Optical coherence tomography (OCT) system

Abstract

One aspect of the present disclosure relates to an optical coherence tomography apparatus for imaging a patient's eye, more particularly to a sweep-source optical coherence tomography apparatus, the apparatus comprising a light source configured to generate a light beam, the light beam being distributed to a two-dimensional region on the patient's retina, and a camera equipped with a 2-D sensor, the apparatus being configured to bundle a plurality of pixels on the sensor such that each data point includes a composite response from a plurality of bundled pixels.

Description

[Technical Field]

[0001] This disclosure relates to an optical coherence tomography (OCT) system for imaging a patient's eye, and more particularly to an OCT system with a sweep source. [Background technology]

[0002] Optical coherence tomography (OCT) is a technique used in clinical practice within ophthalmology for non-invasive imaging of the human eye. Conventional methods for examining the eye involve observing the fundus of the eye using a slit lamp or ophthalmoscope. These methods only show the surface, and often, viewing morphology is necessary or advantageous, which is possible using ultrasound or optical coherence tomography (OCT). OCT provides detailed cross-sectional images of various layers of the eye, enabling ophthalmologists to diagnose and monitor a wide range of eye conditions and diseases. OCT primarily uses optical coherence tomographers based on spectral optical coherence tomography, most widely employed using commercially available superluminescent diode (SLD) and diffraction grating spectrometers. However, many other methods have been studied and commercialized, among which sweep-source optical coherence tomography (SS-OCT), with its advantages of faster imaging speed and longer range, is available for OCT biometrics and diagnostics.

[0003] Commercial optical coherence tomographers primarily use a point scanning method to achieve 2D-3D images by raster scanning a focused beam across the fundus or other parts of the eye, such as the cornea, using a light beam scanning mechanism such as a reflector galvanometer. OCT is also used in other application areas, such as dermatology and industrial inspection. In OCT scanning, the A scan is a one-dimensional depth profile obtained from light backscattered from the tissue as a function of depth. This provides information about the reflectivity or backscatter intensity of the tissue as a function of depth at a particular location. A scan data is often used to evaluate the thickness or dimensions of a particular structure within the tissue.

[0004] A B-scan is a two-dimensional cross-sectional image generated by combining multiple A-scans. It is a series of A-scans acquired along a line or scanning pattern. B-scans provide a two-dimensional representation of tissue structure, showing the internal morphology and organization of different tissue layers. This type of scan is commonly used in ophthalmology to visualize layers of the retina or cornea.

[0005] C-scanning is a three-dimensional representation of an imaged tissue volume. It is constructed by acquiring multiple B-scans at different locations and then combining them to form a complete 3D image. C-scanning provides a comprehensive view of tissue in three dimensions and is particularly useful for visualizing complex structures and evaluating spatial relationships between different tissue layers. C-scanning is commonly used in ocular imaging, such as mapping corneal topography or generating 3D reconstructions of the retina.

[0006] LF-OCT can be based on time-domain OCT (TD-OCT), spectral-domain OCT (SD-OCT), or swept-source OCT (SS-OCT). In line-field OCT, a B scan is recorded, and a C scan is achieved by scanning the irradiated line and further using one of the TD-, SD-, or SS-OCT methods.

[0007] In contrast, in full-field OCT, the en-face OCT scan is acquired immediately with a 2D camera, while the A scan is acquired by SD or SS-OCT method.

[0008] Frontal OCT provides a two-dimensional (2D) image of a specific depth plane within the tissue being imaged. Unlike conventional OCT techniques that generate cross-sectional B-scans, frontal OCT captures a single plane perpendicular to the beam path, providing a "top-down" view of the tissue.

[0009] Frontal OCT technology can utilize full-field illumination of the entire sample using a wide-field light source consisting of broadband wavelengths. The light is split into two paths: a reference path and a sample path. The reference path includes a reference mirror, and the sample path interacts with the tissue. Interference between the reference and sample light waves is used to extract depth information from the tissue. The frontal view can also be reconstructed from 3D data.

[0010] Currently available OCT systems have several drawbacks. For example, high-speed sweep sources are expensive and complex, and are often combined with high-speed 2D mirror scanners, resulting in expensive and complex OCT systems. OCT systems based on spectral range sources require diffraction gratings or spectrometers to resolve wavelengths, which often results in chromatic crosstalk and consequently poor spatial resolution.

[0011] Generally, OCT systems on the market are often very expensive, standalone systems that take up a lot of space in a clinic. [Overview of the Initiative]

[0012] In light of the above, it remains desirable, particularly in ophthalmology, to provide an OCT scanning device that can mitigate one or more of the shortcomings of prior art systems, or at least serve as an alternative.

[0013] In a first aspect, disclosed herein is an optical coherence tomography apparatus for imaging target tissue, particularly the eye of a patient, and more particularly an optical coherence tomography apparatus for a sweep source, the apparatus is (1) A light source that emits light, (2) An optical system configured to direct the sample portion of emitted light as a sample beam of light to a two-dimensional region on a target tissue to be imaged, further comprising an optical system configured to direct the reference portion of emitted light as a reference beam to a reflector, (3) A 2D image sensor comprising an array of sensor pixels, wherein the optical system is further configured to receive the reflective portion of the sample beam reflected by the target tissue and the reflective portion of the reference beam reflected by the reflector, and to orient the received reflective portions onto the image sensor in order to create an interference image of the two-dimensional region of the target tissue, (4) A signal processing circuit configured to process sensor signals from an image sensor to acquire one or more images of a target tissue, particularly depth-resolved images, (5) The signal processing circuit is configured to bundle groups of sensor pixels to each data point such that each data point includes a composite response from multiple bundled pixels and shows interference image information of a portion of the target tissue.

[0014] Bundling multiple pixels into each data point improves the signal-to-noise ratio and reduces the need for optical precision. For the purposes of this explanation, pixel bundling refers to the combination of responses from each group of adjacent pixels in a 2D image sensor. Pixel bundling can be performed by summing, averaging, or otherwise combining the pixel values ​​of the pixels being bundled. Pixel bundling may be performed during or after image sensor readout. Pixel bundling may also be called pixel binning.

[0015] In some embodiments, the signal processing circuit may be configured to select a group of sub-areas within the light-receiving area of ​​the image sensor and to selectively process only pixels from the selected sub-areas. In particular, pixel bundling may be performed on adjacent pixels within each of the selected sub-areas. The sub-areas may not overlap or may overlap. The sub-areas may be in the form of elongated strips, e.g., straight strips, each showing the contour of an elongated rectangular sub-area, or curved strips, particularly annular strips. Annular strips may be contoured as an area between two concentric circles. The selected sub-areas may show the contour of a regular pattern. An example of a regular pattern is a series of parallel, spaced-apart strips. Another example includes a grid formed by two or more sets of strips, where the strips in each set are parallel and spaced apart from one another, and the strips in each set intersect each other, for example, at an angle of 90 degrees. Another example of a regular pattern includes a set of concentric annular strips sized such that the annular strips are radially spaced apart from one another by annular gaps. A strip may define a length along its elongation direction and a width defined in a direction transverse to the elongation direction. For example, a strip may have a width of 2, 3, 4, 5 or more pixels. In some embodiments, pixel bundling may be performed by bundling all pixels, for example, along one line or along multiple adjacent lines of pixels, such that the pixels are bundled along the width of the strip, with one or more lines extending along a direction transverse to, in particular, perpendicular to, the elongation direction of the strip. However, other bundling methods may be used instead. The selected sub-areas together define an area or pattern of interest.

[0016] By defining a region of interest or pattern within the 2D area of ​​the image sensor, responses from pixels outside the region of interest can be ignored, saving data processing time.

[0017] In some embodiments, the 2D area is subdivided into a plurality of strips, and pixel banding is selectively performed within each strip. In this way, the device mimics a more expensive and slower line field scanner while enabling a faster readout speed from the sensor. Further, compared to a line field scanner, the devices of the present disclosure do not utilize a galvanometer scanner, making it less expensive and less complex.

[0018] In various embodiments, the optical system is configured to image an illuminated 2D area of the target tissue onto an image sensor such that an image of the 2D area covers a plurality of pixels of the image sensor.

[0019] For this purpose, the 2D area on the target tissue is at least 1 mm 2 , 2 , 2 , 2 , 2 ,

[0020] , , , for example at least 4 mm 2 , for example at least 10 mm 2 , for example at least 20 mm 2 , for example at least 25 mm 2 , for example 1 mm 2 ~100 mm 2 , for example 4 mm 2 ~75 mm 2 in size. The illuminated 2D area may be rectangular (particularly square), circular (particularly round), or may have another suitable geometric shape. The illuminated 2D area may have a linear extent along each of two mutually perpendicular directions that is at least 1 mm, for example at least 2 mm, for example at least 4 mm, for example at least 5 mm, for example 1 mm to 10 mm.

[0020] The optical system may be configured to illuminate a two-dimensional area in a substantially uniform or non-uniform manner. In a preferred embodiment, the optical system is configured to illuminate only, or primarily, a selected portion of a two-dimensional area, in particular, to project a pattern onto a target tissue, for example, a pattern of one or more illuminated lines, or otherwise, a pattern in which the pattern extends across the two-dimensional area. Thus, the two-dimensional area may be outlined as a lateral range of the illuminated pattern. Therefore, the optical system may be configured to illuminate a two-dimensional area with structured light.

[0021] The optical system can be configured to illuminate the entire two-dimensional region or at least the projection pattern simultaneously, i.e., without scanning the sample beam across the two-dimensional region in a direction perpendicular to the direction of the sample beam.

[0022] Accordingly, in some embodiments, the apparatus is configured to generate a pattern that is projected onto the retina or other target tissue, preferably onto a reflector of a reference path. For this purpose, the optical system may include one or more suitable pattern-generating optical elements, e.g., one or more diffractive optical elements, one or more digital optical processors, one or more spatial light modulators, one or more lens arrays, etc., or a combination thereof. By illuminating with the pattern only the area to be imaged on the selected sub-area, crosstalk is reduced and the required laser power is significantly lower. Accordingly, camera readout may be limited to the sub-area in which the illumination area is imaged. Accordingly, in some embodiments, the signal processing circuit may be configured to select a group of sub-areas within the light-receiving area of ​​the image sensor to correspond to an image of the illumination pattern projected onto the target tissue, in particular, such that the selected sub-area corresponds to the illuminated portion of a two-dimensional area on the target tissue.

[0023] In some embodiments, one or more pattern-generating optical elements are positioned in the optical path in front of the beam splitter, thus providing a compact system.

[0024] In some embodiments, the pattern is generated using one or more diffractive optical elements (DOEs). Using a DOE for pattern generation is a low-cost solution that is easy to insert into a collimated light beam. It is lightweight and does not require additional electronics when using a single fixed pattern. Further, almost all of the input light is put into the output light without absorption or scattering losses.

[0025] In some embodiments, one or more pattern generation optical elements are configured to be selectively inserted into the beam path or otherwise selectively activated. For example, one or more pattern generation optical elements may be mounted on a wheel or slide for selection upon request. In some embodiments, one or more pattern generation optical elements may be configured to selectively generate each pattern. For example, different pattern generation optical elements may be mounted on a wheel or otherwise such that the selected pattern generation optical element can be inserted into the beam path. Since different physicians may have different preferences regarding the patterns used, it is advantageous to enable the physician to select the pattern they desire to use and to enable them to change the pattern they are using.

[0026] In some embodiments, the pattern is generated using a digital light processor (DLP) that can use a digital micromirror device (DMD). The DMD is an array of individually addressable high-reflectivity micromirrors. Compared to prior art line-field OCT scanners that use continuously scanning galvanometers, the DLP pattern is static and thus there is no unwanted movement of the pattern during an OCT scan. Further, using a DLP is more reliable and uses less power than a galvanometer scanner.

[0027] In some embodiments, the pattern is generated using an array of lenses. For example, a 1D array of cylindrical lenses can be used to generate a pattern of multiple lines.

[0028] In some embodiments, the pattern is generated using a spatial light modulator (SLM). Using a spatial light modulator has the advantage of being electrically reconfigurable, so the pattern can be changed dynamically.

[0029] In some embodiments, the pattern includes multiple lines, crosses, line grids, multiple concentric rings, multiple lines intersecting at common intersections like spokes of a wheel, or any other line pattern which is clinically relevant and suitable for efficient image readout and bundling using, for example, CCD or CMOS readout techniques. CCD sensors, in particular, offer a large dynamic range and allow pixel binning. Typically, readout from a CCD sensor is performed line by line, as a CCD sensor will have only one amplifier for all pixels. CMOS sensors, on the other hand, have a smaller dynamic range but allow both binning and partial readout. This is possible because a CMOS sensor has one amplifier per pixel. Furthermore, the selected pattern should preferably convey information to the user, such as defining a relevant cross-section or relevant volume, either readable by a human or by artificial intelligence such as a machine learning algorithm.

[0030] In some embodiments, the light source is a sweep source laser. Therefore, the 2D image sensor may be configured to acquire multiple images during wavelength sweeping of the sweep source laser, thereby obtaining data points for each wavelength of the emitted sample beam. A signal processing unit may be configured to process the 2D images acquired for each wavelength of the emitted sample beam to obtain 3D image information showing the tissue characteristics at each penetration depth. Preferably, the light source is a sweep source laser having an appropriate scanning speed (e.g., 10–100 Hz), an output power of 0.5–100 mW or more, and an output wavelength in the 700–1300 nm wavelength band. However, other types of sweep source lasers may be used. The desired output power and wavelength range may depend on the application.

[0031] For a given camera having a given data extraction rate, several factors can be considered when selecting the scanning rate of the sweep source laser. These factors may include a desired spatial resolution (i.e., a desired number of pixels) and / or, in some embodiments, a desired imaging depth in each A scan, related to the number of spectral readouts during wavelength sweep of the sweep source (i.e., the number of pixel intensities read out within each sweep period).

[0032] In some embodiments, the wavelength of the laser light generated by the light source is an infrared spectrum of 700–1000 nm, for example, about 750–950 nm, preferably about 850 nm. This wavelength range is optimal in terms of attenuation / scattering and is compatible with silicon CCD or CMOS sensors that are sensitive in this wavelength range.

[0033] In some embodiments, the device includes a mounting mechanism for attaching it to a slit lamp. Several advantages can be gained by mounting the OCT system to a slit lamp. The physician can view the patient's eye while scanning is taking place, thereby ensuring that the OCT image is produced from the most interesting areas of the retina. Additionally, the focusing mechanism of the slit lamp can be used to ensure that the OCT system is positioned at the optimal distance from the patient's eye for the best possible OCT image. Finally, a slit lamp-mounted OCT system saves space in a clinic where space is extremely valuable.

[0034] The above and other embodiments will be apparent and will become apparent from the embodiments described below with reference to the drawings. [Brief explanation of the drawing]

[0035] [Figure 1] A schematic diagram of a sweep source light interferometry tomography apparatus according to an embodiment of this disclosure is shown. [Figure 2] An image sensor according to an embodiment of the present disclosure is schematically shown.

[0036] [Detailed explanation] Figure 1 schematically shows a sweep-source optical coherence tomography (OCT) apparatus 101 for imaging a patient's eye according to one embodiment of the present disclosure. The apparatus 101 comprises a light source 102, an optical system 120, an image sensor 106, and a signal processing circuit 130. The optical system 120 may comprise a beam splitter 105, a collimator 108, and optionally one or more pattern-generating optical elements 103 such as a DOE, DLP, DMD, SLM, lens array, and / or equivalent. The beam splitter 105 may be in the form of an unpolarized beam splitter. The light source 102 may be a sweep-source laser having an appropriate sweep speed, e.g., a sweep speed of 1 kHz or a sweep speed of 10 to 100 Hz, and an appropriate output power and wavelength range, e.g., an output power of 0.5 to 100 mW and an output wavelength in the wavelength band of 700 to 1300 nm.

[0037] The beam splitter 105 splits the light from the light source 102 into two paths: a reference path 109 and a sample path 110. The reference path includes a reference mirror 111, and the sample path interacts with the target tissue to be imaged, in this example, the tissue of the subject's eye 107. The light returning from the tissue and the mirror, respectively, is combined by the beam splitter 105 and directed towards the image sensor 106. The combination of the reflected light from the sample path 110 and the reference light from the reference path 109 results in an interference pattern on the image sensor that reflects the tissue characteristics at the tissue surface and / or one or more penetration depths below it. Regions of the sample that reflect a large amount of light back produce greater interference than regions with low reflectivity. Any light outside the coherence length of the light does not cause interference.

[0038] When using time-domain OCT, scanning the reference mirror 111 in the reference path provides interference images of the sample at each penetration depth.

[0039] When using frequency-domain OCT (particularly SD-OCT or SS-OCT), the spectral interferogram can be obtained by the pixels of the image sensor, then processed, and in particular, Fourier transformed, to obtain an axial scan of the reflectance amplitude with respect to depth. In SS-OCT, the spectral interferogram may be acquired sequentially by sweeping the wavelength of the light source 102, which in such embodiments may be a sweep source laser, as described above.

[0040] Therefore, the device enables the capture of interference images of the target tissue at each penetration depth.

[0041] The image sensor 106 may be a charge-coupled device (CCD) or a complementary metal-oxide-semiconductor (CMOS). The image sensor may also be a conventional 2D camera. The apparatus 101 may further include one or more pattern-generating optical elements 103 configured to produce a projected light pattern 104 having precisely defined dimensions. The light pattern may be in the form of multiple lines, crosses, checkerboard patterns, or any other clearly defined geometric form.

[0042] The signal processing circuit 130 can be implemented as any suitable analog and / or digital circuit. It may be partially or fully integrated with the image sensor, or it may be provided as at least partially a separate signal processing unit. The signal processing circuit performs pixel bundling, as described below, and further signal processing, which may depend on the type of OCT used. For example, the signal processing may include Fourier transforms and / or other signal processing operations.

[0043] In some embodiments, the light source is configured to emit a collimated light beam distributed across a 2D region on the patient's retina. The light beam may be collimated along one or both directions and / or configured to illuminate the target tissue to be imaged with structured light, for example, using a pattern-generating optical element 103.

[0044] As described above, in some embodiments, the pattern-generating optical element(s) 103 may include one or more diffractive optical elements. Alternatively, the apparatus may instead include another type of pattern-generating optical element for generating projection patterns, such as a digital mirror device array, an array of lenses, in particular microlenses, or a spatial light modulator. In other embodiments, the DOE or other means for generating projection patterns may be omitted.

[0045] Figure 2 is a schematic diagram of an image sensor, where the reflected light beam is imaged and distributed across a 2D area 201 of the image sensor, i.e., across the light-receiving area of ​​the image sensor. The 2D sensor can be a CCD or CMOS sensor, preferably a low-cost megapixel sensor with a>1, where each horizontal pixel array consists of, for example, 1000 pixels. This sensor replaces the more expensive line scanners typically used. The 2D area illuminated by the beam is then further subdivided. For this purpose, the image sensor and / or associated signal processing circuitry may be configured to selectively read out and / or process only image pixels from selected sub-areas of the 2D area 201. In the illustrated example, the area is subdivided into several linear strips 202, which can be efficiently implemented using either a CCD or CMOS sensor. Other examples of sub-areas, such as rectangles and intersections, can also be considered, but these are often preferable when using a CMOS sensor, where each pixel can be read out independently. Each sub-area, e.g., each strip, consists of a plurality of pixels 207. In the illustrated example, the sensor area is divided into five lines, covering only a portion of the image sensor's illumination area, while information from the remaining pixels is discarded. This has the advantage of reducing the time required for sensor readout.

[0046] In some embodiments, the strip or other type of sub-area read out includes only a fraction of the total number of pixels on the image sensor, for example, 10%, 5%, or even as little as 2% of the pixels on the sensor. This has the advantage of being much faster to read out than using data from all available pixels.

[0047] As described above, in some embodiments, the device may project an illumination pattern onto the irradiated area of ​​the target tissue. Preferably, the projection pattern is aligned with a sub-area, particularly so that the illuminated portion of the projection pattern is imaged onto a selected sub-area of ​​the 2D area 201 of the image sensor, thereby reducing the overall illumination of the tissue and preventing noise from areas that are not illuminated and should not be imaged.

[0048] An enlarged view 203 of one of the strips 202n is shown. This view shows how strip 202n contains a grid of pixels 207. In this example, strip 202n contains 5 rows of pixels, for example, each strip containing 1000 × 5 pixels. Furthermore, when read out, pixels 207 are binned or bundled so that bundled pixels 204a, 204b...204n each correspond to one measurement or data point. Bundling pixels in this way allows each measurement to have a higher signal-to-noise ratio, all strips or sub-areas to be read out simultaneously, and increases the speed of signal acquisition.

[0049] Therefore, the bundled pixels from the sub-area provide a two-dimensional array of data points, which can be considered pixels of a low-resolution 2D image with a resolution smaller than the native pixel resolution of the image sensor.

[0050] By performing depth scanning (using time-domain or frequency-domain OCT), each A scan can be acquired simultaneously at each bundled pixel, resulting in a two-dimensional array of depth-resolved interferometric profiles (A scans). Thus, the array of A scans provides a volumetric representation of the target tissue (C scan). Therefore, a C scan can be obtained within one sweep cycle of the sweep source laser (or accumulated across multiple sweeps).

[0051] Therefore, at least some embodiments of the apparatus described herein may be designed to acquire multiple line-field OCT datasets simultaneously without requiring a light beam to be scanned laterally across the tissue surface.

[0052] Image data acquired by various embodiments of the apparatus disclosed herein provides depth-resolved interferometric image data from which various representations can be created. Accordingly, the signal processing circuits of various embodiments may be configured to process sensor signals from an image sensor to acquire one or more images which may include a volumetric image representation, one or more B-scans of each cross-section, one or more frontal representations of one or more tissue layers at each penetration depth below the tissue surface, and / or other types of images.

[0053] Embodiments of the methods described herein may be implemented by hardware comprising several distinct elements. In apparatus claims that enumerate several means, some of these means may be embodied by one of the same elements, components, or items of hardware. The mere fact that certain means are enumerated in different dependent claims or described in different embodiments does not imply that combinations of these means cannot be used advantageously.

[0054] Where used herein, the term “equipped with” is to be interpreted as identifying the presence of the described feature, element, step, or component, but should not be constrained by the presence or addition of one or more other features, elements, steps, components, or groups thereof.

Claims

1. An optical coherence tomography (OCT) system for imaging target tissue, particularly the retina of a patient's eye, wherein the system is (1) A light source that emits light, (2) An optical system configured to direct a sample portion of emitted light as a sample beam of light onto a two-dimensional region on the target tissue to be imaged, wherein the optical system is further configured to direct a reference portion of emitted light onto a reflector as a reference beam, (3) A 2D image sensor comprising an array of sensor pixels, wherein the optical system is configured to receive the reflective portion of the sample beam reflected by the target tissue and the reflective portion of the reference beam reflected by the reflector, direct the received reflective portions onto the image sensor, and generate an interference image of the target tissue. (4) A signal processing circuit configured to process sensor signals from the image sensor to acquire one or more depth-resolved images of the target tissue, Optical coherence tomography apparatus, wherein each data point includes a composite response from multiple bundled pixels, and is configured to bundle groups of sensor pixels to each data point so as to show interference image information of a portion of the target tissue.

2. The apparatus according to claim 1, characterized in that the signal processing circuit is configured to select a group of sub-areas within the light-receiving area of ​​the image sensor and to selectively process only pixels from the selected sub-areas.

3. The apparatus according to any one of the prior claims, further configured to direct the sample beam onto the target tissue to generate a pattern projected onto the two-dimensional region on the target tissue, in particular a pattern corresponding to a selected group of sub-areas.

4. The aforementioned pattern includes a line pattern, particularly a line pattern selected from a grid formed by multiple parallel lines, crosses, lines intersecting each other, multiple lines intersecting at common intersections, and multiple concentric circles. The apparatus according to any one of the preceding claims.

5. The apparatus according to claim 3 or 4, characterized in that the pattern is generated using one or more pattern-generating optical elements.

6. The apparatus according to claim 3 or 4, characterized in that the pattern is generated by a lens array, particularly a 1-D or 2-D lens array.

7. The apparatus according to any one of the preceding claims, characterized in that one or more of the pattern-generating optical elements are mounted on a wheel or slide for selection as required.

8. The apparatus according to claim 3 or 4, characterized in that the pattern is generated using a digital mirror device (DMD) array.

9. The aforementioned pattern is generated using a spatial light modulator. The apparatus according to claim 3 or 4.

10. The apparatus is a sweep source optical coherence tomography apparatus, and the light source is a sweep source laser. The apparatus according to any one of the preceding claims.

11. Further includes a mounting mechanism for attaching to a slit lamp. The apparatus according to any one of the preceding claims.

Images