JP2007097671A - Apparatus and program for image diagnosis support - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an apparatus for image diagnosis support, etc., which enable diagnostic support using the detailed structure information of a diagnosis subject which is obtained from image data of high resolution. <P>SOLUTION: In a diagnostic image acquired from the image data of the high resolution, a parameter showing the relation between the trait of a profile in the detailed structure of the diagnosis subject and a disease is digitalized and the incidence region and growth manner of a carcinoma, etc., are distinguished from this. Thus, in the case of a lung cancer diagnosis, the small adenocarcinoma of a terminal bronchiole can be distinguished from the small adenocarcinoma of a pulmonary alveoli sector using evaluation about the relation between properties or lesions of a tumor and an existing pulmonary structure with mostly corresponding to a pathology image with a loupe and a cell type classification candidate is automatically extracted from the image findings of the small adenocarcinoma. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an image diagnosis support apparatus and an image diagnosis support program typified by computer-aided diagnosis (CAD) and the like, and in particular, provides image diagnosis support based on a composition form of a malignant tumor. is there.

  Computer-aided diagnosis is based on the premise that the doctor makes the final decision in the diagnosis, and presents the position of the lesion candidate detected from the input digital image and the quantitative results analyzed using the computer in an easy-to-understand manner. Is. This technique is used, for example, in lung cancer screening using chest X-rays or CT images for mass screening, breast cancer mammography automatic screening systems, gastric cancer detection systems, and the like.

  2. Description of the Related Art In recent years, the discovery of small lung cancer and the like by computer-aided diagnosis has increased rapidly with the increase in resolution of image diagnosis apparatuses typified by information processing techniques and X-ray computed tomography apparatuses. In the conventional computer-aided diagnosis, for example, the flow shown in FIG. 18 is executed. Cancer and nodule candidate regions are automatically extracted using threshold processing, pattern matching, a neural network, a filter, a shadow edge shape, and the like. Is being determined.

  Further, an image obtained from recent high-resolution image data has a resolution that substantially corresponds to a pathological image by a loupe. By using such high-resolution images for computer-aided diagnosis, it is possible to evaluate in detail the relationship between the properties and lesions of the tumor and the existing lung structure, and to clearly see the microstructure of the diagnosis target.

  However, the conventional computer-aided diagnosis has the following problems. In other words, the image data is a computer that actively uses these in spite of the high resolution until it can visualize the fine structure information (for example, information about bronchiole, alveolar duct, alveolar sac) of the diagnosis object There is no supportive diagnosis.

  The present invention has been made in view of the above circumstances, and provides an image diagnosis support apparatus and an image diagnosis support program capable of providing diagnosis support using fine structure information of a diagnosis target obtained from high-resolution image data. It is an object.

  In order to achieve the above object, the present invention takes the following measures.

  According to a first aspect of the present invention, there is provided a parameter calculation means for parameterizing a relationship between a feature of a shape in a fine structure and a disease based on an image relating to a diagnostic part acquired using a medical imaging device, and calculating a numerical value thereof. A determination unit that executes determination based on the calculated numerical value of the parameter to generate diagnosis support information related to the diagnosis site; and a display unit that displays the diagnosis support information related to the diagnosis site in a predetermined form. This is a diagnostic imaging support apparatus.

  A second aspect of the present invention is a parameter that causes a computer to parameterize a relationship between a feature of a shape in a fine structure and a disease based on an image relating to a diagnostic site acquired using a medical imaging device, and to calculate the numerical value thereof. A calculation function, a determination function that executes determination based on the calculated numerical value of the parameter, and generates diagnosis support information related to the diagnosis part; and a display function that displays diagnosis support information related to the diagnosis part in a predetermined form; Is a diagnostic imaging support program characterized by realizing the above.

  As described above, according to the present invention, it is possible to realize an image diagnosis support apparatus and an image diagnosis support program capable of providing diagnosis support using the fine structure information of a diagnosis target obtained from high resolution image data.

  Hereinafter, first to third embodiments of the present invention will be described with reference to the drawings. In the following description, components having substantially the same function and configuration are denoted by the same reference numerals, and redundant description will be given only when necessary.

(First embodiment)
FIG. 1 is a block diagram of an image processing apparatus 1 according to this embodiment. As shown in the figure, the image processing apparatus 1 includes an operation unit 10, a display unit 12, a transmission / reception unit 14, a control unit 16, a storage unit 18, an image generation unit 20, a preprocessing unit 22, a candidate area extraction unit 24, A parameter extraction unit 26 and a discrimination unit 28 are provided.

  The operation unit 10 includes a trackball, various switches, a mouse, a keyboard, and the like for incorporating various instructions, conditions, and the like from the operator into the apparatus 1.

  The display unit 12 displays an image, a report input screen, an input screen for performing a predetermined operation, a determination result (described later) executed in the image diagnosis support apparatus, and the like in a predetermined form.

  The transmission / reception unit 14 transmits / receives information that can be used for diagnosis to / from other devices and databases via the network.

  The control unit 16 dynamically or statically controls each unit constituting the diagnostic imaging support apparatus. In particular, the control unit 16 includes a display unit 12, an image generation unit 20, a preprocessing unit 22, a candidate region extraction unit 24, a parameter extraction unit 26, a determination unit 28, and the like in a diagnosis support function using fine structure information described later. Control all over.

  The storage unit 18 includes images acquired by various medical imaging devices (for example, X-ray computed tomography apparatus, magnetic resonance imaging apparatus, ultrasonic diagnostic apparatus, nuclear medicine diagnostic apparatus, X-ray diagnostic apparatus, mammography, etc.) A program for executing a diagnosis support function using structure information, reference image data used for discrimination processing, and the like are stored.

  The image generation unit 20 performs predetermined image processing according to the purpose of diagnosis on images acquired by various medical image devices based on instructions from an interpretation doctor or the like input from the operation unit 2. An image (diagnosis image) used for image diagnosis is generated by image processing in the image generation unit 20.

  The preprocessing unit 22 binarizes the pixel values and the like of the diagnostic image by threshold processing and performs labeling processing. In addition, the preprocessing unit 22 extracts the left and right lungs from the diagnostic image on which the labeling process has been performed, using bone pixel values. Further, the preprocessing unit 22 performs a filtering process using a median filter or the like on the diagnostic image in order to remove noise.

  The candidate area extraction unit 24 extracts lung field areas, abnormal shadow candidate areas, and the like by threshold processing or pattern matching.

  The parameter extraction unit 26 quantifies and calculates a parameter representing the relationship between the feature of the shape in the fine structure in the extracted abnormal shadow candidate region and the disease.

  Based on the parameter value extracted by the parameter extraction unit 26, the determination unit 28 determines a cancer occurrence site, a cancer growth mode, and the like.

(Diagnosis support function using fine structure information)
Next, a diagnosis support function using the fine structure information of the image diagnosis support apparatus 1 will be described. This function parameterizes the relationship between the feature of the shape in the microstructure resulting from the occurrence mechanism (the mechanism of the development process) and the disease, and based on the numerical value, determines the disease occurrence site and the cell type classification of the disease in the diagnosis target To do. In the following, in the present embodiment, for the sake of specific explanation, a case in which diagnosis of a lung region is supported using image data acquired by an X-ray computed tomography apparatus is taken as an example.

  FIG. 2 is a flowchart showing a flow of processing according to the diagnosis support function using the fine structure information (diagnosis support processing using the fine structure information). As shown in the figure, the diagnosis support processing is roughly divided into a high-definition image generation process, an abnormal shadow candidate region extraction process, a parameter extraction process, a generated lung field region / generation pattern determination process, and a determination result display process. be able to. Hereinafter, the contents of each process will be described.

[High-definition image generation process: Step S1]
First, a high-definition image is generated based on image data acquired by an X-ray computed tomography apparatus (step S1). In particular, when acquiring data for generating a high-definition image having a resolution on the order of μm, motion artifacts due to breathing of the subject and heart motion occur. In recent multi-detector CT, the imaging time of the lung field is short. Therefore, for breathing causes, it is sufficient to hold the breath for a short time, and body movement can be sufficiently ignored. On the other hand, motion artifacts caused by the motion of the heart can obtain an image with little motion artifacts by matching the centers of reconstruction in the second half of the diastole.

  In addition, when the electrocardiogram synchronized image reconstruction method is used to reduce motion artifacts, a half reconstruction method and a multi-sector reconstruction method can be used. The half reconstruction method uses 240 degrees (180 degrees + fan angle 60 degrees) data to perform reconstruction. Compared to helical interpolation, the end diastole image is closer to a still image and has fewer motion artifacts. On the other hand, the multi-sector reconstruction method independently collects helical scan and electrocardiogram data, and extracts only data of different view angles that fall in the same cardiac time phase from different detector cycles and different detector rows during reconstruction. Then, linear interpolation is performed, and an image is reconstructed by a conventional half reconstruction algorithm.

  By using these methods, it is possible to reduce motion artifacts that are problematic in high-definition image generation.

[Abnormal Shadow Candidate Area Extraction Process: Steps S2 to S4]
Next, binarization and labeling are performed by threshold processing based on the CT value, and the left and right lungs (lung field region) are extracted from the obtained region using the CT value of the bone. Further, a median filter is applied to the extracted lung field region in order to remove noise (step S2).

  Next, for the volume data of the pre-processed lung field area, take a mask area with a size of m × m × m pixel size, measure the volume with high CT value in the lung field, and the volume is If it is above a certain threshold, it is extracted as an abnormal shadow candidate region (steps S3 and S4).

  As another method for extracting abnormal shadow candidate regions, a mask region of m × m × m pixel size is taken and prepared in advance for volume data that has been preprocessed for lung field region extraction. In addition, pattern matching with a plurality of tumor reference volumes may be performed, and if the correlation value is equal to or greater than a certain threshold value, it may be extracted as a candidate region for an abnormal shadow.

[Parameter extraction process: step S5]
Next, the relationship between the feature of the shape and the disease in the fine structure in the abnormal shadow candidate region is parameterized (parameter extraction), and the numerical value is calculated (step S5).

  FIG. 3 is a diagram showing a flow of processing executed in the parameter extraction process. As shown in the figure, first, the bronchi, bronchiole, and alveolar duct in the candidate area are traced. That is, the region is binarized using a threshold value, and fine line processing is performed to obtain a trace line t (step S51).

  Next, each voxel value is converted to a value d based on air 0 (step S52). This conversion is performed according to the CT value shown in FIG. 4, for example.

  Next, the distance l from each voxel and the trace line t is calculated (step S53). From the anatomical point of view, it is understood that the distance value from the bronchiole epithelium is larger in the distance value from the bronchiole epithelium to the trace line and the distance value from the alveolar epithelium to the trace line. ing.

  Next, the average value of d * l and d * t is calculated (step S54). At this time, the average value of d * l is obtained by calculating d * l of each pixel and calculating the average of d * l in n × n × n voxels.

[Distinction process of lung region / generation pattern: step S6]
Next, based on the parameter value extracted by the parameter extraction unit 26, the lung region where the cancer has occurred, the growth pattern of the cancer, and the like are determined (step S6). In this method, discrimination is performed using an anatomical characteristic that the diameter of the terminal bronchiole of a healthy person is approximately 500 μm and the diameter of the alveoli is 200 to 300 μm.

  FIG. 5 is a diagram showing a flow of processing executed in a discrimination process such as a cancer occurrence site and a cancer growth mode. As shown in the figure, first, a threshold is set using the average value calculated in step S5, and it is determined whether it is a small adenocarcinoma of the terminal bronchiole or a small adenocarcinoma of the alveolar region (step S61). ).

  Next, the growth mode is determined by frequency analysis of the fine structure of the alveolar region (step S62).

  FIG. 6 is a diagram showing a replacement-grown tumor at the lung peripheral site, and FIG. 7 is a diagram showing a non-substituted growth tumor at the lung peripheral site. As can be seen from a comparison between FIG. 6 and FIG. Therefore, it can be determined that the growth is a replacement type when the amplitude value is higher than a certain value, which is a higher-order wavelength, and the non-replacement type growth when the amplitude width is a certain value or less.

  Note that the growth pattern of the alveolar region can also be determined by threshold processing using CT values. That is, in substitutional growth as shown in FIG. 6, tumor cells are present in the alveolar epithelium, and other portions are filled with air. The CT value of air is approximately -1000H.U. On the other hand, in the non-substitution type growth shown in FIG. 7, the cells are filled with tumor cells from the alveolar epithelium to the inside, and the value is about 50 HU. Therefore, a histogram related to the CT value is created, and if the accumulated voxel value or the average voxel value in the n × n × n voxel has a certain threshold width, it is a replacement type growth, and if it exceeds a certain threshold value, the non-replacement type It can be identified as a developing tumor. In the case of the lungs of a healthy person, since there is no tumor portion, almost the entire area is air. Therefore, the integrated voxel value or the average voxel value in the same region is below a certain threshold value.

  By the way, when the mucus is filled in the air portion in the replacement type growth, the image looks very similar to the non-substitution type growth. Therefore, in step S62, even the substitution type growth may be determined as the non-substitution type growth. There is. In this diagnosis support process, in order to prevent such a determination error, the growth style is determined by the following method different from step S62 (step S63).

  That is, there is a difference in the filling process of the mucous membrane in the alveolar region between the substituted type and the non-substituted type. In substitutional growth, mucus fills the entire alveolar region as shown in FIG. 8, whereas in non-substitutional growth, the alveolar epithelium gradually fills into the interior as shown in FIG. Contains some air. In this step, paying attention to this organizational difference, the candidate region is binarized, the distance between the air and the edge is obtained by distance conversion, and the calculated distance value is thresholded. Thereby, even if it is a case where mucus is filled in the air part in substitutional growth, it can be determined as substitutional growth.

[Distinction Result Display Process: Step S7]
The discrimination results obtained by the processes described above are displayed in a predetermined form (step S7).

  FIG. 10 is a diagram showing a display example of the discrimination result obtained in the discrimination process. As shown in the figure, the display unit 12 displays candidate areas and the areas used for discrimination in the candidate areas surrounded by different colors. When an area used for discrimination is clicked, an enlarged image (for example, a volume rendering image with a fine structure) of a fine structure with a fine structure is displayed. Furthermore, in step S6, the region determined to be the generated lung field site is displayed as a generation site candidate, and either substitutional growth or non-substitutional growth is displayed as a cell type classification candidate in a predetermined form.

  Each threshold value used in the diagnosis support process can be changed manually. The discrimination result that changes as the threshold value is changed is displayed on the display unit 12 together with the threshold value as needed. Along with these results, a histogram, frequency distribution, threshold value, etc. used for the analysis are also displayed. Furthermore, by clicking on the displayed lung field region candidates and growth mode candidate characters, it is possible to access the lung cancer database connected to the diagnostic imaging support apparatus via the network, It can be viewed in a separate window.

  According to the configuration described above, the following effects can be obtained.

  According to this diagnostic imaging support apparatus, in a diagnostic image obtained from high-resolution image data, the relationship between the feature of the shape in the fine structure resulting from the occurrence mechanism and the disease is parameterized, and the diagnosis target is based on the numerical value. The disease occurrence site and the cell type classification of the disease are determined. Therefore, in lung cancer diagnosis, it is possible to determine whether a small adenocarcinoma of the terminal bronchiole or a small adenocarcinoma of the alveolar region, and whether the growth is a replacement type or a non-substitution type. As a result, it is possible to contribute to provision of new diagnostic information, and for example, it is possible to support the selection of an operation method, a postoperative prediction, and an anticancer agent.

  Further, according to the image diagnosis support apparatus, the distance between the air and the edge is obtained by distance conversion, the calculated distance value is subjected to threshold processing, and the growth style is determined. Therefore, substitutional growth that is difficult to distinguish from non-substitutional growth can be distinguished with high accuracy.

  Furthermore, according to this image diagnosis support apparatus, the discrimination result is displayed in a predetermined form as an occurrence site candidate and a cell type classification candidate. Along with these results, a histogram, frequency distribution, threshold value, etc. used for the analysis are also displayed. Furthermore, the linked lung cancer database can be accessed by clicking the characters of the lung field region candidate and the growth form candidate. Therefore, the observer can easily visually recognize and understand the result of the diagnosis support process, and can use it for selection of an operation method, a postoperative prediction, and an anticancer agent.

(Second Embodiment)
Next, a second embodiment of the present invention will be described. In the present embodiment, the diagnosis support processing using the fine structure information described above is applied to high-resolution image data obtained from a breast imaging apparatus, and the pathological tissue classification of the breast is performed. In general, in the classification of the histopathological tissue of the breast, the type of tissue subtype is determined in relation to the tendency of intraductal progression and the likelihood of cancer residue, the malignancy of cancer cells, radiosensitivity, and the like. By this discrimination, the tissue subtype can be classified into comedo type and non-comedo type as shown in FIGS.

  FIG. 13 is a flowchart showing a flow of diagnosis support processing using the fine structure information according to the present embodiment. As shown in the figure, the diagnosis support processing according to the present embodiment is similar to the first embodiment in that a high-definition image generation process, abnormal shadow candidate area extraction process, parameter extraction process, discrimination process, and discrimination result display are displayed. It can be divided into processes. Hereinafter, the contents of each process will be described.

[High-definition image generation process: Step S21]
A high-definition image is generated based on the image data acquired by the breast imaging apparatus (step S21).

[Abnormal Shadow Candidate Area Extraction Process: Step S22]
Next, on the high-resolution image of the breast as shown in FIGS. 14 and 15, for example, select (extract) an area of interest (candidate area of abnormal shadow) on the image, such as a mass, calcification, or disorder of the structure. To do. This selection is executed according to the result of CAD that presents a portion that is predicted to have a high degree of malignancy, or is performed manually by an interpretation doctor or the like.

[Parameter extraction process: steps S23 and S24]
Next, a Fourier transform process is performed on the extracted region, and a Fourier transform image of the extracted region is generated. Further, Fourier transform processing is performed on the comedo type reference image and the non-comedo type reference image to generate a Fourier transform image of each reference image (step S23). Note that the X-direction and Y-direction sizes of the reference image are adjusted to the extracted area.

  Next, a correlation value between the Fourier transform image of the region and the Fourier transform image of each reference image is calculated (step S24).

[Distinction process: steps S25, S26]
Next, a type corresponding to a high value among the correlation values calculated in the parameter extraction process is determined as a tissue subtype (comedo or non-comedo) of the extracted region.

  As shown in FIG. 11, in the comedo type, the low frequency in the frequency space (k-space) is significantly higher than the high frequency. On the other hand, as shown in FIG. 12, in the non-comedo type, the high frequency is significantly higher than the low frequency in the frequency space (k-space). Therefore, the tissue subtype of the extracted region is similar to the frequency distribution with the higher correlation value.

[Distinction Result Display Process: Step S27]
The discrimination results obtained by the processes described above are displayed in a form similar to that shown in FIG. 10, for example (step S27).

  According to the configuration described above, effects similar to those of the first embodiment can be obtained also in breast cancer diagnosis. That is, in a diagnostic image obtained from high-resolution image data, a parameter representing the relationship between the feature of the shape of the microstructure to be diagnosed and the disease is quantified, and based on this, the breast tissue subtype is automatically determined. Can do. As a result, it is possible to contribute to provision of new diagnostic information, and for example, it is possible to support the selection of an operation method, a postoperative prediction, and an anticancer agent.

(Third embodiment)
Next, a third embodiment of the present invention will be described. In the present embodiment, the diagnosis support process using the fine structure information described above is applied to high-resolution image data obtained from a breast imaging apparatus, and the tissue morphology of breast cancer is determined. The types of breast cancer are non-invasive duCTal carcinoma, duCTal carcinoma in situ (DCIS) and noninvasive lobular carcinoma, lobular carcinoma. in situ; LCIS). As shown in FIG. 16 (a), non-invasive ductal carcinoma occurs in the epithelial layer of the ductal system and progresses laterally in the duct. On the other hand, as shown in FIG. 16 (b), non-invasive lobular carcinoma arises from the epithelial layer of the terminal gland of the lobule and fills the lobular gland. In addition, microcalcification of breast cancer is caused by the reaction of calcium with a denatured substance of cancer that grows in the breast duct and a part of secretion in the breast duct.
FIG. 17 is a flowchart showing a flow of diagnosis support processing using the fine structure information according to the present embodiment. As shown in the figure, the diagnosis support processing according to the present embodiment includes a high-definition image generation process, an abnormal shadow candidate area extraction process, a parameter extraction process, a discrimination process, and a discrimination result display process, as in the first embodiment. Can be divided into Hereinafter, the contents of each process will be described.

[High-definition image generation process: Step S31]
A high-definition image is generated based on the image data acquired by the mammography device (step S31).

[Abnormal Shadow Candidate Area Extraction Process: Step S32]
Next, on the generated high-resolution image of the breast, an area of interest (candidate area of an abnormal shadow) on the image, such as a tumor, calcification, or disorder of construction, is selected (extracted). This selection is executed according to the result of CAD that presents a portion that is predicted to have a high degree of malignancy, or is performed manually by an interpretation doctor or the like.

[Parameter extraction process: steps S33, S34, S35, S36]
Next, the calcified region in the selected region is segmented using a technique such as a region growing algorithm (step S33). The center-of-gravity position is calculated using the segmented tumor / calcification / construction disorder region pixel values as weights (step S34). In addition, the distance to the edge is calculated for each ε degree (ε is 0-180, for example) from the position of the center of gravity (step S35), and the dispersion value of each distance is a parameter indicating the relationship between the feature of the shape in the fine structure and the disease. And its value is calculated (step S36).

[Determination process: Steps S37 and S38]
Next, the type of breast cancer is determined based on the variance value calculated in the parameter extraction process (step S37). This determination is performed according to the following criteria, for example.

• The variance is very large in noninvasive ductal carcinoma.
・ In non-invasive lobular carcinoma, the variance value falls within a certain range.
・ Determine the threshold value of the variance value to determine whether it is breast cancer or lobular cancer.
[Distinction Result Display Process: Step S37]
The discrimination results obtained by the above-described processes are displayed in a form similar to, for example, FIG. 10 (step S37).

  According to the configuration described above, effects similar to those of the first embodiment can be obtained also in breast cancer diagnosis. That is, in a diagnostic image obtained from high-resolution image data, a parameter representing the relationship between a feature of a shape in a fine structure to be diagnosed and a disease is quantified, and based on this, a tissue form of breast cancer can be automatically determined. it can. As a result, it is possible to contribute to provision of new diagnostic information, and for example, it is possible to support the selection of an operation method, a postoperative prediction, and an anticancer agent.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the components without departing from the scope of the invention in the implementation stage. Specific examples of modifications are as follows.

  Each function according to the present embodiment can also be realized by installing a program for executing the processing in a computer such as a workstation and developing the program on a memory. At this time, a program capable of causing the computer to execute the technique is stored in a recording medium such as a magnetic disk (floppy (registered trademark) disk, hard disk, etc.), an optical disk (CD-ROM, DVD, etc.), or a semiconductor memory. It can also be distributed.

  (2) In each of the above embodiments, an image obtained by an X-ray computed tomography apparatus is used. However, the present invention is not limited to this, and an MR image acquired by another medical image diagnostic apparatus such as a magnetic resonance imaging apparatus may be used.

  In addition, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

  As described above, according to the present invention, it is possible to realize an image diagnosis support apparatus and an image diagnosis support program capable of providing diagnosis support using the fine structure information of a diagnosis target obtained from high resolution image data.

FIG. 1 is a block diagram of an image processing apparatus 1 according to the first embodiment. FIG. 2 is a flowchart showing a flow of processing according to the diagnosis support function using the fine structure information (diagnosis support processing using the fine structure information). FIG. 3 is a diagram showing a flow of processing executed in the parameter extraction process. FIG. 4 is a diagram showing CT values for each representative substance. FIG. 5 is a diagram showing a flow of processing executed in a discrimination process such as a cancer occurrence site and a cancer growth mode. FIG. 6 is a diagram showing a tumor with a replacement-type growth in a peripheral lung region. FIG. 7 is a diagram showing a non-replaceable growth tumor at the lung peripheral site. FIG. 8 is a diagram showing a tumor with a replacement-type growth at a lung peripheral site. FIG. 9 is a diagram for explaining substitutional growth of lung cancer. FIG. 10 is a diagram showing a display example of the discrimination result obtained in the discrimination process. FIG. 11 is a diagram for explaining a comedo type of breast cancer. FIG. 12 is a diagram for explaining a form of breast cancer with a non-comedo type. FIG. 13 is a flowchart showing a flow of a diagnosis support process using the fine structure information according to the second embodiment. FIG. 14 is a diagram illustrating an example of a high-resolution image of a breast acquired by mammography. FIG. 15 is a diagram illustrating an example of a high-resolution image of a breast acquired by mammography. FIG. 16 (a) is a diagram for explaining the form of noninvasive ductal carcinoma. FIG. 16 (b) is a diagram for explaining the form of noninvasive lobular carcinoma. FIG. 17 is a flowchart showing the flow of a diagnosis support process using fine structure information according to the third embodiment. FIG. 18 is a flowchart showing an example of the flow of conventional image diagnosis.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus, 10 ... Operation part, 12 ... Display part, 14 ... Transmission / reception part, 16 ... Control part, 18 ... Memory | storage part, 20 ... Image generation part, 22 ... Pre-processing part, 24 ... Candidate area extraction part, 26 ... Parameter extraction unit, 28 ... Discrimination unit

Claims (14)

Parameterizing means for parameterizing the relationship between the feature of the shape in the microstructure and the disease based on the image relating to the diagnostic region acquired using the medical imaging device, and calculating the numerical value thereof;
A determination unit that performs determination based on the calculated numerical value of the parameter, and generates diagnosis support information related to the diagnosis part;
Display means for displaying diagnosis support information related to the diagnosis part in a predetermined form;
An image diagnosis support apparatus comprising:   When the diagnosis site is the lung, the parameter calculation means includes a length of a trace line obtained by tracing the bronchus, bronchiole, and alveolar duct acquired based on an image relating to the diagnosis site, each pixel in a predetermined region, and 2. The image diagnosis support apparatus according to claim 1, wherein the length of the trace line and each pixel value in a predetermined region are used to parameterize the relationship between the feature of the shape in the fine structure and the disease.   The image diagnosis support apparatus according to claim 2, wherein the determination unit performs a histogram analysis on the pixel values in the predetermined area and sets a threshold value to determine a growth style.   The image diagnosis support apparatus according to claim 3, wherein the display unit displays at least one of a histogram related to pixel values in the predetermined region and the threshold value together with diagnosis support information related to the diagnostic region.   The image diagnosis support apparatus according to claim 2, wherein the determination unit performs a Fourier analysis in the predetermined region and sets a threshold value to determine a growth style.   6. The image according to claim 5, wherein the display means displays at least one of a frequency image distribution in the predetermined region and the threshold obtained by the Fourier analysis together with diagnosis support information related to the diagnosis part. Diagnosis support device.   The image diagnosis support apparatus according to claim 2, wherein the determination unit performs distance conversion in the predetermined area and sets a threshold value to determine a growth style. Further comprising changing means for changing the threshold,
The discriminating means discriminates a growth style using the changed threshold value;
An image diagnosis support apparatus according to any one of claims 3 to 7, The parameter calculation means calculates an average value related to a product of the length of each pixel and the trace line and each pixel value in the predetermined region,
The discriminating means discriminates the cell type classification of the diagnostic site using the average value;
The image diagnosis support apparatus according to claim 1, wherein the image diagnosis support apparatus is an image diagnosis support apparatus.   When the diagnostic site is a breast, the parameter calculation means uses a correlation value between a Fourier transform related to a predetermined region of the image related to the diagnostic site and a reference image, and calculates the feature of the shape and the disease in the microstructure. 2. The diagnostic imaging support apparatus according to claim 1, wherein the relationship is parameterized.   When the diagnostic site is a breast, the parameter calculation means uses the variance value of the center of gravity position of the predetermined region of the image related to the diagnostic site and the distance to the edge for each ε degree to determine the feature of the shape in the microstructure The image diagnosis support apparatus according to claim 1, wherein a relationship between a disease and a disease is parameterized.   The image diagnosis support apparatus according to claim 1, wherein the display unit displays an enlarged image or a volume rendering image related to the predetermined area. Further comprising transmission / reception means for acquiring information relating to the determination from within the database via a network;
The display means displays information related to the obtained determination together with diagnosis support information related to the diagnosis part,
The image diagnosis support apparatus according to claim 1, wherein the image diagnosis support apparatus is an image diagnosis support apparatus. On the computer,
A parameter calculation function for parameterizing a relationship between a feature of a shape in a microstructure and a disease based on an image regarding a diagnostic site acquired using a medical imaging device, and calculating a numerical value thereof;
A discrimination function that performs discrimination based on the calculated numerical value of the parameter and generates diagnosis support information related to the diagnostic site;
A display function for displaying diagnosis support information related to the diagnosis part in a predetermined form;
An image diagnosis support program characterized by realizing the above.

Images