WO2008010305A1

WO2008010305A1 - Image analyzer and image analysis program

Info

Publication number: WO2008010305A1
Application number: PCT/JP2006/318911
Authority: WO
Inventors: Hiroshi Fujita; Toshiaki Nakagawa; Yoshinori Hayashi
Original assignee: Gifu University; Tak Co., Ltd.
Priority date: 2006-07-19
Filing date: 2006-09-25
Publication date: 2008-01-24
Also published as: JP2008022928A

Abstract

[PROBLEMS] To provide an image analyzer and an image analysis program whereby blood vessels taken in an eye-fundus image are efficiently extracted and analyzed at a high accuracy. [MEANS FOR SOLVING PROBLEMS] An image analyzer by which an area to be analyzed, wherein blood vessels suitable for analysis are taken, and another area not to be analyzed, wherein no blood vessel suitable for analysis is taken, are determined in an eye-fundus image and blood vessel analysis is conducted by exclusively using the image of blood vessels located in the area to be analyzed.

Description

Specification

Image analysis apparatus and image analysis program

Technical field

The present invention relates to an image analysis apparatus and an image analysis program that efficiently extract blood vessels captured in a fundus image and analyze them with high accuracy.

Background art

[0002] It is known that analysis of blood vessels taken in a fundus image is effective for diagnosing the health condition and the presence or absence of a disease. For example, it is possible to diagnose arteriosclerotic changes by analyzing the state of the intersection of blood vessels, and by calculating the ratio of the thickness of the fundus artery and vein, that is, the arteriovenous aperture ratio. It is possible to estimate the presence or absence of hypertensive changes. Conventionally, the most widely used method for analyzing fundus images is manual analysis by a doctor or other expert in analysis. However, when analyzing a large amount of fundus images obtained by medical examinations, etc., manual analysis is too time consuming, and a more efficient analysis method has been demanded.

[0003] Various attempts for reducing the load on the expert of analysis and performing analysis more efficiently have been disclosed. Patent Document 1 discloses a measurement method for quantitatively calculating an arteriovenous aperture ratio as an example of blood vessel characteristic analysis. In the measurement method of Patent Document 1, all blood vessels photographed in a concentric region from the optic nerve head are extracted, and a plurality of blood vessel pairs are selected by pairing blood vessels with a short relative inter-vascular distance. ing. Then, from the difference in luminance data between the paired blood vessels, it is determined whether the blood vessel is an artery or a vein, and the arteriovenous aperture ratio is measured.

[0004] However, most of the blood vessels in the fundus extend the optic nerve head force, and when the technique of Patent Document 1 is applied, almost all blood vessels in the fundus image are identified and extracted from the image. For this reason, the blood vessel measurement method of Patent Document 1 has a problem that it takes a very long time to identify and extract blood vessels. In particular, it has been difficult to efficiently analyze the characteristics of blood vessels using a large number of fundus photographs obtained through medical examinations as measurement targets.

[0005] In addition, according to the measurement method of Patent Document 1, the optic papilla force is extended in the direction toward the macula. Blood vessels with a narrow diameter were also uniformly extracted and used for measurement. Such a thin blood vessel is difficult to determine whether it is an artery or a vein by analyzing the luminance data of the image, and therefore it is possible to make an incorrect determination. There was a potential for inaccurate results.

Patent Document 1: Japanese Patent Laid-Open No. 10-243924

Disclosure of the invention

Problems to be solved by the invention

[0006] Conventionally, analysis has been advanced by extracting substantially all blood vessels imaged in the fundus image. As a result, the analysis takes a very long time and it is difficult to increase the efficiency of the analysis. In addition, there is a possibility that the characteristic value is calculated using an image of a blood vessel that is not originally suitable for analysis, and thus the accuracy of the obtained analysis result may not be sufficient.

[0007] The present invention has been made to solve the above-described problem, and it is highly probable that a blood vessel suitable for analysis is present in advance among blood vessels captured in a fundus image. By setting the region force and extracting the blood vessel, the processing speed for extracting the blood vessel is improved, and the image analysis apparatus capable of performing the analysis more efficiently than before is provided. At the same time, it was made in order to provide an image analysis apparatus with improved analysis accuracy by excluding blood vessels that are not suitable for analysis from the analysis.

Furthermore, an object of the present invention is to provide an image analysis program that performs analysis with high accuracy and efficiency by extracting blood vessels suitable for analysis from blood vessels captured in a fundus image. It was made.

Means for solving the problem

[0009] The invention of claim 1 relates to an image analysis apparatus that analyzes a blood vessel imaged in a fundus image. The image analysis apparatus according to the present invention includes, in the fundus image, an analysis target region for analyzing a blood vessel, and a region setting unit for setting a non-analysis target region without performing blood vessel analysis! The blood vessel is analyzed using the image of the blood vessel located in the analysis target region set by the region setting means.

[0010] As a result of analyzing and examining many fundus images, the inventors have a high probability that a blood vessel suitable for analysis exists on the fundus, and there are almost no blood vessels suitable for analysis. The area is In other words, we found that there are blood vessels whose characteristics such as diameter and shape are not suitable for analysis in regions where there are few blood vessels suitable for analysis. Then, do not identify and extract the blood vessels included in the region where there are almost no blood vessels suitable for analysis as non-analysis regions! By setting the high! ヽ region as the analysis target region and using the blood vessel image included in the analysis target region, it is possible to efficiently analyze the blood vessel with high accuracy.

[0011] The invention of claim 2 relates to an image analysis apparatus that analyzes a fundus image and analyzes a blood vessel imaged in the fundus image. The image analysis apparatus of the present invention includes an optic nerve head position detecting means for detecting a position corresponding to the optic nerve head in the fundus image, and a blood vessel to be analyzed for either one of the right eye and the left eye. The position storage means for storing the position information in which the eye is present, and whether the fundus image is an image obtained by photographing the right eye or the left eye, and the position storage means stores the blood vessel position information. If the image is determined to be a side image, the image left / right reversal processing means for horizontally reversing the fundus image, the blood vessel position information stored in the position storage means, and the detection result of the optic disc position detection means And a region setting means for setting a region to be analyzed for analyzing blood vessels and a region not to be analyzed for not analyzing blood vessels in the fundus image. The area setting means of the present invention sets the analysis target area and the non-analysis target area for the image after the reversal processing when the image left / right reversal processing means performs the process of horizontally reversing the fundus image, A blood vessel is prayed using an image of a blood vessel located in the analysis target region set by the region setting means.

[0012] What is the image analysis apparatus of the present invention? A optic disc position detecting means is provided. Since most of the blood vessels in the fundus extend the optic nerve head force, the analysis target area and the analysis target are determined based on the information on the position of the blood vessel to be analyzed by using the position of the optic nerve head as a reference. The outer region can be determined efficiently.

[0013] The image analysis apparatus of the present invention also includes position storage means for storing blood vessel position information suitable for analysis for either the right eye or the left eye, and a fundus image of the right eye and the left eye. It is determined which of the images was taken, and the position information of the blood vessels is stored in the position storage means! An image left / right reversal process that reverses the fundus image when it is determined to be a cunning ヽ side image. Has a step. As a result of various studies, the inventors have found that the position where the blood vessel to be analyzed exists is substantially bilaterally symmetric between the right eye and the left eye, and based on this knowledge, the image left / right reversal processing means of the present invention. Configured. When the position storage means stores the position information of the blood vessel and analyzes the image of the eye on the side, the image is reversed by the image left / right reversing means and the image after reversal is used. As in the case of the fundus image on the side where the position information is stored, the analysis target region and the non-analysis region can be set.

[0014] The invention of claim 3 relates to an image analysis apparatus that analyzes a fundus image and analyzes a blood vessel imaged in the fundus image. The image analysis apparatus of the present invention includes an optic nerve head position detecting means for detecting a position corresponding to the optic nerve head in the fundus image, and one of the right eye and the left eye! Remember the location information of the blood vessels! The position memory means and whether the fundus image is an image of the right eye or the left eye is determined, and the position information of the blood vessel is stored in the position memory means. If the position storage means determines that the position information of the blood vessel stored in the position storage means is reversed horizontally, the position information left / right reversal processing means, the position information of the blood vessel, and the position corresponding to the optic nerve head are detected. Using the results, the fundus image is provided with an area setting means for setting an analysis target area for performing blood vessel analysis and a non-analysis target area for not performing blood vessel analysis. In the region setting method of the present invention, when the position information left / right reversing processing means performs processing for reversing the blood vessel position information to the left / right, the region information is analyzed using the position information of the blood vessel after the reversal processing. A non-target region is set, and a blood vessel is analyzed using an image of a blood vessel located in the analysis target region set by the region setting means.

The image analysis apparatus of the present invention can efficiently determine the analysis target region and the non-analysis target region on the basis of the position of the optic papilla detected by the optic papilla position detection means. In addition, the image analysis apparatus of the present invention stores the position information of the blood vessel suitable for the analysis of either the right eye or the left eye, and the position information is stored! The fundus of the side eye When analyzing an image, the position information of the blood vessel is reversed left and right and used for the analysis. As a result, even if the image analysis apparatus stores the blood vessel position information only on one side, the blood vessel position information is stored! / The side of the fundus image is stored, the position information is stored! /, It is possible to analyze in the same way as an image. [0016] The invention of claim 4 relates to a region setting means of the image analysis device. The region setting means of the present invention is characterized in that a region extending from the optic nerve head to at least one of the ear side and the nose side in the fundus image is set as a non-analysis region. It is clear that the inventors have analyzed many fundus images that there is a low possibility that blood vessels suitable for analysis exist in the region. The shape of the non-analysis region of the present invention is This has been determined based on this finding. The non-analysis region can be defined by a region surrounded by a curve, a region surrounded by a polygon, a circle, or the like.

The invention of claim 5 relates to a region setting means of the image analysis device. What is the region setting method of the present invention? The region between two or more lines extending radially from the optic disc is characterized as the non-analysis region.

[0018] The invention of claim 6 relates to a region setting means of the image analysis apparatus. The region setting means of the present invention is characterized in that a region sandwiched between curves having an inflection point near the optic nerve head is set as a non-analysis region. The curve here can be defined by any curve that can be digitized, such as a quadratic curve, the circumference of an ellipse, and a Bezier curve.

The invention of claim 7 relates to a region setting means of the image analysis device. The region setting means of the present invention is characterized in that a fan-shaped region where the optic nerve head force spreads to at least one of the ear side and the nose side is set as a non-analysis region.

[0020] The invention of claim 8 relates to a region setting means of the image analysis device. The region setting means of the present invention is characterized in that a substantially triangular region extending from the optic nerve head to the ear side and the nose side is set as a non-analysis region.

[0021] The invention of claim 9 relates to a region setting means of the image analysis device. The region setting means of the present invention further includes a macular portion detecting means for detecting a position corresponding to the macular portion in the fundus image. The region setting means sets a region having a shape which is symmetrical with respect to a line connecting the center of the optic nerve head and the center of the macula as a non-analysis region.

[0022] The image analysis device according to claim 10 is characterized in that the content of blood vessel analysis is calculation of an arteriovenous aperture ratio.

[0023] The invention of claim 11 is an image analysis product that analyzes a blood vessel captured in a fundus image. Concerning The image analysis program of the present invention includes a procedure for setting an analysis target region for performing blood vessel analysis processing and a non-analysis target region for performing blood vessel analysis processing in the fundus image, and a region within the analysis target region. The computer is caused to execute a procedure for analyzing a blood vessel using an image of a blood vessel located in the center.

The invention's effect

[0024] The image analysis apparatus and the image analysis program of the present invention have a low probability of existence of a blood vessel suitable for analysis, and do not identify and extract a blood vessel included in the region as a non-analysis region. High probability of having blood vessels suitable for analysis! By setting the heel region as the region to be analyzed and extracting only the images of blood vessels contained therein, blood vessels can be analyzed efficiently and in a shorter analysis time.

[0025] Furthermore, the image analysis apparatus and the image analysis program of the present invention can select only a blood vessel suitable for analysis, extract an image force, and analyze it, so that the analysis accuracy can be improved.

Example

Hereinafter, an embodiment in which the image analysis apparatus of the present invention is applied to an analysis apparatus for the arteriovenous aperture ratio of a blood vessel imaged in a fundus image will be described in detail with reference to FIGS. . (Embodiment 1) FIG. 8 shows a block configuration diagram of the image analysis apparatus 41 of the present embodiment. The image analysis apparatus 41 according to the present embodiment includes a region setting processing unit 52 that sets an analysis target region for analyzing blood vessels and a non-analysis target region for performing blood vessel analysis in the fundus image. Speak with it. Further, the blood vessel extraction processing unit 56, the analysis target blood vessel determination processing unit 54, and the arteriovenous aperture ratio calculation processing unit 58 are included. These processing units 52 to 58 in the present embodiment are stored in the internal storage means 48 of the computer 42 as programs in a format that can be executed by the CPU (central processing unit) 46, and are sequentially executed to perform image analysis. Is called. As the computer 42, an ordinary personal computer or workstation having an input / output unit 44 and a main memory, or a photographing means (not shown) can be applied in addition to the internal storage means 48 and the CPU 46.

[0027] Further, the image analysis device 41 includes position storage means that stores position information on the presence of blood vessels to be analyzed for the arteriovenous caliber ratio for each of the right eye and the left eye and creates a database. The analysis target blood vessel position database 60 that functions as an internal storage means 48 is stored. The analysis target blood vessel position database 60 of the present embodiment stores the numerical values of the blood vessel positions and various characteristics obtained as a result of analysis of many fundus images. In addition, analysis blood vessel distribution data calculated based on these numerical data is stored. Analyzed blood vessel distribution data has characteristics suitable for analysis, and stores data on the numerical distribution of the distribution of blood vessels on the fundus image for analysis of the arteriovenous aperture ratio.

[0028] Among the analyzed blood vessel distribution data of the analysis target blood vessel position database 60, the thick blood vessel suitable for the analysis of the arteriovenous caliber ratio is a direction from 10 o'clock to 2 o'clock from the center of the optic nerve head, It is specified that the lower part is located between the direction of the 4 o'clock direction at 8 o'clock.

[0029] In the analysis target blood vessel position database 60, there is a high possibility that there is a blood vessel branched from a thick blood vessel, if necessary, position data, characteristics of pixel values of blood vessels suitable for analysis, The position and size of the optic disc used as a reference for the position of the blood vessel used in the analysis are also stored.

[0030] The contents of the analysis of the arteriovenous aperture ratio of the blood vessel imaged on the fundus image, which is executed by the image analysis device 41 of the present embodiment, will be described below with reference to the flowchart of FIG. First, in step S2, a fundus image is taken. The fundus image can be obtained by photographing by a photographing unit provided in the image analysis device 41, and can be used by inputting a fundus image photographed by any photographing device from the outside.

[0031] The fundus image analyzed in the present embodiment is taken in a state of being adjusted so that the macular portion is approximately in the center of the image. In the image, identification information for identifying whether the photographed eye is a right eye or a left eye is recorded. However, if it is clear in advance that all fundus images to be analyzed are either the right eye or the left eye, the recording of identification information can be omitted.

The fundus image to be analyzed is a color digital image, and includes two-dimensional coordinate data corresponding to the position of the pixel and pixel value data which is color tone information. If a fundus image taken with an analog input is input and analyzed from an external camera, a two-dimensional coordinate system is set in advance on the image and digitally processed to support coordinate value data and coordinate values. You can obtain the pixel value data of the image to be analyzed and proceed with the analysis. [0033] The region setting processing unit 52 of the image analysis apparatus refers to the analysis target blood vessel position database 60 based on the identification information indicating whether the right eye or the left eye is stored in the fundus image to be analyzed! Then, analysis blood vessel distribution data corresponding to the eye to be analyzed is extracted. Alternatively, if it is clear in advance whether the fundus image to be analyzed is right eye force or left eye, the operator manually inputs the identification information into the image analysis system, and the region setting processing unit is based on the input information! Corresponding analysis blood vessel distribution data can be extracted from the analysis target blood vessel position database 60. Then, the region setting processing unit 52 sets, as the analysis target region, a region with a high probability that the blood vessel to be analyzed is distributed in the fundus image to be analyzed based on the extracted analysis blood vessel distribution data. The region with a low probability of blood vessels being distributed is set as a non-analysis region. (Step S4). At this time, the region setting processing unit determines the boundary between the analysis target region and the non-analysis target region by applying a predefined shape and digitizing the contour shape of the non-analysis target region. The image analysis device does not perform extraction processing of blood vessels included in the set non-analysis region. On the other hand, since the blood vessels included in the analysis target region are likely to be suitable for analysis, processing for extracting the image force is performed in the following steps, and the target is analyzed.

FIG. 5 shows the non-analysis region 6 and the analysis target region 7 set in the fundus image 1 of the right eye by the region setting processing unit 52 of the present embodiment. The region setting processing unit 52 sets the analysis target region 7 and the non-analysis region 6 after adjusting the blood vessel distribution data extracted from the analysis target blood vessel position database 60 according to the resolution of the fundus image 1. Yes. The non-analysis area 6 is an area that extends from the center of the image to the ear side, that is, at 9 o'clock, and avoids an area that has a high probability of having a blood vessel 5 branched from the blood vessel 4 to be analyzed. Is set. The region setting processing unit 52 defines the boundary between the analysis target region 7 and the non-analysis target region 6 with a shape connecting the curves.

[0035] The blood vessel extraction processing unit 56 of the image analysis apparatus uses the fundus image 1 in which the non-analysis region 6 and the analysis target region 7 are set, and extracts blood vessels photographed in the analysis target region 7 ( Step S6). The extraction of blood vessels is performed by identifying, for the analysis target region 7 of the fundus image 1, a region having a pixel value of a color tone unique to the blood vessel portion and a region having a pixel value of a color tone other than the blood vessel portion. . The extraction of blood vessels is focused on the change in pixel values between the blood vessel and its surroundings. A position where the amount of change in the pixel value is large can also be defined as the boundary between the blood vessel part and the surrounding area.

When all blood vessels that can be identified by the analysis of pixel values are extracted from the analysis target region 7, the analysis target blood vessel determination processing unit 54 determines the thicknesses of all the extracted blood vessel portions. Then, a blood vessel having a sufficient thickness not less than the reference value is determined as a blood vessel to be analyzed (step S8). When the analysis target blood vessel is determined, the arteriovenous aperture ratio calculation processing unit 58 analyzes the pixel value of the analysis target blood vessel and the characteristic amount of the blood vessel diameter in more detail to determine whether the blood vessel is an artery or a vein. Identify. Then, the arteriovenous aperture ratio is calculated from the ratio of the obtained arterial to venous blood vessel thickness (step S10), and the process is terminated.

[0037] The image analysis apparatus 41 according to the present embodiment includes numerical data of blood vessel positions and various characteristics obtained from analysis results of a large number of fundus images, and analysis blood vessel distribution data calculated based on these numerical data. The analysis target blood vessel position database 60 is stored. Then, using the analysis blood vessel distribution data, the region setting processing unit 52 sets the non-analysis region 6 and the analysis target region 7 in the fundus image 1. The blood vessel extraction processing unit 56 does not identify and extract blood vessels in the non-analysis target region 6, but extracts only the images of blood vessels included in the analysis target region 7, thereby reducing the analysis time in a shorter time. Blood vessel analysis can be performed. However, since this image analysis apparatus can extract only blood vessels having characteristics suitable for analysis, the value of the arteriovenous aperture ratio using the blood vessels is very accurate.

[0038] (Embodiment 2) The region setting processing unit of the image analysis apparatus according to the present embodiment analyzes an analysis target blood vessel position database storing information on a position where a blood vessel of the fundus to be analyzed exists. Based on this information, the shape of the boundary of the non-analyzed region 8 of the blood vessel is defined by a circle. FIG. 6 shows the non-analysis region 8 and the analysis target region 9 that are set in the fundus image 1 of the right eye by the image analysis apparatus of the present embodiment. The non-analysis region is always defined as a circular region centered around the position near the ear from the center of the fundus image, although its radius and position vary slightly depending on the content of the analyzed blood vessel distribution data. The image to be analyzed, the configuration of the apparatus, the contents of data in the analysis target blood vessel position database, and the flow of processing in this embodiment are the same as those in the first embodiment, and redundant description is omitted.

(Embodiment 3) An image analysis apparatus of this embodiment includes an area setting processing unit, a blood vessel extraction processing unit, a solution In addition to the analysis target blood vessel position database that stores information on the position of the blood vessel of the fundus to be analyzed, the optic nerve head position detection processing And an image horizontal reversal processing unit. These processing units are also stored as programs in a format that can be executed by a CPU (central processing unit), as in the case of the processing units of the first and second embodiments. Done.

[0040] The analysis target blood vessel position database according to the present embodiment stores both numerical data related to the positions of the blood vessels in the fundus of the right eye and numerical data related to the positions of the blood vessels in the fundus of the left eye. . Furthermore, analysis blood vessel distribution data that digitizes the distribution of blood vessels that have characteristics suitable for analysis with respect to the right eye, and analysis blood vessels that quantify the distribution of blood vessels that have characteristics suitable for analysis with respect to the left eye It has analysis blood vessel distribution data that integrates the distribution data. From the analysis results of many fundus images, it is clear that the positions of the blood vessels to be analyzed for the left eye and the right eye are almost symmetrical, and the analysis blood vessel distribution data of the left eye is inverted. Thus, it is possible to integrate analysis blood vessel distribution data of the right eye. By performing such integration, the analysis target blood vessel position database of the present embodiment can increase the number of data about the fundus on one side and improve the accuracy of position information. It provides very reliable data.

The contents of the analysis of the arteriovenous aperture ratio of the blood vessel imaged on the fundus image, which is executed by the image analysis apparatus of the present embodiment, will be described with reference to the flowchart of FIG. First, in step S22, a fundus image is taken. The image analysis apparatus according to the present embodiment can capture and analyze fundus images similar to those analyzed in the first and ^second embodiments.

[0042] In step S24, the image horizontal reversal processing unit of the present embodiment recognizes the identification information recorded in the fundus image and determines whether the fundus image is an image of the left or right eye. . If it is determined that the fundus image is the left eye image, the image is horizontally reversed in step S26. An example of the left eye image determined by the image inversion processing unit is shown in FIG. 4 (a), and the left eye image after the left / right inversion processing is shown in FIG. 4 (b). The horizontal reversal processing of the image is performed by rearranging the arrangement of the pixel value data for each pixel so as to be reversed in the horizontal direction.

[0043] In step S28, the optic disc position detection processing unit of the image analysis device performs pixel detection of the fundus image. The value is analyzed to detect the optic nerve head imaged in the image. The optic nerve head has a unique pixel value like a blood vessel, and its shape is almost circular, so that the fundus image force can be easily identified. Since most of the blood vessels in the fundus extend from the optic nerve head and travel through the retina, the position of the optic nerve head is important for grasping the position of the blood vessel and identifying the power force that is a blood vessel suitable for analysis. Information.

[0044] In step S30, the region setting processing unit extracts the analyzed blood vessel distribution data of the analysis target blood vessel position database. Based on the extracted data, the region with high probability that the blood vessels to be analyzed will be distributed in the fundus image to be analyzed is set as the region to be analyzed, and the region with low probability that the blood vessels to be analyzed will be distributed. Is set as a non-analysis area. At this time, if the image to be analyzed is a fundus image obtained by capturing the right eye, the region setting processing unit sets the non-analysis region and the analysis target region using the image that has been captured and input. When analyzing the fundus image of the left eye, the analysis target area and non-analysis target area are set for the image after inversion processing.

[0045] The area setting processing unit of the present embodiment determines the shape of the non-analysis area as? It is defined by a closed shape approximated by a polygon spreading from the optic nerve head to the ear side and a closed shape approximated by a polygon spreading from the optic nerve power to the nose side, and set on the fundus image 1. it can. The number of sides of this polygon is determined based on the size and shape information of the area indicated by the analyzed blood vessel distribution data. The area setting processing unit according to the present embodiment performs? The position of the optic nerve head detected by the optic nerve head position detection processing unit is used as a region setting reference. In addition, the region setting processing unit uses the detected size of the optic disc to enlarge or reduce the area and shape of the region where the blood vessel to be analyzed indicated by the analysis blood vessel distribution data exists. It is also possible to set an area on the fundus image 1 after performing the operation.

The region setting processing unit of the present embodiment can set the non-analysis region 10 and the analysis target region 11 in the fundus image 1 as shown in FIG. In addition, as shown in FIG. 8, it is possible to set two regions 10 and 10 ′ outside the analysis target. By setting the two non-analysis regions 10 and 10 ', the probability that blood vessels that are not suitable for analysis will be further excluded from the analysis target will be increased, so analysis time will be more efficient and suitable for analysis in less time. Blood vessels are extracted.

[0047] In the blood vessel extraction processing unit of the image analysis apparatus, the non-analysis area 10 and the analysis area 11 are set. For the fundus image 1 in which the fundus image 1 or the non-analysis region 10, 10 'and the analysis region 11' are set, the pixel values of the analysis region 11 or the analysis region 11 are analyzed and photographed. The extracted blood vessel is extracted (step S32). When all the identifiable blood vessels are extracted from the analysis target region, the analysis target blood vessel determination processing unit determines the thickness of all the extracted blood vessel portions. Then, a blood vessel having a sufficient thickness not less than the reference value is determined as a blood vessel to be analyzed (step S34). When the analysis target blood vessel is determined, the arteriovenous caliber ratio calculation processing unit analyzes the pixel value of the analysis target blood vessel and the feature value of the blood vessel diameter in more detail, and identifies whether the blood vessel is an artery or a vein. To do. Then, the arteriovenous caliber ratio is calculated from the ratio of the obtained arterial and venous blood vessel thickness (step S36), and the process is terminated.

[0048] The image analysis apparatus according to the present embodiment focuses on the left-right symmetry of the analysis blood vessel distribution data related to the position of the blood vessel of the right eye and the analysis blood vessel distribution data related to the position of the blood vessel of the left eye. It is integrated as a single data that can be applied to both eyes. By storing analysis blood vessel distribution data that can be applied to both the left and right eyes, data for the right and left can be stored! The base structure is simplified, access to the database can be speeded up, and both eyes can be analyzed with high accuracy and efficiency.

[0049] The image analysis apparatus according to the present embodiment includes an image left / right reversing processing unit capable of horizontally reversing data at the time of photographing for the fundus image of the left eye. Therefore, the region setting processing unit can set the region as it is for the analysis of the fundus image obtained by photographing the right eye, and the analysis of the fundus image obtained by photographing the left eye is not performed for the image after the reversal process. The same process can be applied easily and quickly.

[0050] Further, what is the image analysis apparatus of this embodiment? Equipped with a optic nerve head position detection processing unit that detects the position of the optic nerve head and sets the analysis target area and the non-analysis target area based on that position for more accurate area setting. Thus, it is possible to reliably extract only blood vessels suitable for analysis and calculate the arteriovenous aperture ratio. For this reason, the accuracy of analysis can be further improved.

[0051] (Embodiment 4) The region setting processing unit of the image analysis apparatus of the present invention uses the shape of a closed region corresponding to a non-analysis region of a blood vessel. Radially from the inside of the optic nerve head to the edge of the image It is defined as the region between the extended four line segments and extending from the optic papilla to the ear side! /, To the nose side. In FIG. 9, the image analysis apparatus of the present embodiment sets the non-analysis regions 12, 12 ′ and the analysis target region 13 using the analysis target blood vessel position database data and the optic nerve head position data. The fundus image 1 of the right eye is shown. The image to be analyzed, the configuration of the apparatus, and the processing flow relating to the present embodiment are the same as those in the third embodiment, and a duplicate description is omitted.

[0052] (Embodiment 5) The fundus image analyzed by the image analysis apparatus of the present embodiment is photographed with a square fundus. The region setting processing unit of the image analysis apparatus according to the present embodiment forms the shape of the closed region corresponding to the non-analysis region of the blood vessel radially from the optic papilla to the end of the image 31 as in the fourth embodiment. It is defined as an area between four stretched line segments. In FIG. 10, the region setting processing unit of this example sets the non-analysis regions 14, 14 ′ and the analysis target region 15 using the data of the analysis target blood vessel position database and the data of the position of the optic nerve head. The fundus image 31 of the right eye is shown. The image to be analyzed, the configuration of the apparatus, and the flow of processing related to the present embodiment are the same as those in the third embodiment, and redundant description is omitted.

(Example 6) The image analysis apparatus of this example includes an analysis target blood vessel position database in which a large number of numerical data related to the blood vessel position obtained from the analysis result is stored. Data is updated as new data is added each time an analysis is performed. The analysis target blood vessel position database always stores the analysis blood vessel distribution data updated to the latest state based on the latest numerical data.

In FIG. 11, the region setting processing unit of the present embodiment uses the analysis blood vessel distribution data of the analysis target blood vessel position database and the detected position of the optic nerve head, the non-analysis region 16 and the analysis target region 17. The fundus image 1 of the right eye for which is set. The boundary between the non-analysis region 16 and the analysis target region 17 is defined by a quadratic curve, and this quadratic curve has an inflection point near the optic disc. The image to be analyzed, the configuration of the apparatus, and the processing flow related to the embodiment are the same as those in the third embodiment, and redundant description is omitted.

(Example 7) The analysis target blood vessel position database of the image analysis apparatus of the present example stores a large number of numerical data related to the positions of blood vessels obtained as a result of analysis. The data is updated by adding new data every time analysis is performed. The analysis target blood vessel position database always stores the analysis blood vessel distribution data updated to the latest state based on the latest numerical data. Analyzed blood vessel distribution data is stored as a set of a plurality of feature points representing boundaries of regions where the blood vessels to be analyzed have a high probability of distribution.

[0056] What is the area setting processing unit of this embodiment? A quadratic curve that passes through the optic disc 2 detected by the optic disc location detection processor and is approximated from the feature points of the analyzed blood vessel distribution data is defined as the boundary between the analysis target region and the non-analysis target region. Yes. Then, a region surrounded by the quadratic curve and the end of the fundus image 1 is set as the non-analysis regions 16, 16 ′. FIG. 12 shows a fundus image 1 of the right eye in which the non-analysis regions 16, 16 ′ and the analysis target region 17 ′ set by the image analysis apparatus of the present embodiment are set.

Note that the region setting processing unit of the present embodiment defines the shapes of the non-analysis target regions 16, 16 ′ as regions surrounded by a quadratic curve, but passes through the feature points of the analyzed blood vessel distribution data. When defining the shape of the non-analysis area with a curved line, it is clear that the non-analysis area can be set using a curve such as an ellipse or a Bezier curve.

(Embodiment 8) An image analysis apparatus according to this embodiment includes an area setting processing section, a blood vessel extraction processing section, an analysis target blood vessel determination processing section, an arteriovenous caliber ratio calculation processing section,? It includes a optic disc position detection processing section, a position information left / right inversion processing section, and an analysis target blood vessel position database. These processing units and analysis target blood vessel position database in the present embodiment can be executed by a CPU (Central Processing Unit) in the internal storage means of the computer in the same manner as the configuration of the image analysis apparatus in the embodiment described so far. It is stored as a format program and is executed sequentially for image analysis.

[0059] The analysis target blood vessel position database of the present embodiment stores a large number of numerical data related to the position of the blood vessel in the fundus image of the right eye, and analysis blood vessel distribution data obtained by analysis of the numerical data. Yes. Since the positions of the blood vessels of the right eye and the left eye are almost symmetrical, the analysis blood vessel distribution data for the right eye stored in the analysis target blood vessel position database of this embodiment can be reversed left and right. It can be applied to the analysis of the fundus image of the left eye. The contents of the analysis of the arteriovenous aperture ratio of the blood vessel imaged in the fundus image, which is executed by the image analysis apparatus of the present embodiment, will be described with reference to the flowchart of FIG. First, in step S42, a fundus image is taken. The image analysis apparatus according to the present embodiment can capture and analyze fundus images similar to those analyzed in the third to seventh embodiments. It is not necessary for the fundus image to be analyzed to be taken and memorize the left and right eye identification information.

[0061] The optic nerve head position detection processing unit of the image analysis device analyzes the pixel value of the fundus image and detects the optic nerve head imaged in the image (step S44). In step S46, the position information left / right inversion processing unit determines whether the fundus image is an image obtained by photographing the left or right eye based on the position of the optic nerve head in the image. That is, when the center of the optic nerve head is located on the right side of the image, the right eye image is determined, and when the center of the optic nerve head is located on the left side of the image, the left It is determined that the image is an eye image. If it is determined that the fundus image is an image of the left eye, in step S48, the analyzed blood vessel distribution data stored in the analysis target blood vessel position data base is reversed left and right.

[0062] In step S50, the region setting processing unit sets, as the analysis target region, a region with a high probability that the blood vessel to be analyzed is distributed in the fundus image to be analyzed based on the analysis blood vessel distribution data. The region with low probability of distribution of blood vessels to be analyzed is set as a non-analysis region. At this time, if the image to be analyzed is a fundus image obtained by photographing the right eye, the region setting processing unit sets the non-analysis region and the analysis target region using the analysis blood vessel distribution data in the analysis target blood vessel position database as they are. To do. When analyzing the fundus image of the left eye, the analysis target area and the non-analysis target area are set using the analysis blood vessel distribution data after the inversion process.

[0063] The region setting processing unit of the present embodiment defines the shape of the non-analysis region as a fan shape that spreads to the ear side or the nose side with reference to the optic nerve head, and sets the shape on the fundus image 1. be able to . The size and position of the fan shape are determined based on the information on the size and shape of the region where the analysis target blood vessel is likely to exist, as indicated by the analysis blood vessel distribution data. In addition to the case where the arc part coincides with the edge of the fundus image as in the non-analysis regions 18 and 18 'of Fig. 14, it is included in the fundus image as in the non-analysis region 20 in FIG. May be set in a sector.

[0064] The blood vessel extraction processing unit of the image analysis device is photographed by analyzing the pixel value of the analysis target region with respect to the fundus image 1 in which the fan-shaped non-analysis region and the analysis target region are set. A blood vessel is extracted (step S52). When all the identifiable blood vessels are extracted from the analysis target region, the analysis target blood vessel determination processing unit determines the thickness of all the extracted blood vessel portions. Then, a blood vessel having a sufficient thickness not less than the reference value is determined as a blood vessel to be analyzed (step S54). When the analysis target blood vessel is determined, the arteriovenous caliber ratio calculation processing unit further analyzes in detail the pixel value of the analysis target blood vessel and the feature value of the blood vessel diameter, and identifies whether the blood vessel is an artery or a vein. . Then, the arteriovenous caliber ratio is calculated from the ratio of the obtained arterial and venous blood vessel thickness (step S56), and the process is terminated.

[0065] The image analysis apparatus according to the present embodiment focuses on the symmetry of the position of the blood vessel of the right eye and the position of the blood vessel of the left eye so that only the data of the position of the blood vessel of the right eye is accumulated. Using the blood vessel database, it is possible to analyze the fundus image of both the left and right eyes. Compared with the database storing the blood vessel position data of the left and right eyes, the analysis target blood vessel position database of the present embodiment simplifies the structure of the analysis target blood vessel database, and allows fast access to the database. Therefore, it is possible to analyze the left and right eyes and the shifted eyes efficiently and with high accuracy.

[0066] In the present embodiment, the image analysis apparatus including the analysis target blood vessel position database in which numerical data related to the position of the blood vessel of the right eye has been described in detail, but numerical data of the left eye is stored. It is apparent from the description of this embodiment that an image analysis apparatus including the analyzed blood vessel position database can be configured.

[Embodiment 9] The image to be analyzed, the apparatus configuration, and the flow of processing relating to the present embodiment are the same as those in Embodiment 8, and redundant description is omitted. The region setting processing unit of the image analysis apparatus according to the present embodiment stores the shape of the non-analysis target region of the blood vessel based on the analysis blood vessel distribution data stored in the analysis target blood vessel position database. Define with triangles. For example, as shown in FIG. 15, the region setting processing unit includes a triangular region 22 extending from the optic nerve head toward the ear side with reference to the position of the optic nerve head 2 detected by the optic nerve head position detection processing unit. Set the triangular area 22 'that extends closer to the nose as the non-analysis area . Also, depending on the information of the analyzed blood vessel distribution data, as shown in FIG. 16, two triangles 24 extending from the optic papilla 2 may be set as non-analysis regions.

(Embodiment 10) An image analysis apparatus according to the present embodiment includes an area setting processing unit, a blood vessel extraction processing unit, an analysis target blood vessel determination processing unit, an arteriovenous aperture ratio calculation processing unit, and so on. It includes a optic disc position detection processing unit, a macular position detection processing unit, and an analysis target blood vessel position database. All the processing units and the analysis target blood vessel position database in this embodiment can be executed by the CPU (central processing unit) in the internal storage means of the computer, as in the configuration of the image analysis apparatus in the embodiment described above. Are stored as various types of programs, and are sequentially executed for image analysis.

The macular portion position detection processing unit analyzes the pixel value of the fundus image 1 and detects the position of the macular portion 30 captured in the image. The macula 30 can be easily identified in the fundus image because it has a lower brightness in the image and a darker color than the surrounding area.

[0070] The analysis target blood vessel position database of the present embodiment stores the information of the position of the blood vessel as the analysis target of the arteriovenous aperture ratio for each of the right eye and the left eye as a database. What is the memorized blood vessel position? Numerical values are based on the positions of the optic disc and the macula.

As shown in FIG. 17, the region setting processing unit according to the present embodiment performs processing before setting the analysis target region and the non-analysis target region in the fundus image 1 to detect the center of the detected optic disc 2. Set a reference line 32 that connects the center of the macula 30 with the center. If the reference line 32 is not horizontal, the position for rotating the fundus image 1 is corrected. Then, the non-analyzed region 26 and the analysis target region 27 are set for the corrected fundus image 1 based on the analysis blood vessel distribution data in the analysis target blood vessel position database.

The region setting processing unit determines the boundary between the non-analysis target region 26 and the analysis target region 27 by approximating the shape of the non-analysis target region with a predefined shape. The shape of the non-analysis region 26 shown in FIG. 17 is defined by a region surrounded by a curve extending toward the ear side. However, by defining the shape of the non-analyzed region to be set in advance in the region setting processing unit, even in a region sandwiched between two line segments extending in a circle, polygon, or radial, sector, triangle, etc. It is possible to define this area. Here, the region setting processing unit sets the non-analysis target region 26 so as to be line-symmetric with respect to the reference line 32. This is a setting based on the knowledge that the vascular force suitable for analysis of the arteriovenous caliber ratio 2 and the macular region are positioned almost vertically, and the non-analyzed region 26 is line-symmetric. Thus, the analysis target region 27 is also line symmetric. In this way, the analysis target region 27 set in line symmetry includes all the blood vessels suitable for analysis, and blood vessels are extracted from the analysis target region 27 and extracted from the non-analysis target region 26. By not performing the analysis, the analysis is performed efficiently.

[0074] What is the image analysis apparatus of this embodiment? In addition to the optic disc position detection processing unit, a macular position detection processing unit is provided, and both the optic disc 2 and macular 30 detected in the fundus image 1 to be analyzed are used as the position reference. Thus, it is possible to correct when the fundus image 1 is rotated. As a result, more accurate region setting of the non-analysis region 26 and the analysis target region 27 is performed, and the accuracy of the analysis can be further improved.

As described above, specific examples of the present invention have been described in detail in the embodiments. However, these are merely examples and do not limit the scope of the claims. The technology described in the claims includes various modifications and changes of the specific examples illustrated above. For example, the configuration of the image analysis device in the present embodiment can be individually configured as an external device of the computer by modularizing it in addition to storing it in the computer in the form of a program. In addition, the data structure of the analysis target blood vessel position database described in each embodiment can be arbitrarily combined with the shape of the non-analysis area set by the area setting processing unit.

Brief Description of Drawings

FIG. 1 is a flowchart showing the contents of blood vessel analysis processing executed by the image analysis apparatus according to the first embodiment.

FIG. 2 is a flowchart showing the contents of a blood vessel analysis process executed by the image analysis apparatus according to the third embodiment.

FIG. 3 is a flowchart showing the contents of blood vessel analysis processing executed by the image analysis apparatus of the eighth embodiment.

FIG. 4 is a diagram showing a fundus image of the left eye before and after left / right reversal processing. FIG. 5 is a diagram schematically showing a non-analysis region 6 and an analysis target region 7 set in the fundus image 1 of the right eye by the image analysis apparatus of Example 1.

FIG. 6 is a diagram schematically showing a non-analysis region 8 and an analysis target region 9 set in the fundus image 1 by the image analysis apparatus of the second embodiment.

FIG. 7 is a diagram schematically showing a non-analysis region 10 and an analysis target region 11 set in the fundus image 1 by the image analysis apparatus of the third embodiment.

FIG. 8 is a diagram schematically showing non-analysis regions 10, 10 ′ and analysis target region 11 ′ set in fundus image 1 by the image analysis apparatus of Example 3.

FIG. 9 is a diagram schematically showing the non-analysis regions 12, 12 ′ and the analysis target region 13 set in the fundus image 1 by the image analysis apparatus of Example 4.

FIG. 10 is a diagram schematically showing the non-analysis regions 14, 14 ′ and the analysis target region 15 set in the fundus image 31 by the image analysis apparatus of Example 5.

[11] FIG. 11 is a diagram schematically showing the non-analysis region 16 and the analysis target region 17 set in the fundus image 1 by the image analysis apparatus of Example 6.

FIG. 12 is a diagram schematically showing non-analysis regions 16, 16 ′ and analysis target region 17 ′ set in fundus image 1 by the image analysis apparatus of Example 7.

FIG. 13 is a diagram schematically showing non-analysis regions 18, 18 ′ and analysis target region 19 set in fundus image 1 by the image analysis apparatus of Example 8.

14] FIG. 14 is a diagram schematically showing the non-analysis region 20 and the analysis target region 21 set in the fundus image 1 by the image analysis apparatus of the eighth embodiment.

FIG. 15 is a diagram schematically showing the non-analysis regions 22, 22 ′ and the analysis target region 23 set in the fundus image 1 by the image analysis apparatus of Example 9.

FIG. 16 is a diagram schematically showing the non-analysis regions 24 and 24 ′ and the analysis target region 25 set in the fundus image 1 by the image analysis apparatus of Example 9.

FIG. 17 is a diagram schematically showing a reference line 32, a non-analysis region 26, and an analysis target region 27 set in the fundus image 1 by the image analysis apparatus of the tenth embodiment.

FIG. 18 is a block configuration diagram schematically showing the configuration of the image analysis device according to the first embodiment.

Explanation of symbols 1, 31 Fundus image 2 Optic disc 4 Blood vessel 5 Analysis vessel 5 Blood vessel 4 Branching vessel 6, 8, 10, 12, 14, 16, 18, 20, 22, 24, 26 Non-analysis region 7, 9, 11, 13, 15,, 17, 19, 21, 23, 25, 27 Analysis target area 32 Reference line 41 Image analysis device 42 Computer 44 Input / output unit 46 CPU 48 Internal storage means 52 Area setting processing unit 54 Determination of blood vessel to be analyzed Processing unit 56 Blood vessel extraction processing unit 58 Arteriovenous aperture ratio calculation processing unit 60 Analysis target vessel position database

Claims

The scope of the claims

[1] An image analysis apparatus that analyzes a blood vessel captured in a fundus image, and the image analysis apparatus analyzes an analysis target region for blood vessel analysis and a blood vessel analysis in the fundus image. A region setting means for setting a non-analysis target region to be performed, and performing blood vessel analysis using an image of a blood vessel located in the analysis target region set by the region setting means. An image analysis device.

[2] An image analysis device that analyzes a fundus image and analyzes a blood vessel captured by the fundus image, and the image analysis device corresponds to the optic disc in the fundus image. The optic disc position detecting means for detecting the position, the position storing means for storing the position information of the blood vessel to be analyzed for one of the right eye and the left eye, and the fundus image is the right eye Determine which image was taken of the left eye, and the position information of the blood vessel is stored in the position storage means. If it is determined that the image is the side image, the image is reversed horizontally. Processing means, and the position information of the blood vessels stored in the position storage means! Using the detection result of the optic nerve head position detection means, in the fundus image, an analysis target region for analyzing blood vessels and a region setting means for setting non-analysis target regions without performing blood vessel analysis are provided. The region setting means sets the analysis target region and the non-analysis target region for the image after the reversal processing when the image left / right reversal processing unit performs a process of reversing the fundus image horizontally. An image analysis apparatus characterized in that blood vessel analysis is performed using an image of a blood vessel located in the analysis target region set by the region setting means.

[3] An image analysis device that analyzes a fundus image and is captured in the fundus image, and analyzes the blood vessels. The image analysis device corresponds to the optic disc in the fundus image. The optic disc position detecting means for detecting the position, the position storing means for storing the position information of the blood vessel to be analyzed for one of the right eye and the left eye, and the fundus image of the right eye and the left eye It is determined which image is taken, and when it is determined that the position storage means stores the blood vessel position information, the position storage means stores the information! Using position information left / right reversing processing means for reversing left and right blood vessel position information, the blood vessel position information, and a detection result of a position corresponding to the optic nerve head, The bottom image includes an analysis target region for analyzing the blood vessel and a region setting unit for setting a non-analysis region for which the blood vessel analysis is not performed. The region setting unit includes the position information. When the left / right inversion processing means performs processing to invert the blood vessel position information to the left / right, the analysis target area and the non-analysis area are set using the blood vessel position information after the inversion process, and the area setting is performed. An image analyzing apparatus characterized in that an analysis of a blood vessel is performed using an image of a blood vessel located in the analysis target region set by the means.

[4] The image according to any one of claims 1 to 3, wherein the region setting means sets a region extending from the optic nerve head to at least one of the ear side and the nose side as a non-analysis region. Analysis device.

[5] The method according to any one of claims 1 to 3, wherein the region setting means sets a region sandwiched by two or more line segments radially extending from the optic nerve head as a non-analysis region. The image analysis device described in.

[6] The region setting means according to any one of claims 1 to 3, wherein the region sandwiched by a curve having an inflection point in the vicinity of the optic papilla is set as a non-analysis region. Image analysis device.

[7] The method according to any one of claims 1 to 3, wherein the region setting means sets a fan-shaped region extending from the optic nerve head to at least one of the ear side and the nose side as a non-analysis region. The image analysis device described.

[8] The method according to any one of claims 1 to 3, wherein the region setting means sets a substantially triangular region extending from the optic nerve head to the ear side and the nose side as a non-analysis region. Image analysis device.

[9] The apparatus further includes a macular detection unit that detects a position corresponding to the macular region in the fundus image, and the region setting unit detects a line connecting the center of the optic nerve head and the center of the macular region. The image analysis apparatus according to claim 1, wherein a region having a line symmetrical shape is set as a non-analysis region.

10. The image analysis apparatus according to any one of claims 1 to 9, wherein the content of blood vessel analysis is calculation of an arteriovenous aperture ratio.

[11] An image analysis program for analyzing a blood vessel photographed in a fundus image, wherein the eye In the bottom image, a procedure for setting an analysis target region for performing blood vessel analysis processing and a V ヽ non-analysis target region for performing blood vessel analysis processing, and an image of a blood vessel located in the analysis target region are shown. An image analysis program that allows a computer to execute a procedure for analyzing blood vessels.