JP4878239B2 - Ultrasonic diagnostic equipment - Google Patents

Description

  The present invention relates to an ultrasonic diagnostic apparatus, and more particularly to processing of ultrasonic image data.

  A technique for extracting an image portion such as a tissue to be diagnosed from an ultrasonic image obtained by transmitting and receiving ultrasonic waves is known.

  For example, Patent Document 1 describes a technique for identifying an isolated region by performing a labeling process on an ultrasonic image and determining whether the isolated region is a noise region from the volume value of the identified isolated region. Yes. In general, by using a labeling process, it is possible to specify a collection of connected pixels by determining connectivity between pixels. A technique for determining connectivity between pixels is also described in Patent Document 2.

JP 2004-267484 A Japanese Patent No. 2895414

  By specifying an aggregate of a plurality of pixels using a labeling process, for example, the volume of the aggregate can be obtained. In the technique disclosed in Patent Document 1 described above, it is determined whether noise is generated using the volume value. However, for example, the size of the tissue may be diagnosed based on the volume value. In this way, the development to various applied technologies is expected by using the labeling process.

  However, in the labeling process, for example, data processing is executed for the entire region in the ultrasonic image, and therefore the load for the calculation cannot be ignored. In particular, when performing the labeling process step by step, for example, when performing a temporary labeling process for attaching a temporary label and a relabeling process for converting a temporary label into a final label, the entire area in the ultrasonic image As a result, the processing for data is repeatedly executed, which further increases the load for computation and increases the processing time.

  The present invention has been made in such a background, and an object thereof is to provide an improved technique for labeling processing for an ultrasonic image.

  In order to achieve the above object, an ultrasonic diagnostic apparatus according to a preferred aspect of the present invention is an ultrasonic diagnostic apparatus that processes image data of an ultrasonic image obtained by transmitting and receiving ultrasonic waves, A pixel classifying unit that classifies a plurality of pixels constituting an image into a plurality of types, a labeling processing unit that associates a temporary label with a pixel group that is a local cluster of a plurality of pixels classified into the same type, a plurality of types Based on a label relation table indicating a connection relation between a plurality of temporary labels corresponding to a pixel group, the table analysis means for obtaining a true label obtained by integrating a plurality of temporary labels in a connected state, thereby, A true label corresponding to an aggregate of a plurality of pixels that are classified and connected to the same type is obtained.

  In the above aspect, the image data of the ultrasonic image is, for example, a plurality of pixel data in a two-dimensional plane or a plurality of voxel data in a three-dimensional space. According to the above aspect, since the true label is obtained based on the label relation table, for example, it is not necessary to repeatedly execute the labeling process in order to obtain the true label, and the load of the arithmetic process is reduced.

  In a preferred aspect, the labeling processing unit scans a window that designates a target pixel and its neighboring pixels in an ultrasonic image, and locally detects a plurality of pixels based on a connection relationship between the target pixel and neighboring pixels in the window. A pixel group that is a cluster is obtained, and a label relationship table that indicates a connection relationship between the temporary labels is obtained from temporary labels for a plurality of pixels in the window.

  In a desirable mode, the table analysis means obtains the volume of the aggregate corresponding to the true label based on the label relation table showing the connection relation between the temporary labels and the volume for each temporary label.

  In order to achieve the above object, an image processing method according to a preferred aspect of the present invention is a method for processing image data of an ultrasound image obtained by transmitting and receiving ultrasound, A pixel classifying step for classifying a plurality of constituting pixels into a plurality of types, a labeling process for associating a temporary label with a pixel group that is a local cluster of a plurality of pixels classified into the same type, and a plurality of pixel groups And a table analysis step for obtaining a true label obtained by integrating a plurality of temporary labels in a connected state based on a label relation table indicating a connection relation between a plurality of temporary labels corresponding to A true label corresponding to an aggregate of a plurality of pixels in a classified and connected state is obtained.

  In the image processing method of the above aspect, for example, an ultrasonic diagnostic apparatus is executed in the apparatus. Or the computer which read the program according to the image processing method of the said aspect may perform the said image processing method.

  According to the present invention, an improved technique of labeling processing for ultrasonic images is provided. Thereby, for example, it is not necessary to repeatedly execute the labeling process to obtain the true label, and the load of the arithmetic process is reduced.

  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 transmission / reception unit 12 transmits / receives an ultrasonic wave via the probe 10 in a space including the target tissue. Then, echo data is acquired for each time phase from the space including the target tissue and stored in the three-dimensional data memory (1) 14.

  The probe 10 may be a probe for a three-dimensional ultrasound image that acquires echo data from a plurality of voxels that form a three-dimensional space, or a two-dimensional ultrasound that acquires echo data from a plurality of pixels that form a two-dimensional plane. An image probe may be used. In the following description, voxels and pixels that are structural units for forming a three-dimensional or two-dimensional ultrasonic image are referred to as pixels.

  The noise removal processing unit 16 performs noise removal processing on echo data (ultrasound image data) of each pixel read from the three-dimensional data memory (1) 14. For example, pixels that are spatially isolated and have different luminance values are determined as noise, and the noise is removed by performing processing such as converting the luminance value.

  The binarization processing unit 18 is composed of a comparator and the like, binarizes the ultrasonic image data from which noise has been removed based on a predetermined threshold, and each pixel in the ultrasonic image has a different luminance value. A binarized image is formed by classifying into different types of pixels. For example, a high-luminance pixel is extracted to form a binary image with the low-luminance pixel as the background.

  The region extraction processing block 20 performs a labeling process on the binarized image, and forms a label image in which a set of a plurality of pixels corresponding to the target tissue is labeled from within the binarized image, for example. The image data is stored in the three-dimensional data memory (2) 40. The region extraction processing block 20 obtains a volume value of an aggregate of a plurality of pixels corresponding to the target tissue, for example, and outputs it to the display image forming unit 44. A specific configuration and processing contents in the region extraction processing block 20 will be described in detail later.

  The selector 42 selects the original image data stored in the three-dimensional data memory (1) 14 or the label image data stored in the three-dimensional data memory (2) 40 according to an instruction from the user, and displays the display image. Output to the forming unit 44.

  The display image forming unit 44 forms a display image based on the original image data or the labeling image data. For example, if the original image data or the labeling image data is 3D image data, the display image forming unit 44 performs a volume rendering process or the like on the 3D image data to form a display image.

  The display image formed by the display image forming unit 44 is displayed on the monitor 46. Note that the display image forming unit 44 may display the volume value obtained in the region extraction processing block 20 as a numerical value or a graph.

  FIG. 2 is a functional block diagram showing the internal configuration of the region extraction processing block (reference numeral 20 in FIG. 1). FIG. 3 is an explanatory diagram of image processing in the region extraction processing block. Image processing by the region extraction processing block will be described with reference to FIGS.

  The provisional labeling processing unit 22 shown in FIG. 2 performs provisional labeling processing on the binarized image based on the binarized image data output from the binarization processing unit (reference numeral 18 in FIG. 1). In addition, a label relation table 32 indicating the connection relation between the temporary labels is generated in the process of the temporary labeling process.

  As shown in FIG. 3, the binarized image (a) obtained by the binarization process is classified into two types of pixels, for example, a plurality of high-luminance pixels 52 and other low-luminance background pixels 54. ing. A temporary label image (b) is obtained by applying a temporary labeling process to the binarized image (a).

  The provisional labeling process is a local block of a plurality of pixels based on a connection relationship between a pixel of interest and a neighboring pixel in the window 56 by scanning a window 56 that designates the pixel of interest and its neighboring pixel in the ultrasound image. This is realized by obtaining a pixel group.

  For example, as shown in FIG. 3, a window 56 having a pixel of interest in the lower right of a broken line frame corresponding to 4 pixels of 2 × 2 is used, and the window 56 is arranged for each pixel along the x direction. Move it step by step. Then, at each moving position, when the target pixel in the window 56 corresponds to the high-luminance pixel 52, a temporary label is attached to the target pixel. The temporary labels are attached in the order of 1, 2, 3,. However, if the neighboring pixel in the window 56 has already been assigned a temporary label, a temporary label having the same number as the temporary label attached to the neighboring pixel is attached to the target pixel.

  When the window 56 is moved along the x direction and the labeling process for pixels for one row is completed, the row is moved by one pixel along the y-axis direction, and the window 56 is further moved along the x direction. By repeating this series of movement processes, the window 56 is scanned (raster scanning) over the entire area of the ultrasonic image. In this way, a pixel group that is a local cluster of a plurality of pixels is obtained based on the connection relationship between the target pixel in the window 56 and the neighboring pixels at each movement position. That is, as shown in the temporary label image (b), the pixel group of the temporary label “1”, the pixel group of the temporary label “2”, the pixel group of the temporary label “3”, and the temporary label “4” A pixel group is determined.

  In addition, in the process of executing the temporary labeling process, connection information between temporary labels is obtained. That is, for example, when the pixel of the temporary label “1” and the pixel of the temporary label “2” exist in the window 56 at a certain moving position of the window 56, the temporary label “1” and the temporary label “2” are connected. The connection information (2, 1) indicating that the temporary label “1” and the temporary label “2” are connected is registered in the label relation table LRT (d).

  Incidentally, in the temporary label image (b) of FIG. 3, since the temporary label “1” and the temporary label “3” are also connected, the temporary label “1” and the temporary label “3” are connected. Is also registered in the label relation table LRT (d).

  Thus, the temporary label image and the label relation table 32 are obtained by the temporary labeling processing unit 22 shown in FIG. The image data of the temporary label image is stored in the temporary label image memory 24 and also supplied to the table analysis unit 28.

  The table analysis unit 28 shown in FIG. 2 executes a table analysis process based on the label relation table 32 to generate a label correlation table 34 and further generates a label conversion table 36. That is, as shown in FIG. 3, the label conversion table LTT (e) is obtained by the table analysis process from the label relation table LRT (d) indicating the connection relation between the temporary labels. The label conversion table LTT (e) is a table showing the correspondence between the temporary label and the true label, and is a table showing the true label obtained by integrating a plurality of temporary labels connected in the image.

  As shown in FIG. 3, it can be seen from the label relation table LRT (d) that the temporary label “1” and the temporary label “2” are connected, and the temporary label “1” and the temporary label “3”. "Is in a connected state. In the label conversion table LTT (e), the temporary labels “1”, “2”, and “3” that are connected to each other are integrated into the true label “1”. Since there is no other temporary label in the connected state, the temporary label “4” is changed to the true label “2” by changing only the label number.

  Then, as shown in FIG. 3, a relabeling process using the label conversion table LTT (e) is performed on the temporary label image (b) to obtain a true label image (c). The temporary labels “1”, “2”, and “3” that are connected to each other in the temporary label image (b) are integrated into the true label “1” in the true label image (c). The temporary label “4” in the temporary label image (b) corresponds to the true label “2” in the true label image (c).

  The extraction area determination unit 26 shown in FIG. 2 obtains a true label image based on the data of the temporary label image stored in the temporary label image memory 24 and the label conversion table 36, and corresponds to a predetermined true label from the true label image. The extracted region (for example, a predetermined target tissue) is determined, and the volume value is output.

  Thus, in the present embodiment, the real label image (c) is obtained by performing the relabeling process on the temporary label image (b). In the present embodiment, the calculation load is reduced by using the label conversion table LTT (e) during the relabeling process. That is, in the relabeling process, it is not necessary to perform the process for the entire area in the ultrasonic image as in the case of the temporary labeling process, so that the calculation load is reduced and the calculation time is shortened.

  In the present embodiment, the label conversion table LTT (e) used in the relabeling process is obtained by a relatively easy table analysis process. Next, the table analysis process will be described in detail.

  4 to 6 are diagrams for explaining table analysis processing by the table analysis unit (reference numeral 28 in FIG. 2). In the present embodiment, a label correlation table LCT is created based on the label relationship table LRT, and a label conversion table LTT is created via the label correlation table LCT.

  4A to 5E show how the label correlation table LCT is created from the label relationship table LRT.

  FIG. 4A shows the label relation table LRT. The label relation table LRT is a table in which connection information between temporary labels obtained in the process of executing the temporary labeling process is registered. In the label relation table LRT in FIG. 4A, the connection information (α, β) indicates that the temporary label α and the temporary label β are connected. For example, the connection information (4, 3) indicates that the temporary label 4 and the temporary label 3 are connected, and the connection information (3, 2) indicates that the temporary label 3 and the temporary label 2 are connected. It shows that it is connected.

  FIG. 4B shows an initial state of the label correlation table LCT. The label correlation table LCT is a table in which temporary labels are used as addresses and temporary labels of connection partners connected to the temporary labels are used as data. In the initial state, the data of each temporary label, that is, the temporary label (connection information) of the connection partner is cleared to zero.

  FIG. 4C shows a state in which the connection information from the head (number 0) to number 5 of the label relation table LRT is reflected in the label correlation table LCT. For example, as a result of reflecting the connection information (4, 3) of the label relation table LRT of FIG. 4A, the temporary label 3 of the connection partner is added to the address (temporary label) 4 of the label correlation table LCT of FIG. Is registered.

  Here, consider a case where the sixth connection information (8, 4) of the label relation table LRT of FIG. 4A is registered with respect to the label correlation table LCT in the state shown in FIG. In the state shown in FIG. 4C, the temporary label 7 of the connection partner is already written in the address (temporary label) 8 of the label correlation table LCT. In this case, a temporary label of data 0 is searched from addresses (temporary labels) 8 and 4, respectively.

  That is, as shown in FIG. 5D, a temporary label of data 0 is searched from each of the addresses 8 and 4 toward the smaller address, and addresses 5 and 1 are detected. Then, as shown in FIG. 5E, a process of writing the smaller address to the larger address is performed. That is, the smaller address 1 is written to the larger address 5. By processing the table in this way, it is possible to express that the temporary labels 8 and 4 are connected without overwriting existing connection information.

  Incidentally, the connection relationship is more complicated in the three-dimensional labeling process than in the two-dimensional process. Therefore, in order to shorten the later search time, for example, as shown in FIG. 5F, a process of writing the minimum value of the temporary label to the search start addresses 8 and 4 may be added.

  As described above, the label correlation table LCT is a table representing the connection state (correlation) of the temporary label connection, and the address is a temporary label and the data always has a smaller value than the temporary label of the address. Created in format. Further, in the label correlation table LCT, the data of the address that is the smallest temporary label belonging to the same connected component is left with the initial value “0” in order to clearly indicate that fact.

  FIGS. 6G and 6H show how the label conversion table LTT is created from the label correlation table LCT. The label conversion table LTT is a table in which temporary labels are used as addresses, and true labels corresponding to the temporary labels are used as data.

  First, the label correlation table LCT (FIG. 5E) is scanned from the minimum value side of the address (temporary label), and when the temporary label to which the corresponding address is connected is “0”, “ The number of occurrences of “0” is written as data of the corresponding address of the label conversion table LTT as a true label value. For example, in the label correlation table LCT of FIG. 5 (e), the data of the address (temporary label) 1 (the temporary label to be connected) is “0”, which is “0” appearing for the first time. “1”, which is the number of occurrences of “0”, is written as the true label value at the address (temporary label) 1 of the label conversion table LTT in (g).

  On the other hand, when the label correlation table LCT (FIG. 5E) is scanned from the minimum value side of the address (temporary label) and the temporary label to which the corresponding address is connected is other than “0”, the address of the temporary label to be connected. The data of the label conversion table LTT is read and written as the data of the label conversion table LTT at the corresponding address. For example, in the label correlation table LCT of FIG. 5 (e), the data at address (temporary label) 2 (the temporary label to be connected) is “1” and not “0”, and therefore, as shown in FIG. 6 (g). The data 1 corresponding to the temporary label 1 to be connected is written to the address (temporary label) 2 of the label conversion table LTT.

  Thus, by scanning the label correlation table LCT (FIG. 5E) from the minimum value side of the address (provisional label) to the maximum value of the address, the label conversion table LTT in the state shown in FIG. 6H is obtained. . As described above, the true label image (FIG. 3C) is obtained by the relabeling process using the finally obtained label conversion table LTT.

  Although a specific example of the table analysis processing in the present embodiment has been described with reference to FIGS. 4 to 6, the table analysis processing in the present embodiment can be generalized and described as follows. First, an algorithm for creating a label correlation table (LCT) from the label relation table (LRT) will be described.

  First, as preprocessing for LCT, contents (data) from address 1 to the address of the temporary label maximum value are initialized to 0. In this process, every time a temporary label is generated at the time of primary labeling (at the time of temporary labeling), it is only necessary to write 0 to the address of the temporary label of the LCT. Therefore, when the table analysis process (label integration process) starts. It is also possible to end it.

  Next, the combination of the temporary labels connected is read with reference to the LRT. Hereinafter, the larger one of the combinations of the connection information is referred to as LL, and the smaller one is referred to as LS. If the data of the LCT address LL is “0” (that is, the initial state in which nothing is written), the LS is written as it is. If the data in the LCT address LL is not “0” (already another connection information is already written), if this is overwritten, a part of the connection information will be lost. Consider writing this information at the temporary label minimum value in the connected component that has not been written.

  If it is found that the contents of the LCT address LL are not 0, the LCT connection information is first traced from the LL and LS addresses to the end (searching for the address where the data is 0). Find the label minimum. In addition, what arrived from LL is called LLMIN, and what arrived from LS is called LSMIN.

  When the search result is LLMIN> LSMIN, LSMIN is written to the LLMIN address, and when LLMIN <LSMIN, LLMIN is written to the LSMIN address. When LLMIN = LSMIN, no data is written to the LLMIN and LSMIN addresses, and the value remains 0.

  If the above processing is performed to the end of the valid data of the LRT, creation of the label correlation table (LCT) is completed. In this way, it is always possible to reach the minimum value from any temporary label in the connected component by following the chain of connection information.

  Next, an algorithm for creating a label conversion table (LTT) from the completed label correlation table (LCT) will be described. If it is desired to know only the true number of labels, the number of zeros may be counted up to the maximum value address of the temporary label with reference to the completed LCT. If an LTT for a true label image is to be obtained, the process is performed as follows. However, true labels shall be assigned in ascending order from 1.

  Here, the contents of the LCT address k are represented as LCT [k], and the contents of the LTT address k are represented as LTT [k]. First, as an initial state, k = 1 (k is an address to be processed), c = 1 (c is for obtaining the true label number by counting the number of 0s in the LCT), and the completed LCT Reference is started from address k = 1.

  If LCT [k] = 0, the value of c is written to the address k of the LTT, and c ← c + 1 (c is counted up). If LCT [k] ≠ 0, the contents (LTT [LCT [k]]) are read from the LTT address LCT [k] and written to the LTT address k. This may be performed up to the maximum value address of the temporary label.

  For example, when the temporary labels 1, 2, 2, and 3 are connected, the contents of the LCT are LCT [1] = 0, LCT [2] = 1, LCT [3] = 2. Think about converting to. Since k = 1, LCT [1] = 0, c = 1 is written to LTT [1]. Since LCT [2] ≠ 0 at k = 2, 1 is written to LTT [2] with reference to LTT [1]. Since LCT [3] ≠ 0 at k = 3, 1 is written to address 3 of the LTT with reference to LTT [2]. Since LTT [1] = LTT [2] = LTT [3] = 1 at the end of the processing so far, the temporary labels 1, 2 and 3 are integrated into the true label 1.

  As described above, in this embodiment, the true label is obtained by the temporary labeling process and the table analysis process. However, in this embodiment, the temporary label is further obtained in the table analysis process for obtaining the label conversion table LTT. The volume for each true label value can be obtained from the connection relationship between each volume and the temporary label.

  FIG. 7 is a diagram for explaining volume calculation processing. First, the table shown in FIG. 7A is created from the correspondence between the label correlation table LCT obtained in the process of the table analysis process and the volume for each temporary label.

  The volume for each temporary label can be obtained from, for example, the number of pixels constituting each temporary label and the volume corresponding to each pixel. The table analysis unit (reference numeral 28 in FIG. 2) registers the volume for each temporary label in the temporary label volume value calculation table (reference numeral 38 in FIG. 2).

  Next, as shown in FIG. 7B, the label correlation table LCT is scanned from the temporary label maximum value side, and the volume of each temporary label is added to the volume of the temporary label of the connection partner. However, the addition process is not performed when the temporary label of the connection partner is 0. In this way, when the scan is executed up to the minimum value of the label correlation table LCT, a volume in which the volume of each connected component is stored in the volume of the temporary label minimum value address is obtained after the scan is completed. That is, as shown in FIG. 7B, a value “450” obtained by adding the volumes of the temporary labels 4 to 6 is written to the address (temporary label) 4, and A value “360” obtained by adding the volumes of labels 1 to 3 is written.

  Then, as shown in FIG. 7C, scanning from the temporary label minimum value side and referring to the data of the address of the temporary label to be connected is performed in the same manner as converting the label correlation table LCT into the label conversion table LTT. Then, the value of the volume addition result is rewritten to the volume value of each connected component (the volume value of the connected component to which the temporary label belongs).

  As described above, from the temporary label value of each pixel, it is possible to create a table (FIG. 7C) in which not only the true label value of the connected component to which the pixel belongs but also the volume value can be known at the same time. Since this table creation process is only a table analysis process, the processing time can be greatly shortened compared with the case where the label image (whole ultrasonic image data) is scanned once to obtain the volume of each area. Can do.

  As described above, in the present embodiment, in addition to the raster scanning type temporary labeling process, the volume for each temporary label is stored in the table (temporary label volume value table) during the temporary labeling process. The volume of each region is obtained only by processing. Therefore, it is not necessary to scan the entire ultrasonic image data only to obtain the volume of each region, and for example, it becomes possible to obtain the volume values of a plurality of regions (objects) in a shorter time. The object can be extracted in a shorter time based on the volume value. In addition, a memory for storing a true label image can be omitted.

  Note that the image data handled in the present embodiment may be three-dimensional image data. Therefore, an extension to processing for three-dimensional image data will be described.

  FIG. 8 is a diagram for explaining the concept of three-dimensional labeling. FIG. 8A shows a conceptual diagram of labeling for two-dimensional image data. In the two-dimensional image data, a two-dimensional label image is formed from an input image (for example, a binarized image) by a temporary labeling process. In the two-dimensional label image, a label number is assigned to each aggregate of pixels obtained from a planar connection state between pixels. In FIG. 8A, label numbers are assigned to the three aggregates of labels 1 to 3.

  On the other hand, FIG. 8B shows a conceptual diagram of labeling for three-dimensional image data. In the three-dimensional image data, a label number is assigned to each set of pixels obtained from a three-dimensional connection state between pixels. In other words, in addition to the planar connection state, for example, the connection state within one frame, the connection state between the frames is considered, and three three-dimensional aggregates of labels 1 to 3 as shown in FIG. Is labeled.

  9 and 10 are diagrams for explaining the three-dimensional labeling process. Consider expanding two-dimensional labeling (connection with four surrounding pixels: four connections) to three-dimensional labeling (four connections on a plane and six connections above and below it).

  First, as shown in FIG. 9A, the three-dimensional data is divided into each frame (xy plane), provisional labeling processing is performed for each frame, and a plurality of frames are arranged in the frame arrangement order (along the z-axis direction). A temporary labeling process is performed on the. At that time, the temporary label of each frame starts from a value obtained by adding 1 to the total number of labels up to the immediately preceding frame. That is, the temporary labels are assigned in ascending order of 1, 2, 3,... From the first frame to the final frame.

  Then, as shown in FIG. 9B, paying attention to the temporary label TLd of the pixel of interest d and the temporary label TLu of the same address u in the immediately preceding frame, when TLd ≠ TLu ≠ 0, the combination (TLd, TLu ) As connection information between temporary labels. Thereby, in addition to the four connections in the plane (in the frame), connection information regarding the six connections including the upper and lower connections (between the frames) is obtained. If a table analysis is performed in the same manner as in the two-dimensional case, a three-dimensional label conversion table (LTT) can be obtained. Note that by referring to the temporary label image of the immediately preceding frame, processing for obtaining connection information between frames can be performed while performing temporary labeling of each frame.

  FIG. 10 shows a correspondence relationship between a three-dimensional temporary label image (a), a three-dimensional true label image (b), and connection information (c) between frames of the temporary label image.

  In the example shown in FIG. 10, in the temporary label image (a), the pixel groups of the temporary labels 1, 3, and 6 exist in each of the three frames. The connection information (c) between frames of the temporary label image includes connection information (3, 1) indicating that the temporary labels 1 and 3 are connected, and temporary labels 3 and 6 are connected. Connection information (6, 3) indicating the state is included. Therefore, in the true label image (b), the pixel groups of the temporary labels 1, 3, and 6 of the temporary label image (a) are combined into one to form an aggregate of the true label 1.

  Further, in the true label image (b), the pixel groups of the temporary labels 2, 5, and 8 of the temporary label image (a) are combined into one to form an aggregate of the true label 2, and the temporary label image The pixel groups of the temporary labels 4 and 7 in FIG.

  Note that the connection information between temporary labels between frames generally has a high frequency of occurrence of the same connection information successively in the raster scanning order. Therefore, in order to simplify the table, when the same connection information is generated as compared with the previous connection information, if the connection information is not adopted (not stored in the LRT), the number of connection information is reduced. This can greatly reduce the processing time required for table analysis.

  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. The present invention includes various modifications without departing from the essence thereof.

  For example, in the configuration shown in FIG. 1, the binarization processing unit 18 and the region extraction processing block 20 are configured in a computer, and binarization processing and region extraction processing (labeling processing and table analysis processing) are executed on the computer. You may implement | achieve the form to make. In the case of that form, for example, the computer operates based on a program corresponding to binarization processing and region extraction processing.

1 is a functional block diagram showing an overall configuration of an ultrasonic diagnostic apparatus according to the present invention. It is a functional block diagram which shows the internal structure of an area | region extraction process block. It is explanatory drawing of the image process in an area | region extraction process block. It is a figure for demonstrating the table analysis process by a table analysis part. It is a figure for demonstrating the table analysis process by a table analysis part. It is a figure for demonstrating the table analysis process by a table analysis part. It is a figure for demonstrating the calculation process of a volume. It is a figure for demonstrating the concept of three-dimensional labeling. It is a figure for demonstrating a three-dimensional labeling process. It is a figure for demonstrating a three-dimensional labeling process.

Explanation of symbols

  18 binarization processing unit, 20 region extraction processing block, 22 temporary labeling processing unit, 26 extraction region determination unit, 32 label relation table, 34 label correlation table, 36 label conversion table, 38 temporary label volume value calculation table.

Claims (4)

An ultrasonic diagnostic apparatus that processes image data of an ultrasonic image obtained by transmitting and receiving ultrasonic waves,
Pixel classification means for classifying a plurality of pixels constituting an ultrasonic image into a plurality of types;
Labeling processing means for associating a temporary label with a pixel group that is a local cluster of a plurality of pixels classified into the same type;
Based on a label relation table indicating a connection relation between a plurality of temporary labels corresponding to a plurality of pixel groups, a label correlation table indicating a chain state related to the connection between the temporary labels is created, thereby being in a connected state. A table analysis means for obtaining a true label obtained by integrating a plurality of temporary labels;
Have
Thereby, a true label corresponding to an aggregate of a plurality of pixels that are classified and connected to the same type is obtained.
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 1,
The table analysis means includes
From the label relationship table showing a plurality of connection information combining a large temporary label LL and a small temporary label LS that are connected to each other,
Creating the label correlation table using the value of the temporary label as an address and the value of the temporary label connected to the temporary label as data;
An ultrasonic diagnostic apparatus. The ultrasonic diagnostic apparatus according to claim 2,
The table analysis means includes
Read the combination of the temporary label LL and the temporary label LS with reference to each connection information shown in the label relation table,
In creating the label correlation table,
If the temporary label value is not written in the data of the address LL, the data of the address LL is set to LS,
If the value of the temporary label has already been written in the data of the address LL, the address LLMIN is maintained as it is, and the address LLMIN in which the data is in an initial state is searched from the address LL toward the smaller address. The address LSMIN in which the data is in the initial state is searched from the address LS toward the smaller address,
If LLMIN> LSMIN, the data at address LLMIN is set to LSMIN, if LLMIN <LSMIN, the data at address LSMIN is set to LLMIN, and if LLMIN = LSMIN, the data at address LLMIN = LSMIN is maintained in the initial state.
An ultrasonic diagnostic apparatus. A method of processing image data of an ultrasonic image obtained by transmitting and receiving ultrasonic waves,
A pixel classification step of classifying a plurality of pixels constituting the ultrasonic image into a plurality of types;
A labeling process that associates a temporary label with a pixel group that is a local cluster of a plurality of pixels classified into the same type;
Based on a label relation table indicating a connection relation between a plurality of temporary labels corresponding to a plurality of pixel groups, a label correlation table indicating a chain state related to the connection between the temporary labels is created, thereby being in a connected state. A table analysis process for obtaining a true label by integrating a plurality of temporary labels;
Including
Thereby, a true label corresponding to an aggregate of a plurality of pixels that are classified and connected to the same type is obtained.
An image processing method.

Images