JP5550507B2 - Ultrasonic diagnostic equipment - Google Patents

Description

  The present invention relates to an ultrasonic diagnostic apparatus, and more particularly to a technique for separating a diagnosis object into a plurality of objects.

  2. Description of the Related Art Conventionally, a technique for performing image processing suitable for diagnosis on a tissue or the like in an ultrasonic image has been well known, and the image processing is various according to the type of tissue or the content of diagnosis. And the follicle in the living body is also the object of ultrasonic diagnosis as the tissue or the like.

  In a living body, a plurality of follicles are in close contact with each other and are densely packed. Therefore, in the ultrasonic image, a plurality of follicles are imaged in a state of being connected to each other, and it is difficult to individually confirm the size and shape of each follicle. Therefore, it is desired that each follicle can be separated and individually recognized in the ultrasonic image.

  Watershed is known as a general technique for image processing for dividing a target image, and each follicle may be separated in an ultrasonic image by using this watershed process. However, the processing time of this watershed process is enormous due to the nature of the algorithm. In particular, for example, when applied to a three-dimensional image, the processing time becomes extremely large, and there is a problem in practical use.

  Under such circumstances, the inventors of the present application have conducted research and development on image processing suitable for separation of follicles, for example. In the image processing, in order to individually recognize a plurality of parts such as separated follicles, it is desirable to assign different labels to each of the plurality of parts. Incidentally, Patent Document 1 describes an invention in which a labeling process is used to extract and remove a relatively large noise block in an ultrasonic image.

JP 2004-267484 A

  The present invention has been made in the course of the above-described research and development, and an object thereof is to provide a practical image processing technique for separating a plurality of diagnostic objects in ultrasonic image data. .

  An ultrasonic diagnostic apparatus suitable for the above object includes a probe that transmits and receives ultrasonic waves to a region including a diagnosis target, a transmission / reception unit that collects echo data from the region by controlling the probe, and an echo data In the obtained image data, the contraction process is performed on the diagnosis target to separate the diagnosis target into a plurality of isolated parts, and the plurality of isolated parts in the contracted image data A labeling processing unit that assigns different labels to each of the plurality of isolated portions by performing labeling processing, and in the image data that has been labeled, performs expansion processing on each of the plurality of isolated portions, Restore the size of the diagnostic object while assigning the label of the isolated part to the expanded part obtained from each isolated part An expansion processing section, and having a.

  In the above configuration, the diagnostic object is separated into a plurality of isolated portions by the contraction process, different labels are assigned to each of the plurality of isolated portions, and the isolated portions obtained from each isolated portion are isolated. The size of the diagnosis target is restored while assigning the partial labels. Therefore, it is possible to identify each of the plurality of inflated portions based on the label for the restored diagnosis object. Thereby, for example, the shape, area, and volume of the inflated portion corresponding to the label may be measured for each label, or the measurement amount regarding a plurality of inflated portions may be statistically analyzed, or attention will be paid. For example, it is possible to extract and display only the power expansion part, and the range of practical applications and the high value are immeasurable.

  In a preferred embodiment, the expansion processing unit restores the size of the diagnosis target while forming a boundary at the overlapping portion of the expansion portions that overlap each other by the expansion processing.

  In a preferred embodiment, the contraction processing unit repeatedly performs contraction processing for contracting the diagnosis target in n steps (n is a natural number) in stages, and the expansion processing unit is configured to step-by-step each of the isolated parts. It is characterized by repeatedly executing the expansion process of expanding n times n times.

  In a preferred embodiment, the image processing apparatus further includes a removal processing unit that removes an isolated portion smaller than a threshold value from the plurality of isolated portions separated by the contraction processing from the target of the expansion processing.

  In a preferred embodiment, the diagnostic object restored by the expansion process further includes a volume calculation unit that calculates the volume of the expansion portion corresponding to the label for each label.

  In a preferred embodiment, the diagnosis target is a plurality of follicles closely attached to each other, and the plurality of isolated portions are a plurality of follicles separated from each other by the contraction process.

  In addition, a suitable ultrasonic image processing apparatus that meets the above-mentioned purpose separates a diagnostic object into a plurality of isolated portions by performing a contraction process on the diagnostic object in image data obtained from ultrasonic echo data. A contraction processing unit, a labeling processing unit that assigns different labels to each of a plurality of isolated parts by performing a labeling process on the plurality of isolated parts in the contracted image data, and a labeling process In the image data thus obtained, expansion processing is performed on each of the plurality of isolated portions, and the size of the diagnosis target is restored while assigning the label of the isolated portion to the expanded portion obtained from each isolated portion. And an expansion processing unit.

  For example, a computer may function as the ultrasonic image processing apparatus by using a program for realizing the functions of the above-described contraction processing unit, labeling processing unit, and expansion processing unit.

  According to the present invention, a practical image processing technique for separating a plurality of diagnosis targets in ultrasonic image data is provided.

1 is a diagram illustrating an overall configuration of an ultrasonic diagnostic apparatus that is preferable in the practice of the present invention. It is a figure for demonstrating the various processes by the ultrasonic diagnosing device of FIG. It is a figure for demonstrating the scanning of the filter in a three-dimensional data space. It is a figure for demonstrating the filter process in an expansion process. It is a figure for demonstrating the influence on the expansion process of the small lump accompanying noise etc. FIG.

  FIG. 1 is a diagram showing an overall configuration of an ultrasonic diagnostic apparatus suitable for implementing the present invention. The probe 10 is an ultrasonic probe that transmits / receives ultrasonic waves to / from a region including a diagnosis target. The probe 10 includes a plurality of vibration elements that transmit and receive ultrasonic waves, and transmission of the plurality of vibration elements is controlled by the transmission / reception unit 12 to form a transmission beam. In addition, a plurality of vibration elements receive ultrasonic waves obtained from within the region including the diagnosis target, and a signal obtained thereby is output to the transmission / reception unit 12, and the transmission / reception unit 12 forms a reception beam to form a reception beam. Echo data is collected along.

  The probe 10 is preferably a three-dimensional probe that three-dimensionally collects echo data by scanning an ultrasonic beam (transmission beam and reception beam) in a three-dimensional space. For example, the ultrasonic beam is three-dimensionally scanned by mechanically moving a scanning surface electronically formed by a plurality of vibration elements (1D array transducers) arranged one-dimensionally. Alternatively, the ultrasonic beam may be scanned three-dimensionally by electronically controlling a plurality of two-dimensionally arranged vibration elements (2D array transducers). Of course, a two-dimensional ultrasonic probe that scans the ultrasonic beam in the tomographic plane may be used.

  When echo data is collected by scanning an ultrasonic beam in the three-dimensional space, the echo data (voxel data) for a plurality of voxels constituting the three-dimensional data space corresponding to the three-dimensional space is not shown. Is remembered. Various processes are executed from the binarization processing unit 20 to the expansion boundary processing unit 60 on a plurality of voxels constituting the three-dimensional data space. Therefore, these various processes will be described. In addition, about the part (structure) shown in FIG. 1, the code | symbol of FIG. 1 is utilized in the following description.

  FIG. 2 is a diagram for explaining various processes by the ultrasonic diagnostic apparatus of FIG. FIG. 2A shows the binarization processing in the binarization processing unit 20 (FIG. 1). The binarization processing unit 20 performs binarization processing on a plurality of voxels constituting the three-dimensional data space, thereby forming image data after the binarization processing shown in FIG. As a diagnosis target in the present embodiment, a follicle in a living body is suitable. The binarization processing unit 20 identifies a voxel corresponding to the follicle F and other voxels by comparing a predetermined threshold value with the voxel value (the size of echo data) of each voxel. For example, the voxel value of the voxel corresponding to the follicle F is set to “1”, and the voxel values of the other voxels are set to “0”. In FIG. 2A, the voxel group corresponding to the follicle F is drawn in white, and the other voxel group corresponding to the background is drawn in black.

  In a living body, a plurality of follicles are in close contact with each other and are densely packed. Therefore, in the ultrasonic image, as shown in FIG. 2A, a plurality of follicles F are imaged in a mutually connected state, and it is difficult to individually confirm the size, shape, and the like of each follicle F. Therefore, in the present embodiment, the plurality of follicles F are divided from each other by various processes described later. In FIG. 2, although each image data is drawn two-dimensionally, various processes are executed three-dimensionally in the three-dimensional data space.

  FIG. 2B shows the contraction separation process in the contraction separation processing unit 30 (FIG. 1). The contraction separation processing unit 30 performs contraction processing on a plurality of follicles F in the voxel data after the binarization processing constituting the three-dimensional data space, that is, in the binarized image data shown in FIG. And a plurality of follicles F are separated into follicles F1 to F3 as shown in FIG. The contraction separation processing unit 30 repeatedly executes contraction processing for contracting the follicle F step by step n times (n is a natural number). For the contraction process at each stage, a contraction process filter is used, and the filter is scanned over the entire area of the three-dimensional data space.

  FIG. 3 is a diagram for explaining scanning of the filter 120 in the three-dimensional data space 100. In FIG. 3, the three-dimensional data space 100 is shown in an xyz orthogonal coordinate system. The filter 120 has a three-dimensional configuration in which the lengths in the x-axis direction, the y-axis direction, and the z-axis direction correspond to three voxels, corresponding to a total of 27 voxels. The voxel located at the center of the filter 120 is the target voxel, and the 26 voxels surrounding it are peripheral voxels. Then, the filter 120 is moved in the x-axis direction, the y-axis direction, and the z-axis direction so that all the voxels in the three-dimensional data space 100 become the target voxel, and the entire area in the three-dimensional data space 100 is scanned. Is done.

  In the shrinkage separation process, if at least one voxel value “0” is present in the 26 neighboring voxels in the filter 120 at each scanning position, the voxel value of the target voxel located at the center of the filter 120 is “ 0 ”. For example, if the target voxel is “1 (follicle)” and there is one voxel value “0 (background)” in the surrounding voxels, the target voxel is converted to “0 (background)”. . Then, the filter 120 is scanned over the entire area of the three-dimensional data space 100, and filter processing is performed at each scanning position, thereby completing one-stage contraction processing. Note that the conversion of the voxel value related to the target voxel is executed after the filter 120 is scanned over the entire area in the three-dimensional data space 100. That is, while the filter 120 is being scanned, the voxel value is not converted, and the filtering process is executed on the voxel value before the conversion at any scanning position.

  Thus, when the one-stage contraction process is completed and voxel value conversion is performed based on the result, the second-stage contraction process is performed on the three-dimensional data space 100 composed of the converted voxel values. Is executed. Also in the second stage contraction process, the same filter process as in the first stage is executed. That is, if at least one voxel of the voxel value “0” is present in the 26 neighboring voxels in the filter 120 at each scanning position, the voxel value of the target voxel positioned at the center of the filter 120 is “0”. After the filter 120 is scanned over the entire area of the three-dimensional data space 100, the voxel values are converted.

  The shrinkage separation processing unit 30 repeatedly executes this stepwise shrinkage process n times (n is a natural number). The number of repetitions n is appropriately determined according to the size of each voxel, the size of the filter, etc., and is set to about 10 or less, for example. Of course, the user may be able to adjust the number of times n.

  In the case of the two-dimensional case, instead of the filter 120 shown in FIG. 3, a two-dimensional filter corresponding to a total of nine voxels each having a vertical and horizontal length corresponding to three voxels is used. The voxel located in the region may be the target voxel, and the eight voxels surrounding it may be the peripheral voxels.

  Returning to FIG. 2, the labeling processing unit 40 (FIG. 1) after the contraction processing constituting the three-dimensional data space is separated by the contraction processing into a plurality of follicles F <b> 1 to F <b> 3 as shown in FIG. In the voxel data, that is, in the image data after the contraction process shown in FIG. 2B, a labeling process is performed, and different labels are assigned to the plurality of follicles F1 to F3. A known method can be used as the labeling process. For example, a plurality of voxel clusters having the same voxel value are detected in a three-dimensional data space, and a label number is assigned to each cluster. For example, as shown in FIG. 2 (C), a label 0 is assigned to a background portion that is a block having a voxel value “0”, and follicles F1 to F3 that are blocks having a voxel value “1” are respectively assigned. Labels 1 to 3 are assigned.

  When the labeling process is performed, the expansion boundary processing unit 60 (FIG. 1) in the voxel data after the labeling process constituting the three-dimensional data space, that is, in the image data after the labeling process shown in FIG. In Step 2, an expansion process is performed on each of the plurality of follicles. In the expansion process, a label of the follicle is assigned to the expansion part obtained from each follicle, and further, a boundary is formed in the overlapping part of the expansion parts (expanded follicles) that overlap each other by the expansion process. The size of multiple follicles is restored. 2D, a boundary (background pixel) is formed between follicles corresponding to different labels, and the size of each follicle is the size before the contraction process (immediately after the binarization process). Will be restored.

  The expansion boundary processing unit 60 repeatedly performs expansion processing for expanding the follicle F in stages (n is the same as the number of contraction processing). For the expansion processing at each stage, a filter for expansion processing is used, and the filter is scanned over the entire area in the three-dimensional data space. Also in the expansion process, the three-dimensional filter 120 corresponding to the total of 27 voxels shown in FIG. 3 is used, and the voxel located at the center of the filter 120 is set as the target voxel, and the 26 voxels surrounding the voxel are the surroundings. It is assumed to be a voxel. Then, the filter 120 is moved in the x-axis direction, the y-axis direction, and the z-axis direction so that all the voxels in the three-dimensional data space 100 become the target voxel, and the entire area in the three-dimensional data space 100 is scanned. Is done. However, the filtering process is different between the shrinking process and the expanding process.

  FIG. 4 is a diagram for explaining the filtering process in the expansion process. FIG. 4 shows a condition table regarding voxel value conversion in the process of expanding while forming a boundary (expansion boundary process). In the expansion boundary process, the label value of each voxel is referred to.

  When the target voxel located at the center of the filter 120 (FIG. 3) is label 0 (background), if all of the 26 surrounding voxels are label 0 (background), the target voxel is labeled 0. Further, when the target voxel is label 0 (background), if there are labels other than label 0 (follicle) in 26 surrounding voxels, and all of them are the same label N (same follicle), the attention The voxel is converted to label N. That is, the follicle labeled N is expanded.

  Further, when the target voxel is label 0 (background), there are 26 labels other than label 0 (follicle) in 26 surrounding voxels, and different label numbers (follicles different from each other) are included in them. If so, the target voxel is labeled 0. That is, the target voxel is maintained at label 0 and becomes a boundary between different follicles.

  On the other hand, when the target voxel located at the center of the filter 120 is the label M (follicle), the target voxel is maintained at the label M regardless of the situation of the surrounding voxels.

  Then, the filter 120 shown in FIG. 3 is scanned over the entire area of the three-dimensional data space 100, and the filter processing is performed in accordance with the conditions shown in FIG. Boundary processing ends. Note that the conversion of the label value related to the target voxel is executed after the filter 120 is scanned over the entire area in the three-dimensional data space 100. That is, while the filter 120 is being scanned, the label value is not converted, and the filter process is performed on the label value before conversion at any scanning position.

  Thus, when the one-stage expansion boundary processing is completed and the label value is converted based on the result, the second-stage expansion is performed on the three-dimensional data space 100 formed by the converted label values. Boundary processing is performed. Also in the second stage expansion boundary process, the same filter process as in the first stage is executed. That is, at each scanning position, the filter process is performed according to the conditions shown in FIG. 4, and the label value is converted after the filter 120 is scanned over the entire area in the three-dimensional data space 100.

  The expansion boundary processing unit 60 repeatedly performs this stepwise expansion boundary processing n times. This number of repetitions n is the same as the number of repetitions n of the shrinking process. In this way, as shown in FIG. 2D, the boundary (background pixel) is formed between the follicles corresponding to different labels, and the size of each follicle is restored to the size before the contraction process.

  Even in the expansion boundary processing, in the case of two dimensions, instead of the filter 120 shown in FIG. 3, the vertical and horizontal lengths correspond to three voxels, corresponding to a total of nine voxels. Using a filter, a voxel located at the center may be set as a target voxel, and eight voxels surrounding it may be set as peripheral voxels.

  Returning to FIG. 1, the volume calculation unit 70 calculates, for each label, the volume of each follicle corresponding to the label for the plurality of follicles restored by the expansion boundary process. For example, if the size of a plurality of voxels is uniform, the volume of the follicle corresponding to the label is calculated by multiplying the volume of each voxel and the number of voxels constituting each label. Of course, the volume of the follicle may be calculated more accurately in consideration of the difference in the size of each voxel. Moreover, you may calculate the statistical numerical value about the volume value regarding several follicles.

  The volume of each follicle calculated by the volume calculation unit 70 is displayed on the display unit 90. Note that, based on the voxel values of a plurality of voxels before binarization processing, the image forming unit 80 forms a three-dimensional ultrasonic image in which the follicle is three-dimensionally displayed, and the ultrasonic image is displayed on the display unit 90. May be. As the three-dimensional ultrasonic image, for example, an image by volume rendering is suitable.

  As described above, in the present embodiment, after the plurality of follicles are separated by the contraction process, the size of the plurality of follicles is restored by the expansion process after the labeling process. In the contraction process, if a small lump that is different from the follicle or separated from the follicle remains due to the influence of noise or the like, the shape of the follicle restored by the expansion process may be adversely affected. .

  FIG. 5 is a diagram for explaining the influence on the expansion processing of a small lump due to noise or the like. For example, as shown in FIG. 5I, it is assumed that one follicle is separated by the shrinkage separation process in the shrinkage separation processing unit 30 (FIG. 1), and the label a is assigned to the follicle by the labeling process. In the shrinkage separation process, if a small chunk composed of voxels having a voxel value of “1” is generated due to noise or the like, for example, label b is assigned to the chunk by the labeling process.

  When the expansion boundary process is executed in the expansion boundary processing unit 60 (FIG. 1) in this labeling state, as shown in FIG. 5 (II), the mass of the label b is expanded in addition to the follicle of the label a, Moreover, a boundary is formed between the follicle of label a and the mass of label b by filtering according to the conditions shown in FIG. As a result, it is assumed that the expansion of the follicle of label a is suppressed in the vicinity of the mass of label b, and the original shape of the follicle of label a cannot be restored.

  Therefore, the noise removal processing unit 50 in FIG. 1 removes a lump smaller than the threshold from a plurality of lumps (isolated parts) separated by the shrinking process from the target of the expansion process. For example, among the chunks to which the label is assigned in the labeling processing unit 40 (a plurality of voxel chunks having the same voxel value), one having a voxel number equal to or less than a predetermined threshold is converted to label 0 (background). Thereby, for example, the lump of the label b shown in FIG. 5 is removed as a background, and the follicle of the label a is restored to the original shape in the expansion process (expansion boundary process). In setting a predetermined threshold, the size (number of voxels) is calculated for each label for all labels other than the background, and based on a statistical quantity obtained from the size of all these labels. The size of the follicle may be predicted, and a threshold value for removing a small mass other than the follicle may be determined.

  The ultrasonic diagnostic apparatus according to the preferred embodiment of the present invention has been described above. For example, the binarization processing unit 20, the contraction separation processing unit 30, the labeling processing unit 40, and the noise removal processing unit 50 shown in FIG. In addition, at least one of the expansion boundary processing unit 60 and the volume calculation unit 70 may be realized by a computer, and the computer may function as an ultrasonic image processing apparatus.

  The above-described embodiments are merely examples in all respects, and do not limit the scope of the present invention. The present invention includes various modifications without departing from the essence thereof.

  10 probe, 20 binarization processing unit, 30 contraction separation processing unit, 40 labeling processing unit, 50 noise removal processing unit, 60 expansion boundary processing unit, 70 volume calculation unit.

Claims (7)

A probe that transmits and receives ultrasound to and from a region that includes a diagnosis target;
A transceiver that collects echo data from the region by controlling the probe; and
In the image data obtained from the echo data, by performing a contraction process on the diagnosis target, a contraction processing unit that separates the diagnosis target into a plurality of isolated parts;
A labeling processing unit that assigns different labels to each of the plurality of isolated parts by performing a labeling process on the plurality of isolated parts in the image data that has undergone the shrinkage processing;
In the image data subjected to the labeling process, each of a plurality of isolated portions is subjected to expansion processing, and the label of the isolated portion is assigned to the expanded portion obtained from each isolated portion, and the size of the diagnosis target is set. An expansion processing unit to be restored;
I have a,
The expansion processing unit restores the size of the diagnostic object while forming a boundary at the overlapping portion of the expanded portions that overlap each other by being expanded.
An ultrasonic diagnostic apparatus. A probe that transmits and receives ultrasound to and from a region that includes a diagnosis target;
A transceiver that collects echo data from the region by controlling the probe; and
In the image data obtained from the echo data, by performing a contraction process on the diagnosis target, a contraction processing unit that separates the diagnosis target into a plurality of isolated parts;
A labeling processing unit that assigns different labels to each of the plurality of isolated parts by performing a labeling process on the plurality of isolated parts in the image data that has undergone the shrinkage processing;
In the image data subjected to the labeling process, each of a plurality of isolated portions is subjected to expansion processing, and the label of the isolated portion is assigned to the expanded portion obtained from each isolated portion, and the size of the diagnosis target is set. An expansion processing unit to be restored;
I have a,
The diagnostic object is a plurality of follicles in close contact with each other,
The plurality of isolated portions are a plurality of follicles separated from each other by the contraction process.
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 1 or 2,
The contraction processing unit repeatedly executes contraction processing for contracting the diagnosis object in stages n times (n is a natural number),
The expansion processing unit repeatedly performs expansion processing for expanding each isolated portion in stages n times.
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to any one of claims 1 to 3,
A removal processing unit for removing an isolated portion smaller than a threshold value from the plurality of isolated portions separated by the contraction processing from the target of the expansion processing;
An ultrasonic diagnostic apparatus. In the ultrasonic diagnostic apparatus according to any one of claims 1 to 4,
The diagnostic object restored by the expansion process further includes a volume calculation unit that calculates the volume of the expansion part corresponding to the label for each label.
An ultrasonic diagnostic apparatus. In the image data obtained from the ultrasonic echo data, by performing a contraction process on the diagnosis target, a contraction processing unit that separates the diagnosis target into a plurality of isolated parts,
A labeling processing unit that assigns different labels to each of the plurality of isolated parts by performing a labeling process on the plurality of isolated parts in the image data that has undergone the shrinkage processing;
In the image data subjected to the labeling process, each of a plurality of isolated portions is subjected to expansion processing, and the label of the isolated portion is assigned to the expanded portion obtained from each isolated portion, and the size of the diagnosis target is set. An expansion processing unit to be restored;
I have a,
The expansion processing unit restores the size of the diagnostic object while forming a boundary at the overlapping portion of the expanded portions that overlap each other by being expanded.
An ultrasonic image processing apparatus. In the image data obtained from the ultrasonic echo data, by performing a contraction process on the diagnosis target, a contraction processing unit that separates the diagnosis target into a plurality of isolated parts,
A labeling processing unit that assigns different labels to each of the plurality of isolated parts by performing a labeling process on the plurality of isolated parts in the image data that has undergone the shrinkage processing;
In the image data subjected to the labeling process, each of a plurality of isolated portions is subjected to expansion processing, and the label of the isolated portion is assigned to the expanded portion obtained from each isolated portion, and the size of the diagnosis target is set. An expansion processing unit to be restored;
I have a,
The diagnostic object is a plurality of follicles in close contact with each other,
The plurality of isolated portions are a plurality of follicles separated from each other by the contraction process.
An ultrasonic image processing apparatus.

Images