JP2006263127A - Ocular fundus diagnostic imaging support system and ocular fundus diagnostic imaging support program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an ocular fundus diagnostic imaging support system which can support diagnosis of a disease such as glaucoma by analyzing an ocular fundus photograph and providing a value of a C/D ratio calculated objectively to a medical doctor. <P>SOLUTION: The ocular fundus diagnostic imaging support system 1 uses an ocular fundus diagnostic imaging support computer 2, extracts a diagnostic candidate area from ocular fundus image data concerning the ocular fundus photograph obtained by photographing the ocular fundus of a patient, detects a teat sector and a recessus sector, and respectively calculates a teat approximate circle and a recess approximate circle approximate to the teat sector and the recessus sector. By calculating a ratio of the radius of the teat approximate circle to the calculated recess approximate circle, information of the C/D ratio profitable for diagnosing ophthalmic diseases such as glaucoma is provided to the medical doctor to support the diagnosis. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a fundus image diagnosis support system and a fundus image diagnosis support program, and in particular, uses fundus photographs obtained by photographing the fundus of a subject, and mainly provides useful information related to diagnosis of ophthalmic diseases such as glaucoma. The present invention relates to a fundus image diagnosis support system and a fundus image diagnosis support program that can be provided to a doctor and support the diagnosis.

  Conventionally, the eyeball is irradiated with light from the outside, and the state of the fundus is observed through the eyeball. In addition, the state of the fundus is imaged by an optical device such as a camera, and the fundus photograph is left in medical records. By examining this fundus photograph in detail, a doctor can make a diagnosis of various diseases. This observation of the fundus is a non-invasive diagnostic technique, and the patient (subject) undergoing the examination does not have a physical burden and can be obtained with a relatively simple device. This is one of the commonly used inspection methods. In addition, blood vessels photographed by fundus photographs are a part of the brain's cerebral blood vessels located in the back of the eyeball, and are the only method that allows direct observation of the state of the brain from outside the body. Therefore, it is possible to provide very useful information to doctors and the like. In addition, this fundus photograph can be used not only for diagnosis of ophthalmic diseases such as glaucoma but also for diagnosis of lifestyle-related diseases such as diabetes.

  Here, in the fundus photograph of a healthy subject, the entire fundus is yellowish reddish brown, and is a circle or oval of 1.2 mm to 1.7 mm at a position about 15 degrees outward (ear side direction) of the subject's line of sight. An optic nerve head is observed. This optic disc is anatomically called an optic disc, and the boundary between the fundus and the optic disc is generally clearly observed on the ear side rather than the nose side. is there. In the vicinity of the center of the optic nerve head, an area called a optic nerve head depression or a physiological depression having a circular shape is observed.

  When glaucoma is diagnosed using fundus photographs, the ratios of the diameter or radius of the optic disc and the optic disc recess described above are obtained as one of the criteria for diagnosis. That is, when the value of the diameter ratio of the optic disc depression (C) to the substantially circular optic disc (D) (hereinafter referred to as “C / D ratio”) is 0.6 or more, in other words, the optic nerve If the area of the nipple depression is larger than normal, it is a finding of optic nerve head depression, and generally symptoms of glaucoma are suspected. On the other hand, when the C / D ratio is smaller than 0.6, no nipple depression is found, and it is determined that the patient is in a healthy state. Further, when the difference between the left and right C / D ratio is 0.2 or more, glaucoma is similarly suspected. Diagnosis of glaucoma is not made only with the value of the C / D ratio, but is made in combination with other examinations and doctors' findings, and other suspicious symptoms. Always over 25 mmHg 2) Symptoms due to high intraocular pressure such as “I feel heavy” 3) The optic disc findings change over time 4) The retinal nerve fiber layer defect expands over time , Etc. are also subject to treatment. That is, the value of the C / D ratio is one of useful information for the diagnosis, and glaucoma is not directly judged only by this value.

  At this time, the method of measuring the diameters of the optic disc and the optic disc concavity in order to make a determination is to apply a scale such as a ruler directly on the developed fundus photograph, and the doctors actually set the scale of the ruler. Measured by reading. In addition, a dedicated scale for glaucoma determination for facilitating the measurement and calculation of the C / D ratio has been developed (see, for example, Patent Document 1).

  On the other hand, some fundus photographs taken are converted into electronic data using an image input device such as a scanner, and are input to a computer as image information and used. That is, the image information of the fundus photograph input is displayed on a computer display, and a doctor or the like uses a pointing device such as a mouse, and the boundaries of the fundus region, the optic nerve head region, and the optic nerve head depression region are There are some which determine each area by designating the above. And according to the determined area | region, calculation of C / D ratio is performed by the computer.

JP 2004-49259 A

  However, when using the ruler or the dedicated scale for glaucoma determination described above, it is necessary for a doctor or the like to identify each region with the naked eye according to previous experience, and to manually read the scale on the scale. Therefore, the determination of each region may vary depending on the subjectivity of each measurer such as a doctor, and the measurement error may increase. As a result, the objectivity of the diagnosis result of glaucoma may be questioned. In particular, in the case of a subject whose C / D ratio, which is a criterion for determining whether or not to diagnose glaucoma as described above, near the boundary of 0.6, a case where glaucoma is diagnosed by this measurement error, and glaucoma are diagnosed. There were cases where treatment was unnecessary.

  Furthermore, in the case of a medical institution that takes fundus photographs for a large number of subjects at one time and diagnoses diseases such as glaucoma, such as a medical checkup or a mass screening, in the above-mentioned manual diagnosis method, individual subjects The work time per person required for the diagnosis became longer, and the time required for the entire diagnosis tended to be longer. For this reason, a doctor or other measurer may be burdened with great effort.

  On the other hand, there is already known a method for calculating a C / D ratio by using a scanner or the like to display captured image information of a fundus photograph on a display. According to this, for the fundus image of the fundus photograph displayed on the screen, doctors etc. individually indicate the boundary of each region (region of optic nerve head and optic nerve head depression) with a pointing device such as a mouse or tablet, The C / D ratio can be automatically calculated by the computer after the instruction. However, although the work after designation of each area is simplified, what designates the area of the optic nerve head and the depression of the optic nerve head is performed manually by a doctor, etc. The effect of reducing the labor burden required to specify the area was not so great.

  Therefore, in view of the above circumstances, the present invention analyzes fundus photographs, specifies each area of the optic nerve head and the optic nerve head depression easily and objectively, calculates the C / D ratio, and calculates A fundus image diagnosis support system capable of providing information useful for diagnosis of a disease such as glaucoma by the doctor by presenting the C / D ratio value obtained to the doctor or the like, and a fundus image It relates to a diagnostic support program.

  In order to solve the above-described problem, the fundus image diagnosis support system according to the present invention provides “information for supporting a doctor's diagnosis by analyzing a fundus image obtained by photographing the fundus of a subject using a fundus image diagnosis support computer” A fundus image diagnosis support system capable of the fundus image diagnosis support, wherein the fundus image diagnosis support computer receives fundus image data input receiving means for receiving fundus image data created from the fundus photograph, and the optic disc from the received fundus image data. A candidate area extracting means for extracting a diagnostic candidate area including the diagnostic fundus image data from the fundus image data, and a papillary area detection for detecting a substantially circular papillary area corresponding to the optic disc from the extracted candidate image data indicating the diagnostic candidate area A nipple approximate circle that approximates the shape of the nipple contour indicating the boundary between the means and the detected nipple region and fundus region A head approximate circle calculating means; a recessed area detecting means for detecting a substantially circular recessed area corresponding to an optic disc depression from the candidate image data; and a boundary between the detected recessed area and the nipple area. A recessed approximate circle calculating means for calculating a recessed approximate circle that approximates the shape of the recessed contour shown, and C indicating a ratio of either the radius or diameter of the approximated nipple circle to the calculated recessed approximate circle C / D ratio calculating means for obtaining the / D ratio ”and the like.

  Here, the fundus image data is created by converting the fundus photograph of the subject's fundus into electronic data, and is, for example, a fundus photograph taken with an ordinary silver salt camera and printed, or its positive (negative) Data obtained by converting a film into electronic information using an image input device such as a scanner, or data taken by a digital camera that has recently become widely used may be directly used as fundus image data. Then, the converted fundus image data is input to the fundus image diagnosis support computer through the fundus image data input receiving unit. Here, the fundus image data input accepting unit is not limited as long as it can input the fundus image data described above. For example, the fundus image data input accepting unit takes in the fundus image diagnosis support computer using an image input device such as a scanner or a film scanner. And fundus image data stored in a storage medium such as a flexible disk or a CD-R and taken in through a reader of the storage medium. In the case of data photographed by a digital camera, the digital fundus image diagnosis support computer may be connected with a USB cable or the like, and target fundus image data may be directly captured.

  Further, the candidate area extracting means specifies a candidate diagnosis area including an optic disc and a optic disc depression for calculating a C / D ratio, which will be described later, and is the brightest part of the fundus photograph that is captured most brightly. When the angle of view of the fundus camera is set to 45 degrees and recording is performed with 600 pixels × 600 pixels, for example, a range of 240 pixels × 240 pixels is extracted. Here, the optic disc and the region related to the optic disc depression are generally photographed brighter than the surrounding fundus region and other regions, and the brightest part included in the fundus image data is generally It corresponds to the approximate center of the optic disc depression. Therefore, it is possible to obtain candidate image data including the optic disc that is the measurement target for calculating the C / D ratio by extracting the diagnostic candidate region centering on the brightest part. In addition, it is empirically known that the average diameter size of the general adult optic disc recess is about 80 pixels, and the range specified as the candidate image data includes the entire optic disc. be able to.

  Further, the coordinate position of the brightest part from the fundus image data is determined by converting the color-funded fundus photograph into a monochrome image, and the luminance signal Y recognized as a general method of image analysis is expressed by the following equation (see Equation 1). Is done by asking. Here, G represents the value of each of the green component, B represents the blue component, and R represents the red component.

  Furthermore, each of the nipple region extracting means or the recessed region extracting means detects each region by utilizing a difference in pixel value from the surrounding area. That is, in the nipple area and the fundus area existing around the nipple area, the nipple area has higher brightness than the fundus area, while in the recessed area and the nipple area existing therearound, the nipple area The lightness is higher in the recessed area. Accordingly, the pixel values of the candidate image data are measured, and each region is extracted by using as a boundary a position where a difference in pixel values such that the pixel values of adjacent pixels are greatly different from each other is recognized. The extracted nipple region and the recessed region have a substantially circular shape due to their physiological characteristics.

  Therefore, according to the fundus image diagnosis support system of the present invention, candidate diagnosis image data related to the candidate diagnosis region is extracted from the fundus image data that has received the input according to the luminance signal. As a result, the measurement range for calculating the C / D ratio described later is limited to be narrow. Then, the nipple region and the recessed region are recognized by the difference in pixel values from the surrounding regions. Thereafter, an approximate circle that substantially matches the contour of the extracted region (papill contour or recessed contour) is calculated. Further, the C / D ratio used for determining a disease such as glaucoma is calculated by obtaining the diameter ratio (for example, the ratio of the diameter) of the nipple approximate circle to the calculated recess approximate circle. As a result, the specification of the nipple region and the recessed region and the calculation of the C / D ratio can be performed by the fundus image diagnosis support computer, which greatly simplifies the work manually performed by a conventional doctor or the like. Is possible. Furthermore, since each region is designated according to a standard set in advance by the fundus image diagnosis support computer, variations in the diagnosis result due to the subjectivity of the doctor and individual differences are suppressed, and the doctor provides C according to objective data. The subject can be diagnosed with reference to the value of the / D ratio. In order to more clearly recognize the difference in pixel value in each region, the binary image is binary with a predetermined threshold set for each color component (R component, G component, B component) for candidate image data of a color image. It is also possible to perform the process, and then detect each region.

  Further, the fundus image diagnosis support system according to the present invention includes, in addition to the above configuration, “at least one of the nipple approximate circle calculating means and the recessed approximate circle calculating means approximates the nipple contour or the recessed contour. A candidate circle selecting means for selecting candidate circles to be positioned, and located on the inner circle side of the selected candidate circle, and located on the inner circle side of the candidate circle from a matching region that overlaps the papillary region or the recessed region In addition, the candidate circle coincidence degree is calculated by subtracting the non-overlapping surplus area and the outer circle side of the candidate circle that is not overlapped with the nipple area or the recessed area, and subtracting the missing area overlapping the nipple area or the recessed area And an approximation for determining a matching candidate circle having a maximum candidate circle matching degree as the nipple approximate circle or the recessed approximate circle from a plurality of candidate circles selected by the candidate circle selecting means and the candidate circle selecting means Yen decisive factor It may be those having a when. "

  Therefore, according to the fundus image diagnosis support system of the present invention, when calculating the nipple approximate circle that matches the nipple region, first, candidate circles that approximate the respective regions to be calculated are selected, and the candidate image data is selected. Virtually overlap. At this time, the corresponding portion that is located on the inner circle side of the candidate circle and is superimposed on the target region is set as the matching region B, and the portion that is located on the inner circle side of the candidate circle and is not superimposed on the target region is redundant. A region A (in other words, a region obtained by removing the matching region B from the candidate circle) is defined as a missing region C that is located on the outer circle side of the candidate circle and overlaps the target region. Then, the candidate circle coincidence degree E is calculated by the following equation, and the coincidence candidate circle indicating the maximum value is determined as a nipple approximate circle or a recessed approximate circle (see Formula 2). Here, in Equation 2, x and y indicate the center coordinates of each circle, and r indicates the radius of the circle. At this time, the candidate circle coincidence degree E is maximized when the surplus area A and the missing area C are nearly zero and the coincidence area B is maximized. Note that the same processing as described above is performed when calculating a concave approximate circle that matches the concave region.

  Further, the fundus image diagnosis support system according to the present invention is, in addition to the above configuration, “corresponding to a blood vessel included in the candidate image data and photographed while being superimposed on at least one of the papillary region and the recessed region. A blood vessel region detecting means for detecting a blood vessel region to perform, and a blood vessel region correcting means for correcting and calculating at least one of the approximated nipple circle and the approximated recess circle based on the detected blood vessel region. It may be provided.

  Therefore, according to the fundus image diagnosis support system of the present invention, a blood vessel region that is superimposed on the papillary region and the recessed region and becomes a factor that hinders accurate C / D ratio calculation is detected in advance, and the blood vessel The nipple approximate circle and the recessed approximate circle are calculated by correcting the region. This eliminates the situation in which it is difficult to calculate the C / D ratio due to the superposition of blood vessels, and makes it possible to perform an objective diagnosis with an accurate C / D ratio.

  In the correction of the blood vessel region, the candidate circle coincidence degree E ′ is calculated according to the following formula, and the shape of the nipple region or the recessed region obstructed by the blood vessel region can be recognized as accurate. Here, the region occupied by the blood vessel superimposed in the coincidence region B is the matching blood vessel region Vb, the region occupied by the blood vessel superimposed on the surplus region A is the surplus blood vessel region Va, and the region occupied by the blood vessel superimposed on the defect region C is the defective blood vessel. This is defined as a region Vc.

  On the other hand, the fundus image diagnosis support program according to the present invention provides a “fundus image data input receiving unit that receives an input of fundus image data created from a fundus image obtained by photographing a fundus of a subject, and an optic disc from the received fundus image data. Candidate area extracting means for extracting a diagnostic candidate area including the fundus image data, a papillary area detecting means for detecting a substantially circular papillary area corresponding to the optic disc from candidate image data indicating the extracted diagnostic candidate area; A nipple approximate circle calculating means for calculating a nipple approximate circle that approximates the shape of the nipple contour indicating the boundary between the detected nipple region and the fundus region, and a substantially circular recess corresponding to an optic disc recess from the candidate image data A recessed area detecting means for detecting a recessed area, a recessed area that approximates the shape of a recessed contour that indicates a boundary between the detected recessed area and the nipple area. A recessed approximate circle calculating means for calculating a similar circle, and a C / D ratio calculating means for obtaining a C / D ratio indicating a ratio of either the radius or diameter of the approximated nipple circle to the calculated recessed approximate circle As described above, it is mainly composed of what makes the fundus image diagnosis support computer function.

  Furthermore, the fundus image diagnosis support program according to the present invention includes, in addition to the above configuration, “candidate circle selecting means for selecting a candidate circle that approximates the nipple contour or the recessed contour, the inner circle side of the selected candidate circle Is located on the inner circle side of the candidate circle from a matching region overlapping with the nipple region or the recessed region, and is located outside the nipple region or the recessed region and a non-overlapping surplus region and the candidate circle. From the plurality of candidate circles selected by the candidate circle selection means, the degree of coincidence calculation means for calculating the candidate circle coincidence, which is located on the circle side and subtracts the defective area overlapping the nipple area or the recessed area, The fundus image diagnosis support computer may be caused to function as an approximate circle determining unit that determines a match candidate circle having the maximum candidate circle match degree as the nipple approximate circle or the recessed approximate circle. There.

  Further, the fundus image diagnosis support program according to the present invention, in addition to the above configuration, “corresponds to a blood vessel included in the candidate image data and photographed while being superimposed on at least one of the papillary region and the recessed region. A blood vessel region detecting means for detecting a blood vessel region to be performed, and a blood vessel region correcting means for correcting and calculating at least one of the nipple approximate circle and the recessed approximate circle based on the detected blood vessel region, The fundus image diagnosis support computer may function.

  Therefore, according to the fundus image diagnosis support program of the present invention, by executing the program, the fundus image diagnosis support computer calculates a C / D ratio necessary for diagnosis of a disease such as glaucoma as objective data, It can be provided to doctors.

  As an effect of the present invention, the fundus image diagnosis support system and the fundus image diagnosis support program use a difference in pixel value between a substantially circular papillary region and a recessed region necessary for calculating a C / D ratio from input fundus image data. The C / D ratio can be obtained by obtaining an approximate circle that is detected using the above and further approximately matches the contour of each of the nipple region and the recessed region. Furthermore, by performing binarization processing, division processing of the ear side region and nose side region, correction processing of the blood vessel region, and the like, the nipple region and the recessed region can be correctly detected, and the objectivity can be improved. As a result, the burden on the work of the doctor or the like is reduced, and the doctor can diagnose the disease based on objective facts.

  Hereinafter, a fundus image diagnosis support system 1 (hereinafter simply referred to as “support system 1”) and a fundus image diagnosis support program (hereinafter simply referred to as “support program”) according to an embodiment of the present invention will be described with reference to FIGS. This will be described with reference to FIG. Here, FIG. 1 is a block diagram showing a schematic configuration of the support system 1 of the present embodiment and a functional configuration of a fundus image diagnosis support computer 2 (hereinafter simply referred to as “computer 2”), and FIG. FIG. 3 is an explanatory diagram showing an example of extraction of received fundus image data 3 and (b) candidate image data 16, FIG. 3 is an explanatory diagram showing an example of a detected blood vessel region 6, and FIG. FIG. 5 is an explanatory diagram showing an example of the nipple region 4, and FIG. 5 is an explanatory diagram schematically showing an example of (a) determination of the nipple approximate circle 5 or the recessed approximate circle 33 and (b) detection of the blood vessel region 6. 6 is an explanatory diagram showing an example of the detected recessed area 26, and FIGS. 7 and 8 are flowcharts showing the flow of the calculation process of the C / D ratio by the computer 2.

  As shown in FIG. 1, the support system 1 according to the present embodiment receives input of fundus image data 3 obtained by converting a photographed fundus photograph into electronic data, and corresponds to a portion corresponding to the optic nerve head and a optic nerve head depression. The computer 2 is mainly composed of a computer 2 capable of assisting diagnosis of glaucoma by calculating a value of a C / D ratio indicating a radius ratio of a region to be performed and providing it to a doctor. More specifically, the support system 1 of the present embodiment includes a computer 2, a fundus camera 7 connected to the computer 2 and having a digital camera function for obtaining fundus image data 3 as image input means, and acquisition. The liquid crystal display 8 displays various data such as the fundus image data 3 and the like, and the operation input device 11 including the keyboard 9 and the mouse 10 that receives input of various commands and data to the computer 2 is mainly configured. . Here, as the computer 2 used in the support system 1 of the present embodiment, a commercially available general-purpose computer is used.

  Further, the functional configuration of the computer 2 will be described. The fundus image data input receiving means 12 that receives the input of the fundus image data 3 of the color image acquired by the fundus camera 7, and various types including the received fundus image data 3. Based on the storage means 13 composed of a storage medium such as a hard disk and semiconductor memory for storing image data and analysis data such as the analyzed C / D ratio value, and the luminance signal shown in the received fundus image data 3 A nipple that detects the nipple region 4 using the difference in pixel values between the candidate region extraction means 17 that extracts the candidate diagnosis region 15 and obtains candidate image data 16 and the surrounding fundus region 18 based on the candidate image data 16 The region detection means 19 and a nipple candidate circle 21 that approximates the shape of the nipple contour 20 indicating the boundary between the detected nipple region 4 and the fundus region 18 are selected. Nipple candidate circle selecting means 22 for performing the operation, a nipple coincidence calculating means 23 for calculating the nipple coincidence degree E1 between the selected nipple candidate circle 21 and the nipple region 4, and a nipple coincidence degree from a plurality of selected nipple candidate circles 21 There is provided a nipple approximate circle determining means 25 for determining the nipple coincidence candidate circle 24 having the maximum value E1 as the nipple approximate circle 5. Here, the nipple candidate circle selection means 22 corresponds to the candidate circle selection means in the present invention, the nipple matching degree calculation means 23 corresponds to the coincidence degree calculation means in the present invention, and the nipple approximate circle determination means 25 corresponds to the approximation in the present invention. It corresponds to the circle determination means. The nipple candidate circle selection means 22, the nipple coincidence calculating means 23, and the nipple approximate circle determining means 25 correspond to the nipple approximate circle calculating means in the present invention, and the nipple coincidence E1 is the candidate circle coincidence in the present invention. It corresponds to. Further, the nipple area information 4 a related to the detected nipple area 4 is stored in the storage means 13.

  Further, the computer 2 uses the difference in pixel values of the surrounding nipple area 4 based on the candidate image data 16 to detect the recessed area 26, the detected recessed area 26, A recess candidate circle selecting means 30 for selecting a recess candidate circle 29 that approximates the shape of the recess contour 28 indicating the boundary of the nipple region 4, and a recess between the selected recess candidate circle 29 and the recess region 26. The depression coincidence degree calculating means 31 for calculating the degree of coincidence E2 and the depression coincidence candidate circle 32 having the maximum depression coincidence degree E2 from the selected plurality of depression candidate circles 29 are determined as the depression approximate circle 33. And a recessed approximate circle determining means 34 for performing the above operation. Here, the recessed candidate circle selecting means 30 corresponds to the candidate circle selecting means in the present invention, the recessed matching degree calculating means 31 corresponds to the matching degree calculating means in the present invention, and the recessed approximate circle determining means 34 is the book. This corresponds to the approximate circle determining means in the invention. Further, the recess candidate circle selecting means 30, the recess coincidence degree calculating means 31, and the recess approximate circle determining means 34 correspond to the recess approximate circle calculating means in the present invention, and the recess coincidence degree E2 is the present invention. This corresponds to the candidate circle coincidence degree. Further, the recessed area information 26 a related to the detected recessed area 26 is stored in the storage unit 13.

  In addition, the computer 2 uses a difference between pixel values to detect a blood vessel region 6 that is captured by being superimposed on the candidate image data 16, and a nipple approximation based on the detected blood vessel region 6. Blood vessel region correction means 36 for correcting the shapes of the circle 5 and the recessed approximate circle 33, and a C / D ratio value (C / D ratio) indicating the ratio of the radius of the approximated nipple circle 5 to the calculated recessed approximate circle 33 C / D ratio calculation means 37 for obtaining data 37a) and a calculation result output control means for outputting and displaying the calculated C / D ratio value on the liquid crystal display 8 and presenting it to a doctor or the like so as to be visually recognizable. 38, and an operation signal input control means 39 that is connected to the operation input device 11, receives operation signals such as various commands, and performs C / D ratio calculation processing and the like.

  Next, the processing flow of the computer 2 in the support system 1 of the present embodiment will be described mainly based on the flowcharts of FIGS. Here, Steps S1 to S22 in FIGS. 7 and 8 correspond to the fundus image diagnosis support program of the present invention.

  First, the fundus of the subject is photographed using a fundus camera 7 having a digital camera function, and the computer 2 accepts input of the photographed fundus image data 3 (see step S1: FIG. 2A). Then, the received fundus image data 3 is stored in the storage means 13 (step S2). Here, the fundus image data 3 is obtained by scanning a fundus photo printed on a photographic paper with a scanner, or by scanning a fundus photo positive film with a film scanner, in addition to using the fundus camera 7 described above. May be taken into the computer 2 as the fundus image data 3.

  Then, the luminance signal at each coordinate position of the fundus image data 3 is calculated based on the luminance signal calculation formula (see Equation 1), and the brightest portion 14 indicating the highest luminance signal Y in the fundus image data 3 is calculated. A range of 240 pixels × 240 pixels around the coordinate position is cut out from the fundus image data 3 and extracted and stored as candidate image data 16 (steps S3 and S4). Here, it is shown that the recessed area 26 corresponding to the optic nerve head recessed area has a higher brightness than the surrounding fundus area 18, in other words, is displayed with a higher pixel value. For this reason, it is empirically known that the coordinate position of the brightest portion 14 in the fundus image data 3 corresponds to the recessed region 26. Therefore, by extracting a predetermined range centering on this coordinate position from the fundus image data 3, the extracted candidate image data 16 necessarily includes the nipple region 4 and the recessed region 26 existing inside thereof. (See FIG. 2B). It should be noted that the average size of the general adult nipple region 4 has a diameter of about 80 pixels. As a result, detection of the nipple area 4 and the recessed area 26 and the process of determining each candidate circle, which will be described later, can be performed in a limited range, and the processing load on the computer 2 can be reduced. , Errors during processing can be reduced. Here, in the above example, when the candidate image data 16 is extracted, the range of 240 pixels × 240 pixels is cut out, and the average size of the nipple region 4 is set to about 80 pixels. Of course, the extraction range may be appropriately changed according to the conditions.

  Thereafter, the blood vessel region 6 corresponding to the blood vessel included in the candidate image data 16 is detected (step S5). Here, the blood vessel region 6 is generally photographed darkly with respect to the nipple region 4 and the recessed region 26 existing therearound. Therefore, the pixel values in the blood vessel region 6 and the periphery thereof are greatly different in the candidate image data 16. Accordingly, the position of the blood vessel region 6 in the candidate image data 16 is specified by recognizing a portion where the change in the pixel value is remarkably large as a boundary between the blood vessel region 6 and the nipple region 4 or the recessed region 26 (FIG. 3). reference). In addition, as a function for recognizing a difference in pixel values, it is possible to apply a standard function installed in commonly used image analysis software or the like. 1 can be constructed easily and at low cost. In this embodiment, the binarization process is performed on the candidate image data 16 according to a preset threshold value to extract only the blood vessel region 6 in order to recognize the difference in pixel values more clearly. (See FIG. 3).

  In this case, by setting a threshold value to a pixel value lower than the nipple region 4 and the recessed region 26 existing around the blood vessel region 6, a part having a pixel value lower than the threshold value is recognized as the blood vessel region 6. On the other hand, a portion showing a high pixel value is recognized as the nipple region 4 or the recessed region 26. Note that this binarization processing is performed according to individually set thresholds even when a nipple region 4 and a recessed region 26 described later are detected.

  Then, the blood vessel region information 6a related to the detected blood vessel region 6 is stored in the storage means 13 (step S6). At this time, the stored blood vessel region information 6a includes a coordinate position of a part corresponding to the blood vessel region 6 and binarized image data.

  Next, a substantially circular papillary region 4 corresponding to the optic disc portion is detected (step S7: see FIG. 4). Here, the detection of the nipple region 4 is performed in the same manner as the detection method of the blood vessel region 6 described above, with the fundus region 18 that is present around the nipple region 4 and is displayed with pixel values darker than the nipple region 4. This is performed by using the difference in pixel value. Thereafter, a nipple candidate circle 21 that approximates the substantially circular nipple contour 20 corresponding to the boundary between the detected nipple region 4 and the fundus region 18 is selected (step S8). At this time, an arbitrary coordinate in the nipple region 4 is defined as the center coordinate (x, y) of the nipple candidate circle 21, and an arbitrary value within a predetermined range is set as the radius r of the nipple candidate circle 21. Stipulate. Thereafter, the teat coincidence E1 is calculated (step S9). Here, the nipple coincidence E1 is calculated according to the described calculation formula (see Expression 3) using the blood vessel region information 6a related to the blood vessel region 6 that has already been detected.

  More specifically, the nipple coincidence E1 is located on the inner circle side of the nipple candidate circle 21 and the corresponding portion superimposed on the nipple region 4 is located on the inner circle side of the matching region B and the nipple candidate circle 21. In addition, each non-overlapping portion of the nipple region 4 is located in the surplus region A, the outer circle side of the nipple candidate circle 21 and the corresponding portion overlapping with the nipple region 4 is defined as a missing region C, and a matching region The portion occupied by the blood vessel region 6 superimposed on B is the coincident blood vessel region Vb, the portion occupied by the blood vessel region 6 superimposed on the excess region A is the excess blood vessel region Va, and the portion occupied by the blood vessel region 6 superimposed on the defect region C is the defective blood vessel. It is calculated | required by each area | region defined as the area | region Vc (refer FIG.5 (b)). Here, the values of the areas A, B, C, Va, Vb, and Vc are based on the number of pixels in the candidate image data 16 (in other words, the area occupied in the candidate image data 16).

  When the blood vessel region 6 detected in step S5 is zero or close to zero, that is, when a portion corresponding to a blood vessel is not photographed in the candidate image data 16, the above-described correction of the blood vessel region 6 is performed. The calculation is performed assuming that the values of the blood vessel regions Va, Vb, and Vc in the equation (Equation 3) are zero. That is, the calculation is performed based on the calculation formula (Equation 2) of the candidate circle coincidence not considering the existence of the blood vessel region 6 described above.

  Here, the nipple coincidence E1 shows the maximum value when the selected nipple candidate circle 21 and the nipple region 4 substantially coincide with each other, and the surplus region A and the missing region C are close to zero, that is, the nipple candidate circle. 21 region = a state closest to the nipple region 4.

  When the nipple coincidence E1 of the nipple candidate circles 21 within a predetermined range defined in advance is completely calculated (YES in step S10), the calculated nipple coincidence E1 has the maximum value. The indicated nipple candidate circle 21 is set as a nipple matching candidate circle 24, and the approximate nipple circle 5 is determined (step S11). On the other hand, when the nipple candidate circle 21 for calculating the nipple coincidence degree E1 still remains, a new nipple candidate circle 21 is selected (step S12), the process proceeds to step S9, and the nipple coincidence degree E1 is determined. Recalculation is performed. Here, the reselection of the nipple candidate circle 21 is performed by changing the center coordinate of the nipple candidate circle 21 or changing the value of the radius or diameter.

  Thereafter, a concave approximate circle 33 that approximates the concave contour 28 of the concave region 26 existing inside the nipple region 4 is determined (step S13 to step S18: see FIG. 6). Here, since the process related to the selection and determination of the recessed approximate circle 33 is substantially the same as the process related to the determination of the approximated nipple circle 5 described above, detailed description thereof will be omitted here.

  Then, based on the calculated radius values of the approximate nipple 5 and the approximate recess circle 33, the recess corresponding to the optic disc recess (C) with respect to the approximate nipple 5 corresponding to the optic disc (D). A value of the C / D ratio indicating the ratio of the approximate circle 33 is calculated (step S19). Then, the calculated C / D ratio value is output and displayed through the liquid crystal display 8 so that a doctor or the like can visually recognize it (step S20). Thereby, a C / D ratio value that is useful information for a doctor or the like to make a diagnosis of glaucoma can be obtained by a simple operation of importing the fundus image data 3 into the computer 2, and the diagnosis can be performed efficiently. Can provide support that is done accurately and accurately. At this time, if the value of the C / D ratio is 0.6 or more, it is determined that the suspicion of glaucoma is strong, and if the value is smaller than that, a diagnosis of being healthy is made.

  In particular, since operations such as recognition of the nipple region 4 and the recessed region 26 and measurement of the value of the radius or diameter, which have been performed almost manually in the past, are automated after receiving the input of the fundus image data 3, It is possible to greatly reduce the time required for diagnosis of many patients in a medical institution such as a medical checkup. In addition, the possibility of causing doctor's subjectivity and measurement error in the measurement of each of the areas 4 and 26 is reduced, and the calculation of the value of the C / D ratio is performed according to a predetermined standard. As a result, the objectivity of the calculated C / D ratio value is maintained, and there is no variation in the diagnosis result, so that a doctor or the like can make an accurate diagnosis. In addition, the computer 2 recognizes the region of the blood vessel that is superimposed on the nipple region 4 and the recessed region 26 and is photographed in the fundus photograph, and the nipple approximate circle 5 or the recessed approximate circle 33 is determined in consideration of this. Therefore, the calculation of the C / D ratio is more accurately performed.

  Thereafter, the presence / absence of an instruction to terminate the system is detected (step S21), and when such an instruction is present (YES in step S21), the system is terminated (step S22). On the other hand, if there is no instruction (NO in step S21), the process returns to step S1 and accepts input of new fundus image data 3.

  The present invention has been described with reference to preferred embodiments. However, the present invention is not limited to these embodiments, and various modifications can be made without departing from the spirit of the present invention as described below. And design changes are possible.

  That is, in the support system 1 of the present embodiment, the system that supports the diagnosis of glaucoma using the calculated C / D ratio is shown, but the present invention is not limited to this, and other findings are detected. It is also possible to support diagnosis by presenting it. For example, it is possible to detect “locality expansion”, which is a finding of a portion corresponding to the optic disc depression. In a glaucoma disease, there may be a local enlargement symptom in which the depression region 26 expands uniformly with respect to the papillary region 4 and expands only in one direction. Therefore, when the distance corresponding to the radius from the coordinate position indicating the center of the recessed area 26 to the recessed outline 28 is increased only in one direction, the calculated C / D ratio is a value smaller than 0.6. Can be diagnosed as glaucoma. In general, glaucoma is also suspected when the difference in the C / D ratio between the left and right eyeballs is 0.2 or more. Therefore, a left / right C / D ratio difference calculating means for obtaining a difference between the calculated left / right C / D ratios may be provided to provide information for assisting diagnosis of glaucoma to a doctor or the like.

  Further, in the support system 1 of the present embodiment, the C / D ratio value is calculated based on the radii of the approximated nipple circle 5 and the approximate recessed circle 33. Of course, the value is calculated based on the diameter. It does not matter. In addition, the nipple region 4, the recessed region 26, and the blood vessel region 6 are detected using the difference between pixel values after binarizing the candidate image data, but a threshold is set. The color component may be any one of an R (red) component, a green (G) component, and a blue (B) component. However, for the red (R) component and the green (G) component, the detection of the regions 4 and 26 is relatively easy, but the blue (B) component has a lower pixel value than the surrounding fundus region 18. In some cases, it may be difficult to detect each of the areas 4 and 26.

  Furthermore, in the present embodiment, an example in which the approximate circles 5 and 33 that approximate the respective regions 4 and 26 are determined for one region 4 and 26 is shown, but the present invention is not limited to this. One nipple region 4 (or the recessed region 26) may be divided into two from the center to the left and right, and divided into an ear side region and a nose side region to determine individual approximate circles. That is, since it is empirically known that generally the region on the ear side is clearly captured with respect to the nose side, each region is detected and approximated by applying a different threshold value to each region. Is determined more accurately.

It is a block diagram which shows the schematic structure of the assistance system of this embodiment, and the functional structure of a fundus image diagnosis assistance computer. It is explanatory drawing which shows an example of extraction of (a) received fundus image data and (b) candidate image data. It is explanatory drawing which shows an example of the detected blood vessel area | region. It is explanatory drawing which shows an example of the detected nipple area | region. (A) It is explanatory drawing which shows typically an example of the determination of a nipple approximate circle or a recessed approximate circle, and (b) the detection of a vascular region. It is explanatory drawing which shows an example of the detected recessed area. It is a flowchart which shows the flow of the calculation process of C / D ratio by a computer. It is a flowchart which shows the flow of the calculation process of C / D ratio by a computer.

Explanation of symbols

1 Support system (fundus image diagnosis support system)
2 Computer (fundus image diagnosis support computer)
3 fundus image data 4 nipple area 4a nipple area information 5 nipple approximate circle 6 blood vessel area 6a blood vessel area information 7 fundus camera 8 liquid crystal display 9 keyboard 10 mouse 11 operation input device 12 fundus image data input receiving means 13 storage means 14 brightest part DESCRIPTION OF SYMBOLS 15 Diagnosis candidate area | region 16 Candidate image data 17 Candidate area extraction means 18 Fundus area 19 Nipple area detection means 20 Nipple outline 21 Nipple candidate circle 22 Nipple candidate circle selection means 23 Nipple coincidence degree calculation means 24 Nipple coincidence candidate circle 25 Nipple approximation circle determination Means 26 Recessed area 26a Recessed area information 27 Recessed area detection means 28 Recessed contour 29 Recessed candidate circle 30 Recessed candidate circle selection means 31 Recessed coincidence degree calculating means 32 Recessed coincidence candidate circle 33 Recessed approximate circle 34 Depression approximate circle determining means 35 Blood vessel region detecting means 36 Blood vessel region correcting means 37 C D ratio calculating means 37a C / D ratio data 38 calculated result output control unit 39 an operation signal input control means E1 nipple matching score (candidate circle matching degree)
E2 Depression coincidence (candidate circle coincidence)
A surplus area B coincidence area C deficient area Va surplus blood vessel area Vb coincident blood vessel area Vc deficient blood vessel area

Claims (6)

A fundus image diagnosis support system capable of analyzing a fundus photograph of a fundus image of a subject using a fundus image diagnosis support computer and providing information supporting a doctor's diagnosis,
The fundus image diagnosis support computer includes:
Fundus image data input receiving means for receiving fundus image data input created from the fundus photograph;
Candidate region extraction means for extracting from the fundus image data a diagnostic candidate region including the optic disc from the received fundus image data;
A papillary region detecting means for detecting a substantially circular papillary region corresponding to the optic disc from candidate image data indicating the extracted diagnostic candidate region;
A nipple approximate circle calculating means for calculating a nipple approximate circle that approximates the shape of the nipple contour indicating the boundary between the detected nipple region and the fundus region;
A recessed area detecting means for detecting a substantially circular recessed area corresponding to an optic disc depression from the candidate image data;
A recessed approximate circle calculating means for calculating a recessed approximate circle that approximates a shape of a recessed contour indicating a boundary between the detected recessed region and the nipple region;
Fundus image diagnosis support, comprising: C / D ratio calculating means for obtaining a C / D ratio indicating a ratio of either a radius or a diameter of the approximated nipple circle with respect to the calculated approximate circle for depression. system. At least one of the nipple approximate circle calculating means and the recessed approximate circle calculating means is:
Candidate circle selecting means for selecting a candidate circle that approximates the nipple contour or the recessed contour;
It is located on the inner circle side of the selected candidate circle and is located on the inner circle side of the candidate circle from the coincidence region overlapping the nipple region or the recessed region, and is not located on the nipple region or the recessed region. A degree of coincidence calculating means for calculating a candidate circle coincidence degree, which is located on the outer circle side of the surplus area of the superimposition and the candidate circle and subtracts the defect area overlapping the nipple area or the recessed area;
Approximate circle determination means for determining, from the plurality of candidate circles selected by the candidate circle selection means, a matching candidate circle having the maximum candidate circle matching degree as the nipple approximate circle or the recessed approximate circle. The fundus image diagnosis support system according to claim 1. A blood vessel region detection means for detecting a blood vessel region included in the candidate image data and corresponding to a blood vessel imaged superimposed on at least one of the nipple region and the recessed region;
The blood vessel region correcting means for correcting and calculating at least one of the approximated nipple circle and the approximated recess circle based on the detected blood vessel region. Item 3. A fundus image diagnosis support system according to Item 2.   A fundus image data input receiving unit that receives an input of fundus image data created from a fundus image obtained by photographing the fundus of the subject, a candidate region that extracts a diagnostic candidate region including the optic nerve head from the received fundus image data from the fundus image data An extraction means, a papillary area detection means for detecting a substantially circular papillary area corresponding to the optic disc from the extracted candidate image data indicating the diagnostic candidate area, and a boundary between the detected papillary area and the fundus area A nipple approximate circle calculating means for calculating a nipple approximate circle that approximates the shape of the nipple contour, a recessed area detecting means for detecting a substantially circular recessed area corresponding to an optic disc recess from the candidate image data, A recessed approximate circle calculating means for calculating a recessed approximate circle that approximates the shape of a recessed contour indicating a boundary between the recessed region and the nipple region, and the calculated previous A fundus image diagnostic computer that causes a fundus image diagnosis support computer to function as C / D ratio calculation means for obtaining a C / D ratio indicating a ratio of either a radius or a diameter of the nipple approximate circle to a recessed approximate circle Diagnosis support program.   Candidate circle selecting means for selecting a candidate circle that approximates the nipple contour or the recessed contour, located on the inner circle side of the selected candidate circle, and from a matching region that overlaps the nipple region or the recessed region, A surplus area that is located on the inner circle side of the candidate circle and is not superimposed on the nipple area or the recessed area, and a missing area that is located on the outer circle side of the candidate circle and overlaps the nipple area or the recessed area. A matching degree calculation means for calculating a candidate circle matching degree obtained by subtracting a candidate circle and a plurality of candidate circles selected by the candidate circle selection means, a matching candidate circle having a maximum candidate circle matching degree is determined as the nipple approximate circle or The fundus image diagnosis support program according to claim 4, further causing the fundus image diagnosis support computer to function as an approximate circle determination unit that determines the approximate circle for the depression.   A blood vessel region detecting means for detecting a blood vessel region corresponding to a blood vessel included in the candidate image data and superimposed on at least one of the papillary region and the recessed region; and the detected blood vessel region The fundus image diagnosis support computer is further caused to function as a blood vessel region correction unit that calculates and corrects at least one of the nipple approximate circle and the concave approximate circle based on the circle. Item 6. A fundus image diagnosis support program according to Item 5.

Images