JP5634587B2 - Image processing apparatus, image processing method, and program - Google Patents

Description

  The present invention relates to a technique relating to high image quality of a tomographic image in an eye part.

  An ophthalmic tomographic imaging apparatus such as an optical coherence tomography (OCT) can observe the state inside the retinal layer three-dimensionally. This tomographic imaging apparatus has attracted attention in recent years because it is useful for more accurately diagnosing diseases.

  In ophthalmologic diagnosis, there are cases where a volume image is used to grasp the state of the entire retinal layer and a high-quality two-dimensional tomographic image is used to grasp a layer that does not appear in a low-quality tomographic image.

  The image quality of a tomographic image obtained by OCT depends on the intensity of near infrared light incident on the retina. For this reason, in order to improve the image quality of a tomographic image, it is necessary to increase the intensity of light irradiated to the retina, but from the viewpoint of safety, there is a limit to the intensity of light that can be irradiated to the retina. For this reason, it is desired to generate a high-quality tomographic image while irradiating near-infrared light within a safe intensity range for safety. In response to such a requirement, a technique for generating a cross-sectional image with less noise by superimposing photographed two-dimensional tomographic image groups on each other is disclosed (see Patent Document 1).

  On the other hand, if the thickness of each layer can be measured from a tomographic image of the retina taken by OCT, it is possible to quantitatively diagnose the degree of progression of diseases such as glaucoma and the degree of recovery after treatment. In order to quantitatively measure the thickness of these layers, a technique is disclosed in which the boundary of each layer of the retina is detected from a tomographic image using a computer and the thickness of each layer is measured (see Patent Document 2).

JP 2008-237238 A JP 2007-325831 A

However, in Patent Document 1, one image is generated from images picked up by different scanning lines. Therefore, there is a trade-off relationship between noise reduction and image position misalignment. In addition, since the eyes blink or fixation micromotion occurs, adjacent images are not always similar to each other.
For this reason, when images adjacent to each other are overlapped, an image at a position most desired by a doctor (hereinafter sometimes referred to as an “operator”) is not always obtained with high image quality. In addition, in the case where the boundary of each layer of the retina is detected from the tomographic image using a computer and the thickness of each layer is measured, it is desired to increase the accuracy of the layer detection of the tomographic image.

  The present invention has been made in view of the above problems, and an object thereof is to improve the image quality of a tomographic image. Alternatively, it is an object to improve the accuracy of layer detection.

In order to achieve the above object, the image processing equipment according to an embodiment of the present invention,
Based on a smaller number of two-dimensional tomographic images than the plurality of two-dimensional tomographic images out of the plurality of two-dimensional tomographic images of the fundus acquired when adjusting the position in the depth direction on the fundus of the eye to be examined. Generating means for generating a simple two-dimensional tomographic image;
First layer detecting means for detecting the fundus layer in the new two-dimensional tomographic image;
Based on the detection result of the first layer detecting means, at least one two-dimensional tomographic image among a plurality of two-dimensional tomographic images constituting a three-dimensional image acquired after the plurality of two-dimensional tomographic images are acquired. Second layer detecting means for detecting the fundus layer from the image.
In order to achieve the above object, an image processing method according to one aspect of the present invention includes:
Based on a smaller number of two-dimensional tomographic images than the plurality of two-dimensional tomographic images out of the plurality of two-dimensional tomographic images of the fundus acquired when adjusting the position in the depth direction on the fundus of the eye to be examined. Generating a two-dimensional tomographic image,
A first layer detecting step of detecting the fundus layer in the new two-dimensional tomographic image;
Based on the detection result in the first layer detection step, at least one two-dimensional tomographic image among a plurality of two-dimensional tomographic images constituting a three-dimensional image acquired after the plurality of two-dimensional tomographic images are acquired. And a second layer detection step of detecting the fundus layer from the image.

  According to the present invention, a high-quality tomographic image can be obtained.

It is a figure which shows the structure of an image processing system. It is a flowchart which shows the flow of the tomographic imaging process in an image processing device. It is a figure for demonstrating a volume image and a tomographic image generation process. It is a figure for demonstrating the range which performs imaging | photography, and the position which performs imaging | photography before and after an imaging | photography instruction | indication. It is a figure for demonstrating the imaging position setting of a retinal layer in a tomographic image. It is a figure which shows an example of the scanning line pattern for setting an imaging position. It is a figure which shows the structure of an image processing system. It is a flowchart which shows the flow of the tomographic imaging process in an image processing device. It is a figure which shows an example which applies a layer detection result to another tomographic image.

Example 1
Hereinafter, a first embodiment of the present invention will be described with reference to the drawings. Note that the image processing apparatus according to the present embodiment uses a tomographic image captured to determine the imaging position of the retinal layer to capture a two-dimensional tomographic image with reduced noise. It is characterized by generating. The volume image means a set of two-dimensional tomographic images.

  According to the present embodiment, it is possible to acquire a high-quality two-dimensional tomographic image and a volume image obtained by photographing a wide area by one photographing. Here, high image quality refers to an image in which the S / N ratio is improved as compared with one-time imaging. Or it refers to an image in which the amount of information necessary for diagnosis is increasing.

  Hereinafter, details of the image processing system including the image processing apparatus according to the present embodiment will be described. In the present embodiment, a case will be described in which one high-quality two-dimensional tomographic image and a volume image obtained by photographing a wide range are acquired from a normal two-dimensional tomographic image by one imaging.

  FIG. 1 is a diagram illustrating a configuration of an image processing system 100 including an image processing apparatus 110 according to the present embodiment. As shown in FIG. 1, the image processing system 100 is configured by connecting an image processing apparatus 110 to a tomographic imaging apparatus 120 via an interface.

  The tomographic imaging apparatus 120 is an apparatus that captures a tomographic image of the eye, and includes, for example, a time domain OCT or a Fourier domain OCT. Since the tomographic imaging apparatus 120 is a known apparatus, detailed description thereof is omitted.

  The image processing apparatus 110 includes an image acquisition unit 111, an image storage unit 112, an image processing unit 113, a display control unit 114, and an instruction acquisition unit 115.

  The image acquisition unit 111 acquires a tomographic image captured by the tomographic image capturing apparatus 120 and stores it in the image storage unit 112. In the image storage unit 112, the tomographic images acquired by the instruction acquisition unit 115 and before and after the timing of the imaging instruction by the operator are distinguished and stored. The image processing unit 113 generates a high-quality two-dimensional tomographic image and a three-dimensional tomographic image from the tomographic images distinguished and stored in the image storage unit 112.

FIG. 3A shows a schematic diagram of a macular volume image of the retina imaged by the tomographic image imaging device 120. In FIG. 3A, T1 to Tk are two-dimensional tomographic images of the macula. That is, the volume image is formed by a set of two-dimensional tomographic images obtained by photographing different locations.
Next, a processing procedure of the image processing apparatus 110 according to this embodiment will be described with reference to FIG.

<Step S201>
In step S201, in order to image the retinal layer, a control unit (not shown) performs a planar direction (xy plane in FIG. 3A) and a depth direction (z direction in FIG. 3A) of the eye to be examined. Adjust the position. Here, matching the position in the depth direction corresponds to matching the position of the coherent gate for obtaining the tomographic image.

  While the operator aligns the retinal layer, the tomographic imaging apparatus 120 keeps scanning the same part repeatedly, and displays the two-dimensional tomographic image of the same part on a display unit (not shown). The operator adjusts the position in the plane direction and the depth direction while viewing the two-dimensional tomographic image. FIG. 5 shows a state in which imaging is performed by adjusting the retinal layer to a desired position in the two-dimensional tomographic image. The vertical axis represents time t, and the S / N of the tomographic image is increased by adjusting the position of the retinal layer by a control unit (not shown). In the image storage unit 112, the tomographic images before and after the instruction acquisition unit 115 acquires an instruction to start imaging are distinguished and stored. Usually, the doctor (surgeon) starts imaging when the two-dimensional tomographic image displayed on the display unit is settled at a desired position. Therefore, an image shot before shooting starts approaches a desired position as the shooting start timing approaches. And when it settles down in a desired position, imaging | photography will be started.

  Here, the setting of the imaging position of the retinal layer and the tomographic image acquisition will be specifically described with reference to FIG.

  FIG. 4A is a diagram of the volume image of FIG. 3A viewed in the xy plane. In FIG. 4A, R indicates a range for capturing a volume image, and L indicates a position where the tomographic imaging apparatus 120 repeatedly scans the same part of the eye to be inspected while the operator aligns the retinal layer. Show. The numerical value on the left side of the imaging range R indicates the number of the position where the two-dimensional tomographic image is captured. Here, it is assumed that the vertical resolution (y direction) of the volume image is 128, and the position where repeated shooting is performed is the 64th center. An example will be described in which the operator aligns the retinal layer while viewing the tomographic image that is repeatedly shot. FIG. 4B shows the position number of the volume image to be photographed before and after the photographing instruction. In FIG. 4B, the horizontal axis represents time t, and shows a case where the instruction acquisition unit 115 has acquired a shooting instruction at the timing t1. Until the instruction acquisition unit 115 acquires the imaging instruction, the tomographic imaging apparatus 120 continues to capture the same position (64). When the imaging instruction is acquired, the tomographic imaging apparatus 120 sequentially changes the different positions (1 to 128). Take a picture.

<Step S202>
In step S201, the control unit (not shown) adjusts the position suitable for photographing the retinal layer, and in step S202, the photographing instruction unit (not shown) instructs the start of photographing.

<Step S203>
In step S203, when the operator gives an imaging instruction (timing at t1 in FIG. 4B), a control unit (not shown) sequentially scans the xy plane to capture 128 two-dimensional tomographic images. . The image acquisition unit 111 acquires a tomographic image obtained by repeatedly capturing the same part of the eye to be inspected before an imaging instruction, and acquires a tomographic image by imaging a different part of the eye to be inspected after the imaging instruction. Therefore, in the image storage unit 112, the two-dimensional tomographic images before and after the imaging instruction are distinguished and stored. Here, the image storage unit 112 does not hold all the images before the shooting instruction, but superimposes a predetermined period or a predetermined N images (FIG. 4B) retroactive from the timing of the shooting instruction. It is desirable to store them as candidate images. This is because the tomographic image immediately after the start of alignment is taken at a location different from the desired position, and the image immediately before the imaging instruction is taken is taken at almost the same position as the desired position. Because.

<Step S204>
In step S204, the image processing unit 113 uses the tomographic image stored in the image storage unit 112 to generate a high-quality two-dimensional tomographic image and a volume image obtained by photographing a wide range. First, a process for generating a high-quality two-dimensional tomographic image by the overlay process will be described.

  The image selection unit 116 selects an image satisfying a predetermined condition from the N two-dimensional tomographic images stored in the image storage unit 112 before the start of imaging instruction. First, tomographic images with blinking or vignetting are removed from the overlay candidate images. In this case, it is better to delete the candidate image from the image storage unit 112 during shooting. This is because unnecessary images can be reduced and more useful images can be left. If there are few images that satisfy the predetermined condition, the image selection unit 116 extends the predetermined period retroactive from the timing of the shooting instruction or increases the number of images to be stored. This increases the opportunity to select an image close to the image at the time of shooting, leading to an improvement in image quality.

  As these image removal methods, the S / N ratio of the two-dimensional tomographic image is measured, and the two-dimensional tomographic image below the threshold Th1 is removed. Alternatively, binarization is performed with the threshold Th2, and a two-dimensional tomographic image having an area equal to or larger than the threshold equal to or smaller than a certain number is removed. The threshold is set in advance based on the characteristics of the tomographic imaging apparatus 120. Alternatively, the threshold value may be dynamically set for each eye using a discriminant analysis method, a P-Tile method, or the like. Next, a reference image is selected. The reference image is the latest image taken. When a shooting instruction is given, the latest image is the image (timing t1 in FIG. 4B). However, when the image at that moment is removed by blinking or the like, the tomographic image taken last in the overlay candidate image (for example, the tomographic image taken at t1-1) is used as the reference image. . Thereby, it is possible to improve the image quality of the image being observed before the start of the photographing instruction. In addition, it is possible to achieve both improvement in S / N and suppression of image blur by increasing the number of images used for registration as a predetermined time elapses without performing registration in the initial stage of alignment. Further, the superposition may be performed after the start of the imaging instruction without performing superposition during observation.

  Next, the alignment unit 117 aligns the two-dimensional tomographic images. As the alignment processing, for example, an evaluation function representing the similarity between two tomographic images is defined in advance, and the tomographic image is deformed so that the value of this evaluation function becomes the best. As the evaluation function, for example, a method of evaluating with a pixel value can be cited (for example, a method of performing evaluation using a correlation coefficient can be cited). In addition, as the tomographic image deformation process, for example, a process of performing translation or rotation using affine transformation or changing an enlargement ratio can be cited. In the superimposition process, a tomographic image having an evaluation function value equal to or greater than the threshold Th3 is used. That is, by these processes, M images are selected as the superimposed images among the N overlapping candidate images.

  In the following description, it is assumed that the alignment process between tomographic images has already been completed in the overlay process with a plurality of tomographic images.

  Here, a case where the number of overlapping sheets M in the same cross section is two will be described. FIG. 3B shows a tomographic image generation process in which a plurality of tomographic images are processed and a single tomographic image is generated as a synthesized image (addition average of scanning lines taken on different tomographic images taken at different times). It is a figure for demonstrating a process. FIG. 3C is a high-quality two-dimensional tomographic image (composite image) generated by performing the averaging process using M (two in this embodiment) pixels for each pixel. That is, in FIG. 3C, Aij ″ is a new scanning line calculated by performing the averaging process on the corresponding scanning line.

  In FIG. 3B, Ti ′ and Ti + 1 ′ are tomographic images obtained by photographing the same cross section at different times. Aij ′ and A (i + 1) j ′ represent respective scanning lines (A-scan) in the tomographic images Ti ′ and Ti + 1 ′. The image generation unit 118 calculates Aij ″ in FIG. 3C by performing an averaging process on the scanning lines Aij ′ and A (i + 1) j ′. Note that a high-quality two-dimensional tomographic image (synthesized) The (image) generation process is not limited to the averaging process, and a median calculation process, a weighted averaging process, or the like may be used.

  Next, the image generation unit 118 generates a volume image shown in FIG. 3A from a plurality of binary tomographic images after the imaging instruction stored in the image storage unit 112.

<Step S205>
In step S205, the display control unit 114 displays at least one of the volume image generated in step S204 and the high-quality two-dimensional tomographic image on a display unit (not shown).

  As is clear from the above description, in this embodiment, a high-quality two-dimensional tomographic image is generated using a tomographic image taken to determine the photographing position of the retinal layer, and a volume obtained by photographing a wide range. An image is generated.

  As a result, it is possible to acquire high-quality two-dimensional tomographic images and volume images that capture a wide area without increasing the imaging time, thereby reducing the burden on the subject and the photographer. I can do it. Further, since the image used for alignment is taken while the doctor (surgeon) observes, a two-dimensional image of the same place is acquired as the end of alignment approaches. For this reason, it is possible to superimpose images with little misalignment, leading to high image quality.

  Furthermore, since an image selected from the two-dimensional image used for alignment can be used, higher image quality can be achieved.

  In the present embodiment, a method for generating a volume image and one high-quality two-dimensional tomographic image has been described. However, the present invention is not limited to this. For example, as shown in FIGS. 6A and 6B, there may be a plurality of scanning patterns for aligning the retinal layer. In FIG. 6, R indicates a range for capturing a volume image, and L1 to L6 indicate positions where the tomographic imaging apparatus 120 repeatedly scans the same part of the eye to be inspected while the operator aligns the retinal layer. ing.

  In the image storage unit 112, a two-dimensional tomographic image before an imaging instruction is stored by being distinguished at each scanning position (L1 to L6), and the above-described processing is performed at each position, whereby a plurality of high-quality two-dimensional two-dimensional images are obtained. A tomographic image can be acquired.

(Example 2)
In the first embodiment, a high-quality two-dimensional tomographic image is generated using a tomographic image captured to determine the imaging position of the retinal layer, and a volume image capturing a wide range is generated. The present embodiment is characterized in that segmentation of the retinal layer is performed using the high-quality two-dimensional tomographic image generated above, and the retinal layer in the volume image is segmented using the result.

  According to this embodiment, the segmentation of the volume image is performed using the result of segmentation with a high-quality two-dimensional tomographic image as an initial value (or reference). As a result, the volume image can be segmented with high accuracy.

  FIG. 7 is a diagram illustrating a configuration of an image processing system 700 including the image processing apparatus 710 according to the present embodiment. As illustrated in FIG. 7, the image processing apparatus 710 includes an image acquisition unit 111, an image storage unit 112, an image processing unit 713, a display control unit 714, and an instruction acquisition unit 115. Among these, since functions other than the image processing unit 713 and the display control unit 714 have the same functions as those of the first embodiment, description thereof is omitted here.

  In the image processing unit 713, the first layer detection unit 719 detects a retinal layer from a high-quality two-dimensional tomographic image, and the second layer detection unit 720 detects a retinal layer from the volume image. The alignment unit 717 performs processing for aligning the result of layer detection with a high-quality two-dimensional tomographic image with the volume image.

  Hereinafter, the processing procedure of the image processing apparatus 710 of this embodiment will be described with reference to FIGS. 8 and 9. Since steps other than step S805 and step S806 are the same as steps S201 to S204 of the first embodiment, description thereof will be omitted.

<Step S805>
In step S805, a retinal layer is first detected from a high-quality two-dimensional tomographic image.

  First, the first layer detection unit 719 performs smoothing filter processing on the two-dimensional tomographic image to remove noise components. Then, edge detection filter processing is performed to detect edge components from the tomographic image, and edges corresponding to layer boundaries are extracted. Compared to a low-quality tomographic image, the high-quality two-dimensional tomographic image has a reduced noise component and an enhanced signal component in the retinal layer. Therefore, the edge component corresponding to the layer boundary can be extracted with high accuracy. The first layer detection unit 719 identifies a background region from the two-dimensional tomographic image from which edge detection has been performed, and extracts a luminance value feature of the background region from the two-dimensional tomographic image. Then, the boundary of each layer is determined by using the peak value of the edge component and the luminance value feature between the peaks.

  For example, the first layer detection unit 719 searches for an edge in the depth direction of the fundus from the vitreous body side, and determines the vitreous body and the retinal layer from the peak of the edge component, the upper and lower luminance characteristics, and the luminance characteristics of the background. Determine the boundary. Further, the first layer detection unit 719 searches for an edge in the depth direction of the fundus and determines the boundary of the retinal pigment epithelium by referring to the edge component peak, the luminance characteristic between the peaks, and the luminance characteristic of the background. To do. Through the above processing, the layer boundary can be detected. This is shown in FIG. FIG. 9A shows the boundaries (1 to 6) of the retinal layer detected from the high-quality two-dimensional tomographic image Ti ″. When performing layer detection with a volume image, FIG. Refer to the result of the layer boundary indicated by

In order to apply the layer detection result obtained from the two-dimensional tomographic image to the volume image, the alignment unit 717 performs alignment processing. This alignment process will be described with reference to FIG. FIG. 9B shows a high-quality two-dimensional tomographic image Ti ″ and a tomographic image Tk / 2 obtained by capturing the center of the volume image. In this embodiment, the high-quality two-dimensional tomographic image Ti ″ is a volume. The image is generated from a group of tomographic images in which a position corresponding to the center of the volume image was captured before the image was captured. However, since the eye slightly moves due to fixation fixation or the like, the position is aligned with several tomographic images near the center of the volume image, and an optimal position is selected.
Since the alignment method uses the method shown in step S204 of the first embodiment, description thereof is omitted here. The retinal layer detection results (FIG. 9 (b) 1 to 6) of the two-dimensional tomographic image are superimposed on the tomographic image having the highest evaluation value among several tomographic images near the center of the volume image.

  Next, the second layer detection unit 720 detects the retinal layer from the volume image. First, the second layer detection unit 720 performs a smoothing filter process on the volume image to remove noise components. Then, edge detection filter processing is performed to detect edge components from the tomographic image, and edges corresponding to layer boundaries are extracted. Next, using the respective boundaries of the retinal layer (FIG. 9 (b) 1 to 6) detected by the first layer detection unit 719 as an initial value, an edge is searched in the vertical direction, and one two-dimensional image in the volume image A retinal layer is detected in a tomographic image. Alternatively, with each boundary of the retinal layer detected by the first layer detection unit 719 as an initial value, a dynamic contour method (Snakes, LevelSet, etc.) is applied, and the retina is detected in one two-dimensional tomographic image in the volume image. Detect layer. The retinal layer can be detected with high accuracy in one two-dimensional tomographic image in a volume image that has poor image quality compared to a high-quality two-dimensional tomographic image. Thereby, it is possible to extract the feature amount for each layer region and the feature amount of the background region (average luminance value, variance, standard deviation, etc.) from the volume image. The second layer detection unit 720 performs retinal layer detection from all three-dimensional tomographic images using the feature values obtained there.

  For example, the second layer detection unit 720 searches for an edge in the depth direction of the fundus from the vitreous body side in a tomographic image other than the tomographic image obtained as the initial value. Then, the boundary between the vitreous body and the retinal layer is determined by referring to the peak of the edge component, the luminance characteristics above and below it, and the luminance characteristic of the background obtained from one two-dimensional tomographic image in the volume image.

  Alternatively, as a method for the second layer detection unit 720 to detect the retinal layer from the volume image, a layer boundary may be detected by a method using graph theory such as GraphCut. Since GraphCut requires seed as an initial value, the retinal layer detection result obtained by the first layer detection unit 719 is applied to the volume image, and the inside of the retinal layer region is used as the initial value (seed) of GraphCut. A node corresponding to each pixel of the image and a terminal called sink and source are set, and an edge (n-link) connecting the nodes and an edge (t-link) connecting the terminals and the nodes are set. As the initial value of GraphCut, pixels that are to be the object region O (Object seed) and the background region B (Background seed) are set in advance. In this case, the layer area obtained by the first layer detection unit 719 is set to O, and the other area is set to B in the layer area adapted to the volume image. A layer can be detected by obtaining a minimum cut for a graph created by giving weights to these edges.

  In the tomographic image adjacent to the tomographic image to which the result of the first layer detection unit 719 is applied, the retinal layer is detected using spatial information. In a volume image in which the alignment between adjacent tomographic images has been completed, the similarity between the retinal layer shapes photographed between adjacent tomographic images is high. Therefore, the retinal layer detection result of the tomographic image to which the result of the first layer detection unit 719 is adapted can be applied to the adjacent tomographic image.

  In a tomographic image to which the result of the first layer detection unit 719 is applied, in a pixel whose distance is away from the boundary between the object and the background, the possibility that the object and the background are interchanged is low even in the pixel of the corresponding adjacent tomographic image. That is, for example, when the pixel of the tomographic image to which the result of the first layer detection unit 719 is applied is the background, the pixel of the corresponding adjacent tomographic image is highly likely to be the background. Therefore, the t-link cost is set using information indicating that the pixel is the background in the corresponding pixel of the adjacent tomographic image. In addition, in the tomographic image to which the result of the first layer detecting unit 719 is applied, it is not known whether the pixel of the corresponding adjacent tomographic image is the object or the background in the pixels near the boundary between the object and the background. Therefore, the cost of t-link is set in the pixel of the corresponding adjacent tomographic image using information that the possibility of being the background or the object can occur to the same extent.

  The above processing is applied to all three-dimensional tomographic images to perform retinal layer detection.

<Step S806>
In step S806, the display control unit 714 displays at least one of the volume image created in step S804 and the high-quality two-dimensional tomographic image, and displays the layer boundary detected in step S805 on a display unit (not shown). . As a display method, the layer boundary is superimposed and displayed on the tomographic image. Alternatively, it may be displayed along with the tomographic image as a graph. Furthermore, the thickness of an arbitrary layer may be measured from the layer detection result, and the measurement result may be displayed.

  According to the configuration described above, a high-quality two-dimensional tomographic image is obtained simultaneously with the volume image. Therefore, when performing segmentation of a volume image, segmentation is performed with a high-quality two-dimensional tomographic image, and the detection result is set as an initial value, and the volume image is segmented. As a result, the volume image can be segmented with high accuracy. In addition, as described above, a two-dimensional tomographic image captured for alignment is used for segmentation before the start of volume image capturing. Therefore, since the correlation between the position of the volume image and the two-dimensional captured image is high, the volume image can be segmented with higher accuracy. Furthermore, the two-dimensional tomographic image used for alignment reflects the position that the operator wants to observe, and therefore the operator's intention can also be reflected in the segmentation of the volume image.

  In the present embodiment, the high-quality two-dimensional tomographic image has been described as an image created using a tomographic image photographed for photographing and positioning of the retinal layer, but the present invention is not limited to this. For example, layer detection may be applied with reference to a high-quality two-dimensional tomographic image generated from a tomographic image obtained by photographing a plurality of the same locations after an imaging instruction. Alternatively, layer detection may be applied with reference to a high-quality two-dimensional tomographic image generated by reducing the scanning interval (higher resolution). Furthermore, layer detection may be applied with reference to high-quality two-dimensional tomographic images taken at different times. That is, the present embodiment can be applied to any configuration that can obtain a tomographic image with higher image quality than a volume image.

(Other embodiments)
Each of the above embodiments implements the present invention as an image processing apparatus. However, the embodiment of the present invention is not limited only to the image processing apparatus. The present invention can also be realized as software that runs on a computer. The CPU of the image processing apparatus controls the entire computer using computer programs and data stored in RAM and ROM. In addition, the execution of software corresponding to each unit of the image processing apparatus is controlled to realize the function of each unit.

DESCRIPTION OF SYMBOLS 110 Image processing apparatus 120 Tomographic imaging apparatus 112 Image memory | storage part 113 Image processing part 114 Display control part 115 Instruction acquisition part 116 Image selection part 117 Positioning part 118 Image generation part 719 First layer detection part 720 Second layer detection Part

Claims (16)

Based on a smaller number of two-dimensional tomographic images than the plurality of two-dimensional tomographic images out of the plurality of two-dimensional tomographic images of the fundus acquired when adjusting the position in the depth direction on the fundus of the eye to be examined. Generating means for generating a simple two-dimensional tomographic image;
First layer detecting means for detecting the fundus layer in the new two-dimensional tomographic image;
Based on the detection result of the first layer detecting means, at least one two-dimensional tomographic image among a plurality of two-dimensional tomographic images constituting a three-dimensional image acquired after the plurality of two-dimensional tomographic images are acquired. Second layer detecting means for detecting the fundus layer from an image;
An image processing apparatus comprising:   The second layer detection means has acquired the new two-dimensional tomographic image among a plurality of two-dimensional tomographic images constituting the three-dimensional image based on the detection result of the first layer detection means. The image processing apparatus according to claim 1, wherein the fundus layer is detected from a second two-dimensional tomographic image adjacent to the first two-dimensional tomographic image acquired at a position corresponding to the position. The second layer detecting means is
Based on the detection result of the first layer detecting means, among the plurality of two-dimensional tomographic images constituting the three-dimensional image, the layer of the fundus is detected from the first two-dimensional tomographic image,
The image processing apparatus according to claim 2, wherein the fundus layer is detected from the second two-dimensional tomographic image based on the detection result.   Of the plurality of two-dimensional tomographic images constituting the three-dimensional image, the two-dimensional tomographic image acquired at a position corresponding to the position at which the new two-dimensional tomographic image is acquired is displayed on the display means, and the new The image processing apparatus according to claim 1, further comprising display control means for displaying a layer detected from the two-dimensional tomographic image so as to overlap the displayed two-dimensional tomographic image. A selection means for selecting a smaller number of two-dimensional tomographic images than the plurality of two-dimensional tomographic images;
5. The image according to claim 1, wherein the generation unit generates the new two-dimensional tomographic image by superimposing the plurality of selected two-dimensional tomographic images. 6. Processing equipment.   The image processing apparatus according to claim 5, wherein the selection unit selects a two-dimensional tomographic image having an image quality evaluation value equal to or greater than a threshold value from the plurality of two-dimensional tomographic images. The plurality of two-dimensional tomographic images are images acquired at substantially the same position in the planar direction on the fundus before instructing to acquire a three-dimensional image of the fundus.
The three-dimensional image is an image composed of a plurality of two-dimensional tomographic images acquired at different positions in a planar direction on the fundus after an instruction to acquire the three-dimensional image of the fundus. The image processing apparatus according to any one of 1 to 6.   The depth direction of another two-dimensional tomographic image of the two-dimensional tomographic image immediately before the instruction among the plurality of two-dimensional tomographic images at a position in the depth direction of the two-dimensional tomographic image immediately before the instruction to acquire the three-dimensional image of the fundus The image processing apparatus according to claim 1, further comprising alignment means for aligning the positions of the image processing apparatus.   An acquisition unit that is communicably connected to an imaging device that images the fundus and acquires a plurality of two-dimensional tomographic images of the fundus obtained by imaging the fundus using the imaging device when adjusting the position in the depth direction. The image processing apparatus according to claim 1, further comprising:   A program for causing a computer to execute as each unit of the image processing apparatus according to any one of claims 1 to 9. Based on a smaller number of two-dimensional tomographic images than the plurality of two-dimensional tomographic images out of the plurality of two-dimensional tomographic images of the fundus acquired when adjusting the position in the depth direction on the fundus of the eye to be examined. Generating a two-dimensional tomographic image,
A first layer detecting step of detecting the fundus layer in the new two-dimensional tomographic image;
Based on the detection result in the first layer detection step, at least one two-dimensional tomographic image among a plurality of two-dimensional tomographic images constituting a three-dimensional image acquired after the plurality of two-dimensional tomographic images are acquired. A second layer detection step of detecting the fundus layer from an image;
An image processing method comprising:   In the second layer detection step, the new two-dimensional tomographic image is acquired from a plurality of two-dimensional tomographic images constituting the three-dimensional image based on the detection result of the first layer detecting unit. The image processing method according to claim 11, wherein the fundus layer is detected from a second two-dimensional tomographic image adjacent to the first two-dimensional tomographic image acquired at a position corresponding to the position. In the second layer detection step,
Based on the detection result in the first layer detection step, out of a plurality of two-dimensional tomographic images constituting the three-dimensional image, the layer of the fundus is detected from the first two-dimensional tomographic image,
The image processing method according to claim 12, wherein the fundus layer is detected from the second two-dimensional tomographic image based on the detection result.   Of the plurality of two-dimensional tomographic images constituting the three-dimensional image, the two-dimensional tomographic image acquired at a position corresponding to the position at which the new two-dimensional tomographic image is acquired is displayed on the display means, and the new The image processing method according to claim 11, further comprising a step of displaying a layer detected from a two-dimensional tomographic image so as to overlap the displayed two-dimensional tomographic image. A selection step of selecting a smaller number of two-dimensional tomographic images than the plurality of two-dimensional tomographic images;
The image according to any one of claims 11 to 14, wherein, in the generation step, the new two-dimensional tomographic image is generated by overlaying the selected two-dimensional tomographic images. Processing method.   The image processing method according to claim 15, wherein in the selection step, a two-dimensional tomographic image having an image quality evaluation value equal to or greater than a threshold value is selected from the plurality of two-dimensional tomographic images.

Images