focus. The transmission signal is transmitted to the ultrasonic probe 1 through the transmission / reception changeover switch 5. The ultrasonic probe 1 irradiates a subject (for example, a living body to be diagnosed) with ultrasonic waves according to a transmission signal.

  The ultrasonic probe 1 receives an ultrasonic beam reflected or scattered in the living body and converts it into an electrical signal. This electrical signal is transmitted as a reception signal to the reception beam former 6 via the transmission / reception selector switch 5.

  The receive beamformer 6 is a complex beamformer that mixes two received signals that are 90 degrees out of phase, and performs dynamic focus that adjusts the delay time according to the reception timing according to the control instructions of the control system 4. The RF signal of the real part and the imaginary part is output.

  The envelope detector 7 detects the RF signal output from the receiving beam former 6, converts it to a video signal, and outputs it to the scan converter 8. The scan converter 8 converts the video signal output from the envelope detector 7 into image data (B-mode image data).

  The processing unit 10 first creates a motion vector distribution based on image data of two or more frames output from the scan converter 8. Next, the created motion vector distribution is converted into a scalar distribution. The synthesizer 12 synthesizes the original image data and the corresponding motion vector distribution or scalar distribution. The display unit 13 displays the synthesis result on the screen.

  The parameter setting unit 11 performs processing such as parameters for the processing unit 10 to perform signal processing, selection and setting of images to be displayed by the synthesis unit 12, and the like. These parameters are input by the operator (diagnostic machine operator) of the ultrasonic moving image processing apparatus 100 using the user interface 2. As a moving image display method, for example, an original image and a vector distribution image (or scalar image) are combined into one image and displayed on a display, and two or more moving images are displayed side by side.

  The control system 4, the envelope detection unit 7, the processing unit 10, the parameter setting unit 11, and the synthesis unit 12 can be configured using hardware such as a circuit device that realizes these functions, or can be configured with a microcomputer or CPU ( It is also possible to configure using a computing device such as a Central Processing Unit) and software that defines the operation thereof. Any two or more of these functional units can be configured integrally.

  The configuration of the ultrasonic moving image processing apparatus 100 has been described above. Next, the operation of the ultrasonic moving image processing apparatus 100 will be described.

  Image data is generated by irradiating the subject with ultrasonic waves and detecting ultrasonic signals for at least two frames having different detection timings from the subject. A motion vector analysis is performed on these frames to create a motion vector distribution image. Then, based on the created motion vector distribution image, the luminance value of the pixel of the scalar distribution image is determined. As a method for determining the brightness value, either a process that is robust with respect to a difference in the direction of the tissue boundary or a process that increases the brightness value in response to linear movement of the tissue is applied. Furthermore, a specific operation flow will be described.

  FIG. 2 is a diagram illustrating an operation flow of the processing unit 10 and the combining unit 12. Hereinafter, each step of FIG. 2 will be described.

(FIG. 2: Step S201)
The processing unit 10 receives B-mode moving image data from the scan converter 8.

(FIG. 2: Step S202)
The processing unit 10 extracts two frames, a desired frame and a frame at a time other than that, and calculates a motion vector field from the two frames. The motion vector field is calculated based on a block matching method as described in Patent Document 2, for example.

(FIG. 2: Step S203)
Whether the operator generates an image preferentially clarifying the degree of invasion of a lesion (for example, a tumor) or an image preferentially clarifying the shape (size) via the user interface 2. Select according to the part or case to be diagnosed. For example, when it is desired to diagnose the degree of progression of the tumor based on the degree of infiltration, the invasion degree priority method is selected, and when it is desired to accurately grasp the size of the aneurysm, the shape priority method is selected. If the infiltration degree priority method is selected, the process proceeds to step S204. If the shape priority method is selected, the process proceeds to step S205.

(FIG. 2: Step S204)
The processing unit 10 performs a space vector differentiation process on the motion vector field obtained in step S202. The space vector differentiation process can be expressed by the following (formula 1). The processing unit 10 sets a predetermined small area for the motion vector field after the differentiation process (for example, 3 × 3 pixels), and converts the vector field in the small area into a complex matrix. The calculation for obtaining the differential vector can be easily performed by using a prescribed primary differential filter or the like.

(FIG. 2: Step S205)
The processing unit 10 performs a covariance process on the motion vector field obtained in step S202. The co-dispersion process can be expressed by the following (formula 2).

(FIG. 2: Step S206)
The processing unit 10 performs eigenvalue expansion processing on the complex number matrix generated in step S204 or step S205 to obtain an eigenvalue (scalar value) having the largest absolute value. Instead of the maximum eigenvalue, the absolute value of the eigenvalue having the second largest value or the sum of the absolute values of the eigenvalues can be used.

(FIG. 2: Step S207)
The processing unit 10 converts the result of step S206 to a scalar by assigning it to the image brightness at the center vector position. The processing unit 10 performs this scalarization process over the entire motion vector field. Thereby, a complex matrix can be imaged. The composition unit 12 displays the image on the screen.

(FIG. 2: Steps S206 to S207: Supplement)
In the first embodiment, the method for obtaining a scalarized image using eigenvalues of a complex matrix has been described. However, the method for imaging a motion vector distribution image is not limited to this.

  The operation of the ultrasonic moving image processing apparatus 100 has been described above. Next, the reason why the vector differentiation process is suitable for the infiltration degree priority method will be described.

<Embodiment 1: Evaluation of vector differentiation process>
FIG. 3 is a model diagram showing a state in which a rightward shearing stress is applied to the boundary of the upper end portion of the uniform tissue structure. If the degree of infiltration of tumor tissue or the like is low, the lower part is not displaced even if the upper part is displaced by shear stress (FIG. 3 (a)). On the other hand, when the degree of infiltration increases, the lower part is also pulled and displaced to some extent (FIGS. 3B and 3C).

FIG. 4 is a diagram showing the difference in the amount of displacement in the upper and lower depth directions. When the difference in the amount of displacement in the upper and lower depth directions shown in FIG. 4 is expressed by the displacement gradient parameter D = tan θ, the relationship between the shear stress S and the displacement gradient D is determined by introducing the infiltration parameter H and (Equation 3) Can be expressed as In FIG. 3, the case of (a) H = 0, (b) H = 0.5, (c) H = 0.9 was illustrated.

  If H is a constant value, a linear relationship is established, but generally tissue infiltration is considered to be nonlinear. Therefore, it is assumed that the value change when the value of H is scalarized when the motion vector field is imaged in step S207 of FIG. 2 is also nonlinear. However, when paying attention to a minute region, the value of H needs to be linearly scalarized. If the value of H is not linearly scalarized, the image obtained by scalarizing the motion vector field may not accurately reflect the degree of tissue infiltration. Therefore, the change in the value of H was evaluated using the 3 × 3 pixel vector field model shown in FIG.

  FIG. 5 is a diagram for comparing the change in scalar value when the value of H is changed from 0 to 1 using the 3 × 3 pixel vector field model of FIG. FIG. 5A shows the change in H in the conventional method for scalarizing the maximum eigenvalue of the vector field. FIG. 5B shows a change in H when the vector differentiation process is performed.

  In the maximum eigenvalue method shown in FIG. 5A, as the value of H increases, the value gradually deviates from linearity, and the value tends to increase nonlinearly. In the differential maximum eigenvalue method shown in FIG. 5B, linearity is established even if the value of H increases. Therefore, when paying attention to the degree of infiltration, it can be seen that it is effective to perform vector differentiation as pre-processing for scalarizing the motion vector field. The reason why the vector differentiation process exhibits the above-described effect is considered to be due to the removal of the nonlinearity of the image feature amount by the differentiation process.

  That is, as shown in FIG. 3, there is a non-linear relationship between the magnitude of movement of the altered portion of the tissue on the observed image and the degree of infiltration. However, the non-linearity is removed by performing the space vector differentiation process, The correspondence is linear. Therefore, it is considered that the degree of infiltration can be expressed by image luminance.

  The effectiveness of the vector differentiation process has been described above. Next, the reason why the covariance process is suitable for the shape priority method will be described.

<Embodiment 1: Evaluation of co-dispersion processing>
FIG. 6 is a diagram illustrating an example in which the direction of the motion vector field is shifted in two directions in the 6 × 6 motion vector field. 6A shows a case where the motion vector field is directed in the horizontal direction, FIG. 6B shows a case where the motion vector field faces in the vertical direction, and FIG. 6C shows a case where the motion vector field faces in the oblique direction. An example is shown. The left figure in FIG. 6 shows the direction in which the motion vector is directed by an arrow, and the right figure in FIG. 6 shows an image obtained by scalarizing the left figure by the maximum eigenvalue processing with a matrix size of 3 × 3.

  When imaging is performed by scalarizing the motion vector field, a large luminance difference appears at the boundary portion where the directions of the motion vectors are different. Since this luminance difference looks like a boundary line on the image, the shape and size of the diagnosis target (for example, a tumor) can be grasped by the boundary line. In the example shown in FIG. 6, since the boundary portion has low luminance, the shape to be diagnosed is grasped by the black boundary line.

  In the example shown in FIGS. 6A and 6B, the image luminance at the boundary portion in the vertical and horizontal directions is both 0.015. On the other hand, in the example shown in FIG. 6C, the image brightness of the boundary portion in the oblique direction is 0.022. That is, the image brightness differs by about 47% between the vertical and horizontal boundary portions and the diagonal boundary portions. If the image brightness is greatly different for each direction even though the same tumor boundary portion is represented, it is difficult to identify the boundary portion on the image depending on the direction, which is not preferable for image diagnosis.

  FIG. 7 shows an example in which the same motion vector field as in FIG. 6 is scalarized by the maximum eigenvalue process after the covariance process. Although the image density is reversed due to the difference in processing method, the same boundary portion as that in FIG. 6 is imaged.

  In the example shown in FIGS. 7A and 7B, the image luminance at the boundary portion in the vertical and horizontal directions is 8. On the other hand, in the example shown in FIG. 7C, the image brightness of the boundary portion in the oblique direction is 10. That is, the difference in image luminance is suppressed to 25% between the vertical and horizontal boundary portions and the diagonal boundary portions.

  According to the processing example shown in FIG. 7, by performing the covariance processing, the image luminance of the boundary portion becomes a uniform value (close value) regardless of the direction. It can be said that it is suitable for identifying the size.

  The covariance process has the effect of making the matrix symmetric by making the matrix symmetric. By this action, it is considered that the scalar image at the boundary portion does not change greatly in a specific direction before and after the boundary, and the effect of relaxing the dependence on the direction is exhibited.

  Note that a technique that exhibits an effect of uniformizing matrix elements, for example, a correlation matrixing process, can be used as in the covariance process.

<Embodiment 1: Evaluation using clinical data>
FIG. 8 is a diagram illustrating a result of applying the method according to the first embodiment to an actual tumor. FIG. 8A shows the result of performing the processing by the elastography technique described in Patent Document 1 for comparison. FIG. 8B shows an image that has been scalarized using the maximum eigenvalue after performing vector differentiation. FIG. 8C shows an image that has been scalarized using the maximum eigenvalue after covariance processing has been performed.

  Comparing FIGS. 8 (a) and 8 (b), it can be seen that in FIG. 8 (b), the degree of infiltration at the border of the tumor is represented by the thickness of the border. Comparing FIGS. 8 (a) and 8 (c), it can be seen that the boundary portion of the tumor is clearly extracted by a thin line in FIG. 8 (c).

<Embodiment 1: Summary>
As described above, the ultrasonic moving image processing apparatus 100 according to the first embodiment performs covariance processing on the motion vector field, and then converts the image into an image using a maximum eigenvalue or the like. Thereby, a boundary portion such as a tumor can be extracted with uniform brightness regardless of the direction on the image, which is suitable for determining the shape and size of the tumor on the image.

  In addition, the ultrasonic moving image processing apparatus 100 according to the first embodiment performs vector differentiation on the motion vector field, and then converts the image into an image using a maximum eigenvalue or the like. Thereby, even when tissue infiltration such as a tumor is nonlinear in the depth direction, it is possible to obtain image brightness that accurately reflects the infiltration degree, which is suitable for determining the infiltration degree on the image.

  In addition, the ultrasonic moving image processing apparatus 100 according to the first embodiment receives an instruction specifying whether to observe the shape of the tumor preferentially or to observe the invasion degree of the tumor preferentially. Covariance processing is performed, and vector differentiation processing is performed for the latter. Thereby, since an appropriate image processing method can be selected according to a living body part or a case, an effect of improving the accuracy of image diagnosis can be expected.

  In the first embodiment, only one of the covariance process and the vector differentiation process can be employed. For example, when it is desired to specialize in the process of determining the shape and size of a tumor or the like, only the covariance process can be implemented in the ultrasonic moving image processing apparatus 100. The same applies to vector differentiation processing.

<Embodiment 2>
In the first embodiment, it has been described that the block matching process is performed between two frames in order to obtain the motion vector field. In this block matching process, parameters such as the size of the search range are generally set empirically. However, in practice, an appropriate search range is different for each case. In the second embodiment of the present invention, a method for determining an appropriate search range in the block matching process will be described. The configuration of the ultrasonic moving image processing apparatus 100 is the same as that of the first embodiment.

FIG. 9 is a diagram illustrating block matching performed when obtaining a motion vector. The luminance distribution of a region of interest (ROI: Region of Interest) at a certain m-th frame is represented as P _m (i ₀ , j ₀ ). The processing unit 10 sets a search area around the vicinity including the position of P _m (i ₀ , j ₀ ) in the frame m + Δth (typically Δ = 1), and moves to the movement candidate P _{m + Δ} ( The sum of absolute differences SAD (Sum of Absolute Difference) between i, j) and P _m (i ₀ , j ₀ ) is calculated. SAD is defined by the following (formula 4).

Next, the processing unit 10 obtains the SAD value over the search area, and determines that P _{m + Δ} (i _min , j _min ) that minimizes the SAD value in the SAD distribution is the movement destination. The processing unit 10 determines a vector connecting P _m (i ₀ , j ₀ ) and P _{m + Δ} (i _min , j _min ) as a motion vector.

The processing unit 10 changes P _m (i ₀ , j ₀ ) over the entire B-mode image, and determines each motion vector according to the above processing. Thereby, a motion vector distribution over the entire B-mode image can be obtained.

  The block matching performed when obtaining the motion vector has been described above. Next, a method for determining an appropriate search range in the block matching process will be described.

  FIG. 10 is a diagram illustrating a displacement model when the image is displaced by one pixel. Here, it is assumed that a pixel having luminance m has moved a distance d. FIG. 10B shows the shape of the difference intensity distribution at that time. In FIG. 10B, the value of −m is indicated by the absence of the luminance m at the position before the pixel moves. At a position away from the distance d, a new luminance m is generated and becomes + m.

In order to obtain the displacement distance, the difference between the positive value side luminance m and the negative value side luminance -m position may be obtained. Actual clinical data requires statistical treatment because a plurality of pixels are moved. The average displacement position of the pixels in a certain region of interest can be estimated using the first moment from the reference position. The first moment from the reference position can be expressed by the following (formula 5).

p is a coefficient for normalization and is expressed by the following (formula 6).

  That is, the average displacement position with respect to the same reference position is obtained on the positive value side and the negative value side, and the displacement distance can be statistically estimated as the absolute value of the difference between them.

  FIG. 11 is a diagram illustrating a processing flow for determining a search range in the block matching processing. Hereinafter, each step of FIG. 11 will be described.

(FIG. 11: Steps S1101 to S1102)
The processing unit 10 selects two frames for performing a block matching process between frames, that is, a reference frame and a comparison frame (S1101). The processing unit 10 calculates the difference intensity distribution between the two frames selected in step S1101 (S1102).

(FIG. 11: Step S1103)
The processing unit 10 divides the difference intensity distribution obtained in step S1102 into a positive intensity distribution and a negative intensity distribution.

(FIG. 11: Step S1104)
The processing unit 10 calculates the average displacement position in the positive value difference intensity distribution obtained in step S1103 and the average displacement position in the negative value difference intensity distribution.

(FIG. 11: Steps S1105 to S1106)
The processing unit 10 calculates the absolute difference between the average displacement position in the positive difference intensity distribution and the average displacement position in the negative difference intensity distribution obtained in step S1104, and sets this as the average displacement distance (S1105). The processing unit 10 employs a value equal to or larger than the average displacement distance obtained in step S1105 as a search range for performing the block matching process (S1106).

(FIG. 11: Step S1106: Supplement)
When pattern matching is performed between two frames in which there is a motion, at least a movement source position and a movement destination position must be included in a range in which a matching pattern is searched. Therefore, the search range is set to be greater than the average displacement distance of the pixels. This is because, if a range that is equal to or greater than the average displacement distance is searched, it is considered that the movement source position and the movement destination position of most pixels are included in the search range.

  FIG. 12 is a diagram illustrating a result of applying the method according to the second embodiment to an actual tumor. Hereinafter, each diagram of FIG. 12 will be described.

  FIG. 12A shows the reference frame in step S1101. FIGS. 12B and 12C show the positive value intensity distribution and the negative value intensity distribution calculated using the reference frame and the comparison frame (not shown) shown in FIG. The region of interest (ROI) was 30 × 30 pixels, and the search range was evaluated by moving the ROI in the direction of the dotted line in FIG.

  FIG. 12D shows the average displacement distance for each ROI position. The vertical axis represents the average displacement distance (pixel). According to the figure, it can be seen that although the average displacement distance varies depending on the ROI position, it is generally within 10 pixels or less at any ROI position. In this case, it can be seen that the pattern matching search range may be 10 pixels or more. In order to cover movements more reliably, the pattern matching search range may be set to be equal to or greater than the maximum value of the vertical axis in FIG.

<Embodiment 2: Summary>
As described above, the ultrasonic moving image processing apparatus 100 according to the second embodiment obtains the average displacement distance in the ROI using the difference intensity distribution in the region of interest (ROI) between the two frames, and calculates the motion vector. The search range of block matching to be performed when obtaining is set to be equal to or greater than the average displacement distance. Thus, while ensuring a reasonable search range that can cover the movement, the search range can be reduced as much as possible, and the calculation load can be reduced.

<Embodiment 3>
In the second embodiment, it has been described that the search range for block matching performed when obtaining a motion vector is set to be equal to or greater than the average displacement distance of pixels in the ROI. Thereby, the search range beyond the average displacement distance of each ROI can be ensured.

  However, in the process of calculating the average displacement distance, the displacement distance of pixels that are extremely displaced or the displacement distance of pixels that are hardly displaced is excluded from the target. In the second embodiment, the portion where the average displacement distance is larger than 10 is uniformly excluded, but the determination criterion is not necessarily established. Therefore, in Embodiment 3 of the present invention, a method of introducing a statistical viewpoint when determining the search range and appropriately setting the search range will be described. The configuration of the ultrasonic moving image processing apparatus 100 is the same as in the first and second embodiments.

  In FIG. 12D, the average displacement distance at each ROI position is obtained. This average displacement distance takes a different value for each ROI position within a range of approximately 10 pixels or less. Assuming that this variation distribution follows a normal distribution, the average value and the standard deviation σ of the distribution shown in FIG. 12D are obtained, and the extent to which the variation range is included in the block matching search range is statistically determined. Can be determined.

  For example, when a range obtained by adding 2σ to the average value is included in the search range for block matching, 95% of the distribution shown in FIG. 12D can be covered. When the range obtained by adding 3σ to the average value is included in the search range for block matching, 99.7% of the distribution shown in FIG. 12D can be covered. That is, it is effective to calculate an average displacement distance and its standard deviation in the region of interest, and set a value obtained by adding a value obtained by multiplying the average displacement distance by the standard deviation to a desired constant as a search range.

  In the second to third embodiments, the one-dimensional direction indicated by the dotted line in FIG. 12A is evaluated. However, in order to increase the accuracy, it is desirable to evaluate the entire image in two dimensions.

<Embodiment 4>
In the second to third embodiments, the method for optimally setting the search range for block matching performed when obtaining a motion vector has been described. Thereby, it is possible to set an appropriate search range in accordance with the actual movement of the observation image and optimize the calculation load.

  On the other hand, since the motion vector is obtained over the entire area of the B-mode image, the process of obtaining the motion vector of the next pixel after obtaining the motion vector of a certain pixel is repeated. At this time, if the interval between the motion vectors is set to be small, the resolution of the observation image is increased, and if it is increased, the resolution of the observation image is decreased. However, the optimum vector interval is different depending on the body part and the case. In the fourth embodiment of the present invention, a method for optimally setting the motion vector interval will be described. The configuration of the ultrasonic moving image processing apparatus 100 is the same as in the first to third embodiments.

  FIG. 13 is a diagram illustrating a process of converting a motion vector into a complex matrix. The processing unit 10 sets a matching region (for example, 30 pixels × 30 pixels) including a plurality of pixels in the B-mode image. The processing unit 10 searches the comparison frame for an image area that matches the image in the matching area in the reference frame. However, since it is inefficient to search the entire B-mode image as a search target, the search is performed within the range where the matching region is likely to have moved. This range is referred to as a block matching search range.

  When the processing unit 10 performs block matching in the comparison frame and finds an image region that matches the matching region, the processing unit 10 determines one image based on the difference between the matching region in the reference frame and the moved matching region in the comparison frame. A motion vector is determined.

  The processing unit 10 repeatedly performs the above process over the entire B-mode image while shifting the matching region by a predetermined vector interval (minimum 1 pixel) to obtain a motion vector field of the entire B-mode image. Next, the processing unit 10 assigns the x component and y component of the motion vector to the real part and the imaginary part, respectively, to convert them into complex numbers, and sets a complex matrix having a predetermined matrix length.

  If the vector interval is set small, the image resolution when the motion vector field is imaged becomes high. However, in this case, for an object having a high degree of infiltration and a gentle change in the boundary, the change in the complex matrix is small, and therefore it becomes difficult to distinguish the boundary portion of the entire image. Therefore, when it is desired to identify the infiltration degree with priority, the vector interval may be set coarsely (coarse viewing). Thereby, since the change in a complex matrix becomes large, the boundary part of an infiltrating part can be clarified. That is, it is effective to create a high-resolution motion vector field and a low-resolution motion vector field together.

  More specifically, the processing unit 10 generates a plurality of scalar images by changing the motion vector interval in several stages. That is, at the boundary where the degree of infiltration is low and sharply changed, the vector interval is subdivided to generate a scalarized image with little difference on the image, and at the boundary where the degree of infiltration is high and changes slowly, the coarsened scalarized image Is generated. Thereby, a plurality of scalar images using appropriate motion vector intervals according to the degree of infiltration can be obtained.

  The processing for adjusting the vector interval may be instructed manually by the operator, or may be automatically set by the processing unit 10. In the case of automatic setting, the degree of infiltration is indexed by some numerical value based on information obtained from the image, and the vector interval is set according to the index.

  For example, the boundary portion between the infiltrated region and the normal region of the scalarized image is identified by the method described in the above embodiment, and the half-value width of the luminance change when scanning across the boundary portion is used as an index of the infiltration degree. can do. The half width here refers to the horizontal axis width when the luminance change is half of the peak.

<Embodiment 4: Summary>
As described above, the ultrasonic moving image processing apparatus 100 according to the fourth embodiment performs a plurality of block matching processes with different motion vector intervals, generates a scalarized image created from a high-resolution motion vector field, A scalarized image created from a resolution motion vector field is created. Thereby, since an observation image having a resolution suitable for the degree of infiltration can be obtained, an effect of improving the accuracy of image diagnosis can be expected.

<Embodiment 5>
In the fifth embodiment of the present invention, an example in which the motion vector interval is determined based on spectrum analysis will be described. The configuration of the ultrasonic moving image processing apparatus 100 is the same as in the first to fourth embodiments.

  FIG. 14 is a diagram showing a power spectrum obtained by performing Fourier analysis on the vector field model of 3 × 3 pixels described in FIG. 3 in the depth direction (vertical direction in the drawing). Since there is no change in the horizontal direction in the model shown in FIG. 3, only the vertical direction is analyzed. 14A shows the analysis results when H = 0.1, FIG. 14B shows the analysis results when H = 0.5, and FIG. 14C shows the analysis results when H = 1.0.

  According to FIG. 14, when the degree of infiltration is small, the boundary portion is clear. Therefore, when the power spectrum is analyzed in the vertical direction, the main lobe of the power spectrum spreads and the half width tends to increase. When the boundary portion is clear, the change in the complex matrix generated from the motion vector becomes large. Therefore, it is preferable to increase the image resolution by setting the vector interval small.

  On the other hand, when the degree of infiltration is large, the infiltration spreads and the boundary portion becomes unclear. Therefore, when the power spectrum is analyzed in the vertical direction, the main lobe of the power spectrum is narrowed and the half-value width tends to be reduced. In this case, the vector interval may be set coarsely as described in the fourth embodiment.

  In view of the above, it can be seen that the vector interval has an inverse relationship with the half-value width of the power spectrum obtained by Fourier analysis of the vector field model in the depth direction. That is, it can be said that the processing unit 10 should set the vector interval to the reciprocal of the half width or a constant multiple thereof. Thereby, the vector interval can be set coarsely as the degree of infiltration increases. Further, it is not necessary for the operator to manually set the vector interval.

  In the actual clinical data, the main lobe spreads as a two-dimensional distribution. Therefore, the direction in which the half width is maximum is considered to be a direction orthogonal to the boundary, and the maximum half width may be used.

  Specifically, the motion vector field is first subjected to Fourier analysis in one scanning direction to calculate a two-dimensional power spectrum, and then the half width of the two-dimensional power spectrum is obtained. Then, the scanning direction in which the half width is maximum is determined, and a constant multiple of the reciprocal of the maximum value is set as the vector interval of the motion vectors.

<Embodiment 6>
In the sixth embodiment of the present invention, a method will be described in which an operator designates a range in which a motion vector field is created in a B-mode image, and a vector interval is automatically determined based on the range. The configuration of the ultrasonic moving image processing apparatus 100 is the same as in the first to fifth embodiments.

  The range in which the processing unit 10 creates a motion vector field can be expressed by the following equation assuming a processing procedure for repeatedly obtaining a motion vector as described with reference to FIG.

Creation range = matching region + vector interval × (ROI matrix length−1)
In the sixth embodiment, the operator instructs the range in which the processing unit 10 creates a motion vector field. Then, since the creation range in the above equation is known, if other parameters are obtained, the remaining parameters can be obtained by calculation.

  FIG. 15 is a diagram showing the relationship between the speckle size appearing on the B-mode image and the resolution of the ultrasonic beam. Speckle is granular noise appearing on an image. The speckle size can be roughly predicted based on the characteristics of the subject.

In order to identify a speckle pattern, the matching area size needs to be in a range including two or more speckles that can be separated by an ultrasonic beam. That is, the matching area size is required to be equal to or larger than the sum of the speckle size and the beam half width. This correspondence is expressed by the following (formula 7).

  The first term in Equation 7 corresponds to the speckle size, and the second term corresponds to the beam half width. Since the speckle size is larger than the point spread function (abbreviated as PSF) depending on the density of the minute scatterers of the living tissue, the enlargement ratio is expressed as k times (≧ 1).

  As imaging conditions, the lower limit of the matching region can be determined by substituting the curvature radius, aperture radius, and measurement wavelength of the ultrasonic probe into the above (Equation 7). As the matching area is smaller, the calculation amount of the processing unit 10 can be saved, and the processing can be performed at high speed.

  The creation range and the matching region are determined by the above method, and the ROI matrix length is given to the processing unit 10 in advance. The processing unit 10 can obtain the vector interval by modifying the above calculation formula and using the following formula.

Vector interval = (creation range-matching region) / (ROI matrix length-1)
The specific processing procedure first specifies the observation range size of the image data, and obtains the size of the pattern matching region using the device specification parameters of the ultrasonic probe probe. Then, the size of the pattern matching region is subtracted from the observation range size, and the value is divided by a value obtained by subtracting 1 from the matrix length corresponding to the region of interest of the image data, and the value is set as a vector interval.

If the vector interval is given to the processing unit 10 in advance by the same procedure, the ROI matrix length can be obtained by calculation.

<Embodiment 7>
Embodiments 1 to 6 above can be used in any combination. For example, the method for obtaining the search range described in the second to third embodiments and the method for obtaining the vector interval described in the fourth to fifth embodiments can be used in combination.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiment. However, the present invention is not limited to the embodiment, and various modifications can be made without departing from the scope of the invention. Needless to say.

  In addition, each of the above-described configurations, functions, processing units, etc. can be realized as hardware by designing all or a part thereof, for example, with an integrated circuit, or the processor executes a program for realizing each function. By doing so, it can also be realized as software. Information such as programs and tables for realizing each function can be stored in a storage device such as a memory or a hard disk, or a storage medium such as an IC card or a DVD.

  1: ultrasonic probe, 2: user interface, 3: transmission beamformer, 4: control system, 5: transmission / reception changeover switch, 6: reception beamformer, 7: envelope detector, 8: scan converter, 10: processing unit, 11: parameter setting unit, 12: synthesis unit, 13: display unit, 100: ultrasonic moving image processing apparatus.

Claims (16)

A method for processing an ultrasonic moving image, comprising:
Image data that irradiates a subject with ultrasonic waves, stores a detection result obtained by detecting an ultrasonic signal from the subject in a storage device, and creates image data of at least two frames having different detection timings from the stored detection result Creation steps,
A motion vector distribution image creating step of creating a motion vector distribution image by performing motion vector analysis using a plurality of frames of the image data;
A luminance determining step of determining a luminance value of a pixel of the scalar distribution image using the motion vector distribution image;
Have
In the luminance determination step,
Performing a boundary clarifying step to mitigate the extent to which the luminance change across the boundary portion in the scalar distribution image depends on the direction in the scalar distribution image;
In the boundary clarifying step,
An ultrasonic moving image processing method comprising: performing a covariance step of converting the motion vector distribution image into a complex matrix for each predetermined region of interest and converting the complex matrix into a covariance matrix. In the luminance determination step,
The ultrasonic moving image processing method according to claim 1, wherein an infiltration degree clarification step is performed in which a luminance value of the scalar distribution image is increased in a portion where the movement of the tissue of the subject is large. In the infiltration degree clarification step,
The ultrasonic moving image processing method according to claim 2, wherein a space vector differentiation process is performed on the motion vector distribution image, and a vector differentiation step of converting the obtained differential vector distribution into a complex matrix is performed. . A method for processing an ultrasonic moving image, comprising:
Image data that irradiates a subject with ultrasonic waves, stores a detection result obtained by detecting an ultrasonic signal from the subject in a storage device, and creates image data of at least two frames having different detection timings from the stored detection result Creation steps,
A motion vector distribution image creating step of creating a motion vector distribution image by performing motion vector analysis using a plurality of frames of the image data;
A luminance determining step of determining a luminance value of a pixel of the scalar distribution image using the motion vector distribution image;
Have
In the luminance determination step,
Receiving an instruction designating whether to prioritize the shape of the altered site occurring in the subject or to prioritize the degree of infiltration of the altered site;
If priority is given to the shape of the altered portion, a boundary clarifying step is performed to mitigate the degree to which the luminance change over the boundary portion in the scalar distribution image depends on the direction in the scalar distribution image,
If priority is given to the degree of infiltration of the altered site, an infiltration degree clarification step is performed to increase the brightness value of the scalar distribution image in a portion where the movement of the tissue of the subject is large.
In the boundary clarifying step,
Performing a covariance step of converting the motion vector distribution image into a complex matrix for each predetermined region of interest, and converting the complex matrix into a covariance matrix;
In the infiltration degree clarification step,
A space vector differentiation process is performed on the motion vector distribution image, and a vector differentiation step of converting the obtained differential vector distribution into a complex matrix is performed.
An ultrasonic moving image processing method. The vector differentiation step includes:
For each vector in the motion vector distribution, differentiating the beam direction along the beam axis and the azimuth direction orthogonal to the beam direction to calculate a differential value, and forming a differential vector distribution;
Substituting the beam direction differential value and the azimuth direction differential value of the differential vector with a real component and an imaginary component, respectively, and converting the complex vector into a complex matrix;
5. The ultrasonic moving image processing method according to claim 3, wherein: In the luminance determination step,
Eigenvalue analysis is performed on a covariance matrix converted from the motion vector distribution image, or a complex matrix that is a differential vector,
One of the absolute values of the eigenvalues of the complex matrix, the one of the absolute values of the eigenvalues having the second largest value, or the sum of the absolute values of the eigenvalues is one scalar value. The ultrasonic moving image processing method according to claim 1, wherein the method corresponds to a luminance value of the pixel. The motion vector distribution image creating step includes
Calculating a difference intensity distribution between the image data of the two frames;
Dividing the inside of the region of interest set in the differential intensity distribution into a positive intensity distribution and a negative intensity distribution;
Calculating a first moment average displacement position from a predetermined reference position in the positive intensity distribution;
Calculating a first moment average displacement position from a predetermined reference position in the negative intensity distribution;
Obtaining a difference absolute value between an average displacement position in the positive intensity distribution and an average displacement position in the negative intensity distribution as an average displacement distance;
Setting a search range of block matching processing when creating the motion vector distribution image to be equal to or greater than the average displacement distance;
The ultrasonic moving image processing method according to claim 1, wherein: The motion vector distribution image creating step includes
Calculating a difference intensity distribution between the image data of the two frames;
Dividing the inside of the region of interest set in the differential intensity distribution into a positive intensity distribution and a negative intensity distribution;
Setting a plurality of regions of interest in each of the intensity distributions and calculating an average displacement distance for each region of interest;
Calculating an average value and standard deviation of the average displacement distance for each region of interest;
A value obtained by adding a constant multiple of the standard deviation to the average value is set as a search range for block matching processing when creating the motion vector distribution image;
The ultrasonic moving image processing method according to claim 1, wherein: In the motion vector distribution image creation step,
Block matching processing when creating the motion vector distribution image is performed while shifting by a certain vector interval within the frame,
The motion vector distribution image creating step includes
Creating the high-resolution motion vector distribution image by setting the vector interval to a predetermined subdivision interval;
Creating the motion vector distribution image with low resolution by setting the vector interval to be coarser than the subdivision interval;
Creating a scalar distribution image from each of the high resolution motion vector distribution image and the low resolution motion vector distribution image;
The ultrasonic moving image processing method according to claim 1, wherein: The motion vector distribution image creating step includes
A Fourier analysis of the motion vector distribution image in one scanning direction to obtain a two-dimensional power spectrum;
Obtaining a half width of the two-dimensional power spectrum;
Obtaining the scanning direction in which the full width at half maximum is maximized;
Setting a constant multiple of the reciprocal of the maximum half-value width as the vector interval of the motion vectors;
The ultrasonic moving image processing method according to claim 1, wherein: The motion vector distribution image creating step includes
Receiving an instruction to specify an observation range size of the image data;
Using the device specification parameters of the ultrasonic probe probe, obtaining a size of a pattern matching area to be performed when obtaining the motion vector;
Subtracting the size of the pattern matching region from the observation range size, and dividing the value by a value obtained by subtracting 1 from the matrix length corresponding to the region of interest of the image data;
Setting the result of the division as the vector interval;
The ultrasonic moving image processing method according to claim 1, wherein: In the motion vector distribution image creation step,
The supersonic wave according to claim 1, wherein a size of a pattern matching region to be performed when obtaining the motion vector is set to be equal to or larger than a sum of a speckle size of the image data and a half width of the ultrasonic beam. Wave image processing method. An irradiation unit for irradiating the subject with ultrasonic waves;
A detection unit for detecting an ultrasonic signal from the subject;
A storage unit for storing a detection result detected by the detection unit;
A calculation unit that controls operations of the irradiation unit and the detection unit, and generates an ultrasonic moving image of the subject based on the detection result;
With
The computing unit is
An image data creation step for creating image data of at least two frames having different detection timings from the stored detection results;
A motion vector distribution image creating step of creating a motion vector distribution image by performing motion vector analysis using a plurality of frames of the image data;
A luminance determining step of determining a luminance value of a pixel of the scalar distribution image using the motion vector distribution image;
Run
In the luminance determination step,
Performing a boundary clarifying step to mitigate the extent to which the luminance change across the boundary portion in the scalar distribution image depends on the direction in the scalar distribution image;
In the boundary clarifying step,
An ultrasonic moving image processing apparatus comprising: performing a covariance step of converting the motion vector distribution image into a complex matrix for each predetermined region of interest, and converting the complex matrix into a covariance matrix. In the luminance determination step, the calculation unit includes:
The ultrasonic moving image processing apparatus according to claim 13, wherein an infiltration degree clarification step is performed in which a luminance value of the scalar distribution image is increased in a portion where the movement of the tissue of the subject is large. A program for causing a computer to execute a step of processing an ultrasonic moving image, wherein the computer
Image data that irradiates a subject with ultrasonic waves, stores a detection result obtained by detecting an ultrasonic signal from the subject in a storage device, and creates image data of at least two frames having different detection timings from the stored detection result Creation steps,
A motion vector distribution image creating step of creating a motion vector distribution image by performing motion vector analysis using a plurality of frames of the image data;
A luminance determining step of determining a luminance value of a pixel of the scalar distribution image using the motion vector distribution image;
And execute
In the luminance determination step, the computer
Performing a boundary clarification step to mitigate the extent to which the luminance change across the boundary portion in the scalar distribution image depends on the direction in the scalar distribution image;
In the boundary clarifying step, the computer
An ultrasonic moving image processing program characterized by performing a covariance step of converting the motion vector distribution image into a complex matrix for each predetermined region of interest and converting the complex matrix into a covariance matrix. In the luminance determination step, the computer
The ultrasonic moving image processing program according to claim 15, wherein an infiltration degree clarification step is performed to increase a luminance value of the scalar distribution image in a portion where the movement of the tissue of the subject is large.

Images