JP7237786B2 - Image processing device, image processing method - Google Patents

Description

The present invention relates to an image processing device and an image processing method.

By using an eye tomography apparatus such as an optical coherence tomography (OCT), it is possible to three-dimensionally observe the state inside the retinal layers. This tomography apparatus is widely used in ophthalmology because it is useful for more accurately diagnosing diseases.

In Patent Document 1, a first layer boundary and a second layer boundary are detected in an OCT tomographic image of the fundus, and another layer boundary is set based on the fractional ratio of the distance between the two layer boundaries. and a set layer boundary as a projection depth range to generate a front tomographic image.

U.S. Patent Publication No. 2013/0229621

However, for example, when there is an edema-like layer shape abnormality near the fovea M as shown in FIG. is different, the technique described in Patent Document 1 has a problem that an appropriate projection depth range cannot be set at a site where a layer shape abnormality exists.

SUMMARY OF THE INVENTION The present invention has been made in view of the above problems, and an object of the present invention is to robustly determine a layer boundary regardless of a site in a wide-angle tomographic image of an eye to be examined that includes a layer shape abnormality.

In order to solve the above problems, an image processing device according to one aspect of the present invention includes:
detection means for detecting the position of a first layer boundary candidate and the position of a second layer boundary candidate on a tomographic image of an eye to be inspected;
the position of the boundary candidate of the first layer on the tomographic image in the depth direction of the eye to be examined or the direction intersecting the boundary candidate of the first or the second layer; determining means for determining the position of at least a portion of the region with respect to the first layer based on the position of the estimated boundary of the first layer calculated from the position of the candidate boundary of the layer;
An image processing apparatus characterized by having
Further, an image processing apparatus according to another aspect of the present invention comprises: detecting means for detecting the position of a first layer boundary candidate and the position of a second layer boundary candidate on a tomographic image of an eye to be examined;
position of the first layer boundary candidate in the depth direction of the eye to be examined or in a direction intersecting the first or second layer boundary candidate on the tomographic image; a position of an estimated boundary of said first layer calculated based on a fractional ratio of the distances of a candidate layer boundary and a candidate third layer boundary; a determining means for determining
An image processing apparatus characterized by having
In addition, an image processing method according to another aspect of the present invention comprises:
a step of detecting the position of a first layer boundary candidate and the position of a second layer boundary candidate on a tomographic image of an eye to be inspected;
position of the first layer boundary candidate in the depth direction of the eye to be examined or in a direction intersecting the first or second layer boundary candidate on the tomographic image; determining the position of at least a portion of the region with respect to the first layer based on the position of the estimated first layer boundary calculated from the position of the candidate boundary of the layer;
An image processing method characterized by having

According to the present invention, in a wide-angle tomographic image of an eye to be examined that includes a layer shape abnormality, it is possible to robustly determine the layer boundary regardless of the region.

1 is a block diagram showing the configuration of an image processing system according to first and second embodiments; FIG. 4 is a flowchart of processing that can be executed by the image processing systems of the first embodiment and the second embodiment; FIG. 4 is a diagram for explaining properties of layer boundaries in tomographic images of a healthy eye and a diseased eye; FIG. 4 is a diagram illustrating an example of a frontal tomographic image, a scanning method for OCTA imaging, a projection artifact occurring in a motion contrast image, and a frontal motion contrast image generated in the first and second embodiments; It is a figure explaining the example of the report screen displayed on a display means in 1st Embodiment and 2nd Embodiment. It is a figure explaining the example of the report screen displayed on a display means in 1st Embodiment and 2nd Embodiment. FIG. 4 is a diagram illustrating an example of image quality enhancement processing executed in S203 and S301 of the first embodiment and S203 and S401 of the second embodiment;

[First embodiment]
After detecting a layer boundary candidate for a wide-angle OCT tomographic image, the image processing apparatus according to the present embodiment determines the layer boundary position according to the distance from a predetermined site. It is determined so that the amount of change from is small, and the amount of change is large at a position away from the distance. A case will be described in which layer boundaries are robustly determined irrespective of sites in a wide-angle tomographic image of an eye to be inspected that includes layer shape anomalies.

The image processing system including the image processing apparatus according to this embodiment will be described in detail below.

FIG. 1 is a diagram showing the configuration of an image processing system 100 including an image processing apparatus 300 according to this embodiment. As shown in FIG. 1, in the image processing system 100, an image processing apparatus 300 is connected to a tomographic image capturing apparatus 200, a fundus image capturing section 400, an external storage section 500, a display section 600, and an input section 700 via interfaces. It is configured by

The tomographic image capturing apparatus 200 is an apparatus for capturing a wide-angle tomographic image of the eye, and is composed of an OCT apparatus such as SS-OCT. Not limited to this, SD-OCT may be used. Since the tomographic image capturing apparatus 200 is a known apparatus, a detailed description thereof will be omitted. Here, the tomographic image capturing performed according to an instruction from the image processing apparatus 300 will be described.

In FIG. 1, a galvanomirror 201 is for scanning the fundus with measurement light, and defines the imaging range of the fundus by OCT. Further, the drive control unit 202 controls the drive range and speed of the galvanomirror 201 to define the planar imaging range and the number of scanning lines (scanning speed in the planar direction) on the fundus. Here, the galvanometer mirror is shown as one unit for the sake of simplicity, but it actually consists of two mirrors for X scanning and Y scanning, so that a desired range on the fundus can be scanned with the measurement light. .

A focus 203 is for focusing on a predetermined depth position of the fundus through the anterior ocular segment of the subject's eye. The measurement light is focused on the fundus tissue (the layer forming the retina or choroid) through the anterior ocular segment of the subject's eye by a focus lens (not shown). The measurement light that irradiates the fundus is reflected and scattered by each layer of the fundus and returns.

The internal fixation lamp 204 is composed of a display section 204-1 and a lens 204-2. A plurality of light emitting diodes (LD) arranged in a matrix is used as the display section 204-1. The lighting position of the light-emitting diode is changed according to the part to be photographed under the control of the drive control unit 202 . Light from the display unit 204-1 is guided to the subject's eye via the lens 204-2. The light emitted from the display unit 204-1 has a wavelength of 520 nm, and a desired pattern is displayed by the drive control unit 202. FIG.

The coherence gate stage 205 is controlled by the drive control unit 202 in order to cope with the difference in axial length of the subject's eye. A coherence gate represents a position in OCT where the optical distances of the measurement light and the reference light are equal. Furthermore, by controlling the position of the coherence gate as an imaging method, it is controlled to perform imaging on the surface side of the vitreous body or the retina, or on the deeper side than the outer layer of the retina.

(Configuration of image processing device)
The configuration of the image processing apparatus 300 of this embodiment will be described with reference to FIG.

The image processing apparatus 300 is a personal computer (PC) connected to the tomography apparatus 200 and includes an image acquisition unit 301 , a storage unit 302 , an imaging control unit 303 , an image processing unit 304 and a display control unit 305 . Further, the image processing device 300 implements functions by executing software modules that implement the image acquisition unit 301 , the shooting control unit 303 , the image processing unit 304 and the display control unit 305 by the arithmetic processing unit CPU. The present invention is not limited to this. For example, the image processing unit 304 may be realized by dedicated hardware such as ASIC, and the display control unit 305 may be realized by using a dedicated processor such as a GPU different from the CPU. good too. Also, the connection between the tomography apparatus 200 and the image processing apparatus 300 may be configured via a network.

The image acquiring unit 301 acquires signal data of the SLO fundus image captured by the fundus image capturing unit 400 and the tomographic image captured by the tomographic image capturing apparatus 200 . The image acquisition unit 301 is an example of acquisition means according to the present embodiment. The image acquisition unit 301 also has a tomographic image generation unit 301-1 and a motion contrast data generation unit 301-2. The tomographic image generating unit 301-1 acquires signal data (interference signal) of a tomographic image captured by the tomographic imaging apparatus 200, generates a tomographic image by signal processing, and stores the generated tomographic image in the storage unit 302. . The motion contrast data generation unit 301-2 calculates the degree of difference in signal values between tomographic images belonging to the same cluster (in this embodiment, the decorrelation value of luminance values) to determine the interaction between the displacement of red blood cells and the measurement light. The motion contrast obtained by the action is imaged. The imaging control unit 303 performs imaging control on the tomography apparatus 200 . Imaging control includes instructing the tomographic imaging apparatus 200 regarding the setting of imaging parameters, and instructing the start or end of imaging.

The image processing unit 304 includes an alignment unit 304-1, an image quality enhancement unit 304-2, an image feature acquisition unit 304-3, a layer determination unit 304-4, a correction unit 304-5, a projection unit 304-6, and a measurement unit. 304-7. The layer determination unit 304-4 has a detection unit 304-41, a calculation unit 304-42, and a determination unit 304-43. The image quality enhancing unit 304-2 performs synthesis processing between tomographic images or motion contrast images based on the alignment parameters obtained by the alignment unit 304-1, or combines tomographic images or motion contrast images based on the learning model. image quality enhancement processing. The image feature acquisition unit 304-3 acquires the positions of the fovea fovea and the center of the optic papilla from the tomographic image. The layer determining unit 304-4 acquires the layer boundaries of the retina and choroid, and the boundaries of the anterior and posterior surfaces of the cribriform plate from the tomographic image. The correction unit 304-5 performs processing for two-dimensionally or three-dimensionally suppressing projection artifacts occurring in the motion contrast image (projection artifacts will be described later). A projection unit 304-6 projects a tomographic image or a motion contrast image in the depth range specified by the layer determination unit 304-4 to generate a front tomographic image or a front motion contrast image.

The external storage unit 500 associates information about the subject's eye (patient's name, age, sex, etc.) with captured images (tomographic images, SLO images, OCTA images), composite images, imaging parameters, and parameters set by the operator. holding. The input unit 700 is, for example, a mouse, a keyboard, a touch operation screen, or the like.

Next, the processing procedure of the image processing apparatus 300 of this embodiment will be described with reference to FIG. FIG. 2 is a flow chart showing the flow of operation processing of the entire system in this embodiment.

<Step 201>
By operating the input unit 700 , the operator sets the imaging conditions for the OCTA image to be instructed to the tomography apparatus 200 .

Specifically, the procedure consists of 1) selection or registration of an examination set, 2) selection or addition of a scan mode in the selected examination set, and 3) imaging parameter setting corresponding to the scan mode. Then, in S202, OCTA imaging (under the same imaging conditions) is repeatedly performed a predetermined number of times while intervening breaks as appropriate.
1) Register Ultra Wide examination set 2) Select OCTA scan mode 3) Set the following imaging parameters 3-1) Scanning pattern: Super Large
3-2) Scanning area size: 23x20mm
3-3) Main scanning direction: horizontal direction 3-4) Scanning interval: 0.02 mm
3-5) Fixation light position: macula 3-6) Number of B-scans per cluster: 3
3-7) Coherence gate position: Vitreous body side 3-8) Default display report type: Report for single examination Refers to the default display method for OCT and OCTA images.

As a result, an examination set including an OCTA scan mode set for wide-angle imaging is registered under the name "Ultra Wide". The registered examination set is stored in the external storage unit 500 .

In this embodiment, "Ultra Wide" is selected as the examination set, and "OCTA" mode is selected as the scan mode.

<Step 202>
By operating the input unit 700 and pressing an imaging start button (not displayed) on the imaging screen, the operator starts repeated OCTA imaging under the imaging conditions specified in S201.

The imaging control unit 303 instructs the tomographic imaging apparatus 200 to repeatedly perform OCTA imaging based on the settings instructed by the operator in S201, and the tomographic imaging apparatus 200 acquires the corresponding OCT tomographic images.

Note that in this embodiment, the number of repeated imagings in this step (S202) is set to three. The number of times of repeated imaging is not limited to this, and may be set to any number of times including the case of single (non-repeated) imaging. Further, the imaging time interval between repeated imagings is not limited to being longer than the imaging time interval of tomographic images in each repeated imaging, and both may be substantially the same.

The fundus image capturing unit 400 also acquires an SLO image and executes tracking processing based on the SLO moving image. In this embodiment, the reference SLO image used for tracking processing in repeated OCTA imaging is the reference SLO image set in the first repeated OCTA imaging, and a common reference SLO image is used in all repeated OCTA imaging.

Also, during OCTA repeated imaging, in addition to the imaging conditions set in S201, the same setting values are used (not changed) for selection of left and right eyes and execution/non-execution of tracking processing.

<Step 203>
The image acquisition unit 301 and image processing unit 304 reconstruct the tomographic image acquired in S202 and further generate a motion contrast image.

First, the tomographic image generating unit 301-1 performs wavenumber transformation, fast Fourier transform (FFT), and absolute value conversion (acquisition of amplitude) on the interference signal acquired by the image acquiring unit 301 to generate a tomographic image for one cluster. Generate.

Next, after aligning the tomographic images belonging to the same cluster, the alignment unit 304-1 further aligns the tomographic images between the clusters.

Subsequently, the image acquisition unit 301 and the image processing unit 304 generate motion contrast images using the aligned OCT tomographic images.

A motion contrast data generator 301-2 calculates motion contrast between adjacent tomographic images in the same cluster. In this embodiment, the decorrelation value Mxy is obtained as the motion contrast based on the following equation (1).

Here, Axy indicates the amplitude of the tomographic image data A at the position (x, y), and Bxy indicates the amplitude of the tomographic data B at the same position (x, y). 0≤Mxy≤1, and Mxy takes a value closer to 1 as the difference between the two amplitude values increases. Decrelation arithmetic processing such as Equation (1) is performed between arbitrary adjacent tomographic images (belonging to the same cluster), and the average of the obtained (number of tomographic images per cluster - 1) motion contrast values is An image with pixel values is generated as a final motion contrast image.

Although the motion contrast is calculated here based on the amplitude of the complex number data after FFT processing, the motion contrast calculation method is not limited to the above. For example, motion contrast may be calculated based on the phase information of complex data, or may be calculated based on both amplitude and phase information. Alternatively, motion contrast may be calculated based on the real and imaginary parts of complex data.

Also, in this embodiment, the decorrelation value is calculated as the motion contrast, but the motion contrast calculation method is not limited to this. For example, the motion contrast may be calculated based on the difference between the two values, or the motion contrast may be calculated based on the ratio of the two values.

Furthermore, in the above description, the final motion contrast image is obtained by calculating the average value of a plurality of acquired decorrelation values, but it may be generated by other methods. For example, an image having, as a pixel value, the median value or the maximum value of a plurality of acquired decorrelation values may be generated as the final motion contrast image.

In this embodiment, the image quality enhancing unit 304-2 three-dimensionally aligns and synthesizes a group of motion contrast images obtained through repeated OCTA imaging to generate a high-contrast synthetic motion contrast image. In this embodiment, averaging processing is used as synthesizing processing. Synthesis processing is not limited to simple addition and averaging. For example, the luminance values of each motion contrast image may be arbitrarily weighted and then averaged, or any statistical value including the median value may be used. may be calculated. Also, the alignment process may be performed two-dimensionally. The image quality enhancing unit 304-2 may determine whether or not motion contrast images unsuitable for synthesis processing are included, and then perform synthesis processing excluding the motion contrast images determined to be unsuitable. For example, if the evaluation value (for example, the average value or the median value of decorrelation values) for each motion contrast image is outside a predetermined range, it may be determined that the motion contrast image is unsuitable for synthesis processing.

Note that the image quality improvement process is not limited to superimposing a plurality of motion contrast images. Image quality improvement processing may be performed on the contrast image, or any known image quality improvement processing may be applied. Also, the image quality improvement process is not essential for the image processing of this embodiment, and may be omitted.

In this embodiment, after the motion contrast data generation unit 301-2 synthesizes the motion contrast images three-dimensionally, the correction unit 304-5 performs processing for three-dimensionally suppressing projection artifacts occurring in the motion contrast images.

Here, the projection artifact refers to a phenomenon in which the motion contrast in the retinal surface blood vessels is reflected on the deep layer side, resulting in a high decorrelation value even though the blood vessels do not actually exist. The correction unit 304-5 performs processing to suppress projection artifacts that occur on the 3D synthetic motion contrast image. Although any known projection artifact suppression technique may be used, the present embodiment uses Step-down Exponential Filtering. In Step-down Exponential Filtering, projection artifacts are suppressed by executing the processing represented by Equation (2) for each piece of A-scan data on a 3D motion contrast image.

Here, γ is an attenuation coefficient having a negative value, D(x, y, z) is a decorrelation value before projection artifact suppression processing, and D _E (x, y, z) is projection artifact suppression processing at depth position z. represents the later decorrelation value.

<Step 204>
The detection unit 304-41 detects a plurality of layer boundary candidates or candidates for the anterior surface B8 and posterior surface B9 of the cribriform plate from the superimposed tomographic images, and the image feature acquisition unit 304-3 detects the fovea of the macula as a predetermined portion. Detect the position of M.

Next, the determination unit 304-43 multiplies the weight by which the layer boundary candidate of the first layer detected by the detection unit 304-41 is multiplied, based on the distance from the predetermined part detected by the detection unit 304-41, From the second layer boundary candidate detected by 304-41, the calculation unit 304-42 determines the weight by which the estimated boundary position of the first layer calculated by executing the arithmetic processing is multiplied. Further, the determination unit 304-43 determines the layer boundary as a weighted sum of the layer boundary candidate and the layer boundary estimated position.

Details of the layer determination process will be described in S301 to S304.

<Step 205>
The projection unit 304-6 projects the superimposed tomographic image in the depth range based on the position of the layer boundary or the anterior surface B8 and posterior surface B9 of the cribriform plate detected or determined by the layer determination unit 304-4 in S204, and produces a superimposed frontal tomographic image. to generate As a projection method, any one of average intensity projection (AIP), maximum intensity projection (MIP), and minimum intensity projection (MinIP) can be selected. to project.

Further, the projection unit 304-6 projects the motion contrast image in the projection depth range based on the positions of the layer boundaries detected or determined by the layer determination unit 304-4 in S204 or the anterior cribriform plate B8 and posterior surface B9, and superimposes the front surface. Generate motion contrast images. As a projection method, either MIP (Maximum Intensity Projection) or AIP (Average Intensity Projection) can be selected.

Finally, the image processing apparatus 300 stores the acquired image group (SLO images and tomographic images), the imaging condition data of the image group, the generated three-dimensional and frontal motion contrast images, and the accompanying generation condition data, the inspection date and time, and the eye to be examined. is stored in the external storage unit 500 in association with the information identifying the .

<Step 206>
The display control unit 305 causes the display unit 600 to display information on the tomographic image and the motion contrast image generated in S203, the imaging conditions, and the image generation conditions.

FIG. 5(a) shows an example of a report screen 501 for a single examination imaged in the OCTA mode. The display control unit 305 displays the SLO image 512 on the upper left of the report screen 501, the superimposed frontal tomographic image 509 on the lower left, the superimposed frontal motion contrast image 505-1 of the superficial layer of the retina on the upper center, and the superimposed frontal motion of the deep layer of the retina on the lower center. Each contrast image 505-2 is displayed. In the superimposed frontal motion contrast image 505-1 of the retinal surface layer of FIG. An image is generated and displayed. In addition, in the superimposed frontal tomographic image 509 and the superimposed frontal motion contrast image 505-2 of the deep retina in FIG. - An OCT front image and an OCTA image are generated and displayed, each having a projection depth range up to the estimated position Bc5 with respect to the outer nuclear layer boundary B5.

A list box 510 shown in FIG. 5A, for example, is an example of a front tomographic image generation instruction user interface. As a front motion contrast image generation instruction user interface, for example, the projection depth range specification/change user interface (502, 503, 504, 506, 507, 508) shown in FIG. A B-scan image display area is provided on the right side of the report screen 501, and boundaries 504 and 508 indicating the projection depth range of the tomographic image and the motion contrast image, and motion contrast data (not shown) are superimposed on the B-scan tomographic image. there is

In addition, the projection depth range of the motion contrast image tomographic image and the motion contrast image is the default depth range setting shown below displayed in the GUI such as the list boxes 502 and 506. depth range to the inner plexiform layer)
・Deep retinal observation mode (DCP; depth range from inner reticular layer to outer reticular layer)
・Observation mode of outer retina and choroidal capillary plate (depth range from outer plexiform layer to choroidal capillary plate)
Choroid observation mode (depth range from Bruch's membrane to the boundary between the choroid and sclera)
can be changed by the operator selecting from Alternatively, as shown in 503 and 507 in FIG. 5A, the boundary type and offset amount used to specify the projection range can be selected, or the boundary data 504 and 504 of the layer or cribriform plate superimposed on the B-scan tomographic image can be selected. 508 may be moved by the input unit 700 to change the projection range.

The layer boundaries detected in S302 or determined in S303 are used as the layer boundaries used in these user interfaces. For example, the layer boundary candidate detected in S302 and the layer boundary candidate detected in S302 may be displayed simultaneously in an identifiable manner, or the layer boundary candidate determined in S303 and the layer boundary candidate detected in S302 may be switched and selected. . Alternatively, it is possible to distinguish whether the layer boundary candidate Bd detected in S302 for each layer boundary is used as it is or the layer boundary Bm obtained by applying the determination process described in S303 is used (color/ mark, line type, thickness, character string, etc.). Moreover, when superimposing a front tomographic image and a front motion contrast image, the two images may be superimposed and displayed in an identifiable manner (for example, different colors or transparency).

Further, the layer boundary Bm of the first layer determined in S303 may be changed by the user's operation, and the determination unit 304-43 may be configured to receive such a user's change instruction.

When the image processing apparatus 300 includes the measurement unit 304-7 as in the present embodiment, the layer thickness is measured from the tomographic image within the depth range defined by the layer boundary detected in S302 or the layer boundary determined in S303. . The layer thickness may be measured along the substantially normal direction to the curved surface defined based on the layer boundary, or may be measured along the depth direction of the tomographic image or the optical axis direction of the measurement light. The measurement unit 304-7 detects the blood vessel region from the motion contrast image within the depth range defined by the layer boundary detected in S302 or the layer boundary determined in S303, and binarizes the blood vessel region per unit area (volume). The blood vessel density (Vessel Area Density or Vessel Length Density) may be calculated by measuring the area (volume) of the blood vessel or the length of the central axis of the blood vessel. Measured values such as layer thickness and blood vessel density may be calculated in units of pixels, or representative values of measured values may be calculated within each local region such as the ETDRS grid shown in 517 and 520 in FIG. You may Alternatively, a representative value of the measured values may be calculated within a manually set region of interest as indicated by 518, 519, 521, and 522 in FIG.

Note that the tomographic images and motion contrast images displayed on the display unit 600 are not limited to front images, and may be displayed as cross-sectional images in any direction, such as B-scan images or C-scan images, or may be three-dimensionally rendered. may be displayed as a three-dimensional image. For example, when displaying the motion contrast image as 3D data as shown in FIG. 3D motion contrast data may be displayed.

Also, the image projection method and the presence/absence of projection artifact suppression processing may be selected and changed from, for example, a context menu or a user interface such as a check box as indicated by 527 in FIG. 5B. A motion contrast image after projection artifact suppression processing may be displayed on the display unit 600 not only as a front image but also as a B-scan image or a three-dimensional image. 523 in FIG. 5B shows a display example of a three-dimensional motion contrast image after projection artifact suppression processing.

Further, FIG. 6(a) shows an example of a report screen 501 for a single examination photographed in Custom 3D mode (one of scan modes for photographing a three-dimensional tomographic image). The display control unit 305 displays the layer thickness color map 529 and the layer thickness representative value 530 for each local region on the upper left of the report screen 501, the superimposed front tomographic image 509 and the layer thickness representative value 530 for each local region on the lower left, and the B scan on the upper right. A tomographic image and a layer thickness graph 532 are displayed at the lower right.

FIG. 6B shows a three-dimensional display example of a tomographic image captured in the Custom 3D mode. As in the case of the three-dimensional display of the motion contrast image shown in FIG. 5(b), by specifying the check box indicated by 525 in FIG. A limited three-dimensional tomographic image is displayed.

Further, the details of the processing executed in S204 will be described with reference to the flowchart shown in FIG. 2(b).

<Step 301>
The image quality enhancement unit 304-2 performs image quality enhancement processing on the tomographic image reconstructed in S203. In the present embodiment, the image quality enhancing unit 304-2 uses the alignment parameters calculated by the alignment unit 304-1 to average the tomographic images belonging to a plurality of clusters, thereby generating a superimposed tomographic image. Perform high image quality processing. Note that the image quality improvement process is not limited to superimposition processing for a plurality of tomographic images. Alternatively, any known image quality enhancement process may be applied. Note that the process of this step is not an essential process and may be omitted.

<Step 302>
The detection unit 304-41 detects an inner limiting membrane candidate B1, an inner plexiform layer/inner nuclear layer boundary candidate B4, an outer plexiform layer/outer nuclear layer boundary candidate B5, and a photoreceptor inner segment/outer segment junction candidate as layer boundary candidates from the tomographic image. B6, retinal pigment epithelium/Bruch's membrane boundary candidate B7, cribriform plate anterior surface candidate B8, and posterior surface B9 candidate are detected. In addition, the edge of the detected retinal pigment epithelium/Bruch's membrane boundary candidate B7 (Bruch's membrane opening edge BMO) is specified as the boundary of the optic disc region D. FIG. In this embodiment, a deformable model is used as a method for detecting layer boundaries or candidates for the anterior surface B8 and posterior surface B9 of the lamina cribrosa. may be detected based on That is, any known method may be used as long as it is a method capable of detecting a layer boundary from a tomographic image of the eye. Boundary detection methods using deformable models, graph cuts, and machine learning may be performed three-dimensionally on three-dimensional tomograms, or two-dimensionally on each two-dimensional tomogram. good too.

After detecting a layer boundary candidate from the tomographic image of the eye, the image feature acquisition unit 304-3 detects two depressions in descending order of depression from the shape of the detected internal limiting membrane candidate B1. acquires the fovea centralis M and the optic disc center D as predetermined sites. Here, the shallower recess is acquired as the fovea centralis M, and the deeper recess is acquired as the optic papilla center D. The acquisition method of the predetermined part is not limited to this, and any known acquisition method may be used.

Further, the anterior surface candidate B8 and the posterior surface candidate B9 of the cribriform plate portion may be set manually. For example, it can be set manually by moving the position of a specific layer boundary candidate (for example, inner limiting membrane candidate B1) by a predetermined amount. The types of layer boundaries to be detected are not limited to those described above. For example, nerve fiber layer/ganglion cell layer boundary candidate B2 or ganglion cell layer/inner plexiform layer boundary candidate B3 may be detected. A choroidal boundary candidate B10 may be detected. Also, the part acquired by the image feature acquiring unit 304-3 is not limited to the fovea centralis M or the center D of the optic papilla, and any part may be acquired.

<Step 303>
The determination unit 304-43 performs layer boundary determination processing. In this embodiment, the position detected by the detection unit 304-41 with respect to the layer boundary of a predetermined type of layer, the distance from the predetermined site in the direction perpendicular to the depth direction of the subject's eye or along the boundary, Determine the layer boundary position based on

In this embodiment, the position of the layer boundary candidate of the first layer detected in S302 by the determining unit 304-43 in accordance with the distance from the fovea M in the direction perpendicular to the depth direction of the subject's eye or in the direction along the boundary. and the position of the estimated boundary of the first layer calculated by the calculation unit 304-42 from the boundary candidate of the second layer detected by the detection unit 304-41. decide. Furthermore, the determining unit 304-43 determines the sum of the weighted position of the first layer boundary candidate and the weighted position of the estimated first layer boundary, Bm, as the position of the first layer boundary. Determined as The "position" described here indicates, for example, the coordinates in the tomographic image, and more specifically, the coordinates in the depth direction of the eye to be examined or in the direction intersecting the boundary candidate of the first or second layer. is. In addition, based on the position of the estimated boundary of the first layer calculated from the position of the boundary candidate of the second layer means that the second layer It may be based on the distance of the estimated boundary of the first layer from the first layer boundary candidate calculated from the position of the boundary candidate.

For example, an example of the boundary of the first layer to be calculated is the inner plexiform layer/inner nuclear layer boundary or the outer plexiform layer/outer nuclear layer boundary existing in the intermediate portion of the retina. The boundary of the first layer is not limited to this, and may be any layer boundary. For example, it may be the boundary of the choroid or cribriform plate. Examples of boundaries of the second layer include the ILM, the junction of the inner and outer photoreceptor segments (IS/OS), the outer plexiform layer, and the RPE. The second layer boundary may be any layer boundary other than the first layer boundary. For example, if the first layer boundary is the inner plexiform/inner nuclear layer boundary, the second layer boundary may be the outer plexiform/outer nuclear layer boundary. Further, the first layer boundary is not limited to being calculated using two layer boundary candidates, the first layer boundary candidate and the second layer boundary candidate. Three layers of boundary candidates may be detected and used. The inner limiting membrane candidate B1 is an example of a third layer boundary candidate.

Further, in this embodiment, an example is described in which the layer determination unit 304-4 having the detection unit 304-41, the calculation unit 304-42, and the determination unit 304-43 determines the position of the layer boundary of the first layer. However, it is only necessary to determine the position with respect to at least a partial region of the first layer. For example, the position of the entire layer area may be determined, or the position of the position obtained by calculating the boundary on the surface side and the boundary on the deep layer side of the layer area set) may be determined.

Any known arithmetic processing may be performed as a calculation method, and in this embodiment, the position is calculated on an arbitrary straight line in the depth direction of the subject's eye.

For example, the position of the estimated boundary is calculated as follows.
Outer plexiform layer/outer nuclear layer estimated boundary position Bc5
= Position of photoreceptor inner segment outer segment junction (IS/OS) candidate B6 - 40 μm
Position Bc4 of estimated boundary of inner plexiform layer/inner nuclear layer
= inner limiting membrane candidate B1 + wc4 (Bc5 - B1)
Calculate as An arbitrary value may be set as the weight wc4, but approximately 0.6 or 0.7 is preferable.

Also, the position of the estimated boundary of the first layer may be calculated (calculated) using the position of the boundary candidate of the second layer and the position of the boundary candidate of the third layer.

For example, the fractional ratio of the distance between the position of the boundary candidate in the second layer and the position of the boundary candidate in the third layer (the predetermined distance between the position of the boundary candidate in the second layer and the position of the boundary candidate in the third layer) It may be calculated based on the distance of the ratio of As a more detailed specific example, the position Bc4 of the estimated boundary of the inner plexiform layer/inner nuclear layer as the estimated boundary of the first layer, the boundary candidate B5 of the outer plexiform layer/outer nuclear layer as the boundary candidate of the second layer, Position Bc4 of the estimated boundary between the inner plexiform layer/inner nuclear layer when the inner limiting membrane candidate B1 is used as the third layer boundary candidate
=Position of internal limiting membrane candidate B1+wc4·(Position of outer plexiform layer/external nuclear layer boundary candidate B5−Position of internal limiting membrane candidate B1)
Calculate as

Here, the calculation is performed by setting an arbitrary straight line in the depth direction of the eye to be examined, but the calculation may be performed by setting an arbitrary straight line in the direction intersecting the boundary candidate of the first or second layer. .

Also, the position of the estimated boundary of the first layer may be calculated as a predetermined distance from the candidate boundary of the second layer. For example, when the second layer boundary candidate is the outer plexiform layer/outer nuclear layer boundary candidate B5, the estimated boundary of the inner plexiform layer/inner nuclear layer is defined as a predetermined distance from the outer plexiform layer/outer nuclear layer boundary candidate B5. may be calculated.

Note that the calculation method is not limited to linear calculation, and non-linear calculation processing may be used. Further, the calculation of the position of wc4 may be applied in the substantially normal direction with respect to the layer boundary, or may be applied in the substantially depth direction or the irradiation direction of the measurement light. That is, the calculation process may be performed at a position in the direction intersecting the boundary candidate of the first layer or the second layer of the tomographic image, or at a position in the depth direction of the subject's eye, It may be executed at a position in the A-scan direction.

In addition, in the periphery of the wide-angle tomographic image (PT and PN in FIG. 3(c)), the entire fundus becomes gradually thin, and at the edges of the wide-angle tomographic image, the brightness and contrast of the tomographic image tend to decrease. be. Therefore, in the peripheral portion of the wide-angle tomographic image, the distance between layer boundary candidates (for example, inner plexiform layer/inner nuclear layer boundary candidate B4 and outer plexiform layer/outer nuclear layer boundary candidate B5) detected in a predetermined layer is approximately It tends to be 0, which causes problems when used as the projection depth range. On the other hand, in the posterior pole (macula and optic papilla), layer boundary abnormalities due to edema, traction, and detachment are likely to occur. It will not be possible to include all ranges.

Therefore, in this embodiment, the determination unit 304-43 determines the position of the layer boundary of the predetermined type based on the distance from the predetermined site based on the following formula. Bd is the position of the actually detected boundary candidate of the first layer (hereinafter referred to as actual detection position), and Bc is the position of the estimated boundary of the first layer calculated by calculation using the boundary candidate of the second layer. and the weights to be multiplied are wd and wc, then the position Bm of the boundary of the first layer determined by the determining unit 304-43 is Bm=wd·Bd+wc·Bc
can be expressed as Arbitrary functions or constants may be used as wd and wc, but in this embodiment, wd=1.0-wc and wc is defined as a sigmoid function. Setting one of wd and wc to 1 is also included in weighting.

Of course, wc is not limited to this, and may be set using arbitrary functions such as linear or higher-order functions, rectangular functions, exponential functions, and trigonometric functions. The position here may also be calculated by the position in the direction intersecting the boundary candidate of the first layer or the second layer of the tomographic image, or by the position in the depth direction of the subject's eye. Alternatively, the position in the A-scan direction may be used for calculation.

For example, wc may be set to a sigmoid function that maximizes the slope of the tangent line at a distance of 7 mm from the fovea with a gain of 0.9. By setting the weights wd and wc in this way, the smaller the distance from the fovea (posterior pole), the larger the weight wd for Bd, and the larger the distance from the fovea (periphery), the larger the weight wc for Bc. Can be set large. Therefore, the boundary of the first layer can be accurately and robustly determined for a layer boundary abnormal region such as edema, which tends to occur in the macular region, and for a peripheral region where the layer thickness is thin and tends to have low contrast.

As the distance from a predetermined site (fovea, etc.), a straight line distance in any direction in the tomographic image may be used, or a distance measured along a curved surface (or curve) based on the layer boundary may be used. good. Alternatively, it may be a distance from a predetermined portion in a direction perpendicular to the A-scan direction.

<Step 304>
The layer determination unit 304-4 stores the layer boundary candidate Bd detected in S302 or the layer boundary Bm of the first layer determined in S303 in the storage unit 302 as a layer boundary according to the type of layer boundary. In the present embodiment, the layer boundary Bm determined in S303 for the inner plexiform layer/inner nuclear layer boundary B4 and the outer plexiform layer/outer nuclear layer boundary B5, which are layers in the middle part of the retina, is used as the layer boundary, and the other layer boundaries , the layer boundary candidate Bd detected in S302 is determined as the layer boundary.
The main purpose of using the layer boundary is i) to specify the projection depth range and generate a front image (tomographic image or motion contrast image) within the projection depth range ii) layer thickness measurement (retina, choroid, cribriform plate) Calculation of thickness between constituent layer boundaries)
iii) Specifying a projection depth range, and calculating measurement values based on the vascular region within the projection depth range (e.g., calculation of measurement values based on blood vessel density, blood vessel length, area and volume of the blood vessel region, and blood vessel diameter)
iv) Specifying a projection depth range, and calculating a measurement value (other than layer thickness measurement) based on a tomographic image within the projection depth range (e.g., calculating the size of a cyst or peeling area)
There are four types of

Examples of ii) layer thickness measurements include the thickness of the retinal ganglion cell layer complex (the distance between the inner limiting membrane and the inner plexiform layer/inner nuclear layer boundary).

The measured values listed in ii) to iv) above may be displayed as numerical values or graphs as shown in the lower right of FIG. may Alternatively, the representative values of the measured values for each local area may be displayed as illustrated in 530 in the upper left and lower left of the figure and 517, 518, 519, 520, 521, and 522 in FIG. 5(a). In this embodiment, a value measured based on the layer boundary Bd detected in S302 or the layer boundary Bm determined in S303 is displayed, but the present invention is not limited to this. For example, the layer candidate Bd detected in S302 is also stored in the storage unit 302, and the value measured using the layer boundary Bm obtained by applying the determination process described in S303 (that is, the position of the estimated boundary of the first layer ) and the value measured based on the layer candidate Bd detected in S302 (that is, the second measured value not based on the position of the estimated boundary of the first layer) can be distinguished (color, Marks, character strings, images, numerical values, graphs, etc.) may be displayed, switched, or only one of them may be displayed. Alternatively, for each measured value, it is identified whether the value is measured using the layer boundary candidate Bd detected in S302 as it is or the value is measured using the layer boundary Bm obtained by applying the determination process described in S303. It may be displayed in a possible mode (color, mark, character string, image, etc.).

Further, when displaying values measured based on the layer boundary candidates Bd, at A-scan positions where the distance between the predetermined layer boundaries Bd to be measured is approximately 0, the reliability of the layer boundary is regarded as low and the measured values are displayed. may be displayed in a manner different from the area of , or the measured values may not be displayed.

According to the configuration described above, the image processing apparatus 300 detects a layer boundary candidate for a wide-angle OCT tomographic image, and then determines a layer boundary position based on the distance from a predetermined site. is determined so that the amount of change from the layer boundary candidate is small, and the amount of change is large at a position distant from the distance.

In a wide-angle tomographic image of an eye to be examined that includes a layer shape abnormality, layer boundaries can be robustly determined regardless of the region.

[Second embodiment]
After detecting a layer boundary candidate in a wide-angle OCT tomographic image, the image processing apparatus according to the present embodiment changes the position of the layer boundary candidate according to an evaluation value corresponding to the reliability of the layer boundary candidate. The amount is determined, and the position changed by the change amount is determined as the layer boundary.

A case will be described in which layer boundaries are robustly determined irrespective of sites in a wide-angle tomographic image of an eye to be inspected that includes layer shape anomalies.

The configuration of an image processing system 200 including an image processing apparatus 300 according to the present embodiment and the image processing flow according to the present embodiment are the same as those of the first embodiment, so description thereof will be omitted. Note that S205 to S207 in FIG. 2A in the image processing flow of this embodiment are the same as in the case of the first embodiment, so description thereof will be omitted.

<Step 201>
By operating the input unit 700 , the operator sets the imaging conditions for the OCTA image to be instructed to the tomography apparatus 200 .

Among the imaging conditions, the difference from the first embodiment is that OCTA imaging is not repeatedly performed, that is, the number of clusters acquired in this step is one. i.e.
1) Register Ultra Wide examination set 2) Select OCTA scan mode 3) Set the following imaging parameters 3-1) Scanning pattern: Super Large
3-2) Scanning area size: 23x20mm
3-3) Main scanning direction: horizontal direction 3-4) Scanning interval: 0.02 mm
3-5) Fixation light position: macula 3-6) Number of B-scans per cluster: 3
3-7) Coherence gate position: Vitreous body side 3-8) Default display report type: Single examination report is set as an imaging condition.

<Step 202>
The operator operates the input unit 700 to start OCTA imaging under the imaging conditions specified in S201. The imaging control unit 303 instructs the tomographic imaging apparatus 200 to perform OCTA imaging based on the settings instructed by the operator in S201, and the tomographic imaging apparatus 200 acquires the corresponding OCT tomographic image. Further, the fundus image capturing unit 400 acquires an SLO image and executes tracking processing based on the SLO moving image.

<Step 203>
The image acquisition unit 301 and image processing unit 304 generate a motion contrast image (motion contrast data) based on the OCT tomographic image acquired in S202. After generating the motion contrast image in the same procedure as in S203 of the first embodiment, the correction unit 304-5 executes suppression processing of the projection artifact 405 generated on the motion contrast image.

The image quality enhancement unit 304-2 inputs a low image quality motion contrast image generated from a small number of tomographic images to the machine learning model to obtain a high image quality (low generate motion contrast images (noisy and high contrast). Here, the machine learning model refers to a function generated by performing machine learning using teacher data. In the present embodiment, teacher data composed of pairs of input data, which is a low-quality image acquired under predetermined shooting conditions assumed to be processed, and output data, which is a high-quality image corresponding to the input data. It is a function generated by performing machine learning using

The predetermined imaging conditions include a body part to be imaged, an imaging method, an imaging angle of view, an image size, and the like.

Also, in this embodiment, when the user presses the Denoise button 515 shown in the upper right of the report screen in FIG. .

In this embodiment, the input data used as teacher data is a low-quality motion contrast image generated from a single cluster with a small number of tomographic images, and the output data (used as teacher data) is an addition of multiple aligned motion contrast data. Let the averaged high-quality motion-contrast image be obtained. Output data used as teacher data is not limited to this, and may be, for example, a high-quality motion contrast image generated from a single cluster composed of a large number of tomographic images. The pair of input image and output image used for training the machine learning model is not limited to the above, and any known combination of images may be used. For example, an image obtained by adding a first noise component to a motion contrast image acquired by the tomographic imaging device 200 or another device is used as an input image, and the motion contrast image (acquired by the tomographic imaging device 200 or another device) is used as an input image. The image with the added second noise component (different from the first noise component) may be used as an output image for training a machine learning model.

FIG. 7 shows a configuration example of a machine learning model in the image quality enhancing unit 304-2 according to the embodiment. The machine learning model is a convolutional neural network (CNN), and is composed of a plurality of layer groups for processing input value groups and outputting them. The types of layers included in the above configuration include a convolution layer, a downsampling layer, an upsampling layer, and a merger layer. The convolution layer is a layer that performs convolution processing on an input value group according to set parameters such as the kernel size of filters, the number of filters, the stride value, and the dilation value. Note that the number of dimensions of the kernel size of the filter may also be changed according to the number of dimensions of the input image. That is, if the input is a two-dimensional image (B-scan image, frontal image, cross-sectional image in an arbitrary direction, curved cross-sectional image), the number of dimensions should be two, and if the input is a three-dimensional image, the number of dimensions should be three. Just do it. A down-sampling layer is a process that reduces the number of output value groups to the number of input value groups by thinning out or synthesizing input value groups. Specifically, for example, there is a Max Pooling process. The upsampling layer is a process that makes the number of output value groups larger than the number of input value groups by duplicating the input value groups or adding values interpolated from the input value groups. Specifically, for example, there is linear interpolation processing. The synthesizing layer is a layer that performs a process of synthesizing a group of values such as a group of output values of a certain layer and a group of pixel values forming an image from a plurality of sources and connecting or adding them. In such a configuration, the value group output from the pixel value group forming the input image 701 through the convolution processing block and the pixel value group forming the input image 701 are synthesized in the synthesis layer. The combined pixel values are then shaped into a high quality image 702 in a final convolutional layer. Although not shown in the figure, as an example of changing the configuration of the CNN, for example, a batch normalization layer after the convolution layer or an activation layer using a normalized linear function (Rectifier Linear Unit) is incorporated. You can

In FIG. 7, the image to be processed is one low-quality two-dimensional image for the sake of simplicity, but the present invention is not limited to this. For example, a low-quality three-dimensional image composed of low-quality two-dimensional images at different scanning positions is set as an image to be processed, and the high-quality image processing described with reference to FIG. 7 is applied to each low-quality two-dimensional image. A quality three-dimensional image may be generated. Alternatively, the number of dimensions of the kernel of the filter applied in the machine learning model is set to three dimensions instead of two dimensions, and the three-dimensional low-quality image is input to the image quality enhancement unit 304-2 to output a three-dimensional high-quality image. It is also included in the present invention.

<Step 204>
A detection unit 304-41 detects a layer boundary candidate or a cribriform plate anterior surface B8/posterior surface B9 candidate from the superimposed tomographic images. Next, the determining unit 304-43 calculates an evaluation value corresponding to the reliability of the layer boundary candidate or the anterior surface B8/posterior surface B9 candidate of the cribriform plate and determines the layer boundary based on the evaluation value.

Details of the layer determination process will be described in S401 to S404.

Furthermore, the details of the processing executed in S204 will be described with reference to the flowchart shown in FIG. 2(c). Note that S402 and S404 have the same processing contents as S302 and S304 in the first embodiment, respectively, so details thereof are omitted.

<Step 401>
The image quality enhancement unit 304-2 performs image quality enhancement processing on the tomographic image reconstructed in S203. In this embodiment, the image quality enhancement unit 304-2 inputs a single low-quality tomographic image to the machine learning model shown in FIG. 7, and performs image quality enhancement processing for outputting a high-quality tomographic image. implement. Any known machine learning model can be used as the machine learning model, but in the present embodiment, a machine learning model having the same configuration as that of S203 is used, and detailed description thereof is omitted.

<Step 403>
A detection unit 304-41 detects a layer boundary candidate or a front region B8 and a rear region B9 candidate of the cribriform plate from the superimposed tomographic images. Further, in this embodiment, the determination unit 304-43 calculates an evaluation value corresponding to the reliability of the layer boundary candidate or the candidate of the anterior region B8 and the posterior region B9 of the cribriform plate. Next, the determination unit 304-43 determines the weight by which the first layer boundary candidate is multiplied based on the evaluation value, and the calculation unit 304-42 performs arithmetic processing on several layer boundary candidates to calculate the first weight. Determine the weights to be multiplied for the estimated boundaries of the layers. That is, when the evaluation value is low, the weight to be multiplied by the boundary candidate of the first layer is low, and the weight to be multiplied by the estimated boundary of the first layer is set to be high. Further, the determination unit 304-43 determines the sum of the weighted position of the first layer boundary candidate and the weighted position of the estimated first layer boundary.

Any known evaluation value may be used as the evaluation value corresponding to the reliability of the layer boundary candidate. . Of course, not limited to this, any known evaluation value may be calculated. For example, edge enhancement processing may be performed on the tomographic image, and the evaluation value may be determined based on the edge intensity of the layer boundary in the edge-enhanced image.

Alternatively, the evaluation value corresponding to the reliability may be determined based on the angle that the measurement light makes with respect to the substantially normal direction (or substantially tangential direction) of the curved surface defined by the layer boundary candidate. For example, in the peripheral portion of the fundus, the angle formed by the optical axis direction of the measurement light and the substantially normal direction (substantially tangential direction) of the curved surface defined by the layer boundary becomes large (small). In such a situation, the signal data of the tomographic image detected by the tomographic imaging apparatus 200 is attenuated, and the luminance value of the tomographic image tends to decrease and the luminance contrast between adjacent layers tends to decrease. Assign to value. As a specific evaluation value, for example, (the angle between the irradiation direction of the measurement light and the substantially tangential direction of the layer boundary defined by the layer boundary candidate) [rad]/(π/2) [rad]
should be calculated. In this case, a value from 0 to 1 can be taken as the evaluation value.

Also, the evaluation value is determined based on the shape of the layer boundary candidate (degree of curvature, etc.), the distance between adjacent layer boundary candidates (that is, the layer thickness of the first layer), and the number of layer boundary position candidates at each A-scan position. It is also included in the present invention.

For example, if the layer thickness measured using the layer boundary candidate is zero or a value close to it, there is a high possibility that the layer boundary candidate used for the layer thickness measurement has not been correctly detected (reliability is low). The same applies when the layer thickness measured using the layer boundary candidate becomes an abnormally large value. In such a situation, the estimated layer boundary is often a better layer boundary than the candidate layer boundary. Therefore, the weight w for calculating the weighted sum of the layer boundary candidate and the estimated layer boundary position may be changed according to the layer thickness measured using the layer boundary candidate.

Specifically, when the layer thickness measured using the layer boundary candidate is smaller than a predetermined value, the weight of the estimated layer boundary position may be increased. Alternatively, the weight of the estimated layer boundary position may be increased when the layer thickness is greater than a predetermined value. Also, the weight may be defined as an arbitrary function corresponding to the layer thickness, or the weight may be determined by referring to a lookup table in which the weight corresponding to the layer thickness is stored. .

By doing so, even if the reliability of the layer boundary candidate is low and the layer thickness is close to zero in the periphery of the fundus, etc., a more appropriate layer boundary is determined by increasing the weight of the estimated position of the layer boundary. can do.

Note that the layer boundary determination method in the present invention is not limited to the weighted sum of the layer boundary candidate detected by the detector 304-41 and the estimated layer boundary position calculated by the calculator 304-42. Also, the calculator 304-42 is not an essential component of the present invention. For example, when the evaluation value (distance between layer boundary candidates, etc.) regarding the layer boundary candidate detected by the detection unit 304-41 for a predetermined portion or a predetermined type of layer boundary is less than a predetermined value, the distance between the layer boundary candidates The present invention also includes a case in which the determination unit 304-43 determines the position of the layer boundary candidate such that is equal to or greater than a predetermined value.

According to the configuration described above, the image processing apparatus 300 detects a layer boundary candidate in a wide-angle OCT tomographic image, and then calculates the amount of change from the layer boundary candidate position based on the reliability of the detected layer boundary candidate. A position changed by the change amount is determined as a layer boundary.

As a result, in a wide-angle tomographic image of an eye to be examined that includes a layer shape abnormality, the layer boundary can be determined robustly regardless of the region.

[Other embodiments]
Although each of the above embodiments is implemented as the image processing device 300, the embodiments of the present invention are not limited to the image processing device 300 only. For example, the present invention can be embodied as a system, device, method, program, storage medium, or the like.

Further, in each of the above embodiments, the case of acquiring a wide-angle tomographic image using SS-OCT or SD-OCT has been described. Since the SS-OCT image and the SD-OCT image have different image quality, the detection unit 304-41, the calculation unit 304-42, the determination unit 304-43, and the layer determination unit 304-43 are configured to detect different types of layer boundary groups for both. The present invention also includes the case where the unit 304-4 is configured.

301 image acquisition unit 304-41 detection unit 304-42 calculation unit 304-43 determination unit

Claims (21)

detection means for detecting the position of a first layer boundary candidate and the position of a second layer boundary candidate on a tomographic image of an eye to be inspected;
position of the first layer boundary candidate in the depth direction of the eye to be examined or in a direction intersecting the first or second layer boundary candidate on the tomographic image; determining means for determining the position of at least a portion of the region with respect to the first layer based on the position of the estimated boundary of the first layer calculated from the position of the candidate boundary of the layer;
An image processing device comprising: The determining means
the position of at least a portion of the region with respect to the first layer,
2. The image processing according to claim 1, wherein the determination is based on the sum of the weighted position of the boundary candidate of the first layer and the weighted position of the estimated boundary of the first layer. Device. 3. The image processing according to claim 2, wherein the weighting is calculated according to a distance from a predetermined site in a direction perpendicular to the depth direction of the subject's eye on the tomographic image or in a direction along a boundary. Device. 4. The image processing apparatus according to claim 3, wherein the predetermined site is the center of the optic papilla or the fovea fovea. The determining means
An evaluation value of the boundary candidate of the first layer or an evaluation value of the estimated boundary of the first layer is calculated, and based on the evaluation value, the position of the boundary candidate of the first layer and the position of the first layer 3. The image processing apparatus according to claim 2, wherein the weighting of the position of the estimated boundary is calculated. 6. The image according to any one of claims 1 to 5, wherein the boundary candidate of the second layer is any one of ILM, outer plexiform layer, photoreceptor inner segment outer segment junction, and RPE. processing equipment. 7. The image of any one of claims 1 to 6, wherein the position of the estimated first layer boundary is calculated based on the distance from the position of the candidate boundary of the second layer. processing equipment. 7. The position of the estimated boundary of the first layer is calculated based on the position of the boundary candidate of the second layer and the position of the boundary candidate of the third layer. 1. The image processing apparatus according to claim 1. The determining means
The evaluation value of the boundary candidate of the first layer or the evaluation value of the estimated boundary of the first layer is the luminance contrast between the layers in contact with the boundary candidate or the estimated boundary to be evaluated, the boundary candidate or the estimated boundary edge strength or shape of a boundary, angle formed between a normal or tangent to the boundary candidate or estimated boundary detected by the detection means and the A-scan direction, the first layer calculated from the boundary candidate or estimated boundary 6. The image processing apparatus according to claim 5, wherein the calculation is performed based on at least one of the layer thickness of . 10. The image processing apparatus according to any one of claims 1 to 9, wherein at least a partial area related to said first layer is a boundary of said first layer. 11. The image processing apparatus according to claim 10, wherein the boundary of the first layer is an inner plexiform layer/inner nuclear layer boundary or an outer plexiform layer/outer nuclear layer boundary. The image processing device further has a display unit and output means,
11. The image processing apparatus according to claim 10, wherein the output means superimposes the boundary of the first layer determined by the determination means on the tomographic image displayed on the display unit. The output means is
13. The image processing apparatus according to claim 12, wherein the first layer boundary candidate and the first layer boundary determined by the determination unit are switched and displayed on the display unit. 13. The method according to claim 12, wherein the output means displays the first layer boundary candidate and the first layer boundary determined by the determination means in a identifiable manner on the display unit. image processing device. The image processing device is
further comprising measuring means;
The measuring means
For measurements on the first layer,
15. A first measurement method based on the position of the estimated boundary of the first layer or a second measurement method not based on the position of the estimated boundary of the first layer is used. The image processing device according to any one of . 16. The determining means according to any one of claims 1 to 15, wherein said determining means receives a user's operation to change the position of at least a part of the region related to said first layer determined by said determining means. Image processing device. 17. The image processing apparatus according to any one of claims 1 to 16, wherein said tomographic image is an image acquired by an SS-OCT apparatus. 18. The OCT front image or the motion contrast image of the projection depth range set based on the boundary of the first layer determined by the determination means is generated according to any one of claims 1 to 17. image processing device. detection means for detecting the position of a first layer boundary candidate and the position of a second layer boundary candidate on a tomographic image of an eye to be inspected;
position of the first layer boundary candidate in the depth direction of the eye to be examined or in a direction intersecting the first or second layer boundary candidate on the tomographic image; a position of an estimated boundary of said first layer calculated based on a fractional ratio of the distances of a candidate layer boundary and a candidate third layer boundary; a determining means for determining
An image processing device comprising: 20. The image processing apparatus according to claim 19, wherein at least a partial area related to said first layer is a boundary of said first layer. a step of detecting the position of a first layer boundary candidate and the position of a second layer boundary candidate on a tomographic image of an eye to be inspected;
position of the first layer boundary candidate in the depth direction of the eye to be examined or in a direction intersecting the first or second layer boundary candidate on the tomographic image; determining the position of at least a portion of the region with respect to the first layer based on the position of the estimated first layer boundary calculated from the position of the candidate boundary of the layer;
An image processing method characterized by comprising:

Images