Rapid oct system

Abstract

The invention relates to an OCT system for the retina (4), comprising: an illumination unit (36) which provides SS-OCT source radiation; an illumination beam path running to the object (4); and a measurement beam path coming from the object (4) to a detector (38) having pixels (74), wherein the measurement beam path is coupled-out via a splitter element (42); a filter element (48) which follows the splitter element (42) and makes the numerical aperture of the image larger in a first direction (x) than in a second direction (y) transverse thereto; and a coupling-in device (52, 66) for coupling-in the reference radiation (R) in the pupil and outside the optical axis (6) such that the measurement radiation (M) is superimposed on the reference radiation (R) in an obliquely incident manner on the detector (38); wherein the control device (58) reads out pixels (74) of the detector (38) which correspond to a linear region in the image field (22) which has a line length in the first direction (x) and a line height in the second direction (y), and the control device (58) realises an image field (22), which is higher than the line height, by displacing the linear region over the object (2) along the line height; wherein a depth resolution of the generated 3D image is determined by the spectral range and spectral bandwidth of the tuning, and a lateral resolution of the 3D image is determined by the numerical aperture in the line direction and by the line height.

Description

[0001] Fast OCT system

[0002] The invention relates to a system for optical coherence tomography (OCT), in particular for imaging the retina of an eye.

[0003] Ophthalmic OCT systems have been used for many years to measure the retina and also structures in the anterior chamber of the eye, particularly in the human eye. Most commercially used systems are confocal scanning spectral domain (SD) OCT or swept-source (SS) OCT. Time-domain (TD) systems are used for very few applications due to their lower sensitivity. The axial imaging properties of an OCT system are determined by the wave-optical depth of field and the coherent depth resolution.

[0004] Fundus cameras were established for imaging the retina even before the advent of OCT technology. Modern fundus cameras, such as the system known from WO 2014 / 140256 A2, can capture comparatively large image fields on the retina, corresponding to a field angle in the range of 90° to 130°.

[0005] The field angle of the eye is measured from the pupil. Due to refraction at the front of the cornea and because the eye is filled with water, for a given image field, the angle of incidence of a beam of rays incident on the retina from the pupil is slightly different than the angle of incidence at which the same beam of rays enters the eye from the front, i.e., from air. Because the angle in the eye depends on eye parameters, for precision measurements, the field angle is measured in air without the eye. The above numerical values ​​refer to this measurement of the field angle. Measurement in the eye would produce slightly different values. However, this difference is irrelevant for the present invention. For the sake of completeness, the measuring principle will be briefly explained: For the measurement, the light source is switched on and a point is sought where the light distribution is smallest. This is the pupil. The measurement is then taken at a distance, usually1 m, calculates the field size and converts it into the field angle. This can be verified by any testing center without making any assumptions about the eye.

[0006] For OCT systems for retinal imaging, there is a need to increase the field angle so that image fields comparable to those of fundus cameras can be captured. This requirement impacts the acquisition speed. SD-OCT systems currently allow field angles of approximately 45°. To maintain the lateral resolution achieved by these systems while simultaneously increasing the field angle, significantly more pixels must be scanned. Therefore, acquisition speed becomes a very important optimization parameter for SD-OCT systems with very large image fields. While there are line-scanning SD-OCT systems that combine single-mode line illumination with a 2D spatially and spectrally resolved spectrometer and a 2D camera detector, they are optimized for cost-effective scanning of smaller image fields. The spectral resolution in the spectrometer limits the depth range accessible to such systems and thus indirectly also the field angle.

[0007] The fastest OCT systems currently available are holoscopic systems. Because they operate without confocal suppression, they are sensitive to the detection of multiply scattered signal components, which significantly reduce the in-image contrast of OCT images. When such systems are designed for field angles of 90°, the illumination wave, for physical reasons, has a focus in the anterior chamber of the eye. Due to this focus and the resulting thermal hazards for the eye being examined, only very low laser intensities can be safely used, which limits the system's acquisition speed. DE 10 2018 130 396 A1 describes a method that illuminates with a scattered but fully coherent wave, allowing significantly faster acquisition.Furthermore, this type of illumination makes it possible to suppress multiply scattered light, thus achieving in-image contrast comparable to confocal systems. However, these systems require very expensive high-speed cameras and, even with them, can only capture very small image fields. Such high-speed cameras have resolutions of approximately 1 megapixel, while resolutions of over 10 megapixels are necessary for the preferred viewing angles.

[0008] For SD-OCT systems, the technically achievable maximum spectral resolution in the spectrometer is a key parameter that limits the measurable depth range. Established SD-OCT systems, such as the Cirrus HD from Carl Zeiss Meditec AG, Jena, Germany, operate with approximately 2000 pixels of resolution in the spectrometer for a spectral range of approximately 80 nm around a central wavelength of 840 nm. Since OCT spectrometers must achieve diffraction-limited imaging in all operating conditions, higher resolutions / pixel counts in the spectrometer cannot easily be ensured technically. In SS-OCT systems, however, the achievable depth range is limited by the temporal sampling during the spectral tuning of the source. The depth range that the OCT system must cover increases with the field angle.At a field angle of approximately 90°, sufficient depth resolution must be achieved over a depth range of approximately 6 mm to 8 mm to enable robust imaging even in highly myopic eyes.

[0009] The spectral resolution of the spectrometer is relevant for the detectable depth range in SD systems, and the bandwidth with which the source is tuned and the signal is sampled (so-called tuning bandwidth) in SS systems. The first key parameter determined by the spectral sampling range (spectral range of the SD analysis or the tuned spectral range of the SS source) is the coherent depth resolution with which the depth range can be recorded.

[0010] The wave-optical depth of field is a second essential parameter that must be considered for dimensioning the accessible depth range. The depth of field is proportional to lambda / nA. 2(Lambda = wavelength and nA = numerical aperture). In addition to the depth of field, the numerical aperture also determines the maximum achievable lateral resolution and detection sensitivity, and thus the maximum achievable speed of the system. Doubling the nA improves the lateral resolution by a factor of 2 and increases sensitivity and maximum achievable speed by a factor of 4, but also reduces the accessible depth range by a factor of 4. Commercial systems therefore all operate with a very similar nA of approximately 0.04, which corresponds to a pupil diameter of 1.3 mm with an average eye length of 17 mm in air. If such an OCT system is to be increased in field angle from 45° to 90°, the nA would have to be halved to adjust the accessible depth range accordingly. This would result in an acquisition time four times longer for a full retinal field.

[0011] Against this background of conflicting design criteria, the invention is based on the task of implementing a fast, non-resolution-reduced OCT system with an image field corresponding to a field angle a of at least 90°. In particular, the field angle size should not be achieved at the expense of acquisition time or lateral resolution, and should also be combined with sufficient depth range and depth resolution.

[0012] The invention is defined in the independent claims. The dependent claims relate to preferred developments.

[0013] A system for optical swept-source (SS) coherence tomography of an object, in particular the retina of an eye, images this object in an image field with a numerical aperture onto a detector having pixels. It is designed to implement the swept-source OCT principle. For this purpose, it preferably has an illumination unit that provides source radiation whose wavelength is tunable over a predetermined spectral range and with a predetermined spectral bandwidth and is divided into illumination radiation and reference radiation.

[0014] The system comprises a filter element which influences the numerical aperture in a rotationally asymmetric manner such that it is larger in a first direction (x) than in a second direction (y) lying transversely thereto.

[0015] The system has a control device that reads the detector's pixels, which correspond to a line-shaped region in the image field. The cell-shaped region has a line length along the first direction, which defines the image field width, and a line height along the second direction that is smaller than the line length. The control device shifts the line-shaped region across the object along the line height—either by reading the detector or by controlling a suitably arranged 1D scanner. This creates an image field height that is greater than the line height. The control device reads the detector and generates a 3D image of the object from pixel signals and—according to the SS-OCT principle—an indication of the wavelength tuning. The depth range and depth resolution of the 3D image are determined by the predetermined spectral range and the predetermined spectral resolution of the tuning.The lateral resolution of the 3D image is determined by the numerical aperture in the line direction and by the line height, which acts as at least a partially confocal filter.

[0016] The system further comprises an illumination beam path for illuminating the object and a measurement beam path for imaging the object along an optical axis onto the detector, which has pixels. The illumination beam path and the measurement beam path preferably run together over a front section that exists between a splitter element and the object. The splitter element then decouples the measurement beam path and lies in or in front of, preferably close to, a plane of the measurement beam path that is conjugate to the eye pupil. The filter element is arranged downstream of this splitter element in the imaging direction. It modifies the measurement beam path, which was previously rotationally symmetric with respect to the numerical aperture, such that the object is imaged onto the detector with a non-rotationally symmetric numerical aperture. The numerical aperture is larger in a first direction due to the filter element than in a second direction lying perpendicular to it.To implement the SS-OCT principle, a coupling device is further preferably provided which radiates the reference radiation into the pupil and outside the optical axis, so that the reference radiation is superimposed on the detector with the measuring radiation incident at an angle.

[0017] The invention thus provides an OCT system that is not fully confocal in at least one spatial direction. Particularly preferably, in a first variant, the system is implemented as a line-scan OCT that is confocal only transverse to the line direction, i.e., only in one dimension. A second variant is a holoscopic system with a highly asymmetric image field that, similar to a line-scanner, has the full field width in one direction but only a few to a few tens of pixels in the second direction, and could be referred to as a "wide-strip scanner."

[0018] To achieve the greatest possible depth ranges, the OCT system is designed as a swept-source system (SS-OCT). A characteristic of SS-OCT systems is that they have a slightly different noise behavior than SD-OCT systems and are therefore generally not implemented as single-ended systems, but rather with so-called "balanced detection" or off-axis detection. Both detection methods can suppress the effect known as excess noise, which is caused by the relative intensity noise (RIN) of the source. Furthermore, the constant light component of the reference wave and, in "balanced detection," also the autocorrelation signal of the signal wave can be suppressed.With off-axis detection, only a portion of the autocorrelation signal is suppressed, but only half the number of spectral sampling points needs to be sampled compared to confocal systems, since the complex conjugate signal can be separated from the measurement signal by spatial filtering. Balanced detection can be performed as described in DE 10 2018 130396 A1.

[0019] The filter element can be an aperture stop with a non-rotationally symmetric aperture located in a pupil of the measurement beam path. It is also possible to use a suitably controlled DMD element located in a pupil of the measurement beam path. Both can represent an elliptical aperture. In addition to or as a replacement, the objective lens located between the splitter element and the detector can be designed with a non-rotationally symmetric numerical aperture.

[0020] In all cases, it is preferred that the filter element modifies the measurement beam path with respect to the numerical aperture such that the numerical aperture in the first direction, i.e., along the line, is at least twice, particularly preferably at least three times, the numerical aperture in the second direction perpendicular thereto (line height). The line height is preferably one to four pixels in order to achieve fully confocal filtering in the second direction.

[0021] The preferred detectors are very fast 2D camera detectors with a sufficient number of pixels and frame rates of several hundred fps to several thousand fps. Such camera detectors are currently only available with a few thousand pixels along the long detector side. Furthermore, detectors with a frame overhead time (FOT) of less than 10 ps, ​​especially on the order of 1 ps, are preferred for very fast acquisitions in a narrow, selected region of interest (ROI). Cameras with FOT are often 2D detectors, so that frame rates in the megahertz range in line scan mode, and thus much faster acquisitions, can be achieved with such detectors more easily than with fast 1D line detectors.

[0022] It is particularly advantageous for the off-axis coupling to be offset transversely to the narrow side of the line, i.e., the line height. If balanced detection is performed, the line is also split into two adjacent lines or regions of interest.

[0023] The detector resolution in the long line direction is preferably fully utilized. Furthermore, this significantly simplifies the optical correction of the OCT system's detection lens. The most important reason for this design, however, is the application-specifically desired asymmetric resolution, i.e., numerical aperture, which must be realized in both directions along the line and perpendicular to the line. In the line direction, the numerical aperture is higher than across it because this increases the lateral resolution and sensitivity, and in this direction, no confocal depth suppression occurs due to holoscopic reconstruction. Perpendicular to the line, however, the image is partially or fully confocal, so a lower nA value must be used to achieve the desired large depth range.It turned out that in this direction the resolution is not so crucial, since it is hardly visible in the frequently used B-scan mode of the image.

[0024] Technically, off-axis detection requires three times the sampling density in the off-axis direction. However, if detectors with square pixels are used, off-axis detection has three times the resolution in the off-axis direction. Then, imaging in the OCT system can be realized without a complex anamorphic lens.

[0025] Comparing this property with the asymmetric resolution required by the application, this is the strongest reason to align the off-axis offset perpendicular to the line. A variant with 3-pixel off-axis detection is described below as an example. However, the state of the art also includes 2-pixel off-axis and 4-pixel off-axis arrangements that can be combined with the method described here. The 2-pixel off-axis detection enables, for example, approximately 30% faster imaging, but results in more image artifacts due to the autocorrelation term. The 4-pixel off-axis detection is the slowest of the three variants, but can completely suppress the autocorrelation term.

[0026] All variants of the particularly preferred off-axis detection described below can also be combined with balanced detection.

[0027] Balanced detection is preferably achieved with a crystal polarizer, e.g., a Wollaston polarizer or a Rochon polarizer, which divides the lines / broad bands to be detected into two lines or two spatially non-overlapping regions of interest, which then have the desired phase opposition to each other. Care must be taken to ensure that the reference wave has a polarization that allows it to interfere with both measurement light components.

[0028] Typical line-scan OCT systems described in the prior art are implemented with fully coherent single-mode line illumination. This type of illumination is characterized by a very small, focused illumination line in the anterior segment of the eye, running perpendicular to the line in the retinal plane. Therefore, the laser limits to be observed are dominated by the conditions in the anterior segment. If scattered-wave illumination is used, as described in DE 10 2018 130396 A1, the extent of the illumination wave in the anterior chamber increases significantly, and thus also the laser's safety limits. This allows higher illumination wave powers to be used safely, and high-speed systems with even better sensitivity can be realized.Preferably, therefore, the illumination device provides a line-shaped illumination beam, and the system comprises a scanner which moves the line-shaped illumination beam over the object.

[0029] The illumination device can have a diffusion disk that generates more than one spatial radiation mode in the illumination beam bundle. These modes are spatially and temporally coherent with each other, but have a fixed phase difference (as described in DE 102018 130 396 A1). If a one-dimensional diffusion disk is used, the illumination line in the retinal plane is modulated by illumination speckles. If a complete retinal image is scanned, the image would be streaky. Therefore, it is preferable to use a wide illumination line and a two-dimensional diffusion disk that is approximately 3 to 50 speckle grains wide. The different and uncorrelated intensity modulations in the lines could be used for structured illumination, as is known from microscopic imaging.This allows the lateral resolution to be doubled or the image to be despeckled to a certain extent and the multiply scattered light to be suppressed, thus achieving high-quality images.

[0030] However, motion artifacts must be avoided when imaging the retina of living eyes. It is known in the prior art that no significant motion artifacts occur at A-scan rates of 10 kHz and above, acceptable motion artifacts occur above approximately 1 kHz. For example, near large retinal vessels, image quality is further reduced by high blood flow velocities at an A-scan rate of 1 kHz. For this purpose, it is preferred that the camera detector has a scanning speed such that, with an illumination 10 speckle grains wide and 3 pixels per speckle grain, a 30-pixel-wide ROI can be illuminated and captured at all wavelengths required for SS-OCT within approximately one millisecond. If, for example, 1000 depths are to be recorded, the camera must have ROI frame rates of 1,000,000 frames / s.

[0031] The state of the art (e.g., Hillmann D, et al. (2016) Aberration-free volumetric high-speed imaging of in vivo retina. arXiv:1605.03747) provides methods that can be used to estimate and numerically correct motion artifacts within an A-scan. Therefore, it is particularly preferable to use these algorithms in combination with the method presented here to correct motion artifacts. This can reduce the speed requirements of the camera.

[0032] A key technical advantage of SS-OCT systems with off-axis detection is the fact that no OCT-specific components need to be incorporated into the detection beam path. The only adaptation to OCT imaging is a reference beam irradiated from outside the signal detection pupil and superimposed on the detector. Therefore, it is easily possible to combine such an OCT system with a wide-line scanning fundus camera, such as that described in WO 2014 / 140256 A2, and to capture signals from the fundus camera and the OCT with the same detector, which then enables rapid imaging in an ROI. This combination is therefore particularly preferred.

[0033] It is understood that the features mentioned above and those to be explained below can be used not only in the specified combinations, but also in other combinations or on their own, without departing from the scope of the present invention. The invention is explained in more detail below using exemplary embodiments with reference to the accompanying drawings, which likewise disclose features essential to the invention. These exemplary embodiments are for illustrative purposes only and are not to be interpreted as restrictive. For example, a description of an exemplary embodiment with a large number of elements or components should not be interpreted to mean that all of these elements or components are necessary for implementation. Rather, other exemplary embodiments may also contain alternative elements and components, fewer elements or components, or additional elements or components.Elements or components of different embodiments may be combined with one another unless otherwise stated. Modifications and variations described for one of the embodiments may also be applicable to other embodiments. To avoid repetition, identical or corresponding elements in different figures are designated by the same reference numerals and are not explained more than once. The figures show:

[0034] Fig. 1 shows a schematic representation of the structure of a normal-sighted, average human eye with parameters relevant for OCT imaging,

[0035] Fig. 2 is a representation corresponding to Fig. 1 for a very short-sighted eye,

[0036] Fig. 3 a beam path of a combined fundus and OCT imaging system,

[0037] Fig. 4 shows the beam path of Fig. 3 with further details for coupling light from a

[0038] beam source,

[0039] Fig. 5A and 5B show two variants of the system of Fig. 3, which differ in terms of a scanner arrangement,

[0040] Fig. 6 shows a comparison of the alignment of pupil and readout detector area in the system of Fig. 3 and

[0041] Fig. 7A to 7C Options for shaping an ROI.

[0042] Fig. 1 shows a schematic view of the structure of an average human eye 2 with normal vision and parameters relevant for OCT imaging; Fig. 2 is a corresponding representation for a highly myopic eye 2 with -10 dpt. The retina 4 is to be imaged using an OCT. Relative to an optical axis 6, the eye length of an average human eye with normal vision is 24 mm, whereas the myopic eye shown here is 28 mm. In the area in front of the lens 10, the distance 12 between the corneal vertex and iris is 3 mm. A measurement depth 16 of 2 mm is to be achieved on the retina 4 on the optical axis 6. The field angle α defines the achievable image field. Three image fields 18, 20, 22 are shown, corresponding to field angles α = 40° (image field 18), 60° (image field 20), and 90° (image field 22). At the edges of the respective image fields 18, 20, 22 there is a distance between a surface 24 of the same path length and the retina 4.The area of ​​equal path length 24 is a spherical shell with a radius that is approximately defined by the distance between the iris and the retina 4 on the optical axis 6. In the average emmetropic eye (Fig. 1 ), this distance 14 is 21 mm; in the short-sighted eye shown, it is 25 mm. The area of ​​equal path length coincides with the retina 4 only on the optical axis 6. Otherwise, it lies behind the retina, i.e. more posterior, with respect to the direction of incidence of the ray. This results in a distance 26, 28, 30 at the edges of the image fields 18, 20, 22. The distance 26 for an image field 18 corresponding to a field angle α = 40° is 1 mm for the emmetropic eye and 2 mm for the short-sighted eye. The distance 60, which occurs at the edge of the image field 20 with a field angle a = 60°, has a value of 2 mm for the normal eye and 4.2 mm for the short-sighted eye.The distance 30 for the image field 22, corresponding to a field angle a = 90°, is 4.4 mm for the emmetropic eye and 8 mm for the myopic eye. It is thus almost double that of the emmetropic eye.

[0043] Fig. 3 shows the beam path of an OCT imaging system 32, preferably here as a combined fundus and OCT imaging system. Fig. 4 also shows an (optional) fiber-optic coupling of light from a beam source 36 of the imaging system 32, which serves as reference radiation. The fiber-optic provision of the reference radiation R is not the only possibility, which is why Fig. 3 schematically shows an illumination unit 32 that provides illumination radiation B. From the retina 4 illuminated thereby on the eye 2, measurement radiation M returns and is recorded by a 2D detector 38. In the embodiment shown in Figs. 3 and 4, only the illumination radiation B is deflected via a scanner 40. The image field on the retina 4, imaged with a field angle α of 90° or more, is, in contrast, fixedly imaged onto the 2D detector 38.

[0044] The illumination radiation B is scanned by the scanner 40 and radiated onto the retina 4 via a pupil splitter 42, preferably designed as a polarization splitter, through a scanning lens 44 and an ophthalmoscope lens 46. The illumination unit provides an illumination line, which is scanned by the scanner 40 across the retina 4. Illumination radiation reflected back from the retina is guided to the detector 38 as measurement radiation M. The measurement radiation passes along the optical axis through the ophthalmoscope lens 46 and the scanning lens 44, which together produce a 4f image, so that at the location of an aperture diaphragm 48, a plane is present that is conjugate to the pupil plane of the eye (plane of the iris). This creates an intermediate image between the lenses 46, 44, as shown in Fig. 4.The aperture diaphragm 48 acts as a filter element through a diaphragm opening 50 that is much wider than it is high, influencing the numerical aperture of the image onto the detector 36 in a rotationally asymmetric manner. The numerical aperture decreases toward the short side of the diaphragm opening 50. Preferably, the diaphragm opening 50, as shown in Fig. 3, is elliptical with a width-to-height ratio of at least 2:1, preferably (as shown) at least 3:1. After this filtering, the measuring radiation M is imaged onto the detector 38 by a main objective 54.

[0045] This receives an image of the retina 4 which, due to filtering by the aperture stop 48 in a plane conjugate to the pupil, has a numerical aperture that is larger along the long axis of the aperture 50 than along the short axis. With a long axis to short axis ratio of 3:1, the numerical aperture is also the same, i.e., on the 2D detector 38, the axis parallel to the long axis of the aperture 50 preferably has the full numerical aperture provided by the main objective 54, and perpendicular to this, i.e., parallel to the short axis of the aperture 50, a numerical aperture that is only one-third of the numerical aperture along the long axis.

[0046] The control of the device, in particular the lighting unit 36, as well as the reading of the detector is carried out by a control device 58 which has a processor.

[0047] To implement the SS-OCT principle, the illumination unit 36 ​​has a tunable source 34 that provides not only the illumination radiation B, but also reference radiation R (cf. Fig. 3), which is radiated obliquely through the main objective 54 onto the detector 38 via a reference radiation aperture outside the optical axis 6, i.e., in the off-axis configuration. Fig. 4 shows an example of the embodiment with a fiber-optic illumination unit 36. This comprises the source 34, an output fiber 60, and a splitter 62, which provides the tuned radiation provided at the output fiber 60 into a fiber for the illumination radiation B and an optical fiber 64 for the reference radiation R. The length of this fiber is adapted to the length of the measurement beam path such that the SS-OCT principle is implemented. This adaptation is known to those skilled in the art.At the end of the optical fiber 64, the reference radiation R is guided through the reference radiation aperture 52 by means of a small deflection mirror 66 and directed obliquely, as indicated by the dashed reference radiation R, through the main objective 54 onto the 2D detector 38, where it is superimposed with the measurement radiation M, causing interference. The illumination unit 36 ​​thus comprises the elements 34 and 60-66.

[0048] Fig. 4 shows the effect of the scanner 42 on the illumination radiation B. It was already explained above that the entire image field to be captured on the retina 4 is imaged onto the detector 38. In other words, the scanner 40 does not act on the measuring radiation M. Nevertheless, Fig. 4 shows an imaged beam 70 that is focused onto the detector 38 by the lenses 46, 44 with an intermediate image plane 72 in between, where it is caused to interfere with the reference radiation R. This selection of the beam 70 is achieved by suitable readout of the 2D detector 38. In other words, the control device 58 ensures that only those lines are read out on the detector 38 that correspond to locations on the retina 4 that are also currently illuminated by the line-shaped illumination by the scanner 40. This is the configuration shown in Fig. 5A.

[0049] As an alternative to this, a descanned embodiment according to Fig. 5B is possible, in which both the illumination radiation B and the location from which the measuring radiation M is recorded on the retina are scanned by the scanner 40 across the retina 4 to cover the image field. Compared to the illustration in Fig. 5A, the position of the scanner 40 and the pole splitter 42 are thus reversed. This embodiment has the advantage that illumination and detection are firmly coupled and both are scanned simultaneously across the retina 4. As a result, a selected ROI remains stationary on the optical axis 6. In the embodiment of Figs. 3 to 5A, however, the ROI must be dynamically adapted to the corresponding deflection of the scanner 40.

[0050] For both variants, i.e., for Fig. 5A and Fig. 5B, it is essential that the width of the line-shaped illumination corresponds to the width of the image field imaged on the retina 4. The scanner is therefore preferably a one-dimensional scanner, which shifts the illumination line (in the case of Fig. 5A) or the illumination line and the detected area (in the case of Fig. 5B) across the retina, so that the displacement path of the scanner 40 determines the extent of the image field captured on the retina 4 transversely to the line direction.

[0051] Fig. 6 shows the position of the ROI or the area that is illuminated on the detector 58 in its orientation to the aperture stop 48 or its non-rotationally symmetrical aperture 50, which effects the anamorphic filtering. As can be seen, the ROI 76 extends with its long axis parallel to the long axis of the aperture 50 and has a transverse extension of, for example, three pixels of the pixels 74 of the detector 38. Along the y-axis of the detector 38, confocal filtering is thus achieved, which ensures spatial resolution, although the anamorphic filtering of the non-rotationally symmetrical aperture 50 reduces the numerical aperture in this direction. Transverse to this, i.e. along the y-axis and thus along the long axis of the aperture 50, the image onto the detector 38 is non-confocal.However, the numerical aperture there is also larger (three times as large in the example shown) than along the y-axis, so that confocal filtering is not necessary. This interaction achieves a high imaging speed combined with fast OCT image acquisition. The detector 38 only reads the ROI 76, which is moved along the y-direction by the scanner 40 across the sample (in the case of Fig. 5A) or which is stationary in the center of the detector 38 (in the variant according to Fig. 5B). Since scanning occurs in only one direction, the speed is increased. Since the numerical aperture perpendicular to the scanning direction, in which no confocality of the image can be achieved, is high due to the anamorphic filtering of the aperture stop 48, the wave-optical imaging quality is high in both the y- and x-directions. The wavelength sweep of the SS-OCT principle now only needs to be performed for each line position, ieeach position of the scanner 40, so that overall a fast OCT system is achieved with the SS-OCT technology, which on the one hand is easy to implement and on the other hand can be integrated into an existing fundus camera without any problems, as already explained above and described in more detail below.

[0052] Figures 7A to 7C show various alternatives for selecting the ROI 76. In Figure 6, the ROI is three pixels high (in the y-direction). This achieves diffraction-limited confocal filtering in the y-direction. Figure 7A shows the ROI 76 for off-axis detection with scattered illumination illuminating an area on the retina 4 that is four speckle grains wide. Figure 7B shows a variant for single-line scanning and balanced detection. The two areas 78, 80 on the detector 38 are split by the prism mentioned above and both originate from illuminated sample areas. Fig. 7C also shows the implementation of the “balanced detection” with a single 2D detector, which simultaneously records the mutually antiphase interference signals in two separate areas 78 and 80 and thus feeds them to the “balanced detection” evaluation.

[0053] For illumination, system 32 combines, for example, multicolor LED illumination with swept-source line illumination with or without a two-dimensional diffuser (as in DE 102018 130 396 A, Fig. 8A). Both light sources are scanned with the same scanner 40 to capture the entire image field. As described in WO 2014 / 140256 A2 for Fig. 4 therein, pupil splitter 49 and scanner 40 can be exchanged. The only components that need to be exchanged in the fundus camera to switch to OCT imaging are the pupil splitting components. Intensity splitting or, preferably, polarization splitting is used.

[0054] Furthermore, the preferably elliptical aperture stop 48 with a preferred anamorphic aspect ratio of 1:3 is used to realize the appropriate scanning for off-axis detection. Since the aperture stop 48 lies in the plane of the detection pupil, the reference radiation R required for OCT can be realized, for example, by the single-mode optical fiber 64 and the small deflection mirror 66 for deflecting the reference radiation onto the detector 56. In this configuration, the main objective 54 collimates the reference radiation R and provides a reference wave for the full-frame detector 56.

[0055] The detector 36 is particularly preferably implemented by the camera XIMEA CB160MG-LX-X8G3 with a Luxima Lux160 CMOS sensor, which advantageously has an extremely short frame overhead time of approximately 1 ps, allowing fast recordings of non-fixated, living eyes without motion artifacts.

[0056] Preferably, a preferably motor-driven changing mechanism is provided for changing between a normal pupil splitter for reflection-free fundus imaging, as in the prior art, and the pupil splitter 42 with reference fiber arrangement of the aperture diaphragm 48 for OCT imaging.

[0057] In a typical eye, the illuminating radiation at retina 1 is only backscattered by the top 1 mm of retinal tissue due to its strong scattering coefficient. In most cases, the vitreous has only a low backscattering coefficient, so this region appears dark in a typical OCT image. Therefore, the imaging depth that must be imaged by coherent OCT scanning can be reduced by using a multiple-length reference arm with, for example, 2 mm spacing steps. In a standard SD-OCT system, however, the resulting images would be difficult to interpret due to the superposition of the normal and complex conjugate parts of the signal; mirror artifacts would occur.

[0058] In the above embodiments, the aperture stop with the non-rotationally symmetric aperture opening 50 is provided for anamorphic filtering. This can be replaced by another element that creates a non-rotationally symmetric aperture that is reduced on the axis along which the confocal filtering occurs through the line height (y-direction). An example of this would be a main objective 54 whose numerical aperture is not rotationally symmetric, but rather has a correspondingly different anamorphic design. Another example would be to redirect the radiation in the pupil plane, in which the aperture stop 48 is located in the embodiments of Figs. 3 and 4, in the beam path using a DMD, thereby performing the anamorphic filtering according to the aperture opening 50.The aperture stop 48 is thus only one of possible implementations of an anamorphic filter element that influences the numerical aperture of the image in a non-rotationally symmetric manner. Such a non-rotationally symmetric influence is understood to mean that the numerical aperture is reduced in one direction, while not reduced or reduced to a lesser extent in a direction perpendicular to it. In the embodiment of Figs. 3 and 4, these two directions are the y-direction, in which no reduction occurs, and the x-direction, in which the numerical aperture decreases due to the anamorphic filtering.

[0059] In the spatially resolved SS system with off-axis detection provided here, the complex conjugate part of the signal can be easily filtered out by spatial frequency filtering. Images from System 32, which is extended by a multi-length reference arm, could be displayed much more simply and without artifacts. Furthermore, the imaging speed can be further increased, as fewer images need to be acquired during wavelength tuning of the SS source.

[0060] It is known from the prior art that residual motion artifacts and optical aberrations of the human eye and the imaging system can be estimated in a holoscopic system and coherently corrected during digital post-processing of the image data. The only prerequisite for this is fully coherent or static phase imaging. The line / broadband scanning system implemented in system 32 is capable of such imaging and can therefore be combined with any of these post-processing methods. It is particularly preferred to correct the significant optical aberrations for imaging the peripheral parts of the retina in order to achieve an effective spatial resolution that nearly corresponds to the diffraction limit.

[0061] The OCT imaging system 32 is particularly preferred for use in angiography, as the fast imaging required for this application can be achieved by wavelength sweeping over a reduced tuning bandwidth. To do this, the SS source is simply adjusted so that it only performs sweeps over a reduced tuning bandwidth. The associated reduced depth resolution is sufficient for the then very fast angiography application.

Claims

Patent claims 1 . A system for optical swept-source coherence tomography of an object (4), in particular a retina of an eye (2), in an image field (22), wherein the system images the object (4) onto a detector (38) having pixels (74) with a numerical aperture and comprising: a filter element (48) which influences the numerical aperture rotationally asymmetrically such that it is larger in a first direction (x) than in a second direction (y) lying transversely thereto, and a control device (58) which reads out pixels (74) of the detector (38) which correspond to a line-shaped region in the image field (22) which has a line length defining an image field width along the first direction (x) and a line height which is smaller in comparison along the second direction (y), wherein the control device (58) realizes an image field height which is larger than the line height by shifting the line-shaped region over the object (4) along the line height,and wherein the control device (58) reads the detector (38) and generates a 3D image of the object (4) from signals of the pixels (74) of the detector (38), wherein a depth resolution of the 3D image is determined by a swept-source tuning and a lateral resolution of the 3D image is determined by the numerical aperture in the line direction and by the line height., 2. System according to claim 1, wherein the filter element has an aperture diaphragm (48) arranged in a pupil with a non-rotationally symmetric diaphragm opening (50) or a DMD element arranged in a pupil of the measuring beam path and / or is designed as an objective with a non-rotationally symmetric numerical aperture.

3. The system of claim 2, wherein the aperture stop (48) or the DMD element represents an elliptical aperture.

4. System according to one of the above claims, wherein the line height is one to four pixels (76) to effect confocal filtering in the second direction (y).

5. System according to one of the above claims, comprising a line-shaped illumination beam bundle and has a scanner (40) which scans the line-shaped Illumination beam bundle is moved over the object (4), 6. System according to one of the above claims, comprising a diffusion plate which generates in an illumination beam at least two spatial radiation modes which are spatially and temporally coherent with each other but have a fixed phase difference with each other.

7. System according to one of the above claims, wherein the numerical aperture in the first direction (x) is at least twice, preferably at least three times, the numerical aperture in the second direction (y) lying transverse thereto.

8. System according to one of the above claims, further comprising: an illumination unit (36) which provides source radiation, the wavelength of which is tunable over a predetermined spectral range and with a predetermined spectral bandwidth for implementing the swept-source OCT principle and is split into illumination radiation (B) and reference radiation (R), an illumination beam path running to the object (4) for illuminating the object (4) and a measuring beam path coming from the object (4) for imaging the object (4) along an optical axis (6) onto the detector (38), wherein the illumination beam path and the measuring beam path run together via a front section (44, 46, 72) between a splitter element (42) and the object (4), and the measuring beam path is coupled out via a splitter element (42) which lies in or in front of a pupil of the measuring beam path (54), and a coupling device (52,66) for coupling the reference radiation (R) in the pupil and outside the optical axis (6), so that the reference radiation (R) is superimposed on the detector (38) obliquely incident with the measuring radiation (M), wherein the filter element (48) is arranged downstream of the splitter element (42) in the measuring beam path (M) in the imaging direction.

Images