JP5442542B2 - Pathological diagnosis support device, pathological diagnosis support method, control program for pathological diagnosis support, and recording medium recording the control program - Google Patents

Description

  The present invention relates to a technique for supporting pathological diagnosis by a computer system or the like.

  One pathological diagnosis is biopsy tissue diagnosis in which a part of an organ is collected from a patient as a specimen, and this is made into a tissue slice glass slide (preparation) and observed under a microscope for diagnosis. Such a diagnosis itself must be performed by a doctor, but the amount of work for that is enormous and the burden on the doctor is heavy. Therefore, in order to efficiently perform the observation work for diagnosis and reduce the burden on the doctor, various diagnosis support apparatuses using image processing technology by a computer system have been proposed (for example, Patent Documents 1 to 4). ). In these techniques, an abnormal part in a specimen is detected by extracting features peculiar to the lesion of the target organ from an image obtained by imaging the specimen.

Japanese Patent Laid-Open No. 10-197522 JP 2009-009290 A JP 2009-115598 A JP 2009-180539 A

  In order to efficiently perform diagnosis using this type of apparatus, a function for easily knowing how the cells and tissue findings having certain characteristics are distributed in the specimen is important. However, the pathological diagnosis support apparatus of the prior art has not been able to sufficiently meet such a demand. The main reason is as follows. That is, in the conventional technique, images captured in a specimen are individually evaluated to extract an image having characteristics common to a sample image of a known abnormal tissue, or the extracted image is caused by a lesion. The main focus is on determining whether or not. However, the conventional diagnosis support technology has room for further improvement in terms of improving the efficiency of diagnosis by displaying such extraction / determination results in a manner that makes the distribution in the specimen easy to understand. Was left.

  The present invention has been made in view of the above problems, and provides a technique capable of more efficiently performing pathological diagnosis (for example, stump judgment) in a technical field that supports a pathological diagnosis by a doctor using a computer system or the like. For the purpose.

In order to achieve the above object, the pathological diagnosis support apparatus according to the present invention obtains an original image obtained by imaging a specimen tissue, and a special part having characteristics specific to the lesion of the specimen tissue from the original image Special site detecting means for detecting eosinophilic crystalloid as the visual information for creating a composite image obtained by synthesizing the original image with visual information related to the position of the special site detected by the special site detecting means in the original image A processing means, and a display means for displaying the composite image , wherein the special site detection means includes, for the original image of the specimen tissue stained with hematoxylin and eosin, among structures existing in the original image, The saturation level is surrounded by a low saturation region that satisfies a predetermined low saturation condition, the intensity of the red component satisfies the predetermined high intensity condition, and the intensity of the blue component is predetermined Satisfy the low intensity conditions, and the area is characterized by detecting the structure in a predetermined area range as the special site.

Further, in order to achieve the above object, the pathological diagnosis support method according to the present invention has an image acquisition step of acquiring an original image obtained by imaging a sample tissue, and features unique to the lesion of the sample tissue from the original image. A special part detecting step for detecting eosinophilic crystalloid as a special part, and a visual information processing step for creating a composite image in which visual information related to the position of the detected special part in the original image is combined with the original image; A display step for displaying the composite image, and in the special site detection step, a saturation level of the original image of the specimen tissue stained with hematoxylin and eosin has a saturation level among the structures existing in the original image. Surrounded by a low saturation region that satisfies a predetermined low saturation condition, the intensity of the red component satisfies the predetermined high intensity condition, and the intensity of the blue component satisfies the predetermined low intensity condition. Plus, and the area is characterized by detecting the structure in a predetermined area range as the special site.

  In the invention configured as described above, a special part having a finding peculiar to the lesion of the specimen tissue is detected, and visual information regarding the position of the detected special part is synthesized and displayed on the original image. For this reason, the user can intuitively grasp the distribution state of the special part in the specimen tissue, and can perform pathological diagnosis efficiently. As the display accompanied by “visual information” here, for example, a part of the original image where the special part is detected is displayed in a color or brightness different from other parts, or the position coordinates of the special part in the original image are displayed. The method of doing etc. is possible.

  In these inventions, specific parameters used when detecting a special site include, for example, the saturation level and red color of an original image of a specimen tissue stained with hematoxylin and eosin (hereinafter referred to as “HE staining”). The intensity of the component, the intensity of the blue component, and the geometric features of the structures present in the original image can be used. By detecting a structure in the specimen tissue in which these parameters match or similar to the characteristics of the special part to be detected, the target special part can be detected with high accuracy.

  More specifically, for example, among the structures existing in the original image, the saturation level is surrounded by a low saturation region that satisfies a predetermined low saturation condition, and the intensity of the red component satisfies a predetermined high intensity condition. It is possible to detect as a special part a structure that satisfies this condition and has a blue component that satisfies a predetermined low-intensity condition and has an area in a predetermined area range.

  According to the detection under such conditions, it is possible to detect the eosinophilic crystalloid found in the cancerous tissue in the organ, particularly the gland duct tissue, as a special site with high probability. In HE staining, even if the cytoplasm or cavity of epithelial cells contained in gland duct tissue is stained, the saturation is low. From this, the low-saturation region includes duct tissue with a high probability. Therefore, the acidophilic crystalloid exists in the target image as a structure surrounded by the low saturation region.

  The eosinophilic crystalloid present in the gland duct tissue has a property of being needle-like and dyed red by HE staining and strongly brilliant, and has relatively little variation in shape and size. However, the region showing the property of being strongly dyed in red includes crystalloids similar to erythrocytes and amyloid bodies in addition to eosinophilic crystalloids. It cannot be excluded. On the other hand, compared to these similar substances, the acidophilic crystalloid has a property that it is difficult to be stained blue by HE staining, and the shape of the acidophilic crystalloid can be distinguished from other similar substances by experiments of the present inventors. confirmed.

  Therefore, in the present invention, the region surrounded by the low saturation region has a strong red component, but the blue component is weak, and has a predetermined geometric characteristic, for example, an area, a shape factor, etc. satisfy a predetermined condition. By extracting a thing, it becomes possible to distinguish and detect the eosinophilic crystalloid present in the target image from other substances. Such an effect is useful for effectively supporting a diagnosis for early detection of a disease in which eosinophilic crystalloid is generated as an indication of a lesion, for example, prostate cancer.

  In these inventions, for example, the original image is divided into a plurality of grid images, and the plurality of grid images are displayed in an arrangement based on the arrangement in the original image. The grid image may be displayed with visual information indicating that it is a lesion site. According to such a configuration, the user can overview the position where the special site is detected in the displayed image, and can easily grasp the distribution of the lesion site.

  In this case, for example, not only a grid image including a special part but also a grid image adjacent thereto may be provided with visual information indicating that the grid image is a lesion part. This is because the peripheral part of the special part is often a lesion (shows a lesion image). By using not only the grid image in which the special part is detected but also the adjacent grid image as the lesion part, it is possible to provide support that matches such a use.

  Further, for grid images other than the grid image determined as a lesion site, the grid image is classified into a plurality of classification categories based on the feature amount of each grid image, and the classified grid images are classified for each classification category. You may make it produce the synthesized image which attached | subjected the visual information of a different aspect. As the classification here, classification using a known pathological image as a teacher image, classification based on comparison between grid images, and the like are possible. In this way, not only visual information indicating a special part is displayed, but also a part similar to a known pathological image can be visually displayed, for example, and diagnosis can be more efficiently supported. .

  Further, as another method for displaying visual information, for example, a partial image of an original image and visual information representing the number of special parts detected in the partial image may be combined and displayed. Good. As a user, there are a case where the entire original image is displayed and the overall situation is to be grasped, and a case where a part of the original image is desired to be examined and a partial image that is a part of the original image is displayed. If the number of special parts detected in the partial image is displayed, the partial image to be examined can be easily selected, and the diagnosis efficiency can be further improved. In addition, providing quantitative information can improve the diagnostic accuracy of the user.

The control program according to the present invention, in order to achieve the above object, an image acquisition step of acquiring an original image of the captured analyte tissue, from the original image of the specimen tissues are hematoxylin and eosin stained, the original image Among the structures existing within the low saturation region where the saturation level satisfies a predetermined low saturation condition, the intensity of the red component satisfies the predetermined high intensity condition, and the intensity of the blue component is a predetermined low level. A special site detection step for detecting eosinophilic crystalloid as a special site having characteristics specific to the lesion of the specimen tissue by detecting a structure that satisfies the intensity condition and has an area in a predetermined area range ; A visual information processing step for creating a composite image by combining the original image with visual information related to the position of the detected special part in the original image, and a display for displaying the composite image It is characterized by executing the extent the computer. Furthermore, a recording medium according to the present invention is a computer-readable recording medium, and is characterized by recording a control program for supporting pathological diagnosis described above. According to these inventions, the pathological diagnosis support method according to the above invention can be executed by the computer, and the diagnosis by the user can be performed efficiently.

  According to the pathological diagnosis support technique according to the present invention, it is possible to efficiently perform pathological diagnosis (for example, stump judgment) by a user (diagnostic doctor).

It is a figure which shows one Embodiment of the pathological diagnosis assistance apparatus concerning this invention. It is a block diagram which shows the structure of this pathological diagnosis assistance apparatus. It is a flowchart which shows the diagnostic assistance process in this apparatus. It is a figure which shows the example of a target image. It is a flowchart which shows the 1st example of an acidophilic crystalloid detection process. It is a figure which shows typically the process in which the structure in an image is sorted in the process of FIG. It is a figure which shows the example of a model of a classifier. It is a flowchart which shows the classification process by a classifier. It is a flowchart which shows the display process in this embodiment. It is a figure which shows the example of the image displayed on a screen. It is a flowchart which shows the 2nd example of an acidophilic crystalloid detection process. It is a figure which shows typically the process in which the structure in an image is sorted in the process of FIG. It is a flowchart which shows the 3rd example of an acidophilic crystalloid detection process. It is a figure which shows typically the process in which the structure in an image is selected in the process of FIG.

  FIG. 1 is a diagram showing an embodiment of a pathological diagnosis support apparatus according to the present invention. The pathological diagnosis support apparatus 1 includes a host computer 10 and an imaging unit 20. The host computer 10 has a configuration and functions equivalent to, for example, a known personal computer or workstation terminal, and includes a processor unit 100 that executes various control programs, an input unit 150 that receives operation input from a user, A display unit 160 that displays various types of information, a disk drive 170 that reads out data by accessing an optical disk OD that is an external recording medium on which various types of data such as a control program and image data are recorded, and an interface 180 are provided. . The host computer 10 can be connected to a telecommunication line 30 such as a local LAN or the Internet via an interface 180.

  As the display unit 160, for example, a liquid crystal display can be used. Further, as the input unit 150, for example, a touch panel integrated with the display unit 160 may be used in addition to a pointing device such as a mouse or a keyboard. Further, the recording medium on which the control program and the image data are recorded is not limited to the optical disk, and any hard disk, memory card, USB memory, or the like can be used.

  The imaging unit 20 includes a microscope 202 to which a CCD camera 201 is attached. The microscope 202 forms an enlarged optical image of a preparation prepared from a specimen tissue collected from a patient on a light receiving unit of the CCD camera 201, and the CCD camera 201 captures the image and converts it into digital data to create a so-called virtual slide. . The specimen tissue is stained by an appropriate staining method according to the purpose of observation (for example, hematoxylin / eosin (HE) staining).

  The host computer 10 performs various kinds of image processing on the image of the specimen tissue imaged by the imaging unit 20 and converted into digital data (RGB data) according to the control program read from the optical disk OD, and displays the image on the display unit 160. Thus, the pathological diagnosis by the user (diagnostic doctor) is supported.

  FIG. 2 is a block diagram showing the configuration of the pathological diagnosis support apparatus 1. More specifically, each function block surrounded by a one-dot chain line in FIG. 2 is realized by software in the processor unit 100 by executing a control program read from the optical disk OD by the host computer 10. The contents of the diagnosis support processing realized by these functional blocks will be described in detail later. Here, the outline of each functional block will be briefly described.

  An image (target image) to be processed in the pathological diagnosis support apparatus 1 is, for example, an entire virtual slide obtained by imaging a sample tissue and converting it into data, or an image obtained by dividing the virtual slide into a plurality of images. Also, a plurality of images obtained by individually capturing a plurality of sample tissues collected from the same or a plurality of patients may be set as target images. When one virtual slide is divided into a plurality of target images, information that can specify the position of each target image in the original virtual slide so that the virtual slide can be restored in a later process (for example, coordinates Information) is desirably attached to each target image data.

  Here, a tissue collected from a patient's prostate is used as a sample tissue. RGB image data corresponding to the target image captured by the imaging unit 20 is provided to the grid image generation unit 101 and the special part detection unit 121. The operation of the special part detection unit 121 will be described in detail later.

  The grid image generation unit 101 divides the target image input from the imaging unit 20 into a plurality of grid images with a grid of an arbitrary size that is determined in advance or specified by the user. At this time, information (for example, coordinate information) indicating where each grid image is located in the original target image is given to the grid image data. The input receiving unit 102 functions as an interface for receiving operation inputs from the user via the input unit 150, and receives various operation inputs from the user. The input reception unit 102 transmits control signals corresponding to various operation instructions input by the user to each unit of the apparatus.

  The image feature amount calculation unit 122 has a function of calculating the feature amount of each created grid image. Note that the feature values of all grid images may be calculated, or the feature values of a predetermined grid image selected based on the detection result of the special part detection unit 121 may be calculated. As the feature amount of the image, for example, various things such as the brightness of the image, the texture information, the size of the region satisfying a predetermined condition can be used. What physical quantity is adopted as the feature quantity is determined by a preset parameter or a parameter determined according to an operation input from the user.

  The feature amount of the grid image calculated by the image feature amount calculation unit 122 is input to the classifier 123. The classifier 123 uses, as a reference image, a sample image stored in advance in the sample image storage unit 112, for example, an image in which features specific to cancerous cells appear, or from among captured target images Using the grid image selected by the user as a reference image, each grid image is classified into several classification categories. For the classification, a known classification technique, for example, a classification technique based on the Euclidean distance in the feature amount space can be applied.

  The classification results thus processed by the classifier 123 and the processing results by the special part detection unit 121 described later are accumulated in the classification result accumulation unit 109. The classification results are set by the set calculation unit 110 as necessary, and the result is processed by the display processing unit 111 to generate display data, which is sent to the display unit 160 to display the target image. Displayed on the part 160.

  FIG. 3 is a flowchart showing diagnosis support processing in this apparatus. FIG. 4 is a diagram showing an example of the target image. In the diagnosis support process, first, a target image to be diagnosed is acquired (step S101). As shown in FIG. 4A, the target image is a virtual slide image Iv obtained mainly by imaging the entire specimen tissue S. However, as shown by a broken line in FIG. The image Iv may be divided into a plurality of partial images Ip, and each partial image may be a target image. Further, the virtual slide image Iv can be received from the imaging unit 20, or may be received from another imaging device or storage via the electric communication line 30 or the like.

  From the target image thus acquired (here, the partial image Ip of the virtual slide image Iv), the eosinophilic crystalloid (crystallite) in the gland duct tissue that is characteristically seen in prostate cancer, particularly well-differentiated cancer, is used as a special part. This is detected by image processing performed by the special part detection unit 121. In FIG.4 (b), the code | symbol C has shown the eosinophilic crystalloid distributed in a gland duct tissue.

FIG. 5 is a flowchart showing a first example of the acidophilic crystalloid detection process. FIG. 6 is a diagram schematically showing the process of selecting structures in the image in this process. Eosinophilic crystalloids in prostate tissue
(1) It exists mainly in the duct area,
(2) The sizes are relatively uniform,
(3) It is easy to dye red by HE dyeing,
(4) It is hard to be dyed blue by HE dyeing.
No other tissue or substance in the prostate with all these characteristics is known.

  FIG. 6A shows an example of a target image (partial image Ip acquired from prostate tissue). In the prostate tissue, vascular R2 and glandular structures R4 are scattered in the interstitial region R1. Red blood cells R3 may be seen in the interstitial region R1 or vascular vessel R2. The internal space (gland duct region) of the duct structure R4 may include amyloid bodies R5 and eosinophilic crystalloid R6.

  Regarding the technique for detecting eosinophilic crystalloid in gastric cancer tissue based on the above properties (1), (2) and (3), the example in which the condition “the shape is square” is used for the property (2) is described above. It is described in Patent Document 4 (Japanese Unexamined Patent Application Publication No. 2009-180539). However, according to the present inventors' research on prostate tissue, other substances contained in the sample tissue, such as erythrocytes, amyloid bodies, etc. also have similar properties, and in determination based on these conditions, It is difficult to clearly distinguish between these similar substances and acidophilic crystalloids. On the other hand, the acidophilic crystalloid further has a property different from the above-mentioned other similar substances that it is difficult to stain blue by HE staining. Therefore, in order to detect eosinophilic crystalloid more accurately, it is desirable to make a determination by combining the above four properties.

  Therefore, in order to first extract a gland duct region that may contain eosinophilic crystalloid from the target image, a region in which pixels having a saturation level lower than a predetermined first threshold value are continuously annular in the target image. And the internal closed region surrounded by the ring is extracted as the candidate region 1 (step S201). When the HE-stained prostate tissue is imaged with transmitted light, the cytoplasm of the object of the gland duct region, that is, the epithelial cells constituting the gland duct region and the saturation of the cavity are relatively low. For this reason, the candidate area | region 1 which extracted the closed area | region (for example, cytoplasm and / or cavity) with low saturation as mentioned above has high possibility of including a gland duct area | region.

  Here, the saturation of the image is processed in units of pixels, and the candidate area 1 is determined as an aggregate of pixels that are determined to satisfy the condition of having a saturation lower than the first threshold (low saturation condition). . When an aggregate of pixels satisfying the above condition includes an area that does not satisfy the condition, the entire aggregate including the included area is set as the candidate area 1. That is, the candidate area 1 may include an area having a saturation level higher than the first threshold value. In addition, when there are a plurality of regions that satisfy the condition in the target image, each is set as a candidate region 1. The first threshold is set in advance based on the analysis result of a known image obtained by imaging the gland duct region.

  In FIG. 6B, hatching is added to a region where the saturation level is higher than the first threshold value. That is, a white portion C10 without hatching is a region having a saturation lower than the first threshold value. Then, as shown in FIG. 6C, a white region C11 that combines a region having a saturation lower than the first threshold and a region (higher saturation) included therein is a candidate region 1 described above. It corresponds to.

  In FIGS. 6B to 6F, in the partial image Ip that is the target image, an area extracted as meeting the specified condition is outlined, and an excluded area is hatched. Show. This also applies to FIGS. 12 and 14 described later.

  Next, for each pixel in the candidate area 1, an area composed of pixels having a relatively weak blue component is extracted as the candidate area 2 (step S202). The intensity of the blue component can be quantitatively expressed using the B luminance level or the color difference component of the RGB image signal. For example, the intensity of the blue component that satisfies a condition (low intensity condition) that is lower than a predetermined second threshold value. Extract it. As a result, a region having a relatively strong blue component, for example, a portion including red blood cells, is excluded from the candidate region 1, and the remaining region becomes the candidate region 2 indicated by reference numeral C12 in FIG.

  Next, an area composed of pixels having a relatively strong red component is extracted as candidate area 3 from candidate area 2 (step S203). The intensity of the red component can also be quantitatively expressed using the R luminance level or the color difference component of the RGB image signal. For example, the intensity of the red component that satisfies a condition (high intensity condition) that is higher than a predetermined third threshold value. Extract it. Thereby, as shown in FIG. 6E, a region C13 having a relatively strong red component is extracted from the candidate region 2 as the candidate region 3. Examples of structures that meet the conditions so far in prostate tissue include amyloid bodies and eosinophilic crystalloids.

  As described above, the acidophilic crystalloids are relatively uniform in shape and size. Therefore, the area of each candidate region 3 extracted in this way is calculated, and those within the predetermined area range are extracted (step S204). In this case, not the pixel unit but the entire area which is an aggregate of pixels is handled as one processing target. The area range is set in advance based on the image analysis result of the eosinophilic crystalloid contained in the known pathological image. Here, it is based on the knowledge that the variation in the size of the eosinophilic crystalloid is relatively small, but since the eosinophilic crystalloid is often a needle-like crystal that is long in a uniaxial direction, it depends on its shape. You may make it judge. Thereby, as shown in FIG. 6 (f), amyloid bodies are excluded from the candidate regions, and the remaining region C14 can be determined to be eosinophilic crystalloid with high accuracy (step S205).

  When the acidophilic crystalloid is detected in the target image in this way, information related to the acidophilic crystalloid from the processing results so far, for example, the size of each acidophilic crystalloid, the position in the target image, the detected number, etc. Is acquired (step S206). The obtained information is stored in the classification result accumulation unit 109. The above is the process for detecting eosinophilic crystalloid.

  Returning to FIG. 3, the description of the diagnosis support process will be continued. Subsequently, the target image (virtual slide image Iv or its partial image Ip) is divided into a plurality of grid images Ig by an arbitrarily sized grid as shown in FIG. 4B (step S103). From the grid image Ig divided in this way, the grid image containing the eosinophilic crystalloid and the grid image adjacent to the grid image are extracted from the positional information of the eosinophilic crystalloid, and these may be cancerated. The lesion is classified into a very high lesion (step S104). Subsequently, each grid image Ig that has not been determined to be a lesion site is classified by a classifier as described below.

  FIG. 7 is a diagram showing a model example of the classifier. In the classifier 123 in the present embodiment, as in the classifier 5 shown in FIG. 7A, a reference image and an input image are input, and it is determined whether they are similar to each other or dissimilar. Output. More specifically, the classifier 5 obtains both Euclidean distances in the feature amount space from the feature amounts of the reference image and the input image, and determines that the two are similar if the distance is smaller than a predetermined threshold. If the distance is greater than the threshold, it is determined that the distance is not similar.

  By replacing each grid image in the target image and sequentially inputting it to such a classifier 5, each grid image can be classified into those that are similar to the reference image and those that are not. The classification result changes each time the classification rule such as how to select the reference image and the feature amount to be used is changed. Therefore, the classifier 5 in this embodiment can be expressed as shown in FIG. In other words, N grid images # 1, # 2,..., #N are input to the classifier 5 as input images, and the grid images are classified by giving classification rules thereto. For example, a classification result A is obtained for a certain classification rule, and classification results B, C, and D are obtained for another classification rule. Various models such as a neural network and discriminant analysis have been proposed as such a classifier algorithm, and the present invention is not limited to the above, and such a known classifier (for example, JP-A-2003-317082) is proposed. It is possible to appropriately select and use those described in the publication.

  FIG. 8 is a flowchart showing the classification process by the classifier. First, the feature amount of each grid image Ig is calculated (step S301). When the feature amount of each grid image is calculated, the grid image is subsequently classified by the above-described classifier, and prior to that, a reference image serving as a judgment criterion for classification is set (step S302). As the reference image, either a known pathological image stored in advance in the sample image storage unit 112 or a grid image selected from the captured target image may be used. When a known pathological image is used, it is possible to determine the presence or absence of a portion similar to the known pathological image in the target image. In addition, when a grid image selected from the target image is used, a portion having a similar structure in the target image can be extracted, so that erroneous determination due to individual differences can be reduced. The reference image setting rule may be determined in advance, or may be specified by a user through an operation input from the input unit 150.

  Based on the feature amounts of the reference image selected in this way and each grid image in the target image, classification is performed by the classifier (step S303). The classification result is stored in the classification result accumulation unit 109 (step S304). The above is the classification process in this embodiment. In addition, although all the grid images are made into the classification | category object here, based on the detection result of the special site | part detection part 121, you may exclude from the classification | category object the grid image from which the eosinophilic crystalloid was already detected. This is because it can be determined as a lesion by the presence of eosinophilic crystalloid without classification.

  Returning to FIG. 3 again, the description of the diagnosis support process will be continued. As a result of the above processing (acidophilic crystalloid detection processing, classification processing), each grid image has a lesion site determined to be cancerous due to detection of acidophilic crystalloid, and acidophilic crystalloid is detected. Although it was not, it is classified into several categories such as a part containing another abnormal element and a normal part. Next, a specific method for effectively supporting the diagnosis by the doctor by reflecting the classification result thus obtained on the display screen will be described.

  In this embodiment, as described above, the grid image is classified based on the detection result of the eosinophilic crystalloid and the grid image is classified by the classifier, and the classification result is accumulated and recorded in the classification result accumulation unit 109. These classification results are displayed alone or on the display unit 160 as a composite image, for example, in combination as appropriate according to the user's request, thereby improving the efficiency of diagnosis by the user. In order to display a combination of a plurality of classification results, a set operation of the classification results (step S106) is performed. Here, a plurality of classification results can be displayed simultaneously by obtaining a union or intersection of various classification results.

  FIG. 9 is a flowchart showing display processing in this embodiment. FIG. 10 shows an example of an image displayed on the screen. In the display processing unit 111, each grid image is color-coded based on the classification result (step S501). This color coding can be performed, for example, such that the grid image determined to be a lesion site is red and the other grid images have the same color as the original image. Thereby, visual information reflecting the classification result is attached to each grid image. Note that the visual information is not limited to such color coding, and as described above, for example, various information such as light and dark for each grid image and a frame attached to the image can be used.

Next, by arranging each grid image based on the coordinate information in the original target image attached to each grid image, the target image is reconstructed as a partial image Ip of the virtual slide image Iv (step). S502). In this embodiment, the user is allowed to select whether to display the entire virtual slide image or a part thereof. The selection content is judged (step S503), and if the entire display is not selected, the partial image Ip obtained as described above is displayed on the screen of the display unit 160 (step S506).
The partial image Ip reconstructed and displayed in this way is combined with the virtual slide image captured by the imaging unit 20 or a part thereof, with the color coding according to the detection result by the special part detection unit 121 and the classification result by the classifier 123. Is displayed. Therefore, the user can grasp at a glance how cells and tissues having predetermined characteristics are distributed in the displayed virtual slide image.

  The virtual slide image Iv is composed of a plurality of partial images Ip. Therefore, when the entire display is selected, the virtual slide image Iv is reproduced by arranging the partial images constituting the virtual slide image Iv based on the coordinate information indicating the position in the virtual slide image Iv. Configure (step S504). Then, reduced image data is created so that the entire reconstructed virtual slide image Iv fits on the screen (not shown) of the display unit 160 (step S505), and an image corresponding to the image data is displayed (step S505). S506). As a result, the entire virtual slide image Iv is displayed on the screen of the display unit 160.

  At this time, since each grid image constituting the virtual slide image Iv reflects the classification result by color coding or the like, the user can see how the cells and tissues corresponding to the classification rule specified earlier are displayed in the virtual slide. You can see at a glance whether it is distributed. Further, the image can be enlarged or reduced by switching between the whole image and the partial image as necessary. Further, by switching the partial images to be enlarged and displayed, it is possible to perform observation while changing the position in the virtual slide.

  When the display of only the classification result based on the eosinophilic crystalloid detection result is obtained, as shown in FIG. 10A, the grid image in which the eosinophilic crystalloid C is detected and the grid image adjacent thereto are other It is displayed with visual information different from the grid image (here, darker than other grid images). When the display of the union of the acidophilic crystalloid detection result and the classification result by the classifier is obtained, as shown in FIG. 10B, the grid image in which the acidophilic crystalloid C is detected and adjacent thereto The grid images classified into several classification categories by the classifier and the classifier are displayed in a state of being separately colored with different colors (or densities). FIG. 10 shows the display contents when the display of the partial image Ip is selected by the user.

  Further, as indicated by a symbol Tx in FIG. 10C, for example, information related to the position and the number of detected eosinophilic crystalloids may be displayed in text together with an image obtained by superimposing the original image and the classification result. . In this case, when the partial image Ip is selected as a display target, the number of eosinophilic crystalloids detected in the currently displayed partial image Ip may be displayed in text, or in the entire virtual slide image Iv. The number of detected acidophilic crystalloids may be displayed together. By displaying such quantitative information together, the user can more intuitively grasp the distribution status of the eosinophilic crystalloid in the virtual slide image Iv, and the diagnosis can be performed efficiently.

  Although the case where the partial image Ip is displayed has been described here, the same applies to the case where the entire virtual slide image Iv is displayed.

  As described above, in this embodiment, the virtual slide image obtained by imaging the specimen tissue is image-processed, and a special site characteristic to the lesion of the tissue (here, eosinophilic crystalloid in prostate cancer) is detected. The detected position is displayed together with the original image so that it can be seen visually. Specifically, the original image is divided into a plurality of grid images, and among them, the grid image including the acidophilic crystalloid is displayed with visual information (color, shading, etc.) different from the other grid images. By doing so, the user (diagnostic doctor) does not need to find the special site by visual observation of the image, and can intuitively grasp the distribution of the special site in the sample tissue. As a result, accurate diagnosis can be efficiently performed with a small amount of work, and the burden on the diagnostician is greatly reduced.

  In addition, for grid images other than the grid image in which the special part is detected, an abnormal part can be detected by performing classification using a known pathological image or a grid image selected from the specimen tissue as a reference image. In addition, by displaying the classification result superimposed on the original image and the detection result of the special part, various diagnoses by the user can be effectively supported.

  Further, when detecting a special part, a region having all the properties (1) to (4) described above is extracted from the target image. By doing in this way, it becomes possible to detect the eosinophilic crystalloid in the gland duct tissue accurately from other similar substances. Thereby, for example, well-differentiated cancer in the prostate in which eosinophilic crystalloid is a sign of lesion can be easily found, and the work efficiency for diagnosis can be greatly improved.

  As described above, in this embodiment, the imaging unit 20 functions as the “image acquisition unit” of the present invention, and the display unit 160 functions as the “display unit” of the present invention. In this embodiment, the special part detecting unit 121 and the display processing unit 111 function as the “special part detecting unit” and the “visual information processing unit” of the present invention, respectively. The classifier 123 functions as the “classification unit” of the present invention. Therefore, in this embodiment, the processor unit 100 also has functions as “special part detecting means”, “visual information processing means”, and “classification means” of the present invention. Furthermore, in this embodiment, the virtual slide image Iv corresponds to the “original image” of the present invention.

  In the pathological diagnosis support process (FIG. 3) of the above embodiment, step S101 corresponds to the “image acquisition process” of the present invention, and step S102 corresponds to the “special part detection process” of the present invention. . Steps S103, S104 and S106 correspond to the “visual information processing step” of the present invention. Further, step S105 corresponds to the “classification process” of the present invention, and step S107 corresponds to the “display process” of the present invention.

  The present invention is not limited to the above-described embodiment, and various modifications other than those described above can be made without departing from the spirit of the present invention. For example, the detection of eosinophilic crystalloid based on the above properties (1) to (4) is not limited to the above-described mode, and can be performed as follows, for example.

  FIG. 11 is a flowchart showing a second example of the acidophilic crystalloid detection process. FIG. 12 is a diagram schematically showing the process of selecting structures in the image in this process. In this example, based on the same conditions as in the above example, a closed region with a low saturation level, a region with a weak blue component, and a region with a strong red component are individually designated as candidate regions 1, 2, and 3 from the target image. (Steps S701, S702, and S703).

  A product set of these areas, that is, a common area that satisfies all the conditions is extracted as a candidate area 4 (step S704). Those that satisfy all of the conditions so far are likely to be amyloid bodies or eosinophilic crystalloids. Similar to the first processing example, by extracting a candidate area 4 having a predetermined area (step S705), amyloid bodies can be excluded. The extracted region can be determined as an acidophilic crystalloid (step S706), and necessary information can be acquired (step S707).

  FIG. 13 is a flowchart showing a third example of the acidophilic crystalloid detection process. FIG. 14 is a diagram schematically showing the process of selecting structures in the image in this process. Also in this example, based on the same conditions as in the above-described example, a closed region with a low saturation level, a region with a weak blue component, and a region with a strong red component are designated as candidate regions 1, 2, and 3 from the target image, respectively. Individually extracted (steps S801, S802, and S803). Among these, the candidate region 3 having a strong red component is extracted as a candidate region 4 only in a predetermined area range (step S804), and amyloid bodies are excluded from the candidate region 4.

  The candidate area 4 includes red blood cells other than the eosinophilic crystalloid, but by extracting the common area between the candidate areas 1 and 2 and the candidate area 4 (step S805), the red blood cells are excluded. Similarly to the above two examples, only eosinophilic crystalloid can be detected (steps S806 and S807).

  As described above, according to the present invention, the structure is in a closed region surrounded by the low-saturation pixel region, and the red component and the blue component are in the predetermined intensity range, and the predetermined area is reduced. By extracting only those possessed, it is possible to detect eosinophilic crystalloid with high accuracy, and the application order of these conditions is arbitrary.

  In the acidophilic crystalloid detection process of the above embodiment, only one of the upper limit value and the lower limit value is set for the saturation, red component, and blue component thresholds. Both may be set and the extraction condition may be defined by the range specified by these threshold values. In addition to the area described above, a shape factor, circularity, or the like may be used as a parameter related to the geometric feature of the structure.

  In the above embodiment, the imaging unit 20 including the CCD camera 201 and the microscope 202 is provided as an “image acquisition unit”. However, the imaging unit is not essential in the present invention. As described above, it is also possible to receive an original image from the telecommunication line 30 via the interface 180 or from an external recording medium via an appropriate interface. In this case, these interfaces serve as “image acquisition means”. Will work. In this sense, the “step for acquiring an image” is not essential in the control program of the present invention.

  Moreover, in the said embodiment, in addition to the special site | part detection part 121 which detects the eosinophilic crystalloid in gland duct tissue, it has the classifier 123 which classify | categorizes a grid image based on a reference image, However, It has a classifier. Instead, even if only the detection result of the special part is displayed, the user's diagnosis can be sufficiently supported.

  In this embodiment, processing by each functional block is realized by the host computer 10 executing the control program recorded on the optical disk OD. However, the recording medium for recording the control program is limited to the optical disk. Any other recording medium can be used as described above. The control program may be distributed through the telecommunication line 30.

  Moreover, although the pathological diagnosis support apparatus 1 of the said embodiment is comprised combining the general purpose host computer 10 and the imaging part 20, it is made to comprise a pathological diagnosis support apparatus combining an imaging part and a dedicated terminal. Also good. Further, by incorporating the control program according to the present invention into an existing virtual slide creation device, the virtual slide creation device may function as a pathological diagnosis support device.

  Further, for example, the classification result and the set operation result can be distributed to the outside through the telecommunication line 30 or recorded on an external recording medium instead of or in addition to being displayed on the display unit 160. You may do it.

  The present invention can be suitably applied to a technical field in which various pathological diagnoses by a doctor are supported by a computer system or the like.

20 Imaging unit (image acquisition means)
100 processor unit (special part detection means, visual information processing means, classification means)
111 Display processing unit (visual information processing means)
121 Special part detection unit (special part detection means)
123 Classifier (Classification means)
160 Display unit (display means)
Ig Grid image Ip Partial image Iv Virtual slide image (original image)
S101 Image acquisition process S102 Special part detection process S103, S104, S106 Visual information processing process S105 Classification process S107 Display process

Claims (12)

Image acquisition means for acquiring an original image obtained by imaging a sample tissue;
From the original image, special site detection means for detecting eosinophilic crystalloid as a special site having characteristics peculiar to the lesion of the specimen tissue,
Visual information processing means for creating a composite image obtained by combining the original image with visual information related to the position in the original image of the special part detected by the special part detection means;
Display means for displaying the composite image ,
The special part detecting means is a low saturation region where a saturation level satisfies a predetermined low saturation condition among structures existing in the original image of the specimen tissue stained with hematoxylin and eosin. A structure in which the intensity of the red component satisfies a predetermined high intensity condition, the intensity of the blue component satisfies a predetermined low intensity condition, and the area is in a predetermined area range is detected as the special portion. A pathological diagnosis support apparatus characterized by the above.   The low saturation condition is smaller than a predetermined first threshold, the low intensity condition is smaller than a predetermined second threshold, and the high intensity condition is larger than a predetermined third threshold. The pathological diagnosis support apparatus according to claim 1. The visual information processing means divides the original image into a plurality of grid images, displays the plurality of grid images on the display means in an arrangement based on the arrangement in the original image, and the special part is included. The pathological diagnosis support apparatus according to claim 1, wherein the visual information indicating that the grid image is a lesion site is attached to the grid image. The pathological information according to claim 3 , wherein the visual information processing unit attaches the visual information indicating that the grid image is a lesion site to the grid image including the special site and a grid image adjacent thereto. Diagnosis support device. Classifying means for classifying the grid image into a plurality of classification categories based on the feature amount for each grid image for the grid image other than the grid image that is the lesion site,
The pathological information according to claim 3 or 4 , wherein the visual information processing unit creates the composite image with the visual information in a different form for each of the classification categories with respect to the grid image classified by the classification unit. Diagnosis support device. 2. The visual information processing unit generates the composite image by combining a partial image of the original image and the visual information representing the number of the special parts detected in the partial image. The pathological diagnosis support apparatus according to any one of 5 . An image acquisition step of acquiring an original image obtained by imaging a sample tissue;
From the original image, a special site detection step of detecting eosinophilic crystalloid as a special site having characteristics peculiar to the lesion of the specimen tissue,
A visual information processing step of creating a composite image obtained by combining the original image with visual information related to the position of the detected special part in the original image;
A display step of displaying the composite image,
In the special part detection step, for the original image of the specimen tissue stained with hematoxylin and eosin, among the structures existing in the original image, a low saturation region where the saturation level satisfies a predetermined low saturation condition A structure in which the intensity of the red component satisfies a predetermined high intensity condition, the intensity of the blue component satisfies a predetermined low intensity condition, and the area is in a predetermined area range is detected as the special portion. A pathological diagnosis support method characterized by the above.   The low saturation condition is smaller than a predetermined first threshold, the low intensity condition is smaller than a predetermined second threshold, and the high intensity condition is larger than a predetermined third threshold. The pathological diagnosis support method according to claim 7. In the visual information processing step, the original image is divided into a plurality of grid images, the plurality of grid images are displayed in an arrangement based on the arrangement in the original image, and the grid image including the special part is included. The pathological diagnosis support method according to claim 7 or 8 , wherein the visual information indicating that the grid image is a lesion site is attached. For the grid image other than the grid image determined as the lesion site, the method further comprises a classification step of classifying the grid image into a plurality of classification categories based on the feature amount for each grid image,
The pathological diagnosis support according to claim 9 , wherein in the visual information processing step, the composite image in which the visual information having a different aspect for each classification category is added to the grid image classified in the classification step. Method. An image acquisition step of acquiring an original image obtained by imaging a sample tissue;
From the original image of the specimen tissue stained with hematoxylin and eosin , among the structures existing in the original image, the saturation level is surrounded by a low saturation region that satisfies a predetermined low saturation condition, and the red component By detecting a structure in which the intensity satisfies a predetermined high intensity condition, the intensity of the blue component satisfies a predetermined low intensity condition, and the area is in a predetermined area range, characteristics specific to the lesion of the specimen tissue can be obtained. A special site detection step for detecting eosinophilic crystalloid as a special site having,
A visual information processing step of creating a composite image obtained by combining the original image with visual information related to the position of the detected special part in the original image;
A control program for supporting pathological diagnosis, characterized by causing a computer to execute a display step of displaying the composite image. The computer-readable recording medium which recorded the control program of Claim 11 .