JP4644145B2 - Ultrasonic diagnostic equipment - Google Patents

Description

  The present invention relates to an ultrasonic diagnostic apparatus, and more particularly to an ultrasonic diagnostic apparatus that forms a three-dimensional image of a subject.

  As an ultrasonic diagnostic apparatus that forms a three-dimensional image, an apparatus that forms a volume rendering image is known (see Patent Documents 1 and 2). In this method, a predetermined voxel operation is sequentially performed on individual echo data (voxel data) existing on each perspective line (ray), and a result of the voxel operation is calculated for each perspective line. A three-dimensional image (volume rendering image) is formed as a set of result values of the voxel operation regarding the line. The volume rendering image formed in this way greatly contributes to the diagnosis of, for example, a fetus.

Japanese Patent No. 2883584 JP 2001-145631 A

  The fetus is in amniotic fluid with a relatively low echo value. In the volume rendering method, a voxel having a small echo value is not easily reflected in the result of the voxel calculation. Therefore, echoes obtained from the fetus are strongly reflected in the volume rendering image, and as a result, an image of the fetus floating in the amniotic fluid is clearly displayed.

  However, if floating substances such as fragments of the uterine wall are present in the amniotic fluid, the echoes of the floating objects are reflected in the voxel calculation, and in the volume rendering image, for example, an irregular shape is formed on the fetal surface. It may appear. In addition, echoes of structures in the fetus's head or in the body are reflected in the voxel calculation, which may appear as an uneven shape on the surface of the fetal image.

  In general, smoothing processing is known as a technique for removing noise and the like on an image. The smoothing process is also effective for the volume rendering image. For example, the above-described irregularities on the fetal surface can be removed by the smoothing process.

However, if the smoothing process is simply used, the image data is distorted (discolored).
Therefore, in the case where the state of the inside of the fetus is clearly observed by a tomographic image or the like, the warp is adversely affected. That is, the simple smoothing process may improve the image quality of the volume rendering image, but may deteriorate the image quality of the tomographic image.

  The present invention has been made in such a background, and an object thereof is to provide a new technique for improving the image quality of an ultrasonic image.

  In order to achieve the above object, an ultrasonic diagnostic apparatus according to a preferred aspect of the present invention includes a transmission / reception unit that transmits and receives ultrasonic waves in a three-dimensional space including a subject to acquire a plurality of voxel data; A surface detector that detects surface voxel data corresponding to the surface of the subject from a plurality of voxel data, and a three-dimensional view of the subject based on the plurality of voxel data obtained by smoothing the surface voxel data A three-dimensional image forming unit that forms an image and a display unit that displays the formed three-dimensional image are provided.

  In the above configuration, surface voxel data corresponding to the surface of the subject is detected, and the surface voxel data is smoothed. Therefore, when a volume rendering image is formed as a three-dimensional image, for example, uneven noise on the surface of the subject is removed. Further, since the surface voxel data is detected, for example, a mode in which the smoothing process is not performed on the voxel data obtained from within the subject is possible. That is, for example, when forming a tomographic image in the subject, it is possible to form a clear tomographic image without performing smoothing processing on the voxel data obtained from the subject. Thus, with the above configuration, the image quality of the three-dimensional image can be improved without degrading the image quality of the tomographic image.

  In a preferred aspect, the surface detection unit detects surface voxel data by obtaining a region of voxel data in which the voxel data value changes greatly from comparison with neighboring voxel data. In a preferred aspect, the surface detection unit calculates a standard deviation related to a voxel data group in the local region, compares the standard deviation with a threshold value, extracts a surface region composed of a plurality of local regions having a large standard deviation, and extracts surface voxel data. Is detected.

  In a preferred aspect, the three-dimensional image forming unit forms a volume rendering image as the three-dimensional image. In a preferred embodiment, the surface voxel data after the smoothing process is subjected to an amplification process that amplifies the data value by a certain amount, and the three-dimensional image forming unit performs the smoothing process and the amplification process on the surface voxel data. A volume rendering image is formed based on a plurality of voxel data applied.

  In a desirable mode, the image processing apparatus further includes a cross-sectional image forming unit that forms a plurality of cross-sectional images of the subject based on a plurality of voxel data in which only the surface voxel data is subjected to smoothing processing and amplification processing. In a preferred aspect, the cross-sectional image forming unit forms three cross-sectional images substantially orthogonal to each other.

  The present invention provides a new technique for improving the quality of ultrasound images. Thereby, for example, the image quality of the three-dimensional image can be improved without degrading the image quality of the tomographic image.

  Hereinafter, preferred embodiments of the present invention will be described.

  FIG. 1 shows a preferred embodiment of an ultrasonic diagnostic apparatus according to the present invention, and FIG. 1 is a functional block diagram showing the overall configuration thereof. The ultrasonic diagnostic apparatus according to the present invention forms a three-dimensional image of a subject based on echo data obtained from the subject. The subject to be diagnosed is, for example, a tissue in a living body or a fetus in a mother body. Therefore, in the present embodiment, the fetus in the mother body will be described as a diagnosis target. However, the diagnostic object of the ultrasonic diagnostic apparatus according to the present invention is not limited to the fetus.

  The three-dimensional image formed in the present embodiment is a volume rendering image formed based on the volume rendering method. Therefore, first, the volume rendering image will be outlined.

  FIG. 2 is a diagram for explaining the principle of the volume rendering method. The three-dimensional data space 50 means a set of a plurality of voxel data acquired by scanning an ultrasonic beam, and is virtually constructed in the ultrasonic diagnostic apparatus. Here, the three-dimensional data space 50 has xyz orthogonal coordinate axes, and voxel data exists at each coordinate in the three-dimensional data space 50. When the three-dimensional ultrasonic transmission / reception space is configured as a polar coordinate space of rθφ, the coordinates of (r, θ, φ) that each voxel data has are (x, y, z). The process of converting to the coordinates is performed.

  In volume rendering, a viewpoint is generally set outside the three-dimensional data space 50, and a screen 100 as a two-dimensional plane is virtually connected to the opposite side of the viewpoint through the three-dimensional data space 50. Is set. Then, a plurality of rays (perspective lines) 60 are defined from the viewpoint side. The ray 60 penetrates the three-dimensional data space 50. For this reason, the ray 60 corresponds to a voxel data string composed of a plurality of voxel data. When a voxel operation based on the volume rendering method is sequentially executed for each voxel data from the viewpoint side along the ray 60, a pixel value is determined as a result of the final voxel operation. The pixel value is mapped to coordinates corresponding to the ray 60 on the screen 100.

  Two-dimensional orthogonal coordinate axes are associated on the screen 100, and a ray 60 is set for each coordinate on the screen 100. Then, the pixel value obtained for each ray 60 is mapped on the screen 100, whereby a three-dimensional image (volume rendering image) is formed on the screen 100. In the volume rendering method, various types of calculation formulas for voxel calculation for each voxel data are known. Basically, in any arithmetic expression, opacity (opacity) is used as a parameter for each voxel calculation of each voxel data. Using such parameters, an output light amount (output value) is obtained for each voxel operation, and this is used as an input light amount (input value) in the next voxel operation. This is repeated, and the output light amount at the time when the calculation end condition is satisfied is converted into a pixel value. That is, it is based on a model in which light propagates through a medium while being scattered and attenuated. In the present embodiment, for example, the following expression is used as an arithmetic expression for voxel arithmetic.

  The voxel calculation executed for each ray 60 is performed, for example, when the target coordinate exceeds the three-dimensional data space 50 or the accumulated addition value of opacity used in each voxel calculation is a predetermined value (for example, 1). The process is terminated when a predetermined condition is satisfied, for example, when the number is exceeded. The output light amount at the end of the calculation is associated with the pixel value of the ray 60.

  The voxel calculation for each ray 60 is greatly influenced by the voxel data on the surface of the fetus that is the subject. That is, the voxel data corresponding to the amniotic fluid does not greatly contribute to the voxel calculation because the voxel data corresponding to the amniotic fluid is small in the process in which the voxel calculation proceeds along each ray 60 to the amniotic fluid, the fetal surface, and the inside of the fetus. . When the voxel calculation proceeds to the fetal surface, the voxel data obtained from the fetus is greatly reflected in the voxel calculation because the voxel value is large. The accumulated value of the opacity of the voxel calculation on the fetal surface becomes a value close to a predetermined value (for example, 1). Therefore, even if the voxel calculation proceeds to the inside of the fetus, the voxel data inside the fetus greatly contributes to the voxel calculation. No longer. Thus, the voxel data on the fetal surface has a great influence on the voxel calculation.

  Due to the large influence of the voxel data on the fetal surface, the fetus floating in the amniotic fluid is displayed relatively clearly in the volume rendering image. Therefore, the volume rendering image is extremely effective particularly when observing the appearance of the fetus. On the other hand, since the influence of the voxel data on the fetal surface is large, the fetus surface is easily affected by the state of the voxel data on the fetal surface. For example, when there are variations or fluctuations in the voxel data on the fetal surface due to some influence, the fluctuations and fluctuations are easily affected.

  FIG. 3 is a diagram showing the state of the voxel data on the fetal surface. FIG. 3A shows a state in which there are variations and fluctuations in the voxel data on the surface of the fetus 90. If the voxel calculation proceeds along the ray 60 in a state where the voxel data on the surface of the fetus 90 has variations and fluctuations, the voxel data on the surface of the fetus 90 has a great influence on the voxel calculation, resulting in variations in the voxel data. It is reflected in the voxel calculation and may appear as fine uneven noise on the surface of the fetus 90 image of the volume rendering image, for example.

  Further, as shown in FIG. 3B, if a floating substance 70 such as a uterine wall fragment exists on the ray 60 in front of the fetus 90, the voxel data of the floating substance 70 greatly affects the voxel calculation. In some cases, irregularities associated with the influence of the floating material 70 may appear on the surface of the fetus 90 image of the volume rendering image. Further, as shown in FIG. 3C, although the structure (bone, brain, organ, etc.) inside the fetus 90 has a relatively low influence on the voxel calculation, the influence is not zero. May affect the surface.

  A smoothing process is known as a technique for removing uneven noise appearing in an image. Smoothing processing is also effective for volume rendering images. For example, it is possible to remove unevenness on the surface of the fetus by performing smoothing processing on a plurality of voxel data in a three-dimensional space. However, if the smoothing process is simply used, the image data becomes distorted (unclear). Therefore, when the state of the inside of the fetus is clearly observed by a tomographic image or the like, the smoothing process is warped and has an adverse effect. Will affect. That is, the simple smoothing process may improve the image quality of the volume rendering image while degrading the image quality of the tomographic image.

  In this embodiment, the image quality improvement of the volume rendering image and the maintenance of the image quality of the tomographic image are achieved by the method described in detail below. Therefore, returning to FIG. 1, an embodiment of the ultrasonic diagnostic apparatus according to the present invention will be described in detail.

  The ultrasonic probe 10 is an ultrasonic probe that transmits and receives ultrasonic waves to and from a fetus in the mother's body. The ultrasonic probe 10 includes a plurality of vibration elements (not shown), and forms an ultrasonic beam in a three-dimensional space. Multiple vibration elements are controlled electronically or mechanically, and reflected waves (echoes) are acquired from the entire area of the three-dimensional space (in the scanning space) by scanning the ultrasonic beam in the three-dimensional space. To do.

  The transmission / reception unit 12 controls a plurality of vibration elements in the ultrasonic probe 10 to form a transmission beam, and scans it in a three-dimensional space. Further, a plurality of received signals obtained from a plurality of vibration elements are phased and added to form a reception beam, and voxel data is acquired from the entire scanning space. Thus, the transmission / reception unit 12 has functions of a transmission beamformer and a reception beamformer.

  The three-dimensional data memory 14 is a memory that stores voxel data supplied from the transmission / reception unit 12. The three-dimensional data memory 14 stores a plurality of voxel data obtained from the entire area in the three-dimensional space at addresses corresponding to the positions of the respective voxel data. For example, a three-dimensional space is represented by an xyz orthogonal coordinate system, and the voxel value (luminance value) of each voxel is stored at an address corresponding to the coordinate value of that voxel. Incidentally, when the address of each voxel data is designated by the read control unit 15, the voxel data of the designated address is output to each processing unit in the subsequent stage.

  The smoothing processing unit 16 performs a smoothing process on a plurality of voxel data. For example, the voxel values of a certain voxel are averaged using the voxel values of voxels around the voxel. Specifically, for example, a 5 × 5 × 5 window is set in the xyz orthogonal coordinate system with the target voxel as the center, the average value of the voxel values of 125 voxels in the window is calculated, and the window The voxel value of the target voxel located at the center is set as the average value. Then, while moving the window over the entire area in the three-dimensional space, the average value is obtained using all voxels in the three-dimensional space as the target voxel. Thereby, all the voxel data in the three-dimensional space are smoothed. The smoothed voxel data is output to the weight circuit 18 and the standard deviation calculator 20.

  The weighting circuit 18 performs weighting processing (amplification processing) on the data after smoothing processing by multiplying each smoothed processing voxel data by a predetermined weighting coefficient. The amplified voxel data is output to the selector 24. The standard deviation calculator 20 calculates a standard deviation based on the smoothed voxel data. Further, the calculated standard deviation is compared with a predetermined threshold value in the determination circuit 22. The standard deviation calculator 20 and the determination circuit 22 function as a surface detector that detects the surface voxel data by extracting the surface area of the fetus that is the subject.

  By performing the smoothing process, the voxel data becomes data from which fine noise and fine structures are removed. Therefore, the image formed by the voxel data after the smoothing process is a smooth image as a whole. Even if it becomes smooth as a whole, the change in the voxel data is greatest at the boundary between the amniotic fluid and the fetus, so that the boundary can be extracted from the voxel data after the smoothing process. That is, in the voxel data after the smoothing process, the boundary portion between the amniotic fluid and the fetus can be extracted from the change of the voxel data in a state where fine noise and fine structures are removed. For the boundary extraction, a technique of differential processing using a differential filter such as Prewit, Sobel, Roberts, or Raprasian may be used, but in this embodiment, a standard deviation is used. That is, the standard deviation calculation unit 20 calculates the standard deviation σ using the following formula.

  The standard deviation calculation unit 20 calculates standard deviations regarding a plurality of voxel data groups. For example, a 3 × 3 × 3 window is set in the xyz orthogonal coordinate system centering on the target voxel, and the standard deviation of the voxel values of 27 voxel data in the window (N = 27 in Equation 2) is calculated. . Then, while moving the window over the entire area in the three-dimensional space, the standard deviation is obtained using all the voxels in the three-dimensional space as the target voxel.

  Further, the judgment circuit 22 compares the standard deviation with a threshold value for each window, and extracts a window having a large standard deviation. A window with a large standard deviation is a portion where the voxel value changes drastically. Therefore, a window with a large standard deviation can be regarded as a boundary portion between the amniotic fluid and the fetus. By the way, the threshold for judging the standard deviation is suitable for extracting the boundary between the amniotic fluid and the fetus in advance according to the number of gradations of each voxel data, the size of the amniotic fluid and the luminance value of the fetus, etc. Value is set. Further, a threshold value may be set by the user via the operation panel 30, or the user may correct the threshold value.

  The selector 24 selects one of the two types of voxel data based on the determination result of the determination circuit 22. That is, for the voxel data whose standard deviation is larger than the threshold value and is regarded as the boundary portion between the amniotic fluid and the fetus, the smoothed and amplified data output from the weight circuit 18 is selected. On the other hand, for voxel data other than the boundary portion where the standard deviation is smaller than the threshold value, the original data of the voxel data read directly from the three-dimensional data memory 14 is selected.

  FIG. 4 is a diagram for explaining voxel data for each processing stage of the present embodiment. FIG. 4 shows a schematic diagram of an image formed from the voxel data for each processing stage. FIG. 4 shows a schematic diagram using a two-dimensional image for the sake of illustration, but the actual voxel data handled in this embodiment is obtained from the three-dimensional space as described in detail above. Thus, a three-dimensional data space (reference numeral 50 in FIG. 2) is formed.

  FIG. 4A shows a schematic diagram of an image obtained from the original data. This corresponds to voxel data stored in the three-dimensional data memory (reference numeral 14 in FIG. 1). On the other hand, FIG. 4B shows a schematic diagram of an image obtained from the data after the smoothing process. That is, it corresponds to the voxel data subjected to the smoothing process by the smoothing processing unit (reference numeral 16 in FIG. 1). As a result of the smoothing process, the uneven noise on the surface of the fetus is removed, and fine noise and fine structures in the image are removed.

  FIG. 4C shows a schematic diagram of an image in which the boundary of the fetal surface is extracted. That is, the determination circuit (reference numeral 22 in FIG. 1) and the selector (reference numeral 24 in FIG. 1) select the voxel data (B) that has been subjected to smoothing processing on the fetal surface boundary portion shown in FIG. On the other hand, voxel data (A) that has not been subjected to smoothing processing is selected for portions other than the fetal surface.

  FIG. 4D shows a schematic diagram of an image that has been smoothed only on the fetal surface (near the body surface). That is, FIG. 4D corresponds to an image formed by voxel data after data selection by the determination circuit and the selector. In FIG. 4D, weighting processing (see the weighting circuit 18 in FIG. 1) is applied to the voxel data on the fetal surface in addition to the smoothing processing.

  Thus, in the present embodiment, three-dimensional data (a plurality of voxel data) in which the smoothing process is performed only on the fetal surface is formed.

  Returning to FIG. 1, the image forming unit 26 is based on a plurality of voxel data output from the selector 24, that is, three-dimensional data (see FIG. 4D) in which only the fetal surface is smoothed. A three-dimensional image is formed.

  For the formation of a three-dimensional image, the volume rendering method described in detail above (see FIG. 2) is used, and a volume rendering image is formed as a three-dimensional image. Since the volume rendering image formed in the present embodiment is formed based on voxel data obtained by smoothing the fetal surface, there is no fine uneven noise on the surface of the fetal image, and it is smooth and natural. An image is formed. In the present embodiment, weighting processing is applied to the voxel data on the fetal surface in addition to the smoothing processing. By this weighting process, the volume rendering image becomes smoother. As the weighting coefficient, for example, a predetermined set value (1.25) is used. Of course, the user may set the weighting factor via the operation panel 30, or the user may correct the weighting factor.

  Further, the image forming unit 26 forms a tomographic image based on the plurality of voxel data output from the selector 24. The image forming unit 26 forms three orthogonal cross sections composed of three cross sections orthogonal to each other in the three-dimensional data space. For example, a top view, a side view, a front view, and the like correspond to them. The image forming unit 26 extracts voxel data on each of the three orthogonal cross sections from a plurality of voxel data in the three-dimensional data space, and forms three cross-sectional images. The tomographic image formed in this embodiment is formed based on voxel data in which only the fetal surface is smoothed. Therefore, it is possible to form a fine image that is not subjected to the smoothing process on the inside of the fetus other than the fetal surface.

  The volume rendering image and the orthogonal three-section image formed by the image forming unit 26 are displayed on the monitor 28. For example, the display area of the monitor 28 is divided into four, and three cross-sectional images of an orthogonal three cross-sectional image and one volume rendering image are displayed simultaneously. Further, based on a user instruction input via the operation panel 30, only one of the volume rendering image and the orthogonal three-section image may be displayed.

  Incidentally, in the present embodiment, the image forming unit 26 forms both types of images, a three-dimensional image and a tomographic image. For this reason, for example, when the function of the image forming unit 26 is realized by a CPU, the voxel data output from the selector 24 is temporarily stored in one CPU memory, and only one CPU memory is used for three-dimensional processing. A configuration for forming both types of images, tomographic images, becomes possible. That is, it is not necessary to separately provide two CPU memories, that is, a CPU memory for tomographic images and a CPU memory for three-dimensional images. Therefore, the present embodiment is excellent in terms of the manufacturing cost of the apparatus.

  As mentioned above, although preferred embodiment of this invention was described, embodiment mentioned above is only a mere illustration in all the points, and does not limit the scope of the present invention.

1 is a functional block diagram showing an overall configuration of an ultrasonic diagnostic apparatus according to the present invention. It is a figure for demonstrating the principle of a volume rendering method. It is a figure which shows the state of the voxel data of the fetus surface. It is a figure for demonstrating the voxel data for every process step of this embodiment.

Explanation of symbols

  14 three-dimensional data memory, 16 smoothing processing unit, 18 weight circuit, 20 standard deviation calculation unit, 22 determination circuit, 24 selector, 26 image forming unit.

Claims (7)

A transmission / reception unit for acquiring a plurality of voxel data by transmitting / receiving ultrasonic waves in a three-dimensional space including a subject;
A surface detector that detects surface voxel data corresponding to the surface of the subject from a plurality of voxel data;
A three-dimensional image forming unit that forms a three-dimensional image of a subject based on a plurality of voxel data composed of surface voxel data and other voxel data, wherein only surface voxel data is subjected to smoothing processing and amplification processing ; ,
A display unit for displaying the formed three-dimensional image;
Having
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 1,
The surface detection unit detects surface voxel data by obtaining a region of voxel data in which the voxel data value greatly changes from a comparison with neighboring voxel data.
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 1,
The surface detection unit calculates a standard deviation related to the voxel data group in the local region and compares it with a threshold value, extracts a surface region composed of a plurality of local regions having a large standard deviation, and detects surface voxel data.
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 3.
The three-dimensional image forming unit forms a volume rendering image as the three-dimensional image.
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 4,
Wherein Ru amplification process is performed to amplify a constant amount of data values to the surface voxel data after the smoothing process,
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to any one of claims 1 to 5 ,
A cross-sectional image forming unit that forms a plurality of cross-sectional images of a subject based on a plurality of voxel data composed of surface voxel data and other voxel data, in which only surface voxel data is subjected to smoothing processing and amplification processing. In addition,
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 6,
The cross-sectional image forming unit forms three cross-sectional images substantially orthogonal to each other;
An ultrasonic diagnostic apparatus.

Images