Optical Coherence Tomography System for Ophthalmology

Abstract

Data representing structures in the eye, in particular cross-sectional images, are recorded by generating multiple A-scans (A1, A2, A3) using swept-source OCT at different locations in the eye, each scan generating multiple reflection values (r) at multiple points along the light trajectory. A combined value (r) is then calculated from the several reflection values (r) at different locations in the eye. C To improve data quality, a model is fitted to the reflectance values at at least one curved structure in the eye (Ca, Cp, La, Lp), which is then used to identify points on the A-scans (A1, A2, A3) that will be merged. This allows the data to be projected tangentially to the curved structure, reducing noise and achieving improved resolution perpendicular to the structure.

Description

Technical Field

[0001] The present invention relates to a method for obtaining data representing structures in the eye and an ophthalmic apparatus for performing such a method. Background Art

[0002] Optical coherence tomography (OCT) systems for ophthalmic applications are known, for example, from EP 3572765. They include an optical coherence tomography interferometer. This interferometer is used to record multiple A-scans of the eye. Each A-scan consists of a series of reflection values for multiple points along the light path through the eye. For example, these reflection values can be used to generate a one-dimensional, two-dimensional, or three-dimensional image of the eye. Summary of the Invention

[0003] The problem underlying the present invention is to provide a method and a device of the type mentioned above, which generate data with more information, for example with better resolution and / or less noise.

[0004] This problem is solved by the method and device of the independent claims.

[0005] The present invention thus relates to a method for obtaining data representing structures in the eye, in particular data representing cross-sectional images, by means of optical coherence tomography.

[0006] The data obtained by the present technique may, for example, represent a one-dimensional or two-dimensional cross-section of the eye and / or various parameters of the eye.

[0007] The method comprises at least the following steps:

[0008] - A collection of recorded A-scans: at least some of these A-scans are recorded at different locations of the eye. Each A-scan is used to generate a plurality of reflectance values, for example as a function of the optical penetration depth in the eye.

[0009] - Calculate multiple combined values. Each such combined value is obtained from the reflection value at a different position in the eye.

[0010] By combining the reflection values obtained at different locations in the eye into said "combined value", additional or more accurate information can be obtained as described below.

[0011] Advantageously, the method further comprises the step of modeling at least one curved structure of the eye using the reflectance values. Such a curved structure may be any structure discernible from the reflectance values, such as the front or back surface of the cornea or lens or retina.

[0012] Advantageously, the modeled curved structure can be used to identify points on a subset of the A-scans that will be used to calculate a given combined value. For example, when combining reflection values into a combined value, this allows the A-scan data to be "projected" along a tangent direction to the curved structure, thereby improving spatial resolution in a direction perpendicular to the curved structure and / or improving the signal-to-noise ratio of the data.

[0013] The subset of A-scans used advantageously comprises several A-scans recorded at different positions of the eye.

[0014] In order to calculate the combined value at a curved structure, the method may advantageously comprise at least the following steps:

[0015] - In the subset of A-scans to be used, determine for each A-scan i an estimate r of the reflectance value at the intersection of the A-scan i with the modeled structure e (i). For example, this estimate may be the reflectance value of the measured point on A-scan i closest to the intersection point, or it may be an interpolation of the reflectance value of A-scan i at the location of the intersection point.

[0016] - Use the estimated r e (i) to calculate the combined value.

[0017] In one embodiment, to calculate the combined value along a given axis, the method may further include the following steps:

[0018] At least first and second curved structures in the eye are modeled using reflectance values, wherein the first structure is located at a first position along the axis and the second structure is located at a second, different position along the axis. For example, the first structure may be the anterior surface of the lens (or cornea) and the second structure may be the posterior surface of the lens (or cornea).

[0019] In that case, advantageously, the method may comprise at least one of the following two steps:

[0020] To obtain a combined value at a first location, a plurality of first reflection values from different A-scans at points tangential to the first curved structure are combined. In the example above, this would mean that tangential reflection values along the anterior surface of the lens are combined to generate a combined value at the location where the axis intersects the anterior surface.

[0021] To obtain a combined value at a second location, a plurality of second reflection values from different A-scans at points tangential to the second curved structure are combined. In the example above, this would mean that the tangential reflection values along the posterior surface of the lens are combined to generate a combined value at the location where the axis intersects the posterior surface.

[0022] Since the first and second structures are different structures, the geometry used to combine the reflection values at the first and second locations will follow different curves within the eye, each curve optimized for the respective structure to be represented.

[0023] To obtain a combined value at a third position between the first and second positions, the method may further include the step of obtaining a combined value at a third position between the first and second positions. This is achieved by combining a plurality of third reflection values from points in different A-scans, which are located in the region between and / or near the first and second structures. In this context, "near" is advantageously understood to mean a given point that is less than 500 μm, particularly less than 100 μm, and even more preferably less than 10 μm from the respective structure.

[0024] When the data represents a two-dimensional cross-section of the eye, it can be used to show an image of a modeled structure, such as a cross-section of the cornea or lens. In that case, a combined value calculated by combining multiple reflection values along a tangent direction of the structure can be used to generate an image point of the structure in the cross-sectional view.

[0025] Advantageously, the combined value is calculated by combining a plurality of reflection values from different A-scans outside the two-dimensional cross-sectional plane.

[0026] As mentioned, the "combined value" is calculated by combining the reflectance values of different A-scans. This operation may include at least one of the following operations:

[0027] - calculate the average, in particular the weighted average, of the reflection values to be combined,

[0028] - calculate the quantiles of the reflection values to be combined, in particular the median,

[0029] - calculate the minimum or maximum of the reflection values to be combined, and / or

[0030] - Calculates the interpolation of the reflection values to be combined.

[0031] When calculating the combined value, the reflection values of A-scans that exhibit specular reflection are advantageously not used. Rather, adjacent A-scans that do not exhibit specular reflection can be used. In this context, an A-scan "exhibiting specular reflection" is understood to be an A-scan having reflection values derived from specular reflection recorded by an OCT measurement.

[0032] The present method allows to suppress such specular A-scans since it can rely on a combination of adjacent A-scans.

[0033] Color can be used to provide additional information about the structure of the eye. For example, the "data" calculated by the method can represent a two-dimensional image in the image plane, and the "combined value" can be the color value of a pixel in the image. The method can then include at least the following steps:

[0034] - For each pixel, identify at least one reflectance value at the point corresponding to the pixel.

[0035] -Determines the color of a pixel, the color depends on the distance of the point from the image plane.

[0036] This allows the reflecting structure to be assigned different colors depending on its position relative to the cross-sectional image.

[0037] The present invention also relates to an ophthalmic device comprising

[0038] -Optical Coherence Tomography Interferometer: This OCT interferometer is used to record the A-scans.

[0039] - A control unit configured and adapted to perform the method described herein: this control unit is provided with suitable software and hardware for performing the steps of the invention. It may also include a display, storage and / or data interface for displaying, storing and / or transmitting data determined by the present technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The present invention will be better understood and other objects besides those set forth above will become apparent when the following detailed description is considered. This description refers to the accompanying drawings, in which:

[0041] Figure 1 shows a schematic arrangement of an embodiment of an ophthalmic device,

[0042] Figure 2 shows an embodiment of a scanning mode,

[0043] Figure 3 Shown Figure 2 The reflection value of the A-scan position is shown in bold,

[0044] Figure 4 represents a series of A-scans in a plane,

[0045] Figure 5 are three A-scans and a schematic diagram of the anterior and posterior surfaces of the cornea and lens.

[0046] Figure 6 Shown Figure 4 an image in which the A-scan is shifted in the z direction so as to place the anterior lens surface on a single line,

[0047] Figure 7 Shown Figure 4an image in which the A-scan is shifted in the z direction so as to place the posterior lens surface on a single line,

[0048] Figure 8 is a schematic diagram of an embodiment for calculating a combined value at position x, y, z for an A-scan i,

[0049] Figure 9 represents the reflectance values of a single A-scan and the combined values along the visual axis of the eye,

[0050] Figure 10 represents the combined values of a one-dimensional cross section along the visual axis of the eye, magnified on the cornea and lens and depicting the effect of projecting a single A-scan along the surfaces of the cornea and lens,

[0051] Figure 11 shows a refraction-corrected cross-sectional image using the techniques described herein,

[0052] Figure 12 The brightness values of the image showing the "frontal" view of the eye, and

[0053] Figure 13 Shown for Figure 13 The various parts of the image are color-coded.

[0054] (Note: All grayscale images in the figures are halftoned to improve reproducibility. Halftoning is not typically used when representing images on electronic displays.) DETAILED DESCRIPTION

[0055] Device Overview

[0056] Figure 1 An ophthalmic device is for example an ophthalmic microscope with OCT capability.

[0057] It includes an optical coherence tomography interferometer 10-26.

[0058] The interferometer has a light source 10 which, in the present embodiment, is a swept source light source, ie it generates a narrow band of light whose wavelength can be adjusted.

[0059] Light from a light source 10 passes through a beam splitter 12 , in particular a fiber optic beam splitter, and is sent to two interferometer arms 14 , 16 .

[0060] The first arm is the reference arm 14, which comprises at one end a mirror 18. The light impinging on the mirror 18 is sent back to the beam splitter 12 and from there at least partially to a light detector 20.

[0061] The second arm is the sample arm 16. It comprises collimating optics 22 for collimating the probe light coming from the beam splitter 12. The light is then fed through two scanning mirrors 24a, 24b and an objective lens 26 for generating a probe beam 28. Depending on the position of the scanning mirrors 24a, 24b, the probe beam 28 may be displaced laterally in the xy plane perpendicular to the optical axis z of the device.

[0062] In this embodiment, an interferometer is used that generates a telecentric probe beam 28, i.e., a probe beam 28 for various x and y coordinates (such as Figure 1 The light beams 28 and 28' in FIG. 2 are parallel to each other. This simplifies the analysis in the context of the technology described below.

[0063] In the illustrated embodiment, the probe beams are shown as being focused on the anterior surface of the cornea, but they may also be focused on any other portion of the eye 30 of particular interest.

[0064] Probe beam 28 enters eye 30 where it is reflected or scattered by structures of the eye. Light reflected from such structures returns to beam splitter 12 and from there at least partially to photodetector 20 where it may interfere with light from reference arm 14.

[0065] Figure 1 The device operates by recording multiple A-scans. For each such A-scan, the probe beam 28 is brought to the desired x- and y-positions by means of scanning mirrors 24a, 24b. The central wavelength of the light source 10 is then tuned within a given wavelength range, which is typically much wider than the spectral width of the light from the light source 10. The light at the light detector 20 is measured as a function of the central wavelength.

[0066] Spectral analysis, in particular a Fourier transform, of the signal from detector 20 can then be used to generate a reflectance value for the eye 30 along axis z for a given A-scan. The reflectance value is meant to relate to the reflected and scattered light described above. Following OCT imaging conventions, the reflectance value can be represented by a value proportional to the reflected intensity, or a value proportional to the logarithm of the reflected intensity, or other range-compressed values, for example. In more general terms, a "reflectance value" indicates the amount of light returned from a certain location along the A-scan. Advantageously, it can be linearly related to the amount of light, or its logarithm, or any other function thereof.

[0067] This type of OCT measurement is known to the person skilled in the art and is described, for example, in EP 3 572 765 and the references cited therein.

[0068] The device also includes a control unit 32, which may be provided with, for example, a microprocessor 34, a memory 36, and a display 38. The memory 36 may hold data and program instructions required to perform the steps of the method. For example, the display 38 may be used to display the data determined thereby and, in particular, to display any cross-sectional images obtained by means of the techniques described herein.

[0069] Advantageously, the measurement range of the OCT interferometer 10-26 (for a single A-scan) extends at least from the cornea to the retina of a typical eye. In other words, with a single A-scan (i.e., for SS-OCT with a single light source scan), depth-resolved information of at least 40 mm (in air) can be obtained. This allows the techniques described below to be applied over the entire axial length of the eye without, for example, having to apply stitching to combine different measurements.

[0070] Figure 2 An example of a scan pattern used in the measurement is shown, i.e., it shows the position of the probe beam 38 in the xy plane during various A-scans. This type of pattern is described in EP 3021071. Other scan patterns, such as those described in EP 3217144 or US 8705048, may also be used.

[0071] Figure 3 The reflection value r (raw data) is shown as the scanning point index and Figure 2 The function of the z-coordinate of the fat scan point is shown in . In this figure, higher reflectance values are represented by darker spots. As can be seen, this data is quite noisy.

[0072] Eye movement correction

[0073] A-scans recorded in this manner can optionally be corrected for eye movement, for example, by using at least the following steps:

[0074] 1. Identify the reflection of at least one given eye structure (e.g., the anterior corneal surface) in the A-scan.

[0075] 2. Fit a model describing the shape and motion of the structure to the positions of the identified reflections. For example, such a model may have geometric parameters of the structure (such as curvature) and motion parameters (such as three-dimensional position and velocity in x, y, and z coordinates).

[0076] The parameters obtained in fitting step 2 can then be used to transform the OCT measurements into a coordinate system that is fixed to the frame of the eye.

[0077] Suitable motion correction techniques are described, for example, in WO 2013 / 107649 or US 7452077.

[0078] Combined A-scan

[0079] The above steps generate a set of A-scans recorded at different xy positions in the eye. Each A-scan comprises a plurality of reflection values r(zij) along the light path in the eye, where i is the index of the A-scan (identifying the x and y position of the A-scan) and zi1...ziN specifies the z position along the depth of the eye as recorded in A-scan i.

[0080] As mentioned, the reflection values r(zij) are advantageously, but not necessarily, motion corrected.

[0081] The present technology is based on obtaining data describing the structure of the eye. The data may, for example, represent a one-dimensional or two-dimensional cross-section of the eye, or it may, for example, represent a three-dimensional volume image of the eye.

[0082] To calculate data representing the eye at a given position, the technique calculates a “combined value” r C , wherein at least some of these combined values are each obtained from several reflection values r(zij) recorded for different positions of the eye.

[0083] Some methods for calculating these combined values are described below.

[0084] Projection along the eye structure

[0085] One of the techniques described here is based on modeling at least one curved structure of the eye and then combining the reflectance values of different A-scans tangentially along this curved structure.

[0086] This is Figure 4 This is illustrated in Figure 1, which shows a series of A-scans in a plane. The image clearly shows the anterior and posterior surfaces Ca, Cp of the cornea and the anterior and posterior surfaces La, Lp of the lens. It also shows the retina R. (Because—for telecentric A-scans extending parallel to the visual or optical axis of the eye (outside the eye)—the A-scans intersect at a single point on the retina, the retina appears at essentially the same z-coordinate for all A-scans.)

[0087] For example, using edge detection techniques on each A-scan, the z position (axial position along the light trace) of any of these surfaces can be estimated to obtain, for example, four sets of points in xyz space corresponding to each of these surfaces.

[0088] In a next step, the parameters of the model for each of these surfaces may be calculated, for example using conventional fitting techniques.

[0089] For example, surfaces can be modeled by spherical surfaces defined by a center and a radius, or they can be modeled by polynomial surfaces, for example of second order.

[0090] Once the parameters of the model have been calculated, a refined "combined value" can be calculated on one of the surfaces for a given xy coordinate, i.e., by projecting the reflection values of several A-scans along the surface and combining them mathematically, for example, by averaging or by another mathematical operation described further below.

[0091] This process is Figure 5 This is schematically illustrated in FIG, which shows three A-scans A1, A2, A3 and the four surfaces Ca, Cp, La and Lp obtained by fitting. The points / circles along the A-scans represent the z positions zij at which the reflection values r(zij) have been measured.

[0092] For example, to obtain a fine combined reflection value r at the intersection of the surface Ca and the A-scan A2 C , the reflection values of the three A-scans A1 , A2, A3 along the surface Ca can be combined.

[0093] For example, a simple algorithm may identify for each A-scan the measurement positions (ie, z positions) LCa1, LCa2, LCa3 closest to the surface Ca and combine their reflection values r(LCa1), r(LCa2), r(LCa3).

[0094] A more sophisticated algorithm can not only identify the positions LCa1, LCa2, LCa3 closest to the surface Ca, but also identify the second closest positions and interpolate the reflection values of these closest positions for each A-scan to obtain a better estimate of the reflection value at the intersection of the A-scan with the surface Ca, and then mathematically combine these estimates.

[0095] More generally, to estimate the reflectance value r at a given position x, y on a structure of the eye (such as one of the surfaces Ca, Cp, La, Lp) C , you can perform the following steps:

[0096] - Step 1 (Modeling). Model at least one curved structure of the eye based on the reflectance values measured by the A-scans. For this modeling step, only a subset of the A-scans of step 2 below may be used, or a larger set of A-scans may be used.

[0097] - Step 2 (subset selection): Select a subset of A-scans around position (axis) x, y for calculating the combined value r CThis subset contains A-scans that are close to x,y, i.e., A-scans that are closer than a value R to x,y. If R is chosen large, noise suppression improves, but x- and y-resolution suffers. If R is chosen small, noise increases, but resolution improves. Typically, R can be, for example, 1 mm or 0.5 mm.

[0098] - Step 3 (Determine Individual A-Scan Reflectance Estimates): For each A-scan i in the subset, calculate an estimate of the reflectance value at the intersection of the A-scan with the modeled structure, re(i). For example, this step may comprise calculating the intersection of the A-scan i with the modeled structure and then using, for example, the measured reflectance value closest to the intersection as the estimate re(i). Alternatively, it may comprise calculating an interpolated value of the reflectance value of the A-scan at the intersection (which will also indirectly use at least the reflectance value at the scan position closest to the intersection and at least one additional reflectance value).

[0099] 4. Combine the estimated re(i) of A-scan i in the subset into a combined value r C This combination can use various mathematical operations, which will be described in more detail below.

[0100] The above steps describe how to calculate the combined value r at a given eye structure C These steps can be generalized to estimate the reflectance r at any given location x, y, z C , also away from any structures of the eye, e.g. Figure 5 At position x, y, z1.

[0101] The first method is Figure 6 and 7 Shown in.

[0102] Figure 6 Shown Figure 4 A-scans, but each A-scan has been shifted along the direction z so that their reflection peaks corresponding to the front surface La of the lens are located at the same z position. Similarly, Figure 7 Shown Figure 4 A-scans, but each A-scan has been shifted along the direction z so that their reflection peaks corresponding to the front surface Lp of the lens are located at the same z position.

[0103] In that case, to calculate the combined value r at a given z position between the anterior and posterior surfaces La, Lp of the lens C (z), the estimated value re(i) for a given A-scan i can be, for example, Figure 6 The reflection value r'(i,z) of A-scan i and Figure 7The weighted average of the reflection values r”(i,z) of the A-scan i. The weights are linearly selected according to the distances d1 and d2 from point z to La and Lp, respectively. The distances, i.e.,

[0104] r e (z)=(d1·r”(i,z)+d2·r'(i,z)) / (d1+d2). (1)

[0105] In other words, two sets of A-scans are calculated by offsetting the A-scans along z, where the first set is aligned at the first structure and the second set is aligned at the second structure. Then, to calculate the estimate re(i) for a given A-scan i at a given position z between these structures, the reflection values r' and r" of A-scan i in the two sets are combined using equation (1).

[0106] The second method is Figure 8 Here it is assumed that the combined value for a given position x, y, z is calculated from a subset of A-scans close to x, y, z. The figure illustrates which position z(i) along A-scan i (denoted as Ai) will be used. That is, re(i) for A-scan i is determined as follows:

[0107] - Calculate the z positions z1 and z2 of the two eye structures (here structures La and Lp) at the xy position, and the z positions z1(i) and z2(i) of the two eye structures at the xy position of the A-scan i.

[0108] - Position z(i) on A-scan i is calculated from z1 and z2, z1(i) and z2(i), taking into account that z(i) should be between z1(i) and z2(i) in a similar way to z being between z1 and z2. For example, z(i) can be calculated using the following relationship:

[0109] z(i)=z1(i)+(z2(i)-z1(i))·(z-z1) / (z2-z1). (2)

[0110] Then, the estimate re(i) at position z(i) can be calculated. For example, it can be calculated using the reflection value of A-scan i at the measured point closest to z(i), or it can be calculated by interpolation of the reflection values of A-scan i at distance z(i).

[0111] More generally, in order to calculate a combined value at a position z between positions z1 and z2, the method can combine "a plurality of third" reflection values (r) of A-scan points located in a region between and / or near two modeled structures of the eye. As mentioned above, "near" is advantageously understood as a specified point having a distance of less than 500 μm, in particular less than 100 μm, and in particular less than 10 μm from the corresponding structure. For example, the "near" situation can apply when using the technique defined in b) below, in which case some of the "plurality of third" values can be located outside the two modeled structures (typical radii of curvature of such structures can be 8 mm or more, and the distances of the points corresponding to the plurality of third reflection values can be up to almost 3 mm, possibly less, resulting in a maximum definition of "near" of 500 μm.)

[0112] "Plural thirds" can be defined, for example, in one of the following ways:

[0113] a) When using the technique of equation (1), this "third plurality" of reflection values may for example be defined as comprising the reflection values of the first and second subsets of points on the A-scan used:

[0114] - The first subset of points is along a line parallel to the first modeled structure ( Figure 8 A first (advantageously three-dimensional) curve C1 of La in is arranged tangentially and this first curve is offset to intersect the point at the coordinates x, y, z.

[0115] - The second subset of points is along a line parallel to the second modeled structure ( Figure 8 A second (advantageously three-dimensional) curve C2 of Lp in φ is arranged tangentially and this second curve is also offset to intersect the point at coordinates x, y, z.

[0116] In other words, the first curve C1 and the second curve C2 intersect at point z.

[0117] b) When using the technique of equation (2), the "third" reflection values can be defined as, for example, the values along the structure modeled from the first and second on the A-scan used ( Figure 8 The reflection value of the point extending tangentially from the interpolated curve C3 (La, Lp in ) (advantageously a three-dimensional curve).

[0118] The techniques described in this section "Projections along the structure of the eye" are useful for computing the combined value r at xy positions away from the eye vertex or visual or optical axis. C And / or it is particularly advantageous for computing combinations of geometrically wide A-scan subsets (ie, for large values of R), since in these cases the curvature and / or slope of the structure can introduce large errors in computing the combined value.

[0119] Calculating combined values

[0120] Once the estimates re(i) for all A-scans i in the subset of A-scans have been determined, the combined value r at position x, y, z can be evaluated. C As mentioned above, for example, the combined value r C Can be one of the following values:

[0121] - An average of the estimates re(i) of the individual A-scans. Advantageously, a weighted average is used, wherein the weight of a given estimate re(i) depends on the distance of the A-scan i from the position x, y, for example using Gaussian weights.

[0122] - Estimate a quantile of re(i), such as the median.

[0123] - Estimate the minimum or maximum value of re(i).

[0124] - Estimate the interpolation value of re(i), for example, by fitting a first-order or second-order polynomial r to the estimated re(i) C (x,y) uses the x and y coordinates of A-scan i as the independent variables.

[0125] Advantageously, the device comprises a user interface allowing the user to select how to combine the estimates re(i).When the user wants to display quantiles, for example, the user may be able to select values between a minimum, a median and a maximum, either continuously or in a graded manner.

[0126] Suppressing specular reflections

[0127] Typically, A-scans close to the vertex of the eye show strong specular reflections, which tend to overwrite the information obtained from other A-scans in the considered subset.

[0128] Therefore, in one embodiment, the reflectance values of any A-scans that exhibit specular reflections can be suppressed when calculating the combined values. For example, this can be achieved by not using any A-scans whose distance from the corneal vertex is less than a threshold Rs (e.g., Rs is 0.1 mm).

[0129] Cross-section calculation

[0130] For example, the above techniques can be used to calculate a one-dimensional or two-dimensional cross-section of the eye.

[0131] For example, a particularly interesting cross-section is a one-dimensional cross-section along the visual or optical axis of the eye, i.e., a central A-scan. For example, it can be obtained by using a subset of all A-scans having a lateral distance from the apex of, for example, less than 1 mm, optionally excluding any A-scan less than 0.1 mm from the apex in order to suppress specular reflection signals as described above.

[0132] As an example, Figure 9 Such a cross section is shown in FIG, with a combined value r C The points in this figure show individual A-scans, while the continuous line is through Figure 6 and 7 The combination r obtained by the method shown in C (z) value, used for Figure 3 A-scans with a mid-range distance between 0.1 and 1.0 mm. As can be seen, the combined value r C The noise is much lower than that of individual A-scans.

[0133] Figure 10 is another one-dimensional cross section along the visual axis of the eye, showing only the front of the eye. This graph contains two curves:

[0134] - The dotted curve shows the combined values r obtained from the A-scans in the subset C (z), without projecting the reflectivity values of the A-scan along the curved surfaces of the cornea and lens. More precisely, the combined value r C (z) is obtained by combining (eg, by using the median) only the corresponding values of the A-scans at the same z coordinate (ie, the A-scans in the subset are projected orthogonally onto the visual axis of the eye).

[0135] - The solid curve shows the combined value r obtained from the A-scans in the subset using a combination of the projection technique along the surface and using equation (1) above C (z).

[0136] For example, from the two surfaces of the cornea ( Figure 10 points Ca and Cp in the retina) and the posterior surface of the lens ( Figure 10 As can be seen from point Lp in the figure, the solid curve reveals additional details and structure.

[0137] In addition, the dotted lines are offset to the right for surfaces Ca, Cp, and La, and to the left for surface Lp. This reflects the fact that "orthogonal" projection generates systematic offset errors when used at the vertices of the surfaces. The corrected (solid line) data does not show these errors.

[0138] Figure 9 and 10 The cross-sectional data can be used, for example, to measure the depth of the cornea, the depth of the anterior chamber of the eye, the thickness of the lens, the total length of the eye, and / or the dimensions of other structures of the eye, such as the thickness of the anterior lens capsule complex or the corneal epithelium. A good signal-to-noise ratio allows accurate results to be obtained. For example, peak fitting can be used to locate the maxima of individual scattering structures.

[0139] Additional one-dimensional scans of this type can be obtained for points outside the visual or optical axis of the eye, i.e., for arbitrary xy positions, and for example they can be used to generate color-coded depth maps of corneal thickness, corneal epithelial thickness and / or anterior chamber depth.

[0140] Several one-dimensional scans of this type can also be combined to generate two-dimensional cross-sectional data for a cross section parallel to the visual or optical axis of the eye, for example, to generate a meridional cross-sectional image.

[0141] The two-dimensional cross-sectional data can be used to generate a cross-sectional image, for example, by combining the values r C Encoded as grayscale values.

[0142] Advantageously, when generating such cross-sectional data, the combined value r is calculated from the maximum of the reflection values of the corresponding positions of the A-scans in the used subgroup, even when other techniques described in the section "Calculating combined values" may also be used. C .

[0143] Figure 11 Such a (Jarvis halftone) cross-sectional image using averaging is shown.The image has been refraction corrected to take into account refraction of the A-scan as it passes through the eye.

[0144] When generating two-dimensional cross-sectional data, the selection of subsets of A-scans to be combined can be asymmetric with respect to the cross-section. For example, when generating a horizontal cross-section (i.e., y = 0), a subset of A-scans within the region x = x0 ± Δx and y = ± Δy can be used to calculate the combined value r at a certain coordinate x0. C , where Δy>Δx, for example, Δy=2 mm, Δx=0.5 mm.

[0145] More generally, when calculating cross-sectional data for a given cross-sectional plane, the combined value r at a given coordinate x, y, z is calculated. C The subset of A-scans includes only A-scans within a volume of space that is larger in a direction perpendicular to the plane than in any direction parallel to the plane.

[0146] Deep Coding

[0147] Yet another aspect of the invention relates to generating a color-coded image, wherein color is used to encode the distance of structures from the image plane.

[0148] In the following, this method is described for "frontal" images of the eye, ie images representing the eye as seen from the front (ie from the outside along direction z). However, it can also be used to generate any type of cross-sectional images.

[0149] Without limiting the generality of the concept, the "image plane" of an image may be any plane perpendicular to the viewing direction.

[0150] In this respect, similar to the above, the method uses an A-scan, or a subset of data points thereon, for each point in the image and combines their reflectance values to generate a combined value, in this case a color value.

[0151] exist Figure 12 and 13 An example of the resulting frontal image is shown in . In practice, that is, as displayed on the electronic display 38 of the device, this image will be a color image. However, for the purpose of representing it herein, Figure 12 shows its gray level (i.e., brightness or lightness), while Figure 13 Possible colors used in some of its areas are shown (in text form).

[0152] This image in the xy plane (ie perpendicular to the axis of the A-scan) can be calculated, for example, as follows:

[0153] 1. Set up three two-dimensional arrays representing the pixels of the image and initialize them to zero. The first array, D(m,n), stores the distance values. The second array, R(m,n), stores the reflectance values. The third array, W(m,n), stores the weight values.

[0154] 2. For each A-scan of interest, i, do the following:

[0155] 2a. Determine the z position z(i) and amplitude r(i) of the maximum reflection value on A-scan i.

[0156] 2b. Iterate over at least a subset of the image pixels. For each image pixel m,n (corresponding to position x(m), y(m) in the xy plane), compute the weight w, e.g.,

[0157]

[0158] where x(i), y(i) are the x and y coordinates of the center of A-scan i, and g(r,σ) is a Gaussian function with a given variance σ. The variance σ depends on the diameter of A-scan i.

[0159] The weights w(m,n) are added to the third array W(m,n). w(m,n) z(i) are added to the first array D(m,n). w(m,n) r(i) are added to the second array R(m,n).

[0160] 3. Once all A-scans i (or a desired subset thereof) have been processed in this manner, perform the following operations on each pixel m,n of the image:

[0161] 3a) For example, the average reflectance is determined using the following formula:

[0162]

[0163] where ε is a small number used to avoid division by zero.

[0164] 3b) For example, the average depth is determined using the following formula:

[0165]

[0166] 3c) Average depth For example, it can be used to determine the color or hue. For example, this can be achieved by means of a lookup table that delivers the color as a function of the average depth. In the RGB color space, this color can be represented by its RGB components, for example.

[0167] 3d) Average reflectivity For example, it is used to set the brightness of a color or hue. In the RGB color space, this can be achieved, for example, by multiplying the RGB components by the average reflectance and then normalizing the RGB values.

[0168] The color is assigned to pixel n,m as in steps 3c and 3d.

[0169] For example, in Figure 12 and 13 In the example above, the structures at the top of the image are reflections from the eyelids. Since they are closest to the viewer, they will have the first color 1 (or range of colors). Reflections originating from the iris are farther from the viewer and will appear as the second color 2. For example, reflections visible in the pupil could originate from the cornea, the front surface of the lens, the back surface of the lens, or structures deeper within the eye. Depending on where they originate, they will have different colors 3, 4, and so on.

[0170] therefore, Figure 12 and 13 The images shown in can easily distinguish different scattering structures in the eye.

[0171] The same technique can be used not only for frontal views of the eye, but also for virtual views from other directions. In that case, step 2 above is not performed for an A-scan, but for a cylindrical or prismatic region extending, for example, perpendicular to the image plane.

[0172] Note: In addition to using the average reflectance, in step 3 above, the method can also use any other combination, such as using the maximum reflectance and its depth.

[0173] Advantageously, the determination of the colour (in step 3c above) involves dividing the average depth by The positions of the surfaces modeled in the eye (i.e., surfaces Ca, Cp, La, and Lp, and the retina R) are compared and color is attributed to the function based on this determination. This allows pixels to be colored based on the structure from which the reflection originates.

[0174] More generally, the "data" computed by the method may represent a two-dimensional image defining an image plane, and the "combined value" may be a color value attributed to a pixel in the image. The method then comprises at least the following steps:

[0175] - For each pixel, at least one reflection value is identified at the point corresponding to the pixel (e.g., because the point is located in a volume of space extending perpendicular to the image and having a cross-section corresponding to the pixel, optionally taking into account refraction at the interface of the eye).

[0176] -Determines the color of a pixel, the color depends on the distance of the point from the image plane.

[0177] Advantageously, for each pixel, the point with the strongest reflection value corresponding to this pixel is used in determining the color.

[0178] Additionally, the magnitude of the reflectance value of the point(s) corresponding to a pixel may be used to determine the brightness or lightness of the pixel.

[0179] The step of determining the color of a pixel advantageously comprises the following sub-steps:

[0180] - Comparing said distances with the positions of several structures Ca, Cp, La, Lp that have been modeled as reflectance values. This allows performing "attribution" on a point in one of the structures or the space between, before, or after the structures. For example, a point can be attributed to the space in front of the cornea, the cornea itself, the space between the cornea and the lens, the front surface of the lens, the inside of the lens, the back surface of the lens, the space between the lens and the retina, or the retina. Other attributions are also possible, for example, by parsing the cornea more finely (attributing the point to the front surface of the cornea, the inside of the cornea, or the back surface of the cornea) or by parsing the lens more coarsely (attributing the point to the entire lens).

[0181] -Select a color based on this attribute.

[0182] This allows, for example, to attribute the same color to all reflections from a given structure even if this structure is curved with respect to the image plane.

[0183] notes

[0184] The techniques described herein are particularly suitable for determining data representing a cross-section parallel to the visual or optical axis of the eye. This cross-section can be a one-dimensional cross-section, for example, a linear cross-section along the visual or optical axis of the eye, or it can be a two-dimensional cross-section, for example, represented as a cross-sectional image.

[0185] In another embodiment, as shown, the data may represent a two-dimensional cross-section or image perpendicular to the visual or optical axis of the eye.

[0186] However, the data may also represent other parts of the eye, such as a single point of the eye, the entire volume of the eye or certain parameters (such as the length of the eye, etc., see the examples above).

[0187] These techniques can be used with any type of OCT, but in particular with time-domain OCT and frequency-domain OCT. However, frequency-domain OCT, and in particular swept-source OCT, has an advantage due to its ability to rapidly acquire A-scans.

[0188] While there are shown and described presently preferred embodiments of the invention, it is to be distinctly understood that the invention is not limited thereto but may be otherwise variously embodied and practiced within the scope of the appended claims.

Claims

1. A method for obtaining data representing structures in the eye by means of optical coherence tomography interferometry, the method comprising the following steps: − recording a set of A-scans (A1, A2, A3), wherein at least some of said A-scans (A1, A2, A3) are recorded at different locations of the eye, and wherein each A-scan (A1, A2, A3) is used to generate a plurality of reflectance values r for a plurality of points along a light path through the eye, − modeling at least one curved structure of the eye using said reflectance value r, − Calculate multiple combination values r C , where each combination value r C is obtained from several reflectance values r at different locations in the eye, where the modeled curved structure is used to identify the subset of A-scans (A1, A2, A3) that will be used to calculate the combined value r C point, It is characterized in that in order to calculate the combined value r of the modeled curved structure C , the method further comprises the following steps: − For each A-scan i in the subset, determine an estimate r of the reflection value r at the intersection of this A-scan i with the modeled curved structure e (i) − Use the estimated r e (i) To calculate the combined value r C ,as well as The method comprises the following steps: − To calculate the combined value r along a given axis C , modeling at least a first curved structure and a second curved structure in the eye using the reflectance value r, wherein the first curved structure is located at a first position along the axis and the second curved structure is located at a second position along the axis, − To obtain the combined value r at the third position between the first position and the second position C , combining a plurality of third reflection values r of points from different A-scans (A1, A2, A3), wherein the points are in the region between the first curved structure and the second curved structure, Among them, the points corresponding to the third reflection values − arranged along a curve (C3) interpolated from the first curved structure and the second curved structure, or − comprising a first subset of points arranged along a first curve (C1) parallel to the first curved structure and a second subset of points arranged along a second curve (C2) parallel to the second curved structure, wherein the first curve (C1) and the second curve (C2) intersect at the third position.

2. The method of claim 1 , wherein each of the first curved structure and the second curved structure is one of an anterior surface of the cornea (Ca), a posterior surface of the cornea (Cp), an anterior surface of the lens (La), a posterior surface of the lens (Lp), and a retina (R).

3. The method according to claim 1 or 2, comprising the steps of: An estimate r of the A-scan i in the subset for a given depth coordinate is calculated by interpolating at least two reflection values r on the A-scan i that are closest to the given depth coordinate. e (i).

4. The method according to claim 1 or 2, further comprising at least one of the following steps: − To obtain the combined value r at the first position C , combining a plurality of first reflection values r from different A-scans (A1, A2, A3) at points along said first curved structure, and / or − To obtain the combined value r at the second position C , combining a plurality of second reflection values r from different A-scans ( A1 , A2 , A3 ) at points along the second curved structure.

5. A method as claimed in claim 1 or 2, wherein the data represents a two-dimensional cross section showing a cross-sectional view of the structure, the method comprising using a combined value r C Steps to generate image points of curved structures in cross-section.

6. The method according to claim 5, wherein: For the combination value r C At least some of the combined values r C is calculated by combining multiple reflection values r from different A-scans (A1, A2, A3) out of the plane of the cross section.

7. The method according to claim 1 or 2, wherein a plurality of combined values r are calculated C The steps include at least one of the following: − Calculate the mean value of the reflection values r to be combined, − Calculate the quantile of the reflection values r to be combined, − Calculate the minimum or maximum value of the reflection values r to be combined, − Calculates the interpolated value of the reflection value r to be combined.

8. The method according to claim 1 or 2, wherein a plurality of combination values r are calculated C The steps include calculating the combined value r C The reflectance values of any A-scans that exhibit specular reflection are not used when φ(R) is set to φ(R). This is achieved by not using A-scans whose distance from the corneal vertex is less than a threshold value Rs.

9. The method according to claim 1 or 2, wherein the method for calculating the combined value r at the position (x, y) is C The position of the reflection value r is from an A-scan that is within a distance of less than 1 mm from the position (x, y).

10. A method as claimed in claim 1 or 2, comprising the step of using the magnitude of the reflectance value r of the point to determine the brightness or lightness of a pixel.

11. The method of claim 1 or 2, wherein the data represents data of an A-scan of the eye.

12. The method of claim 1 or 2, wherein the data represents an A-scan or two-dimensional cross-sectional data parallel to the visual or optical axis of the eye.

13. The method of claim 1 or 2, wherein the optical coherence tomography is spectral domain OCT.

14. The method of claim 1 or 2, wherein the interferometer has a measurement range of at least 40 mm for a single A-scan.

15. The method of claim 1 or 2, wherein the interferometer generates a telecentric probing beam.

16. The method of claim 1, wherein: The data representative of structures in the eye includes data representing cross-sectional images.

17. The method of claim 7, wherein: The average of the reflection values r to be combined is a weighted average of the reflection values r to be combined, or The quantile of the reflection values r to be combined is the median of the reflection values r to be combined.

18. The method of claim 9, wherein: Used to calculate the combined value r at position (x,y) C The position of the reflection value r is from an A-scan that is within a distance of less than 0.5 mm from the position (x, y).

19. The method of claim 13, wherein: The optical coherence tomography is swept-source OCT.

20. An ophthalmic device comprising − Optical Coherence Tomography Interferometer, and − A control unit (32) configured and adapted to carry out the method according to claim 1 or 2.

Images