JP2010019572A - Image processing device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To calculate a fractal dimension without depending on the difference in concentration between pixels of an image to be an object. <P>SOLUTION: This image processing device 1 includes: an original image storage part 11 for storing image data on a predetermined original image; a mesh processing part 133 which performs in mesh units, binarizing processing for partitioning into meshes a binarized image obtained by binarizing the original image, and for binarizing each mesh into white or black, as to a plurality of different mesh sizes respectively; a fractal dimension calculation part 16 for calculating a fractal dimension on the basis of the processing results in the processing part 133; and an output part 20 for outputting the fractal dimension. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to an image processing technique for calculating a fractal dimension.

  Japanese Patent Application Laid-Open No. 2004-133867 discloses a technique for performing image processing of cerebral blood flow to perform predetermined processing, calculating a fractal dimension, and evaluating the non-uniformity of the cerebral blood flow distribution. In this Patent Document 1, a SPECT (Single Photon Emission Computed Tomography) image is binarized using various threshold values of a predetermined ratio with respect to the maximum radioactive activity value of the whole brain, and a fractal dimension is obtained from the relationship between each binarized image and the threshold value. Is calculated.

In addition, in the examination of liver function using RI (Radio Isotope) medicine by the conventional method, a region of interest (ROI) is set in the liver and heart by an operator's manual operation from images showing the liver and other organs. Then, after preparing the time-radioactivity curve, it is necessary to determine the severity of liver damage by analyzing the disappearance of the RI drug in the blood, the index of liver accumulation, etc. from the count in the ROI.
JP 2002-28141 A

  In the method of Patent Document 1, since a fractal dimension is calculated using a density difference generated between pixels of a target image, good results may not always be obtained depending on conditions at the time of shooting. .

  In addition, in the conventional liver inspection, the operator manually designates the liver region, so there is a problem that the error due to this manual operation and the variation among operators cannot be avoided, and the inspection result is not stable. It was.

  Accordingly, an object of the present invention is to provide a technique for calculating a fractal dimension without depending on a density difference between pixels of an image to be processed.

  Another object of the present invention is to improve the accuracy of liver examination using images and reduce the burden on the subject.

  An image processing apparatus according to an embodiment of the present invention includes a storage unit that stores image data of a RI (Radio Isotope) image of a liver, the RI image of the liver is divided into meshes, and each mesh is binarized into white or black Mesh processing means for performing binarization processing in units of meshes for each of a plurality of different mesh sizes, fractal dimension calculation means for calculating a fractal dimension based on the processing result of the mesh processing means, and the fractal dimension calculation Output means for outputting the fractal dimension calculated by the means.

  In a preferred embodiment, the apparatus further comprises means for accepting selection of a predetermined area of the RI image of the liver, and binarization means for binarizing the image of the selected area to generate a binarized image. The mesh processing means may perform binarization processing in units of meshes using the binarized image.

  In a preferred embodiment, the binarization unit may binarize each pixel by comparing a value of each pixel of the RI image of the liver with a predetermined threshold value.

  In a preferred embodiment, the apparatus may further include a determination unit that determines a disease state based on the fractal dimension calculated by the fractal dimension calculation unit, and the output unit may output a determination result by the determination unit. .

  In a preferred embodiment, the mesh processing means performs a binarization process for each of the plurality of different mesh sizes, and then forms a contour of a closed region formed by a black mesh at each mesh size. The number of meshes may be counted, and the fractal dimension calculating unit may calculate the fractal dimension based on the number of meshes constituting the outline of the closed region for each mesh size.

  In a preferred embodiment, the mesh processing means performs binarization processing in units of mesh for each of the plurality of different mesh sizes, and then the number of meshes constituting a closed region formed by a black mesh at each mesh size. And the fractal dimension calculation means may calculate the fractal dimension based on the number of meshes constituting the closed region for each mesh size.

  The image processing apparatus according to an embodiment of the present invention may be an image processing apparatus that performs processing of an ultrasonic echo image. That is, this image processing apparatus includes a storage unit that stores image data of an ultrasonic echo image, binarization in units of meshes, in which the ultrasonic echo image is divided into meshes and each mesh is binarized into white or black. Mesh processing means for performing processing for each of a plurality of different mesh sizes, and fractal dimension calculation means for calculating a fractal dimension based on the processing result of the mesh processing means.

  In a preferred embodiment, the apparatus further comprises means for accepting selection of a predetermined area of the ultrasonic echo image, and binarization means for binarizing the image of the selected area to generate a binarized image. The mesh processing means may perform binarization processing in units of meshes using the binarized image.

  Hereinafter, an image processing apparatus according to an embodiment of the present invention will be described with reference to the drawings.

  The image processing apparatus according to the present embodiment will be described by taking as an example an apparatus having a function of performing processing on a medical image obtained by photographing a human body and supporting diagnosis based on the result.

  FIG. 1 is a functional configuration diagram of an image processing apparatus 1 according to the first embodiment.

  The image processing apparatus 1 according to the present embodiment includes, for example, a general-purpose computer system including a computer main body having a processor and a memory, an input device, a display device, and the like. These components or functions are realized, for example, by executing a computer program.

  The image processing apparatus 1 includes an original image data storage unit 11, an image processing unit 13, a binarized image data storage unit 14, and a count data storage unit 15 that stores count data that is a processing result in the image processing unit. A fractal dimension calculation unit 16 that calculates a fractal dimension based on count data, a pathological condition determination unit 17 that determines a pathological condition based on the fractal dimension, a classification pattern storage unit 19 that stores a pattern for pathological classification, and a fractal And an output unit 20 that outputs a fractal dimension calculated by the dimension calculation unit 16 and a pathological condition determination result by the pathological condition determination unit 17.

  The original image data storage unit 11 stores original image data to be processed by the image processing unit 13. For example, in the present embodiment, the original image data storage unit 11 stores an RI image obtained by photographing the liver using RI (Radio Isotope). In the following embodiments, a liver SPECT image will be described as an example of the RI image, but a liver PET image may be used. The original image may be a CT image other than the RI image, an MRI image, an X-ray image, an ultrasonic echo image (particularly, an ultrasonic echo image of the liver), or the like.

  The image processing unit 13 includes a region extraction unit 131 that extracts a target region of a processing target image, a binarization processing unit 132 that binarizes the extracted image, and a binarized image divided into meshes to obtain a black mesh And a mesh processing unit 133 that counts the number.

  The region extraction unit 131 acquires a SPECT image from the original image data storage unit 11 and extracts a region designated by the user. For example, when a user designates a region of interest in a SPECT image using a display device and an input device (not shown), the region extraction unit 131 extracts an image of the designated region.

  The binarization processing unit 132 binarizes the SPECT image of the region extracted by the region extraction unit 131 and stores it in the binarized image data storage unit 14. For example, the binarization processing unit 132 binarizes each pixel into a “white pixel” or a “black pixel” according to the pixel value. In the case of a SPECT image, each pixel value is a count value of RI (Radio Isotope) detected by the imaging device of the SPECT image, so that the binarization processing unit 132 uses, for example, a maximum count value as a binarization threshold value. One is selected from 30%, 40%, 50%, 60%, 70%, etc. The threshold for binarization is stored in the count data storage unit 15. In the case of an ultrasound echo image of the liver, each pixel value is a gray scale detected by the ultrasound echo image imaging device, so that the binarization processing unit 132 uses, for example, the maximum gray scale as the binarization threshold. One is selected from 30%, 40%, 50%, 60%, 70%, etc. of the value. The threshold for binarization is stored in the count data storage unit 15.

  The mesh processing unit 133 acquires the binarized image data stored in the binarized image data storage unit 14, divides the entire binarized image into meshes of a predetermined mesh size, and will be described below. Perform mesh processing.

  First, FIG. 2 shows how the mesh 6 of the binarized image 50 is divided by the mesh processing unit 133. The mesh processing unit 133 divides the binarized image 50 into a plurality of different mesh sizes, and performs the following processing for each mesh size. FIG. 2A illustrates a case where the screen is divided into vertical and horizontal 512 × 512 mesh (mesh size 512), and FIG. 2B illustrates a case where the screen is partitioned into vertical and horizontal 256 × 256 mesh (mesh size 256).

  3 and 4 are enlarged views when the binarized image 50 is divided into meshes.

  FIG. 3 is an enlarged view of the case of FIG. 2A, for example. In the binarized image 50, each pixel is binarized into a white pixel 52 or a black pixel 53. FIG. The binarized image 50 is divided into meshes 601 to 616 with a mesh size “512”. In the case of FIG. 3, one mesh includes 9 pixels.

  FIG. 4 is an enlarged view of FIG. 2B, for example, and the binarized image 50 is divided into meshes 651 to 654 with a mesh size “256”. In the case of FIG. 4, 36 pixels are included in one mesh.

  Based on the states of the pixels 52 and 53 in the meshes 601 to 616 and 651 to 654, which are divided into predetermined mesh sizes as described above, the mesh processing unit 133 performs the meshes 601 to 616 and 651 to 651, respectively. 654 is binarized to “white” or “black”. As a rule when binarizing the meshes 601 to 616, 651 to 654, for example, when the number of black pixels in the meshes 601 to 616, 651 to 654 is a predetermined number or more, the mesh is set to “black”. To do. Here, the predetermined number serving as the threshold may be, for example, “1”, or may be equal to or more than half of all the pixels included in the mesh.

  In this embodiment, when a majority of the pixels in each mesh is “black”, the mesh is “black”. If the mesh is binarized according to this, in the example of FIG. 3, the meshes 604, 607, 608, and 610 to 616 become “black”, and in the example of FIG. 4, the meshes 652, 653, and 654 become “black”. .

  Thus, after binarizing each mesh 6 to white or black, the mesh processing unit 133 counts the number of “black” meshes. There are the following two methods for counting the number of “black” meshes.

  A method for counting the number of black meshes will be described with reference to FIG. That is, as shown in FIG. 5A, the closed region 80 is formed by the “black” mesh in a state where the mesh binarization processing for performing binarization in units of the mesh 6 is completed. At this time, as shown in FIG. 5B, the first counting method extracts the contour 81 of the closed region 80 and counts the number of meshes constituting the contour 81. In the second counting method, as shown in FIG. 5C, the contour 81 of the closed region 80 and the inside thereof, that is, the total number of meshes of the closed region 80 are counted.

  The mesh processing unit 133 stores the count value of the closed region outline 81 and the entire closed region 80 in the count data storage unit 15 for each mesh size.

  FIG. 6 shows an example of the data structure of the count data storage unit 15.

  As shown in the figure, the count data storage unit 15 includes, as data items, a mesh size 151, a total number of meshes 152 at that mesh size, and binarization when binarized by the binarization processing unit 132 The threshold value 153, the count number 154 of the contour of the closed region counted by the mesh processing unit 133, and the count number 155 of the entire closed region are included.

  The fractal dimension calculation unit 16 refers to the count data storage unit 15 and calculates a fractal dimension. For example, as shown in FIG. 7, the fractal dimension calculation unit 16 displays a count log data storage unit in a logarithmic graph in which the natural logarithm of the mesh size is abscissa and the natural logarithm of the count number of the contour and the closed region is ordinate. Fifteen data are plotted to obtain a regression line, and the fractal dimension is calculated from the slope. In this embodiment, the fractal dimension is calculated for each of the closed region count and the contour count.

  Returning to FIG. 1, the classification pattern storage unit 19 stores a classification pattern for classifying a patient's pathology based on the fractal dimension and the patient's attributes. For example, depending on the combination of the patient's age, sex, physique, presence or absence of smoking, and the fractal dimension, the pathological condition of the site shown in the image (eg, tumor or inflammation, primary if it is a tumor, or metastasis) Or the like, etc.) is stored.

  The pathological condition determination unit 17 refers to the classification pattern storage unit 19 and determines the pathological condition of the part shown in the image based on the fractal dimension.

  The output unit 20 outputs at least one of the fractal dimension and the pathological condition determination result to a display device or printing device (not shown).

  FIG. 8 is a flowchart showing a processing procedure of the image processing apparatus 1 having the above configuration.

  First, a processing target image is acquired from the original image data storage unit 11, and the region extraction unit 131 extracts a target region according to an instruction from the user (S11).

  The binarization processing unit 132 binarizes the extracted target area using a predetermined threshold (S12).

  Next, the mesh processing unit 133 determines the mesh size and further binarizes each mesh obtained by dividing the binarized image into mesh sizes (S13). Then, the mesh processing unit 133 further counts the number of black meshes in the contour and the closed region for the binarized mesh image, and stores it in the count data storage unit 15 (S14).

  Here, the mesh processing unit 133 determines whether or not the counting process has been completed for all the predetermined mesh sizes (S15). Then, when the count process is not completed for all mesh sizes (S15: No), the process returns to step S13 to continue the process.

  When the processing is completed for all mesh sizes (S15: Yes), the fractal dimension calculation unit 16 refers to the count data storage unit 15 and calculates the fractal dimension (S16).

  The pathological condition determination unit 17 refers to the classification pattern stored in the classification pattern storage unit 19 based on the fractal dimension calculated in step S16, and determines a highly likely pathological condition (S17).

  Then, the output unit 20 outputs the fractal dimension calculated in step S16 and the pathological condition determination result in step S17 (S18).

  Accordingly, diagnosis support can be performed by analyzing image data and predicting a disease state based on the analysis result.

  Next, the configuration of the image processing apparatus 2 according to the second embodiment of the present invention will be described. In the present embodiment, differences from the first embodiment will be mainly described, and configurations or functions common to the first embodiment may be denoted by the same reference numerals and description thereof may be omitted.

  FIG. 9 is a functional configuration diagram of the image processing apparatus 2 according to the second embodiment.

  The image processing apparatus 2 according to the present embodiment includes, for example, a general-purpose computer system including a computer main body having a processor and a memory, an input device, a display device, and the like. These components or functions are realized, for example, by executing a computer program.

  The image processing apparatus 2 includes an original image data storage unit 11, an image processing unit 23, a count data storage unit 25 that stores count data that is a processing result in the image processing unit 23, and a fractal dimension based on the count data. A fractal dimension calculation unit 26 to calculate, a pathological condition determination unit 17 that classifies a pathological condition based on the fractal dimension, and a classification pattern storage unit 19 that stores a classification pattern are provided.

  In the present embodiment, the image processing unit 23 includes an area extraction unit 131 and a count processing unit 235.

  The count processing unit 235 according to the present embodiment binarizes the image of the region extracted by the region extraction unit 131 using a plurality of binarization threshold values, and counts the number of pixels at each threshold value. For example, the binarization threshold is set to 30%, 40%, 50%, 60%, 70%, etc. of the maximum count value, similarly to the binarization processing unit 132 of the first embodiment. In addition, the count processing unit 235 counts the number of pixels forming the outline of the closed region and the number of pixels in the entire closed region, as in the first embodiment.

  Then, the binarization threshold value related to the above-described binarization process and the count value at that time are respectively stored in the count data storage unit 25.

  FIG. 10 shows an example of the data structure of the count data storage unit 25.

  As shown in the figure, the count data storage unit 25 includes, as data items, a threshold value 253 when binarization processing is performed, a count number 254 of the contour of the closed region counted by the count processing unit 235, and the entire closed region. Includes a count of 255.

  The fractal dimension calculation unit 26 calculates the fractal dimension with reference to the count data storage unit 25 as in the first embodiment. At that time, the fractal dimension calculation unit 26 counts the logarithmic graph with the natural logarithm of the binarization threshold as the horizontal axis and the natural logarithm of the count number of the contour and the closed region as the vertical axis as shown in FIG. The data in the data storage unit 25 is plotted to obtain a regression line, and the fractal dimension is calculated from the slope.

  The processing procedure of the image processing apparatus 2 according to the present embodiment will be described using the flowchart shown in FIG.

  First, a processing target image is acquired from the original image data storage unit 11, and the region extraction unit 131 extracts a target region according to an instruction from the user (S21).

  The count processing unit 235 sets a binarization threshold and binarizes the extracted image of the target area (S22). Then, the count processing unit 235 counts the number of pixels serving as the outline of the black pixels constituting the closed region and the number of pixels of the entire closed region, and stores the counted number in the count data storage unit 25. (S23).

  Here, the count processing unit 235 determines whether or not the count processing has been completed for all of the predetermined binarization thresholds (S24). And when the count process is not completed about a binarization threshold value (S24: No), it returns to step S22 and continues a process.

  When the processing is completed for all the binarization threshold values (S24: Yes), the fractal dimension calculation unit 26 refers to the count data storage unit 25 and calculates the fractal dimension (S25).

  The pathological condition determination unit 17 refers to the classification pattern stored in the classification pattern storage unit 19 based on the fractal dimension calculated in step S25, and determines a highly likely pathological condition (S26).

  Then, the output unit 20 outputs the fractal dimension calculated in step S25 and the pathological condition determination result in step S26 (S27).

  Accordingly, diagnosis support can be performed by analyzing image data and predicting a disease state based on the analysis result.

  The above-described embodiments of the present invention are examples for explaining the present invention, and are not intended to limit the scope of the present invention only to those embodiments. Those skilled in the art can implement the present invention in various other modes without departing from the gist of the present invention.

  For example, in the first embodiment, the pixel is simply binarized and then binarized in units of mesh. However, binarization in units of pixels is omitted and binarization is performed directly from the original image in units of mesh. You may make it make it.

  According to the present invention, there is no variation among operators, the examination time can be shortened, and an accurate liver function examination can be performed. In addition, since the correlation with existing testing methods has been obtained, the existing data can be used effectively, making it possible to perform common functional evaluations such as statistical analysis between facilities, and to unify diagnostic criteria that eliminate differences between facilities It becomes possible.

It is a functional lineblock diagram of image processing device 1 concerning a 1st embodiment of the present invention. The method of dividing the mesh of the binarized image is shown. An enlarged view when a binarized image is divided into meshes is shown. An enlarged view when a binarized image is divided into meshes is shown. It is explanatory drawing of the counting method of the number of black meshes. An example of the data structure of the count data storage unit 15 is shown. It is explanatory drawing of a fractal dimension. 3 is a flowchart showing a processing procedure of the image processing apparatus 1. It is a functional block diagram of the image processing apparatus 2 which concerns on the 2nd Embodiment of this invention. An example of the data structure of the count data storage unit 25 is shown. It is explanatory drawing of a fractal dimension. 4 is a flowchart showing a processing procedure of the image processing apparatus 2.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1, 2 Image processing apparatus 6 Mesh 11 Original image data storage part 13, 23 Image processing part 14 Binary image data storage part 15, 25 Count data storage part 16, 26 Fractal dimension calculation part 17 Pathological condition determination part 19 Classification pattern storage Unit 20 output unit 50 binarized image 52 white pixel 53 black pixel 53 number of black pixels 80 closed region 81 contour 131 region extraction unit 132 binarization processing unit 133 mesh processing unit

Claims (8)

Storage means for storing image data of a RI (Radio Isotope) image of the liver;
Mesh processing means for dividing the RI image of the liver into meshes, and performing binarization processing in units of meshes for binarizing each mesh into white or black for each of a plurality of different mesh sizes;
A fractal dimension calculating means for calculating a fractal dimension based on the processing result of the mesh processing means;
An image processing apparatus comprising: output means for outputting the fractal dimension calculated by the fractal dimension calculation means. Means for accepting selection of a predetermined region of the RI image of the liver;
Binarization means for binarizing the image of the selected area to generate a binarized image; and
The image processing apparatus according to claim 1, wherein the mesh processing unit performs binarization processing in units of meshes using the binarized image.   The image processing apparatus according to claim 2, wherein the binarization unit binarizes each pixel by comparing a value of each pixel of the RI image of the liver with a predetermined threshold value. Based on the fractal dimension calculated by the fractal dimension calculating means, further comprising a determination means for determining a disease state
The image processing apparatus according to claim 1, wherein the output unit outputs a determination result by the determination unit. The mesh processing means performs binarization processing in units of mesh for each of the plurality of different mesh sizes, and then counts the number of meshes constituting the contour of the closed region formed by the black mesh at each mesh size. And
5. The image processing according to claim 1, wherein the fractal dimension calculating unit calculates the fractal dimension based on the number of meshes constituting the outline of the closed region for each mesh size. apparatus. The mesh processing means, for each of the plurality of different mesh sizes, after performing binarization processing in units of mesh, each count the number of meshes constituting a closed region formed by a black mesh at each mesh size,
The image processing apparatus according to claim 1, wherein the fractal dimension calculation unit calculates the fractal dimension based on the number of meshes constituting the closed region for each mesh size. Performing a binarization process for each of a plurality of different mesh sizes by dividing a RI (Radio Isotope) image of the liver into meshes and binarizing each mesh into white or black;
Calculating a fractal dimension based on the result of the binarization processing in units of mesh for each mesh size;
Outputting the calculated fractal dimension. A computer program for image processing,
Performing a binarization process for each of a plurality of different mesh sizes by dividing a RI (Radio Isotope) image of the liver into meshes and binarizing each mesh into white or black;
Calculating a fractal dimension based on the result of the binarization processing in units of mesh for each mesh size;
A computer program for image processing to perform the step of outputting the calculated fractal dimension.

Images