JP6042573B2 - Image processing apparatus, image processing method, and image processing program - Google Patents

Description

  The present invention relates to an image processing apparatus, an image processing method, and an image processing program for processing a tomographic image captured by a tomographic imaging apparatus or the like to generate an image suitable for a diagnostic image.

  As one of ophthalmologic diagnosis apparatuses, a tomographic imaging apparatus using optical interference called OCT (Optical Coherence Tomography) for capturing a tomographic image of the fundus has been put into practical use. With such a tomographic imaging apparatus, signal data in the z direction (A scan) obtained by scanning the fundus along the x direction, with the left and right direction of the fundus in the x direction, the vertical direction in the y direction, and the depth in the z direction. By arranging data) groups in the scanning direction (x direction), a tomographic image (B scan image) in the xz direction can be acquired. If general OCT imaging is performed, for example, a tomographic image is captured at a speed of 40 images / second, and a group of 100 or more retinal tomographic images can be acquired by one examination (imaging of a part of the retina). .

  However, since these tomographic images are B-scan images created by simply processing the acquired A-scan data, they contain a lot of noise and so on, so that each image is suitable for interpretation. Absent. In view of this, various image processing methods have been proposed in order to generate high-quality images suitable for interpretation. For example, an addition process is performed on images of a captured tomographic image group to obtain a tomographic image. A process of creating an image is performed. Patent Document 1 discloses a technique for generating a tomographic image with less noise by adding and averaging the entire photographed two-dimensional tomographic image.

  By the way, the photographed tomographic image may include an image in which distortion or positional deviation has occurred due to the influence of fixation fine movement. If a tomographic image in which such distortion or misalignment occurs is included in the processing target and the averaging process is performed, a small amount of lesion information may be erased by the averaging process, or the tomographic image for interpretation obtained after the averaging process May not be as good as expected.

  Therefore, instead of using the entire captured tomographic image group as a target for the averaging process, the obtained tomographic images are frequently selected from the viewpoint of similarity and used for the averaging process. For example, in Patent Document 2 and Patent Document 3, when generating a composite image from a plurality of tomographic images taken at the same position, only a tomographic image having a similarity higher than a threshold value is selected, and a tomographic image having a large positional deviation. A technique for avoiding the use of is disclosed.

  In addition, in order to improve the image quality of the tomographic image obtained after the averaging, an averaging process is also performed while correcting a local positional deviation that occurs during imaging. For example, in Patent Document 4, one image is taken out from a plurality of tomographic images as a reference image, a predetermined number of tomographic images are selected, and the reference image and the selected image are divided into the same number of regions, respectively. A technique for correcting a positional shift between selected images for each divided region is disclosed. Non-Patent Document 1 discloses a technique for directly aligning A scan data obtained by measurement one by one. Further, Patent Document 5 discloses a technique for performing alignment in consideration of information on a region around a region to be aligned when performing local alignment.

JP 2008-237238 A JP 2009-005789 A JP 2009-291313 A Japanese Patent No. 5199031 International Publication No. WO2014 / 103501A1

"Enhancing the signal-to-noise ratio in ophthalmic optical coherence tomography by image registration-method and clinical examples", Thomas Martini Jorgensen et al., Journal of Biomedical Optics, Vol. 12 (4), 041208

  As described above, a predetermined number of images having a higher degree of similarity are selected from the obtained tomographic image group and used for the averaging process, or the averaging process is performed while correcting a local positional deviation that occurs during imaging. This makes it possible to generate a high-quality additive average tomographic image suitable for interpretation. However, the data processing amount for that purpose is considerable, and there is a problem that it takes time to generate an image suitable for a diagnostic image.

  The present invention has been made in view of the above points, and from the A scan data group obtained by scanning substantially the same position of the object a plurality of times, the high quality suitable for interpretation while suppressing the data processing amount. It is an object of the present invention to provide an image processing apparatus, an image processing method, and an image processing program capable of generating a tomographic image.

  In order to achieve the above object, first, the present invention provides a plurality of sets of A scan data obtained by scanning substantially the same position of an object a plurality of times in a tomographic imaging apparatus that captures a tomographic image of the fundus of the eye to be examined. An image processing apparatus for processing a group to generate a tomographic image for interpretation, wherein a plurality of B scan images are generated from the plurality of sets of A scan data groups and are simply added to form a simple addition B scan image. Generating a B scan image having the highest correlation with the simple addition B scan image among the plurality of B scan images, and specifying a group of A scan data corresponding to the highest correlation B scan image as reference A scan data From the reference data selection means for selecting as a group and the B scan image generated from the plurality of sets of A scan data groups other than the reference A scan data group, the reference A A predetermined number of B scan images having a high correlation with the B scan image generated from the can data group are specified, and a predetermined number of sets of A scan data groups corresponding to the specified predetermined number of B scan images are targeted A Target data selection means for selecting as a scan data group, a reference B scan image is generated from the reference A scan data group, and the A scan data included in the target A scan data group is respectively obtained for the reference B scan image. An image processing apparatus is provided that includes an averaging means for generating a tomographic image for interpretation by averaging (Invention 1).

  According to the above invention (Invention 1), a plurality of sets of A-scan data groups obtained by scanning substantially the same position of an object a plurality of times are directly used as object data for subsequent averaging processing, A predetermined number of B scan images having a high correlation with the B scan image generated from the A scan data group are specified, and only a predetermined set number of A scan data groups corresponding to the specified predetermined number of B scan images are obtained. By setting the target A scan data group as the target of the averaging process, it is possible to remove the A scan data group in which distortion or misalignment has occurred due to the influence of fixation micromotion and perform the averaging process. It becomes possible to generate an additive average tomographic image of quality.

  In the said invention (invention 1), it is preferable that the said reference | standard data selection means and the said object data selection means produce | generate a B scan image using some A scan data of an A scan data group (invention 2).

  Each B scan image used in the process of selecting a target A scan data group to be subjected to the averaging process from a plurality of sets of A scan data groups obtained by scanning the substantially same position of the target object a plurality of times It is sufficient that the A scan data included in the A scan data group to be arranged is simply arranged, and it is sufficient that each B scan image can be correlated by similarity determination or the like. According to the above invention (invention 2), since the B scan image is generated by using a part of the A scan data of the A scan data group, the generated B scan image is a partially thinned image. The amount of data processing can be suppressed as compared to using all the A scan data of the A scan data group. Even in the case of a partially thinned B-scan image, it can be used without any problem in most cases in order to see the correlation by similarity determination or the like. The amount of data processing can be suppressed without substantially affecting it.

  In the above inventions (1, 2), the averaging means continuously outputs A scan data at corresponding positions in each of the predetermined number of sets of A scan data groups included in the target A scan data group. It is preferable to perform the treatment (Invention 3).

  According to the above invention (Invention 3), even if the A scan data has a different number of scans, the A scan data at the corresponding position is likely to be similar A scan data. Data processing volume can be reduced by processing certain A scan data continuously.

  Second, the present invention processes a plurality of sets of A-scan data groups obtained by scanning a substantially same position of a target object a plurality of times in a tomographic imaging apparatus that captures a tomographic image of the fundus of the eye to be examined, and interprets tomographic images. An image processing method for generating an image, wherein a plurality of B scan images are generated from the plurality of sets of A scan data groups and are simply added to generate a simple addition B scan image, and the plurality of B scan images are generated. A reference data selection step of identifying a B scan image having the highest correlation with the simple addition B scan image among the scan images, and selecting an A scan data group corresponding to the highest correlation B scan image as a reference A scan data group Generated from the reference A scan data group from B scan images generated from the plurality of sets of A scan data groups other than the reference A scan data group. A predetermined number of B scan images having a high correlation with the B scan image to be selected, and a predetermined set number of A scan data groups corresponding to the specified predetermined number of B scan images are selected as target A scan data groups A target data selection step, a reference B scan image is generated from the reference A scan data group, and the A scan data included in the target A scan data group is averaged with respect to the reference B scan image. There is provided an image processing method including an addition averaging step for generating a tomographic image (Invention 4).

  According to the above invention (invention 4), since a plurality of sets of A scan data groups obtained by scanning substantially the same position of the object a plurality of times are used as reference data for each subsequent process, the obtained A scan data The amount of data processing can be reduced as compared with the case where a B-scan image is once generated from a group and used as reference data for each subsequent process. On the other hand, a predetermined number of B scan images having a high correlation with the B scan image generated from the reference A scan data group are specified, and a predetermined number of sets of A corresponding to the specified predetermined number of B scan images. By using only the scan data group as the target A scan data group as the target of the averaging process, it is possible to perform the averaging process by removing the A scan data group that is distorted or misaligned due to the influence of fixation movement. It becomes possible to generate a high-quality additive average tomographic image suitable for interpretation.

  In the said invention (invention 4), it is preferable to produce | generate a B scan image using the A scan data of a part of A scan data group in the said reference | standard data selection process and the said object data selection process (invention 5).

  Each B scan image used in the process of selecting a target A scan data group to be subjected to the averaging process from a plurality of sets of A scan data groups obtained by scanning the substantially same position of the target object a plurality of times It is sufficient that the A scan data included in the A scan data group to be arranged is simply arranged, and it is sufficient that each B scan image can be correlated by similarity determination or the like. According to the above invention (invention 5), since the B scan image is generated by using a part of the A scan data of the A scan data group, the generated B scan image is a partially thinned image. The amount of data processing can be suppressed as compared to using all the A scan data of the A scan data group. Even in the case of a partially thinned B-scan image, it can be used without any problem in most cases in order to see the correlation by similarity determination or the like. The amount of data processing can be suppressed without substantially affecting it.

  In the above inventions (4, 5), in the addition averaging step, A scan data at a corresponding position in each of the predetermined number of sets of A scan data groups included in the target A scan data group is continuously generated. It is preferable to process (Invention 6).

  According to the above invention (invention 6), even if the A scan data has a different number of scans, the A scan data at the corresponding position is likely to be similar A scan data. Data processing volume can be reduced by processing certain A scan data continuously.

  Third, the present invention is for causing a computer to function as an image processing apparatus according to any one of inventions 1 to 3, or for causing a computer to execute the image processing method according to any one of inventions 4 to 6. An image processing program is provided (Invention 7).

  According to the image processing apparatus, the image processing method, and the image processing program of the present invention, a high-quality tomographic image suitable for interpretation while suppressing the data processing amount from the A scan data group obtained by scanning the same portion a plurality of times. Can be generated.

1 is an optical diagram showing an overall configuration of a tomographic imaging apparatus according to an embodiment of the present invention. It is explanatory drawing which shows the relationship between the A scan data obtained by a tomography apparatus, the B scan image produced | generated from the said A scan data, and a scanning direction. It is explanatory drawing which shows the A scan data storage state to the data storage area of a data storage part. It is a flowchart which shows the flow of the process in this embodiment. It is a schematic diagram explaining the positional deviation correction and the addition averaging process of A scan data. It is a schematic diagram explaining the processing order of A scan data.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. As shown in FIG. 1, the tomographic imaging apparatus according to the present embodiment captures a tomographic image of a desired region of the fundus using the fundus of the eye E to be imaged as an object to be imaged. A portion denoted by reference numeral 10 is a demultiplexing / multiplexing optical system. This optical system emits light having a wavelength of 700 nm to 1100 nm and a temporal coherence length of about several μm to several tens of μm, for example, a superluminescent diode. A broadband low-coherence light source 11 made of (SLD) is provided.

  The low-coherence light generated by the low-coherence light source 11 is adjusted in light quantity through the light quantity adjustment mechanism 12 and is incident on the optical coupler 13 through the optical fiber 13a, and then split optical elements through the optical fiber 13b and the collimating lens 14. To the beam splitter 15. In addition, you may make it branch and multiplex using an optical circulator instead of the optical coupler 13. FIG.

  The light incident on the beam splitter 15 is divided into reference light and measurement light. The measurement light enters the focus lens 31 and the measurement light is focused on the fundus of the eye E. The measurement light focused on the fundus is reflected by the mirror 32, passes through the lens 33, and is scanned in an arbitrary direction by the x-axis scanning mirror (galvano mirror) 34 and the y-axis scanning mirror (galvano mirror) 35. The measurement light scanned by the x-axis and y-axis scanning mirrors 34 and 35 passes through the scan lens 36, is reflected by the dichroic mirror 37, passes through the objective lens 38 and enters the fundus oculi, and the fundus is measured by the measurement light. Are scanned in the x and y directions. The measurement light reflected from the fundus returns to the beam splitter 15 by reversing the above path.

  In such an optical system, the focus lens 31, mirror 32, lens 33, x-axis scanning mirror 34, y-axis scanning mirror 35, scan lens 36, dichroic mirror 37 and objective lens 38 after the beam splitter 15 are tomographic images. A measurement optical system 30 of the photographing apparatus is configured. This measurement optical system is provided with optical components such as a mirror and a lens as appropriate in addition to the illustrated optical components, but is omitted in order to avoid complexity.

  On the other hand, the reference light split by the beam splitter 15 is reflected by the mirror 41 and then passes through the objective lens dispersion compensation glass 42 and the lenses 43 and 44. After that, the light is reflected by the mirror 45 and passes through the eye dispersion compensation glass 50 that compensates the refractive index dispersion of the eye E to be examined. Then, the light is reflected by the dichroic mirror 46 to adjust the condenser lens 47 and the amount of light. It passes through the variable aperture 48 and reaches the reference mirror 49. In order to adjust the optical path length, the condensing lens 47, the variable aperture 48, and the reference mirror 49 move together in the optical axis direction as shown by a double arrow in FIG. The reference light reflected by the reference mirror 49 returns to the beam splitter 15 by reversing the above optical path.

  In such an optical system, the mirror 41, the objective lens dispersion compensation glass 42, the lenses 43 and 44, the mirror 45, the eye dispersion compensation glass 50, the dichroic mirror 46, the condenser lens 47, and the reference mirror 49 as a reference object A reference optical system 40 of the tomographic imaging apparatus is configured. This reference optical system is provided with optical components such as a mirror and a lens as appropriate in addition to the illustrated optical components, but is omitted in order to avoid complications.

  The measurement light and the reference light that have returned to the beam splitter 15 are superimposed to become interference light, which passes through the collimating lens 14, the optical fiber 13b, and the optical coupler 13 and enters the spectroscope 16 through the optical fiber 13c. The spectroscope 16 includes a diffraction grating 16a, an imaging lens 16b, a line sensor 16c, and the like. The interference light is split into a spectrum corresponding to the wavelength of the low coherence light by the diffraction grating 16a and is lined by the imaging lens 16b. An image is formed on the sensor 16c.

  The signal from the line sensor 16c is subjected to signal processing including Fourier transform by the signal processing means 18 realized by the CPU of the computer 17, and a depth signal indicating information in the depth direction (z direction) of the fundus is generated. The The depth signal (A scan data) at the sampling time is obtained by the interference light at each sampling time of the fundus scanning. Therefore, when one scan is completed, a plurality of A scan data corresponding to a plurality of positions along the scanning direction can be acquired, and a two-dimensional tomographic image (B scan image) can be generated therefrom.

  Hereinafter, in the description of the present embodiment, it is assumed that scanning with N samplings is performed M times. In this case, the M scans are performed at substantially the same position unless the influence of fixation micromotion is considered. Although the scanning direction is the same in this embodiment every time, the even-numbered scanning may be performed in the reverse direction, and the sampling order (scanning direction) is not particularly limited. If the number of samplings in one scan is N, N A scan data can be acquired when one scan is completed, and a B scan image as shown in FIG. 2 can be generated from the N A scan data. it can. If the scanning of a desired region of the fundus is repeated M times in this way, a total of M sets of A scan data groups composed of N A scan data are generated. Since one B scan image can be generated from each A scan data group, M B scan images can be generated from M sets of A scan data groups. The N × M A scan data groups are sequentially stored in the data storage unit 19 in the computer 17.

FIG. 3 is an explanatory diagram showing how the A scan data group is stored in the data storage area R of the data storage unit 19. The data storage area R is divided into vertical N rows and horizontal M columns, and one data storage area R _ij (i = 1 to N, j = 1 to M). When the first scan starts, A scan data is generated for each sampling, and A _ij (i = 1 to N, j = 1 to 1) is obtained by j scans (i samplings per scan). expressed as _{M), R} 21, _R 31 from the upper left of the data storage area _{R 11} sequentially, · · · _{R N1} to the a scan data _{a 11, a 21, a 31} , ··· a N1 is stored, respectively Go. Then turn the second scanning starts from a data storage area _{R 12} on the second column top from left to right _{R 22, R 32,} ··· and A-scan to _{R N2} data _{A 12, A 22,} A ₃₂ ,..., A _N2 are respectively stored, and then the third, fourth,..., M−1th scan is sequentially performed, and in the Mth scan, the upper right data storage area R _1M is scanned. A scan data A _1M , A _2M , A _3M ,..., A _NM are respectively stored in R _2M , R _{3M ,.} In the present embodiment, each A scan data group composed of N A scan data A _ij obtained in the j-th (j = 1 to M) scan is represented as AG _j (j = 1 to M). For example, the A scan data group obtained by the _first scan becomes an A scan data group AG1 as shown in FIG.

  As described above, the computer 17 generates A scan data (group) by the signal processing means 18 and stores it in the data storage unit 19, and also uses the stored A scan data (group) to provide high quality suitable for interpretation. It also has a function as tomographic image generation means 20 for generating a tomographic image of the above. That is, in the present embodiment, the computer 17 corresponds to the image processing apparatus according to the present invention. The tomographic image generation means 20 includes a reference data selection step for selecting a reference A scan data group (reference A scan data group) and a target for selecting an A scan data group (target A scan data group) to be added and averaged. A tomographic image for interpretation is generated through a data selection step and an addition and averaging step of performing an averaging process based on the target A scan data group and the reference A scan data group to generate a tomographic image for interpretation. The flow of processing by the tomographic image generation means 20 will be described below.

  FIG. 4 is a flowchart showing the flow of processing by the tomographic image generation means 20, which includes a reference data selection step (S101 to 105), a target data selection step (S201 to 203), and an addition averaging step (S301 to 305). Consists of

First, M B scan images B _j (j = 1 to M) are generated from M sets of A scan data groups AG _j (j = 1 to M) obtained as a result of performing M scans ( S101). The B scan image B _j generated at this time does not use all the N pieces of A scan data A _ij constituting each set of A scan data group AG _j , but is equally spaced at a rate of one in two. Generated by arranging the thinned N / 2 A scan data in the scanning direction. Thus, the amount of data processing can be suppressed by thinning out the A scan data and using it for B scan image generation.

  When the A scan data is thinned and used to generate the B scan image, it is not always necessary to thin out one scan every two at equal intervals. For example, two scans of three are thinned at equal intervals. N / 3 pieces of A scan data may be used, or may be used by being thinned out at equal intervals at other ratios. Further, when the B scan image is generated, the A scan data to be used may be determined in consideration of which part of the B scan image is the A scan data. In many cases, the most desired place in the B-scan image is near the center. For example, in the A-scan data that forms the vicinity of the center of the B-scan image, one is used for every two, and the other places are configured. In the A scan data to be used, the number of A scan data to be used may be changed between the vicinity of the center and the others, for example, one in three. Alternatively, when looking at the effects of fixation disparity, it is desirable to pay attention to the start and end of shooting. For example, in the A scan data constituting the both ends and the center of the B scan image, the ratio is one in two. In the A scan data constituting other locations, it may be used at a rate of one in three.

  As described above, the tomographic image generation means 20 according to the present embodiment adopts only a part of A scan data from the N A scan data constituting the A scan data group as necessary, or N pieces of A scan data. The use of all A scan data can also be selected. Also, when only a part of A scan data is adopted, how much A scan data is adopted (how much A scan data is thinned out) and how it is adopted (how thinned out). It can be set freely.

After the M B scan images B _j (j = 1 to M) are generated, the M B scan images B _j are simply added to generate one simple added B scan image B _A ( S102). Subsequently, the similarity of each of the M B scan images B _{j with} respect to the simple addition B scan image B _A is calculated, and which B scan image B _j has a higher similarity is compared (S103). As the evaluation function indicating the similarity, for example, the correlation coefficient shown in Equation 1 can be used.

Here, B (k) is a set of pixel values (number of pixels n), and B (upper horizontal line) is an average of pixel values. The similarity with each image B _j is obtained by translating, rotating, enlarging / reducing each image and aligning with the simple addition B-scan image B _A to obtain a correlation function. Alternatively, the correlation function may be obtained. Further, the similarity can be calculated not by using the correlation coefficient but by using the difference (positional deviation amount) between the corresponding pixel values between each image B _j and the simple addition B-scan image B _A. Further, the correlation coefficient or the positional deviation amount may be obtained for the entire image, or may be obtained for a partial region (for example, a region having a lesioned part or a characteristic part).

The similarities with respect to the simple addition B-scan image B _A obtained for each B-scan image B _j are compared, and the B-scan image with the highest similarity is specified as the selected B-scan image B _S (S104). Furthermore, the A scan data group corresponding to the selected B-scan image _{B S,} i.e. the A scan data groups from which to generate the selection B-scan image _{B S} is selected as the reference A-scan data group AG _S (S105).

Subsequently, the similarity of M-1 B scan images other than the selected B scan image B _{S with} respect to the selected B scan image B _S is calculated, and which B scan image has a higher similarity is compared (S201). The calculation of the similarity can be carried out in the same manner as step S103 and S104 to identify the selected B-scan image B _S above.

Result of the comparison, identifies the B-scan image of m sheets in the order of higher similarity to selected B-scan image _{B S (m <M) (} S202), A -scan data group corresponding to the m pieces of B-scan image, i.e. The m sets of A scan data groups from which these m B scan images are generated are selected as target A scan data groups AG _Tk (k = 1 to m) (S203). The target A scan data group AG _Tk (k = 1 to m) is an A scan data group to be added in the subsequent averaging process. For example, assuming that the number of samplings is N = 1,800, the number of scannings is M = 100, and the number of sets to be added is m = 30, 1,800 × 100 = 180,000 A stored in the data storage unit 19. The data aggregate composed of the scan data is reduced to a data aggregate composed of 1,800 × 30 = 54,000 A scan data, and the 54,000 A scan data are subjected to the averaging process. It will be.

In the selection of the target A scan data group AG _Tk , the B scan images are not simply specified in the descending order of similarity, but a predetermined similarity threshold is set in advance and the similarity exceeds the threshold. A method of extracting all the B scan images and selecting all the A scan data groups from which the extracted B scan images are generated as the target A scan data group AG _Tk may be employed.

In the reference data selection process (S101 to 105) and the target data selection process (S201 to 203) so far, the M sets of A scan data groups AG _j (j = 1 to M) obtained by M scans are obtained. among which data group or to the reference a-scan data group AG _S as a reference for the averaging process, and how the object data set are used to averaging process a-scan data group AG Tk _{(k =} 1~m) It is selected whether to do. In other words, object A scan data group AG _Tk used for averaging the reference A-scan data group AG _S as a reference for processing and averaging processing, obtained in many times the scanning of the scan was performed M times A It is determined whether it is a scan data group.

Thus, far steps The generated B-scan images are both images used only for selection of the reference A-scan data group AG _S and the target A-scan data group AG _Tk, subsequent averaging step It is not used for (S301 to 305). For this purpose, if the correlation is viewed by similarity determination or the like, even a partially thinned B-scan image can be used without any problem, and the image quality of the finally obtained averaged tomographic image The amount of data processing can be suppressed without substantially affecting.

Subsequently, the single reference B-scan image _{T R} is generated using all N A scan data constituting the reference A-scan data group AG _S (S301). Reference B-scan image T _R to be generated at this time is an image serving as a reference for subsequent averaging process, each A scan data A _T included in the target A-scan data group AG Tk _{(k =} 1~m) by going to averaging the reference B-scan image T _R while the positional deviation correction, and generates a tomographic image for interpretation. Since each A scan data group is composed of N A scan data, there are a total of N × m A scan data _AT to be added. Thereafter, the i-th sampled addition target A scan data A _Tik (i = 1 to N, k = 1 to m) constituting the k-th set of target A scan data group AG _Tk (k = 1 to m). the flow going arithmetic mean relative positional offset correction while the reference B-scan image T _R will be described.

First, as shown in FIG. 5, the addition target A scan data A _Tik is selected, and E which is a portion used for alignment in the addition target A scan data is set (S302). In this case, E is set at a height of, for example, 300 pixels so as to include data of a region where the retinal tissue in the depth direction of the fundus is imaged.

On the other hand, the reference B-scan image T _R, in order to evaluate the correlation between portions of the E of the addition target A scan data around the aforementioned E, so as to include a peripheral portion thereof, for example, a width of 31 pixels × height A search area F of 400 pixels is set (S303). Then, the addition target A-scan data, the reference B-scan image T _R in the search region F, x as the correlation value is maximized, based on the corrected position of the addition target A scan data is shifted in the z-direction B-scan image to search from the T _R. And by the correlation value determining the shift amount such that maximum, x, positional displacement amount Δx in the z-direction, Delta] z are calculated for the reference B-scan image T _R of addition target A scan data A _Tik (S304). The correlation value can be calculated using, for example, a formula similar to the formula (1).

Shifting Thus, x, positional deviation amount [Delta] x in the z-direction, when seeking Delta] z, reference B-scan image _T positional displacement amount addition target A scan data _{A Tik} against _R [Delta] x, by Delta] z component in the x, z-direction correcting the positional deviation by, will the addition target a scan data _{a Tik} added and averaged in the reference B-scan image _{T R} (S305).

The steps from S302 to S305 are repeated for all A scan data (N × m) constituting the target A scan data group AG _Tk , and when the averaging process is completed for all A scan data, the interpretation tomography The image is completed.

In addition, when iteratively performs misalignment correction and addition averaging processing for A scan data constituting the target A scan data group AG _Tk , it is possible to further reduce the data processing amount by repeating the processing in the following order. It becomes.

A total of m sets of A scan data are targeted for addition averaging. Here, the i-th sampled addition target A scan data A _Tik constituting the kth set of target A scan data group AG _Tk is subjected to positional deviation correction and addition averaging, and then the kth set of target A scan data A _The addition target A scan data _{AT (i + 1) k} sampled next to _Tik is not set as the next positional deviation correction target, but in the k + 1 set of target A scan data group AGT _{(k + 1)} . If the addition target A scan data A _{Ti (k + 1)} located at the position corresponding to the currently processed addition target A scan data A _Tik is set as the next positional deviation correction target, the setting of the search region is not changed. Therefore, the data processing amount can be suppressed.

  Referring to FIG. 6, for example, it is assumed that there are two sets of A scan data groups AG and AG ′ to be added and averaged, each of which is composed of 20 A scan data. The A scan data constituting the A scan data group AG is A1 to A20, and the A scan data constituting the A scan data group AG ′ is A1 ′ to A20 ′. Target B scan images T and T ′ can be generated from the A scan data groups AG and AG ′, respectively.

Here, when the A scan data currently subjected to the positional deviation correction and addition averaging processing is A4 of the A scan data group AG, in FIG. 6A, the A scan data to be processed next is the target B scan image. It is assumed that A5 is located next to A4 being processed in T. The A scan data constituting the target B scan image T is processed in order in the scanning direction, A6 following A5, A7 following A6 and A7. On the other hand, in FIG. 6B, after A4 of the A scan data group AG is processed, the next is A scan data at a position corresponding to A4 in the A scan data group AG ′ (target B scan image T ′). A4 'is processed, and then the process returns to the A scan data group AG again. Next to A5 and A5, A5' and the data at the corresponding position are processed together, jumping over the A scan data group. In this way, if the A scan data at the corresponding positions are processed at once, similar A scan data is continuously aligned, so that the data processing amount can be reduced. Further, leading to even suppress significant amount of data processing that needs to reset the search area in the reference B-scan image T _R of each A-scan data to be processed is changed is eliminated.

  As described above, the image processing apparatus according to the present embodiment generates a high-quality tomographic image suitable for interpretation while suppressing the data processing amount from the A scan data group obtained by scanning the same portion a plurality of times. can do. In the present embodiment, a plurality of tomographic images (B scan images) are generated from the A scan data group obtained by a plurality of scans, and the averaging process is not performed based on the tomographic images. A major feature is that an A scan data group obtained by a plurality of scans is stored as it is, and an averaging process is performed based on the A scan data group. As a result, it is possible to reduce the data processing amount rather than once generating a B scan image from the obtained A scan data group and using the B scan image as reference data for each subsequent process.

  The tomographic imaging apparatus according to the present invention has been described above with reference to the drawings. However, the present invention is not limited to the above-described embodiment, and various modifications can be made.

For example, in the above embodiment, setting the search area F based B-scan image T _R, addition target A by aligning is part E of the addition target A scan data A _Tik the search area F A method of calculating the positional deviation amount of the scan data, correcting the positional deviation amount, and averaging is adopted, but the technique of the alignment and the averaging process is not limited to this. For example, as disclosed in Patent Document 5, an evaluation area is set around the addition target A scan data A _Tik, and the alignment of the addition target A scan data is combined with the preceding and following A scan data, The correlation value may be evaluated and aligned with the search area. Alternatively, without setting the search area F based B-scan image T _R, as disclosed in Non-Patent Document 1, direct addition target A-scan for each A-scan data constituting the reference B-scan image T _R Data A _Tik may be aligned one by one. In this case, setting of the evaluation region, it is possible to omit the like step of setting a search area F based B-scan image T _R, using the position deviation amount obtained by aligning the evaluation area to the search area F Therefore, it is not necessary to perform misalignment correction of the addition target A scan data A _Tik , so that the image quality of the finally obtained tomographic image may be slightly lowered, but substantially sufficient image quality is obtained. The amount of data processing can be greatly reduced, and the time for creating a tomographic image for interpretation can be greatly shortened.

E Eye to be examined 10 Demultiplexing / combining optical system 11 Low coherence light source 12 Light quantity adjusting mechanism 13 Optical coupler 14 Collimating lens 15 Beam splitter 16 Spectroscope 17 Computer 18 Signal processing means 19 Data storage section 20 Tomographic image generating means 30 Measurement optical system 31 Focus lens 34 x-axis scanning mirror 35 y-axis scanning mirror 36 Scan lens 37 Dichroic mirror 38 Objective lens 40 Reference optical system 42 Dispersion compensation glass for objective lens 46 Dichroic mirror 47 Condensing lens 48 Variable aperture 49 Reference mirror 50 Eye dispersion to be examined Compensation glass

Claims (7)

An image processing apparatus that generates a tomographic image for interpretation by processing a plurality of sets of A-scan data groups obtained by scanning substantially the same position of an object a plurality of times in a tomographic imaging apparatus that captures a tomographic image of the fundus of the eye to be examined Because
A plurality of B scan images are generated from the plurality of sets of A scan data groups and are simply added to generate a simple addition B scan image, and among the plurality of B scan images, the simple addition B scan image A reference data selection means for identifying a B scan image having the highest correlation and selecting an A scan data group corresponding to the B scan image having the highest correlation as a reference A scan data group;
From a B scan image generated from the plurality of sets of A scan data groups other than the reference A scan data group, a predetermined number of B scan images highly correlated with the B scan image generated from the reference A scan data group A target data selection means that identifies and selects a predetermined set number of A scan data groups corresponding to the specified predetermined number of B scan images as a target A scan data group;
A reference B scan image is generated from the reference A scan data group, and the A scan data included in the target A scan data group is added to the reference B scan image while correcting misalignment for each A scan data. An image processing apparatus comprising: averaging means for generating a tomographic image for interpretation on average.   The image processing apparatus according to claim 1, wherein the reference data selection unit and the target data selection unit generate a B scan image by using a part of A scan data of the A scan data group.   The addition averaging means continuously processes A scan data at corresponding positions in each of the predetermined number of A scan data groups included in the target A scan data group, The image processing apparatus according to claim 1. An image processing method for generating a tomographic image for interpretation by processing a plurality of sets of A-scan data groups obtained by scanning a substantially same position of a target object a plurality of times in a tomographic imaging apparatus that captures a tomographic image of the fundus of the eye to be examined Because
A plurality of B scan images are generated from the plurality of sets of A scan data groups and are simply added to generate a simple addition B scan image, and among the plurality of B scan images, the simple addition B scan image A reference data selection step of identifying a B scan image having the highest correlation and selecting an A scan data group corresponding to the B scan image having the highest correlation as a reference A scan data group;
From a B scan image generated from the plurality of sets of A scan data groups other than the reference A scan data group, a predetermined number of B scan images highly correlated with the B scan image generated from the reference A scan data group A target data selection step of identifying and selecting a predetermined number of sets of A scan data groups corresponding to the specified predetermined number of B scan images as a target A scan data group;
A reference B scan image is generated from the reference A scan data group, and the A scan data included in the target A scan data group is added to the reference B scan image while correcting misalignment for each A scan data. An image processing method comprising: an averaging process for averaging to generate a tomographic image for interpretation.   5. The image processing method according to claim 4, wherein in the reference data selection step and the target data selection step, a B-scan image is generated using a part of A-scan data of the A-scan data group.   In the addition averaging step, the A scan data at a corresponding position in each of the predetermined number of A scan data groups included in the target A scan data group is continuously processed, The image processing method according to claim 4 or 5.   An image processing for causing a computer to function as the image processing apparatus according to any one of claims 1 to 3, or for causing a computer to execute the image processing method according to any one of claims 4 to 6. program.