JP2006149654A - Support of diagnosis about lesion of eye fundus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide technique for effectively supporting diagnosis about a lesion of the eye fundus by a medical doctor using an eye fundus image. <P>SOLUTION: An eye fundus image diagnosis supporting apparatus is provided with a time series image data base 62 and a diagnostic information data base 64. In the time series image data base 62, concerned area images imaged in time series from the initial stage to the final stage of the lesion by each name of the lesion and each plurality of patients are stored. In the diagnostic information data base 64, patient information and attribute information of the lesion are stored. A retrieval part 34 retrieves time series images including a similar image similar to an image to be diagnosed based on the image to be diagnosed inputted to an image data input part 31 and the patient information input to a patient information input part 32. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a technique for supporting diagnosis of a fundus lesion using a fundus image.

  Conventionally, when diagnosing a fundus lesion using a fundus image, a doctor images the fundus of a patient with a fundus camera, refers to the fundus image, the patient's past history, various examination results, Diagnosis was based on the knowledge and experience of doctors. However, because there are individual differences in doctors' knowledge and experience, when a plurality of doctors diagnose a fundus lesion of a single patient, not all doctors always make an accurate diagnosis.

  In recent years, referring to a case database including an image database that records a plurality of images and a diagnostic database that records diagnostic data related to each image, a similar image that is similar to the image to be diagnosed is searched and the search is performed. A system for displaying a result on a monitor has been proposed (see Patent Document 1 below). With this technique, it is possible to search for cases similar to the image to be diagnosed from among a large number of images, and support the diagnosis of the doctor.

JP 2004-5364 A

  However, in the technique described in Patent Document 1, each image recorded in the image database is an image at a certain point captured for each of a large number of cases. Accordingly, it is not always possible to search for a similar image depending on the progress of the lesion in the diagnosis target image. Further, according to the technique described in Patent Document 1, future changes in lesions could not be predicted and diagnosed. In other words, there was room for improvement in support of lesion diagnosis.

  The present invention has been made to solve the above-described problems, and an object of the present invention is to provide a technique that more effectively supports diagnosis of a fundus lesion by a doctor using a fundus image than before. To do.

  In order to solve at least a part of the above-described problems, the present invention employs the following configuration in a fundus image diagnosis support apparatus that supports diagnosis of a fundus lesion using a predetermined fundus image. That is, the fundus image diagnosis support apparatus of the present invention is a time-series image database in which a plurality of at least a part of the fundus image captured in time series for each of a plurality of types of the lesions is recorded for each lesion group. A storage unit that stores information, an image input unit that inputs a diagnosis target image that is at least a part of the fundus image to be diagnosed, the time-series image database, and based on the diagnosis target image, The gist includes a search unit that searches for the group that includes similar images similar to the diagnosis target image, and an output unit that outputs at least one of a plurality of fundus images included in the searched group. To do.

  In the fundus image diagnosis support apparatus of the present invention, the fundus image to be handled is a digital image. This fundus image may be a fundus image captured by a so-called digital fundus camera, or may be an image obtained by capturing a fundus photograph captured by a fundus camera using a film with an image scanner. As the “lesion group”, for example, a group classified based on the name of a lesion such as hard vitiligo, an aneurysm, and punctate bleeding, a group for each patient, and a classification based on the progression rate of the lesion. Group. “Based on the diagnosis target image” means “based on characteristics such as color, size, shape, etc. of each part in the diagnosis target image”. The output unit may output the fundus image to a display device such as a monitor or may output it to a printer.

  In the present invention, the time-series image database records a plurality of fundus images captured in time series for each of a plurality of types of fundus lesions. The search can be performed with high accuracy. Further, the output unit is not limited to the similar image, and can output a fundus image captured after the similar image is captured in the searched group, for example. Can also be predicted and diagnosed. That is, the fundus image diagnosis support apparatus of the present invention can effectively support diagnosis of a fundus lesion by a doctor using the fundus image.

  In the fundus oculi image diagnosis support device, the time-series image database records at least a part of the fundus image captured in time series for each group, and further records for each patient, The storage unit further stores a diagnosis information database corresponding to at least a part of the fundus image captured in time series and recording predetermined diagnosis information about the patient, and the fundus image diagnosis support apparatus includes: And a diagnostic information input unit for inputting predetermined diagnostic information related to the patient corresponding to the diagnostic target image, wherein the search unit refers to the time-series image database and the diagnostic information database, and The search may be performed based on the image and the input diagnostic information.

  In the present specification, “predetermined diagnostic information about a patient” includes, for example, the patient's age, sex, medical history, test results such as blood pressure, visual acuity, intraocular pressure, and viewing angle. The type of diagnostic information input to the diagnostic information input unit can be arbitrarily set.

  In the present invention, the search is performed based not only on the diagnosis target image but also on the diagnosis information on the patient corresponding to the diagnosis target image. Therefore, the search result is obtained rather than performing the search based only on the diagnosis target image. You can narrow down.

  In the fundus image diagnosis support apparatus of the present invention, the time series image database is extracted from each of the time series imaged fundus images as at least a part of the time series imaged fundus images. An image of a predetermined region of interest including a lesion portion is recorded, and the image input unit inputs the entire fundus image to be diagnosed as the diagnosis target image, and the fundus image diagnosis support apparatus A region-of-interest image generation unit that extracts the region of interest from the entire inputted fundus image and generates an image of the region of interest; and the search unit is based on the generated image of the region of interest. The search may be performed.

  The fundus image includes various tissues such as the optic disc and retinal blood vessels in addition to the lesioned part. Since the color and shape of tissues such as the optic nerve head and retinal blood vessels are unique to each person, it may be difficult to perform the above search using the entire fundus image.

  In the present invention, since the above search is performed based on an image of a predetermined region of interest including a lesion portion, the influence of an image of a region other than the region of interest can be eliminated and the search can be performed with high accuracy. The region-of-interest image generation unit may extract a region of interest specified by the user and generate an image of the region of interest, or may extract a region of interest using a known image processing technique. An image of the region of interest may be generated.

  In the fundus oculi image diagnosis support device, the one selected by the user or the fundus image diagnosis support device from the diagnosis target image and images of a plurality of regions of interest included in the searched group. And a synthesized image generating unit that generates a synthesized image by synthesizing the image of the region of interest, and the output unit may output the generated synthesized image.

  By doing so, for example, it is possible to output a fundus image (synthetic image) indicating a state of a lesion estimated in the past of a patient to be diagnosed or a fundus image indicating a state of a lesion predicted in the future.

  In the fundus oculi image diagnosis support device, the composite image generation unit is further picked up after the image of the diagnosis target image and the image of the selected one region of interest included in the searched group. The image of the region of interest may be combined to generate a plurality of combined images, and the output unit may sequentially output the plurality of combined images in the order of the time series at a predetermined timing. .

  The images of the plurality of regions of interest that are combined with the diagnosis target image are not necessarily images that are continuously captured in time series. For example, when an image of a region of interest is captured every month, the images of the region of interest captured every year are combined with the diagnosis target images to generate a plurality of composite images. Also good. The predetermined timing may be a timing instructed by the user, or may be a preset timing such as a fixed period.

  According to the present invention, a fundus image showing the course of a lesion predicted in the future can be sequentially output in time series, so that the convenience of the fundus image diagnosis support apparatus can be improved.

  In the fundus image diagnosis support apparatus, the time-series image database records the image of the region of interest for each group and further for each patient, and the storage unit further includes the region of interest. A diagnostic information database in which predetermined diagnostic information related to the patient and predetermined attribute information related to the progression of the lesion are recorded, and the fundus image diagnosis support apparatus further includes the diagnostic object A diagnostic information input unit configured to input predetermined diagnostic information related to a patient corresponding to an image, wherein the search unit refers to the time-series image database and the diagnostic information database, the diagnosis target image, and the input The search is performed based on the diagnosed information, and the composite image generation unit further refers to the diagnosis information database, Based on the serial attribute information, it may be generated the synthesized image.

  Here, the “predetermined attribute information relating to the progression of the lesion” is information relating to the influence of the lesion on other portions other than the lesion portion on the fundus. The attribute information includes information such as the influence of the lesion on the periphery, the conditions under which the influence occurs, the behavior at the time of contact with retinal blood vessels, components of a healthy fundus such as the optic nerve head, and the like. As this attribute information, for example, “when a lesion comes into contact with a retinal blood vessel, the lesion passes under the retinal blood vessel and blocks the retinal blood vessel”, “inducing vitreous opacification after the middle stage of the lesion” And the like.

  In the present invention, since a composite image is generated based on the attribute information, a more accurate composite image can be generated.

  In the fundus oculi image diagnosis support apparatus, the search unit further includes the time of capturing an image of the region of interest as the similar image from the diagnosis information of the patient corresponding to the image of the region of interest included in the searched group. The diagnostic information corresponding to the image of the region of interest captured by the user or selected by the fundus image diagnosis support apparatus is extracted, and the output unit further extracts the extracted diagnostic information from the diagnostic target image. May be output as prognostic information estimated in the future of the patient corresponding to.

  By doing so, the doctor can know the prognostic information estimated in the future of the patient to be diagnosed, so the convenience of the fundus image diagnosis support apparatus can be improved.

  In the fundus image diagnosis support apparatus, the search unit further extracts a plurality of pieces of the diagnostic information, and the output unit uses the plurality of pieces of diagnostic information as the prognosis information in the order of the time series in a predetermined timing. Thus, the data may be output sequentially.

  The plurality of pieces of diagnostic information to be extracted do not necessarily have to be diagnostic information continuous in time series. For example, when diagnostic information is recorded every month, diagnostic information every year may be extracted. The predetermined timing may be a timing instructed by the user, or may be a preset timing such as a fixed period.

  According to the present invention, prognostic information predicted in the future can be sequentially output in time series, so that the convenience of the fundus image diagnosis support apparatus can be further improved.

  Further, in the fundus image diagnosis support apparatus, when outputting the prognosis diagnosis information, a simulation unit that simulates the prognosis diagnosis information based on the diagnosis information of the patient, the attribute information, the progress of the lesion, the progress speed, etc. You may make it prepare. By doing so, the convenience of the fundus image diagnosis support apparatus can be further improved.

  The present invention does not necessarily have all the various features described above, and may be configured by omitting some of them or combining them appropriately. The present invention can be configured as an invention of a support method that supports diagnosis of a fundus oculi lesion in addition to the configuration as the above-described fundus image diagnosis support device. Further, the present invention can be realized in various modes such as a computer program that realizes these, a recording medium that records the program, and a data signal that includes the program and is embodied in a carrier wave. In addition, in each aspect, it is possible to apply the various additional elements shown above.

  When the present invention is configured as a computer program or a recording medium that records the program, the entire program for controlling the operation of the fundus image diagnosis support apparatus may be configured, or only a portion that performs the function of the present invention. It is good also as what comprises. The recording medium includes a flexible disk, a CD-ROM, a DVD-ROM, a magneto-optical disk, an IC card, a ROM cartridge, a punch card, a printed matter on which a code such as a barcode is printed, a computer internal storage device (RAM or Various types of computer-readable media such as a memory such as a ROM and an external storage device can be used.

Hereinafter, embodiments of the present invention will be described in the following order based on examples.
A. Configuration of fundus image diagnosis support device:
B. Time series image database:
C. Diagnostic information database:
D. Function block:
E. Fundus image diagnosis support processing:
F. Variations:

A. Configuration of fundus image diagnosis support device:
FIG. 1 is an explanatory diagram showing a hardware configuration of a fundus image diagnosis support apparatus 100 as an embodiment of the present invention. The fundus image diagnosis support apparatus 100 is an apparatus for assisting a doctor in diagnosing a fundus lesion using a fundus image.

  As shown in the figure, a fundus image diagnosis support apparatus 100 according to the present embodiment includes a digital fundus camera 10 for capturing a fundus image of a patient, a CPU 20, a memory 30, an input device 40 such as a keyboard and a mouse, A display device 50 and a hard disk 60 are provided. The hard disk 60 stores a computer program for controlling the operation of the fundus image diagnosis support apparatus 100, a time series image database 62, and a diagnosis information database 64, which are executed by the CPU 20 on the memory 30. Yes.

B. Time series image database:
FIG. 2 is an explanatory diagram showing an example of the contents of the time-series image database 62. In this embodiment, the time-series image database 62 includes n types of lesions such as hard vitiligo, vascular aneurysm, and punctate hemorrhage for each lesion name and for each of a plurality of patients from the initial stage of the lesion. An image of a region of interest including a portion of a lesion imaged in time series (hereinafter referred to as a region-of-interest image) is stored in the last stage. In the present specification, for each patient, a plurality of region-of-interest images captured in time series from the initial stage to the end stage of a lesion are also referred to as time series images.

C. Diagnostic information database:
3 and 4 are explanatory diagrams showing an example of the contents of the diagnostic information database 64. FIG. In the present embodiment, patient information and lesion attribute information are stored in the diagnostic information database 64.

  FIG. 3 is an explanatory diagram showing an example of patient information. This patient information includes the patient's name, gender, age, blood pressure, intraocular pressure, visual acuity, viewing angle, and other test results, the patient's history, examination date, and treatment records. . Although not shown, this patient information is associated with each region of interest image in the time-series image database 32 shown in FIG. 2, and the patient information of each patient corresponds to the region of interest image. It is recorded in time series. This patient information corresponds to diagnostic information in the present invention.

  FIG. 4 is an explanatory diagram showing an example of lesion attribute information. As shown in the figure, this attribute information includes the name of the lesion, the effect that the lesion has on the surroundings, the conditions under which the effect occurs, and the contact with healthy retinal components such as retinal blood vessels and the optic disc. This includes information such as behavior. From this attribute information, for example, it can be seen that the lesion A passes under the retinal blood vessel and closes the retinal blood vessel when the lesion contacts the retinal blood vessel.

D. Function block:
FIG. 5 is an explanatory diagram illustrating functional blocks of the fundus image diagnosis support apparatus 100. As illustrated, the fundus image diagnosis support apparatus 100 includes an image data input unit 31, a diagnostic information input unit 32, an image processing unit 33, a search unit 34, and an output unit 35. The image processing unit 33 includes a region-of-interest image generation unit 332 and a composite image generation unit 334. Each of these functional blocks is configured by software by loading a predetermined computer program stored in the hard disk 60 into the memory 30. You may comprise at least one part of each of these functional blocks by hardware.

  The image data input unit 31 inputs the entire fundus image of the patient to be diagnosed imaged by the digital fundus camera 10 (hereinafter, “the entire fundus image of the patient to be diagnosed” is also referred to as “diagnosis object image”). The patient information input unit 32 inputs patient information of a patient to be diagnosed. The patient information may be manually input by the user using the input device 40, or may be read from the patient's electronic medical record prepared separately and input. The patient information input to the patient information input unit 32 is used as a key for narrowing down the search for time series images in the search unit 34, as will be described later. 64 may be a part of the items of patient information recorded in 64, and can be arbitrarily set. The patient information input unit 32 corresponds to the diagnostic information input unit in the present invention.

  The region-of-interest image generation unit 332 extracts a region of interest including a lesion from the entire diagnosis target image input to the image data input unit 31, and generates a region-of-interest image. The region-of-interest image generation unit 332 may generate a region-of-interest image by extracting the region of interest specified by the user on the diagnosis target image displayed on the screen of the display device 50, or may be known. A region of interest image may be generated by automatically extracting a region of interest using an image processing technique.

  FIG. 6 is an explanatory diagram showing how a region-of-interest image is generated. The fundus image 150 as the diagnosis target image input to the image data input unit 31 is shown on the left side of the figure, and the region-of-interest image 200 generated by the region-of-interest image generation unit 332 is shown on the right side of the figure. In the illustrated example, the fundus image 150 shows a state in which the optic disc 151, retinal blood vessels 152 and 153, and a lesion 154 are captured. In the fundus image 150, a region surrounded by a broken line is a region of interest (ROI). As illustrated on the right side of the figure, the region-of-interest image generation unit 332 extracts a region of interest (ROI) from the fundus image 150 and generates a region-of-interest image 200.

  The search unit 34 (see FIG. 5) refers to the time-series image database 62 and the diagnostic information database 64 stored in the hard disk 60, inputs the region-of-interest image generated by the region-of-interest image generation unit 332, and patient information. Based on the patient information input from the unit 32, the type of lesion including a similar image similar to the region-of-interest image generated by the region-of-interest image generation unit 332 and the patient are identified from the time-series image database 62. Search for similar images. The search for similar images is performed based on features such as the color, size, and shape of the lesion, and is narrowed down based on patient information. At this time, a plurality of similar images may be searched.

  FIG. 7 is an explanatory diagram showing a state of searching for similar images. Here, an example in which the time series images of “lesion n” and “patient: 5” are specified from the time series image database 62 shown in FIG. As shown in the figure, it is assumed that the time-series image database 62 stores ten region-of-interest images 62n-5a to 62n-5j for “lesion n” and “patient: 5”. In the illustrated example, the region-of-interest image 62n-5e is shown as a similar image.

  The composite image generation unit 334 (see FIG. 5) selects one interest selected by the user from the entire diagnosis target image input to the image data input unit 31 and the time-series image specified by the search unit 34. A composite image is generated by combining the region image. At this time, the composite image generation unit 334 generates a composite image using the attribute information described above. Generation of a composite image using attribute information will be described later. In this embodiment, the composite image generation unit 334 selects one image selected by the user from the entire diagnosis target image input to the image data input unit 31 and the time-series image specified by the search unit 34. The region-of-interest image is combined to generate a combined image, but the entire diagnosis target image and one region-of-interest image automatically selected by the fundus image diagnosis support device are combined, A composite image may be generated.

  FIG. 8 is an explanatory diagram showing how a composite image is generated. The fundus image 150 as the diagnosis target image input to the image data input unit 31 is shown on the upper left side of the figure. In the lower left part of the figure, a region-of-interest image 200A designated by the user is shown. Further, the composite image generated by the composite image generation unit 334 is shown on the right side of the figure. As illustrated on the right side of the figure, the composite image generation unit 334 generates a composite image 300 by combining the fundus image 150 and the region-of-interest image 200A.

  FIG. 9 is an explanatory diagram showing various composite images generated using attribute information. FIG. 9A shows the fundus image 150 input to the image data input unit 31. In this fundus image 150, an optic disc, retinal blood vessels 152 and 153, and a lesion 154 are captured. Here, the attribute information of the lesion 154 includes: “When the lesion comes into contact with the retinal blood vessel, the lesion passes under the retinal blood vessel, and the retinal blood vessel is blocked or necrotic. , “Vitreous vitreous turbidity occurs” is described.

  FIG. 9B shows a synthesized image 300A obtained by synthesizing the fundus image 150 and the region-of-interest image including the lesion 154A selected from the time-series images. In the illustrated example, the lesion 154 </ b> A has not yet contacted the optic nerve head 151 or the retinal blood vessels 152 and 153.

  FIG. 9C shows a synthesized image 300B obtained by synthesizing the fundus image 150 and the region-of-interest image including the lesion 154B selected from the time-series images. In the illustrated example, the synthetic image 200B shows that the lesion 154B is in contact with the retinal blood vessel 153 and passes below the retinal blood vessel 153.

  FIG. 9D shows a synthesized image 300C obtained by synthesizing the fundus image 150 and the region-of-interest image including the lesion 154C selected from the time-series images. In the illustrated example, the lesion 154C is in contact with the retinal blood vessel 153, passes under the retinal blood vessel 153, and the retinal blood vessel 153 is necrotized.

  FIG. 9E shows a composite image 300D obtained by combining the fundus image 150 and the region-of-interest image including the lesion 154D selected from the time-series image. In the illustrated example, the lesion 154D is in contact with the retinal blood vessel 153, passes under the retinal blood vessel 153, and the retinal blood vessel 153 is blocked and deformed.

  FIG. 9F shows a composite image 300E obtained by combining the fundus image 150 and the region-of-interest image including the lesion 154E selected from the time-series image. In the illustrated example, the lesion 154E is in contact with the retinal blood vessel 153, passes under the retinal blood vessel 153, and further shows that vitreous opacification has developed.

  In response to an instruction from the user, the output unit 35 (see FIG. 5) selects one region-of-interest image selected from the time-series images shown in FIG. 7, a synthesized image shown in FIG. The prognosis information to be output is output to the display device 50. The output unit 35 may output the region-of-interest image selected by the user, the composite image, and the prognosis information to the printer.

  FIG. 10 is an explanatory diagram showing an example of a display screen 500 for prognosis diagnosis information displayed on the display device 50. This display screen 500 is generated by the image processing unit 33. In the example shown in the figure, it is shown that prognosis information after five years is displayed. The display screen 500 includes a composite image 300 in which the fundus image 150 input to the image data input unit 31 and a region-of-interest image predicted after five years are combined, and a simulation of an estimated visual field at that time. An image 400, estimated clinical data 502, and estimated data 504 relating to the visual field are displayed. The display screen 500 is updated synchronously every time the user selects an image of interest to be combined with the fundus image 150 from the region of interest image shown in FIG.

  In the simulation image 400, an image that simulates the appearance of the reference object 402 is displayed as illustrated. In the simulation image 400, a region 404 indicates a region with a clear field of view, and a region 406 and a region 408 indicate a state in which the field of view is unclear in this order. An area 410 indicates that the visual field is lost due to a lesion.

  The estimated clinical data 502 and the estimated data 504 relating to the visual field are data corresponding to the region-of-interest image of the lesion that is read from the diagnostic information database 64 and predicted five years later.

  By displaying such prognostic information, the doctor can know the diagnostic information estimated in the future of the patient to be diagnosed, so that the convenience of the fundus image diagnosis support apparatus 100 can be improved. .

  In addition, by showing the progress of future symptoms (prognostic information) to the patient, it can be used for patient education, promotion of informed consent, implementation of doctor accountability, and the like.

E. Fundus image diagnosis support processing:
FIG. 11 is a flowchart showing the flow of fundus image diagnosis support processing. This process is a process executed by the CPU 20 of the fundus image diagnosis support apparatus 100.

  First, the CPU 20 inputs a diagnosis target image and patient information through the image data input unit 31 and the patient information input unit 32 (step S100).

  Next, the CPU 20 generates a region-of-interest image by the region-of-interest image generation unit 332 as described above (step S110).

  Next, the CPU 20 refers to the time-series image database 62 and the diagnostic information database 64 by the search unit 34 (step S120), the region-of-interest image generated in step S110, and the patient input in step S100. Based on the information, a time-series image including a similar image similar to the region-of-interest image generated in step S110 is searched (step S130).

  Next, the CPU 20 determines a region-of-interest image to be combined with the diagnosis target image from the time-series images searched in step S130 according to the user's instruction, and the combined image generation unit 334 previously explains the region-of-interest image. As described above, a composite image is generated (step S140).

  Next, the CPU 20 reads out diagnostic information corresponding to the region-of-interest image determined in step S140 by the search unit 34 from the diagnostic information database 64, and generates prognosis information by the image processing unit 33 (step S150). .

  And CPU20 outputs the prognosis diagnosis information produced | generated by step S150 by the output part 35 (step S160).

  Then, the CPU 20 determines whether or not an instruction to end the fundus image diagnosis support process has been input by the user (step S170). When the process is continued without inputting the fundus image diagnosis support process end instruction (step S170: NO), the process returns to step S140. When an instruction to end the fundus image diagnosis support process is input by the user (step S170: YES), the fundus image diagnosis support process ends.

  According to the fundus image diagnosis support apparatus 100 of the present embodiment described above, the time series image database 62 records a plurality of digital fundus images captured in time series for each of a plurality of types of fundus lesions. Therefore, it is possible to search for similar images and time-series images with higher accuracy than before, regardless of the progress of the lesion. Further, the output unit 35 can output not only the similar image but also the fundus image, the composite image, and the prognosis information captured after the image of the similar image in the searched time series image. The physician can predict and diagnose future changes in the lesion. That is, the fundus image diagnosis support apparatus 100 of the present invention can effectively support a diagnosis of a fundus lesion by a doctor using the fundus image.

F. Variations:
As mentioned above, although several embodiment of this invention was described, this invention is not limited to such embodiment at all, and implementation in a various aspect is possible within the range which does not deviate from the summary. It is. For example, the following modifications are possible.

F1. Modification 1:
In the above embodiment, the time series image database 62 records the region of interest image, and the search unit 34 searches for the time series image based on the region of interest image generated by the region of interest image generation unit 332. It was supposed to be, but it is not limited to this. The time series image database 62 records the entire fundus image, and the search unit 34 searches the time series image based on the entire fundus image to be diagnosed input to the image data input unit 31. Also good. However, the entire fundus image includes various tissues such as the optic nerve head and retinal blood vessels in addition to the lesion part, and the color and shape of these tissues are unique to each person. It may be difficult to perform the search using the entire image. Therefore, according to the above embodiment, it is possible to eliminate the influence of the image of the region other than the region of interest and perform the search with high accuracy.

F2. Modification 2:
In the above embodiment, prognosis information is output in the fundus image diagnosis support process, but the present invention is not limited to this. For example, steps S150 and S160 in FIG. 11 may be omitted, and only the composite image generated in step S140 may be output.

F3. Modification 3:
In the above-described embodiment, the composite image generation unit 334 generates a single composite image by combining the diagnosis target image and the single region-of-interest image selected by the user. Absent. The synthesized image generation unit 334 further synthesizes the diagnosis target image and the region of interest image captured after the region of interest image selected by the user is captured, generates a plurality of synthesized images, and outputs the output unit 35. Thus, a plurality of composite images may be sequentially output at a predetermined timing in time series order. By doing so, it is possible to sequentially output a fundus image showing the course of a lesion predicted in the future in time series, and thus the convenience of the fundus image diagnosis support apparatus can be improved. In this case, the plurality of region-of-interest images combined with the diagnosis target image do not necessarily have to be images continuously captured in time series. For example, when the region-of-interest image is captured every month, the region-of-interest image captured every year may be combined with the diagnosis target image to generate a plurality of composite images. . The predetermined timing may be a timing instructed by the user, or may be a preset timing such as a fixed period.

F4. Modification 4:
In the above embodiment, when outputting the prognosis diagnosis information, the search unit 34 extracts the diagnosis information corresponding to one region-of-interest image selected by the user from the diagnosis information database 64, and the image processing unit 33 Based on this diagnostic information, one prognostic diagnostic information is generated and output by the output unit 35. However, the present invention is not limited to this. The search unit 34 further extracts a plurality of pieces of diagnostic information, generates a plurality of prognostic diagnosis information based on the plurality of pieces of diagnostic information by the image processing unit 33, and outputs a plurality of pieces of prognostic diagnosis information by the output unit 35. The data may be output sequentially at a predetermined timing in the sequence order. By doing so, the prognostic information predicted in the future can be sequentially output in time series, so that the convenience of the fundus image diagnosis support apparatus 100 can be improved. Note that the plurality of pieces of diagnostic information to be extracted do not necessarily have to be diagnostic information continuous in time series. For example, when diagnostic information is recorded every month, diagnostic information every year may be extracted. The predetermined timing may be a timing instructed by the user, or may be a preset timing such as a fixed period.

F5. Modification 5:
In the above embodiment, the fundus image diagnosis support apparatus 100 is configured as a single apparatus, but is not limited thereto. For example, the fundus image captured by the server having the functional blocks shown in FIG. 5, the time series image database 62, the diagnostic information database 64, etc., and the digital fundus camera 10, and patient information are transmitted to the server. Alternatively, the fundus image diagnosis support system may be configured by connecting a client capable of displaying information received from the server via a network such as the Internet. By doing so, the function of the server can be shared among a plurality of doctors and a plurality of hospitals.

F6. Modification 6:
In the above-described embodiment, the fundus image diagnosis support apparatus that supports diagnosis of a fundus lesion is described, but the present invention is not limited to this. The present invention is also applicable to an image diagnosis support apparatus that supports the diagnosis of other lesions that change in time series, such as cancer, using images.

It is explanatory drawing which shows the hardware constitutions of the fundus image diagnosis assistance apparatus 100 as one Example of this invention. It is explanatory drawing which shows an example of the content of the time series image database. It is explanatory drawing which shows an example of patient information. It is explanatory drawing which shows an example of the attribute information of a lesion. It is explanatory drawing which shows the functional block of the fundus image diagnosis assistance apparatus. It is explanatory drawing which shows the mode of the production | generation of a region-of-interest image. It is explanatory drawing which shows the mode of the search of a similar image. It is explanatory drawing which shows the mode of the production | generation of a synthesized image. It is explanatory drawing which shows the various composite images produced | generated using attribute information. It is explanatory drawing which shows an example of the display screen of prognostic information. It is a flowchart which shows the flow of a fundus image diagnosis assistance process.

Explanation of symbols

100 ... Fundus image diagnosis support device 10 ... Digital fundus camera 20 ... CPU
30 ... Memory 31 ... Image data input unit 32 ... Patient information input unit 33 ... Image processing unit 34 ... Search unit 35 ... Output unit 40 ... Input device 50 ... Display device 60 ... Hard disk 62 ... Time series image database 64 ... Diagnostic information database 150 ... Fundus image 151 ... Optic nerve head 152, 153 ... Retinal blood vessels 154, 154A to 154E ... Lesions 200, 200A ... Region-of-interest images 300, 300A-300E ... Composite image 332 ... Region-of-interest image generation unit 334 ... Composite image generation unit 400 ... Simulation image 500 ... Display screen 502 ... Estimated clinical data 504 ... Estimated data on visual field ROI ... Region of interest

Claims (11)

A fundus image diagnosis support apparatus that supports diagnosis of a fundus lesion using a predetermined fundus image,
For each of the plurality of types of lesions, a storage unit that stores a plurality of time-series image databases in which at least a part of the fundus image captured in time series is recorded for each group of the lesions;
An image input unit that inputs a diagnosis target image that is at least a part of a fundus image to be diagnosed;
A search unit that references the time-series image database and searches for the group including similar images similar to the diagnosis target image based on the diagnosis target image;
An output unit that outputs at least one of a plurality of fundus images included in the searched group;
A fundus image diagnosis support apparatus. The fundus image diagnosis support apparatus according to claim 1,
The time-series image database records at least part of the fundus image captured in time series for each group, and further records for each patient,
The storage unit further stores a diagnostic information database in which predetermined diagnostic information about the patient is recorded corresponding to at least a part of the fundus image captured in time series.
The fundus image diagnosis support apparatus further includes a diagnosis information input unit that inputs predetermined diagnosis information about a patient corresponding to the diagnosis target image,
The search unit refers to the time-series image database and the diagnosis information database, and performs the search based on the diagnosis target image and the input diagnosis information.
Fundus image diagnosis support device. The fundus image diagnosis support apparatus according to claim 1,
The time-series image database includes a predetermined region of interest including a portion of the lesion extracted from each of the time-series imaged fundus images as at least a part of the time-series imaged fundus image. Records an image,
The image input unit inputs the entire fundus image to be diagnosed as the diagnosis target image,
The fundus image diagnosis support apparatus further includes a region-of-interest image generation unit that extracts the region of interest from the entire input fundus image and generates an image of the region of interest.
The search unit performs the search based on the generated image of the region of interest.
Fundus image diagnosis support device. The fundus image diagnosis support apparatus according to claim 3, further comprising:
The diagnostic target image is synthesized with a single region of interest image selected by the user or the fundus image diagnosis support apparatus from among a plurality of regions of interest images included in the searched group. A composite image generation unit for generating a composite image;
The output unit outputs the generated composite image;
Fundus image diagnosis support device. The fundus image diagnosis support apparatus according to claim 4,
The synthesized image generation unit further synthesizes the diagnosis target image and an image of the region of interest included in the searched group and captured after the image of the selected single region of interest is captured. And generating a plurality of the composite images,
The output unit sequentially outputs the plurality of composite images in a time sequence in a predetermined timing;
Fundus image diagnosis support device. The fundus image diagnosis support apparatus according to claim 4,
The time-series image database records the image of the region of interest for each group, and further records for each patient,
The storage unit further stores a diagnostic information database corresponding to the image of the region of interest, in which predetermined diagnostic information related to the patient and predetermined attribute information related to the progression of the lesion are recorded.
The fundus image diagnosis support apparatus further includes a diagnosis information input unit that inputs predetermined diagnosis information about a patient corresponding to the diagnosis target image,
The search unit refers to the time-series image database and the diagnosis information database, performs the search based on the diagnosis target image and the input diagnosis information,
The composite image generation unit further refers to the diagnostic information database and generates the composite image based on the attribute information.
Fundus image diagnosis support device. The fundus image diagnosis support apparatus according to claim 6,
The search unit is further imaged after imaging of the region of interest image as the similar image from the diagnosis information of the patient corresponding to the image of the region of interest included in the searched group, and the user, or Extracting diagnostic information corresponding to the image of the region of interest selected by the fundus image diagnosis support apparatus;
The output unit further outputs the extracted diagnostic information as prognostic information estimated in the future of a patient corresponding to the diagnosis target image.
Fundus image diagnosis support device. The fundus image diagnosis support apparatus according to claim 7,
The search unit further extracts a plurality of the diagnostic information,
The output unit sequentially outputs the plurality of pieces of diagnostic information as the prognostic information at a predetermined timing in the order of the time series.
Fundus image diagnosis support device. A support method for supporting diagnosis of a fundus lesion using a predetermined fundus image,
As a process executed by the computer,
(A) A step of storing, in a predetermined storage device, a time-series image database in which at least a part of a fundus image captured in time series for each of the plurality of types of lesions is recorded for each lesion group. When,
(B) obtaining a diagnostic object image that is at least a part of the fundus image to be diagnosed;
(C) referring to the time-series image database, and searching the group including a similar image similar to the diagnosis target image based on the diagnosis target image;
(D) outputting at least one of a plurality of fundus images included in the searched group;
A support method comprising: A computer program for supporting diagnosis of a fundus lesion using a predetermined fundus image,
A function of acquiring a diagnosis target image that is at least a part of a fundus image to be diagnosed;
For each of the plurality of types of lesions, refer to a time-series image database in which at least a part of a fundus image captured in time series is recorded for each group of lesions, and similar images similar to the diagnosis target image A function of searching for the group including:
A function of outputting at least one of a plurality of fundus images included in the searched group;
A computer program for realizing a computer.   A recording medium in which the computer program according to claim 10 is recorded so as to be readable by a computer.

Images