Systems, methods, and media for multiple-reference arm spectral domain optical coherence tomography

Abstract

To provide systems, methods, and media for multiple-reference arm spectral domain optical coherence tomography.SOLUTION: An embodiment includes: a sample arm connected to a light source 202; a first reference arm having a first path length; a second reference arm having a longer second path length; a first optical coupler 222 that couples light from the sample arm and light from the first reference arm together; a second optical coupler 224 that couples light from the sample arm and light from the second reference arm together; and an optical switch 230. The optical switch comprises: a first input port connected to the first optical coupler; a second input port connected to the second optical coupler via an optical waveguide 226 that induces a delay at least equal to an acquisition time of an image sensor of a spectrometer 124; and an output port connected to the image sensor of the spectrometer.SELECTED DRAWING: Figure 2A

Description

[Technical field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application is based on and claims the benefit of and priority to U.S. Provisional Patent Application No. 62 / 571,787, filed October 12, 2017, which is incorporated herein by reference in its entirety.

[0002] STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT: Not applicable. [Background technology]

[0003] Spectral domain optical coherence tomography (SD-OCT) is an optical coherence tomography technique that has become increasingly useful due to its ability to generate data with relatively high sensitivity, high speed, and phase stability. However, conventional SD-OCT suffers from depth-dependent sensitivity falloff that reduces the usefulness of SD-OCT in imaging certain types of samples. For example, such depth-dependent sensitivity falloff can reduce the usefulness of SD-OCT in imaging tissues and / or large luminal organs with non-uniform surface topologies where the distance between the probe and the tissue surface often varies widely.

[0004] In general, the sensitivity roll-off observed in an SD-OCT system is determined by the spectral resolution of the spectrometer used to capture the spectral domain image data, which is in turn limited by the finite size and number of pixels in the linear detector array used to read the image data.

[0005] In addition, taking the Fourier transform of the real interference signal generated by an SD-OCT system produces mirror images on either side of the zero delay. This makes it even more difficult to easily distinguish between negative and positive image depths. To avoid this ambiguity, the zero delay is typically set at or outside the tissue surface, since the empty space between the SD-OCT probe and the surface usually does not produce much information. This makes it easier to generate an image of the tissue, but reduces depth sensitivity, since the empty space is sampled at high resolution and the tissue away from the surface is sampled at low resolution or not sampled at all.

[0006] 1A and 1B show examples of conventional single-reference arm spectral domain optical coherence tomography systems 100 and 150. FIG. 1A shows a commonly used configuration of an SD-OCT system, which uses a Mach-Zehnder interferometer for SD-OCT. FIG. 1B shows another commonly used configuration of an SD-OCT system, which uses a Michelson interferometer for SD-OCT. In these conventional SD-OCT systems, the imaging depth range is limited by the number of spectral acquisitions. It is determined by both the number of pixels in the linear detector used and the width of the pixel. Although the number of pixels is usually sufficient to achieve a relatively longer imaging depth range, the width of the spectrum acquired by each pixel is determined by the width of the detector pixel, so the effective imaging depth range is limited by the finite width of the detector pixel. Generally, the maximum imaging depth range is half the coherence length of the light acquired by each pixel. In addition, due to conjugate symmetry, the zero delay position is usually set outside the sample to avoid ambiguity between mirror images in the real signal. Therefore, only half of the imaging range can usually be used for images without mirror artifacts. Furthermore, since the degree of coherence decays with the path length difference, the sensitivity of the SD-OCT system rolls off relatively rapidly, making the measurement sensitivity insufficient for imaging samples close to the edge of the available imaging depth range.

[0007] As shown in FIG. 1A, the light source 102 can provide light to the sample arm and the reference arm via a polarization control 104, an optical amplifier 106, and a fiber coupler 108. A portion of the light (e.g., 80%) is directed to the sample arm, and a second portion of the light (e.g., 20%) is directed to the reference arm. An optical circulator 110-1 directs the light received from the fiber coupler 108 to the sample 112 (in the sample arm), and a second optical circulator 110-2 directs the light to a reference reflector 114 (in the reference arm). The light in the sample arm can be directed to the sample via collimating optics 116-1 and focusing optics 118 (e.g., lenses) that can cast a beam with a focal depth tuned to near the surface of the sample 112. Portions of the beam may be reflected at various depths in the sample depending on the reflectivity of the sample, and are then received by collection optics 118 and directed via collimating optics 116-1 to optical circulator 110-1, which directs the reflected light to fiber coupler 120 for combination with light from the reference arm. Collimating optics 116-2 directs the beam from the reference arm to reference reflector 114, which reflects the beam back via collimating optics 116-2 to optical circulator 110-2, which directs the light reflected by reference reflector 114 to fiber coupler 120 via polarization control 122 for combination with light from the sample arm. Fiber coupler 120 combines light from both the sample arm and the reference arm and directs the light to spectrometer 124, which generates a wavelength-dependent signal representative of the structure of the sample along the depth direction.

[0008] 1B operates on a similar principle to system 100 shown in FIG 1A, but instead of using optical circulators in both the sample and reference arms, it includes only one optical circulator 110-3 before fiber coupler 108. In systems 100 and 150, the length of the reference arm can be set by adjusting the position of reference reflector 114 to set the depth of the zero delay point relative to sample 112.

[0009] Therefore, new devices, systems and methods for multi-reference arm spectral domain optical coherence tomography are desirable. Summary of the Invention

[0010] According to some embodiments of the disclosed subject matter, devices, systems and methods are provided for multi-reference arm spectral domain optical coherence tomography.

[0011] According to some embodiments of the disclosed subject matter, a system for spectral domain optical coherence tomography is provided, the system comprising: a light source; an image sensor; a sample arm coupled to the light source, the sample arm configured to project light from the light source towards a sample; a first reference arm having a first path length, the first reference arm coupled to the light source and the sample arm; a second reference arm having a second path length longer than the first path length, the second reference arm coupled to the light source and the sample arm; a first optical coupler configured to combine light from the sample arm and the first reference arm; a second optical coupler configured to combine light from the sample arm and the second reference arm; and an optical switch, the optical switch having a first port coupled to an output of the first optical coupler, a second port coupled to an output of the second optical coupler via a length of optical waveguide that creates a delay at least equal to an acquisition time of the image sensor; and a third port coupled to the image sensor, The optical switch is configured to selectively provide light from one of the first port and the second port, and block light from the other of the first port and the second port.

[0012] In some embodiments, the system further comprises a processor programmed to: cause the optical switch to output light received at the first port during a first time window; cause the image sensor to generate a first vector of data representative of a state of the sample in the first time window that includes a first time based on the light received during the first time window; cause the optical switch to output light received at the second port during a second time window; cause the image sensor to generate a second vector of data representative of the state of the sample in the first time window based on the light received during the second time window that includes a second time; cause the optical switch to output light received at the first port during a third time window that is after the first time window and the second time window, the third time window not including either the first time or the second time; and cause the image sensor to generate a third vector of data representative of the state of the sample in the third time window based on the light received during the third time window.

[0013] In some embodiments, the first vector and the second vector correspond to a first lateral position on a surface of the sample, and the third vector corresponds to a second lateral position on the surface of the sample.

[0014] In some embodiments, the processor is further programmed to generate a matrix of data, the first vector being a first row of the matrix, the second vector being a second row of the matrix, and the third vector being a third row of the matrix.

[0015] In some embodiments, the first vector comprises at least N elements, each of the N elements corresponding to a pixel of the image sensor, and the matrix comprises at least N columns.

[0016] In some embodiments, a second optical path length from the second optical coupler to the second port of the optical switch is at least 5 optical microseconds longer than a first optical path length from the first optical coupler to the first port of the optical switch.

[0017] In some embodiments, the second optical path length from the second optical coupler to the second port of the optical switch is at least 1 kilometer long.

[0018] In some embodiments, the system further comprises a fiber splitter having an input coupled to the light source, a first output coupled to the sample arm, and a second output coupled to the first reference arm and the second reference arm, the fiber splitter configured to provide at least half of the light received at the input to the first output.

[0019] In some embodiments, the fiber splitter is configured to provide three-quarters of the light received at the input to the first output.

[0020] In some embodiments, the fiber splitter is a first fiber splitter and the system further comprises a second fiber splitter having an input coupled to the second output of the first fiber splitter, a first output coupled to the first reference arm, and a second output coupled to the second reference arm, the second fiber splitter configured to provide half of the light received at the input to the first output.

[0021] In some embodiments, the system further comprises a first optical circulator coupled to the light source, the sample arm, the first optical coupler, and the second optical coupler, a second optical circulator coupled to the light source, the first reference arm, and the first optical coupler, and a third optical circulator coupled to the light source, the second reference arm, and the second optical coupler.

[0022] In some embodiments, the system further comprises a variable delay line coupled between the light source and the first reference arm.

[0023] In some embodiments, the sensitivity of the system is at least 98 dB throughout the imaging depth of the system.

[0024] In some embodiments, the second path length is longer than the first path length by half a maximum imaging depth of the system.

[0025] According to some embodiments of the disclosed subject matter, there is provided a method for spectral domain optical coherence tomography comprising: receiving, by a processor, a matrix of data comprising: a first vector of data representative of a state of a sample in a first time window including a first time, generated using a first reference arm having a first path length; a second vector of data representative of the state of the sample in the first time window, generated using a second reference arm having a second path length; a third vector of data representative of the state of the sample in a second time window not including the first time, generated using the first reference arm; and a fourth vector of data representative of the state of the sample in the second time window, generated using the second reference arm; and generating, by the processor, first image data based on the first vector of data and the third vector of data. generating, by the processor, second image data based on the second vector of data and the fourth vector of data; calculating, by the processor, a spatial offset between the first image data and the second image data based on a comparison of a portion of the first image data and a portion of the second image data; adding, by the processor, a first plurality of vectors to the portion of the first image data in a number based on the spatial offset to generate a zero padded image; removing, by the processor, a second plurality of vectors from the second image data in a number based on the spatial offset to generate a cropped image; and combining, by the processor, the zero padded image and the cropped image to generate an image of the sample having a reduced sensitivity roll off.

[0026] In some embodiments, the first vector is a first row of the matrix and the second vector is a second row of the matrix.

[0027] In some embodiments, the first vector comprises at least N elements, each of the N elements corresponding to a pixel of the image sensor, and the matrix comprises at least N columns.

[0028] In some embodiments, the spatial offset corresponds to a particular number of columns out of the N columns.

[0029] In some embodiments, each entry in the first vector corresponds to a wavelength, and the method further comprises mapping the wavelength for each element to a wavenumber k.

[0030] In some embodiments, the method further comprises the steps of: converting each element of the first vector into a depth value; performing a discrete Fourier transform.

[0031] In some embodiments, the method further comprises generating a real image of the sample based on the first image data by further dividing the matrix into two sub-matrices by selecting vectors on the positive side of zero delay for inclusion in a first sub-matrix and selecting vectors on the negative side of the zero delay for inclusion in a second sub-matrix.

[0032] In some embodiments, the portion of the first image data corresponds to the first sub-matrix.

[0033] In some embodiments, the method further comprises generating a binary mask based on the second sub-matrix.

[0034] In some embodiments, the method further comprises generating a weighting matrix W comprising a plurality of weighting coefficients Cm based on the binary mask and the function.

[0035] In some embodiments, the function is Cm=(tanh(x)+1) / 2, where x is a value in the range of -2π to 2π.

[0036] In some embodiments, the first vector and the second vector correspond to a first lateral position on a surface of the sample, and the third vector and the fourth vector correspond to a second lateral position on the surface of the sample.

[0037] According to some embodiments of the disclosed subject matter, a non-transitory computer readable medium is provided that includes computer executable instructions that, when executed by a processor, cause the processor to perform a method for spectral domain optical coherence tomography, the method including the steps of receiving, by the processor, a matrix of data comprising: a first vector of data representative of a state of a sample in a first time window that includes a first time, generated using a first reference arm having a first path length; a second vector of data representative of the state of the sample in the first time window, generated using a second reference arm having a second path length; a third vector of data representative of the state of the sample in a second time window that does not include the first time, generated using the first reference arm; and a fourth vector of data representative of the state of the sample in the second time window, generated using the second reference arm; generating, by the processor, first image data based on the vector and the third vector of data; generating, by the processor, second image data based on the second vector of data and the fourth vector of data; calculating, by the processor, a spatial offset between the first image data and the second image data based on a comparison of a portion of the first image data to a portion of the second image data; adding, by the processor, a first plurality of vectors to the portion of the first image data in a number based on the spatial offset to generate a zero padded image; removing, by the processor, a second plurality of vectors from the second image data in a number based on the spatial offset to generate a cropped image; and combining, by the processor, the zero padded image and the cropped image to generate an image of the sample having a reduced sensitivity roll off.

[0038] Various objects, features and advantages of the disclosed subject matter will become more fully apparent from the following detailed description of the invention when taken in conjunction with the accompanying drawings, in which like elements are designated with like reference numerals and in which: [Brief description of the drawings]

[0039] [Figure 1A] 1 shows an example of a conventional single-reference arm spectral domain optical coherence tomography system. [Figure 1B] 1 illustrates another example of a conventional single-reference arm spectral domain optical coherence tomography system. [Figure 2A] 1 illustrates an example of a system for multi-reference arm spectral domain optical coherence tomography in accordance with some embodiments of the disclosed subject matter. [Figure 2B] 1 illustrates another example of a system for multi-reference arm spectral domain optical coherence tomography in accordance with some embodiments of the disclosed subject matter. [Diagram 3] 1 illustrates example images generated using multiple reference arms of a multi-reference arm Spectral Domain Optical Coherence Tomography system embodied in accordance with some embodiments of the disclosed subject matter. [Figure 4] 1 illustrates an example process for generating spectral domain optical coherence tomography image data using multiple reference arms according to some embodiments of the disclosed subject matter. [Diagram 5] 1 shows example timing diagrams illustrating the collection of image data according to some embodiments of the disclosed subject matter. [Figure 6] 13 illustrates an example process for combining image data generated using different reference arms according to some embodiments of the disclosed subject matter. [Figure 7A] 11A-11C show example images illustrating various operations that can be performed during the combination of image data generated using different reference arms according to some embodiments of the disclosed subject matter. [Figure 7B] 11A-11C show further example images illustrating various operations that can be performed during the combination of image data generated using different reference arms according to some embodiments of the disclosed subject matter. [Figure 7C] 13A-13C show examples of weighted combining coefficients that may be used to combine image data generated using different reference arms according to some embodiments of the disclosed subject matter. [Figure 8] 1 shows examples of sensitivity roll-off observed for a first reference arm, a second reference arm, and a combination of the first and second reference arms of a Spectral Domain Optical Coherence Tomography system embodied in some embodiments of the disclosed subject matter. [Figure 9] FIG. 1 illustrates an example of hardware that can be used to implement imaging and / or computing devices that can be used in conjunction with some embodiments of the multi-reference arm spectral domain optical coherence tomography mechanism embodied by some embodiments of the disclosed subject matter. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0040] According to some embodiments of the disclosed subject matter, mechanisms (which may include devices, systems and methods) are provided for multi-reference arm spectral domain optical coherence tomography.

[0041] According to some embodiments of the disclosed subject matter, the mechanisms described herein can be used to implement an SD-OCT system with reduced sensitivity roll-off (e.g., compared to conventional SD-OCT systems) by using multiple reference arm delays. In some embodiments, the optical delays between the multiple reference arms can be adjusted such that the region of maximum sensitivity for each image generated using the multiple reference arms is located within the sample near the surface of the sample for the first reference arm and approximately half the ranging depth of the first reference arm. In some embodiments, the mechanisms described herein can use a fiber delay line to time delay the interferogram from one reference arm such that interference originating from the same sample location within the same time window can be detected by one detector. In some embodiments, the mechanisms described herein can combine images resulting from multiple reference arm delays to create a composite image with reduced sensitivity roll-off and extended imaging depth range.

[0042] As described below, an SD-OCT system embodied using the mechanisms described herein has a maximum sensitivity of over 105 dB and a minimum sensitivity of 95 dB over a depth range of 6 mm, and images of tissue obtained in vitro have demonstrated the ability of such a system to more clearly visualize tissue at the edges of a depth range.

[0043] In some embodiments, the electromagnetic radiation reflected from the sample can be split and combined with different reference beams having different path lengths. For example, the electromagnetic radiation in one reference arm can be delayed by half the imaging range of the system relative to the other reference arm and combined with half of the electromagnetic radiation reflected from the sample to generate an interference pattern. In another example, the electromagnetic radiation from the second reference arm can be combined with the other half of the electromagnetic radiation reflected from the second reference arm to generate a separate interference pattern representing another portion of the sample. In such an example, the two interference patterns can correspond to backscattered electromagnetic radiation from the sample interfering with reference electromagnetic radiation having two different path lengths.

[0044] In some embodiments, the path length (also referred to herein as delay) of one or more reference arms can be adjusted using an optical circulator that directs light from the light source to a reference mirror, the position of which is adjustable. Additionally or alternatively, one or more reference arms can omit the reference mirror, and the reference beam can be provided by directing light from the light source to the reference arm through an optical system having a path length approximately the same as the path length to the sample. For example, light from the light source can be split into two reference arms, where the reference arm light from a reference arm having a fixed path length is directed to a fiber coupler (e.g., a 99 / 1 fiber coupler), and another reference arm can be configured to have an adjustable variable path length (e.g., by extending the fiber, increasing the length in free space, etc.).

[0045] In some embodiments, after the electromagnetic radiation reflected from the sample is combined with the electromagnetic radiation from the reference arm, and before the electromagnetic radiation corresponding to the interference pattern created by the combination of the sample arm electromagnetic radiation and the reference arm electromagnetic radiation reaches an interferometer (e.g., via an optical switch), the combined electromagnetic radiation corresponding to one (or more) reference arms can be delayed by passing the electromagnetic radiation through a length of fiber optic waveguide (e.g., a single mode optical fiber) to delay the electromagnetic radiation in time relative to the combined electromagnetic radiation corresponding to another reference arm. For example, the length of optical fiber can be configured to delay the electromagnetic radiation from a particular reference arm in time by the time it takes for a linear detector to obtain information from the electromagnetic radiation of the other reference arm.

[0046] In some embodiments, the spectrum of the interfering electromagnetic radiation from each reference arm can be detected by scattering the interfering electromagnetic radiation in one dimension, according to wavelength, towards a detector array. For example, the scattering signal can be detected by a spectrometer or a line scan camera. It should be noted that, although a linear array of CCD pixels is used to detect such electromagnetic radiation in this specification, this is merely an example and that in some cases, a portion of a two-dimensional array may be used to detect such electromagnetic radiation.

[0047] In some embodiments, an optical switch may be used to alternately pass the interfering electromagnetic radiation from each reference arm in a sequential manner to alternately detect multiple interfering electromagnetic radiation signals representative of the sample within a particular time window. In some embodiments, sequential detection of multiple spectra of interfering electromagnetic radiation corresponding to multiple reference arms can ensure that the obtained interference spectra correspond to the same time window, since the electromagnetic radiation from one reference arm is delayed in time by the same amount of time used to detect each one of the spectra.

[0048] In some embodiments, the sequentially acquired spectra can be processed to combine information representing interference with light from each reference arm. For example, the spectra can be processed by subtracting background and fast Fourier transforming to generate depth information (e.g., representing the reflectance profile of the sample in the axial / depth direction). In such an example, the resulting depth information can be used to generate image data with depth information corresponding to both reference arms. In some embodiments, the mechanisms described herein can be used to generate images with reduced sensitivity roll-off and increased imaging depth range (e.g., compared to conventional SD-OCT systems), thereby mitigating one of the biggest drawbacks of conventional SD-OCT systems.

[0049] Various other schemes have been explored to mitigate the sensitivity roll-off of SD-OCT. For example, one scheme uses reference or sample arm phase shifting to introduce a carrier frequency to facilitate quadrature detection. The complex spectral interference signal from the quadrature detector allows positive and negative delay signals to be more easily distinguished, which makes it easier to place the zero delay in the middle of a range of depths, increasing the depth at which high-resolution signals can be detected with higher sensitivity. However, such quadrature detection approaches usually require multiple additional optical components that introduce losses, thus reducing the overall maximum achievable OCT sensitivity.

[0050] As another example, one scheme uses an optical switch in the reference arm of an interferometer to alternate between transmitting delays in two different reference arms. In such an example, a single reference arm can contain multiple reflectors at different path lengths, and the light in the reference arm can be passed alternately along each path. The interference spectra corresponding to each reference delay are acquired sequentially using a single spectrometer and then combined by concatenating the most sensitive parts of the cropped images generated from each reference arm. Although this approach can increase sensitivity at depth, it reduces the transverse resolution for images acquired using a high-speed scanning SD-OCT system because the images are acquired at different times. It can be expected to be small.

[0051] As yet another example, one scheme uses dual reference arms with separate light sources and corresponding separate interferometers with line scan cameras used to capture information from each reference arm. This scheme introduces additional complexity (e.g., due to the introduction of different light sources, different sensors, etc.) and is also bulkier and more costly than a scheme that uses only one spectrometer.

[0052] As yet another example, another scheme multiplexes multiple sequentially acquired spectra with slightly offset frequency combs to attempt to reduce the sensitivity roll-off. Although this approach can reduce the sensitivity roll-off, the sequential acquisition of data can be expected to result in reduced lateral resolution for images obtained using SD-OCT systems.

[0053] 2A illustrates an example system 200 for multi-reference arm spectral domain optical coherence tomography according to some embodiments of the disclosed subject matter. In some embodiments, the system 200 can include an electromagnetic radiation source 202 (for convenience herein referred to as a light source 202). The light source 202 can be any suitable light source or multiple light sources. For example, the light source 202 can be embodied using a superluminescent diode.

[0054] In some embodiments, the light source 202 emits electromagnetic radiation (for convenience herein referred to as “light”). 206. Additionally, in some embodiments, the polarization controller 204 can change the polarization state of the light emitted towards the optical amplifier 206, which can be used to affect the gain of the optical amplifier 206.

[0055] In some embodiments, light from the light source 202 is received by a fiber splitter 208 that can be used to split the input light into two output lights, a first output that can be directed to a sample arm and a second output that can be directed to multiple reference arms. The fiber splitter 208 can split the input light into any suitable portion. For example, the fiber splitter 208 can be a 75 / 25 fiber splitter that directs 75% of the input light into the sample arm and 25% of the input light into the reference arm.

[0056] In some embodiments, the optical circulator 210-1 can receive light output from one port of the fiber splitter 208 and direct the light to a sample arm of the system 200. In some embodiments, the sample arm can include an optical system 216-1 that can be used to direct a light beam to the sample 112. In some embodiments, the optical system 216-1 can include any suitable optical system for generating a beam with sufficient focal depth to generate data that can be used to generate an optical coherence tomography signal. In some embodiments, the optical system 216-1 can be configured to direct the light to a particular location. For example, the optical system 216-1 can include one or more components that can be actuated to change the direction of the output beam.

[0057] In some embodiments, a portion of the light directed to the sample 112 is reflected from the sample 112 and received by the optical system 216-1, which directs the reflected light back to the optical circulator 210-1. The optical circulator 210-1 can direct the light received from the optical system 216-1 to another fiber splitter 218. In some embodiments, the fiber splitter 218 can split the input light and direct a portion of the reflected light to a fiber coupler that can be used to combine the light reflected from the sample with the light from each reference arm. The fiber splitter 218 can split the input light to any suitable portion. For example, the fiber splitter 218 can be a 50 / 50 fiber splitter (e.g., a 1×2 or 2×2 3 dB fiber splitter) that directs 50% of the input light to a first fiber coupler associated with a first reference arm and 50% of the input light to a second fiber coupler associated with a second reference arm.

[0058] In some embodiments, the portion of the light directed by fiber splitter 208 to the reference arm can be received by another fiber splitter 220, which can again split the light to direct a portion of the light from the light source to each reference arm. The fiber splitter 220 can split the input light into any suitable portion. For example, the fiber splitter 220 can be a 50 / 50 fiber splitter (e.g., a 1×2 or 2×2 3 dB fiber splitter) that directs 50% of the input light to a first reference arm and 50% of the input light to a second reference arm.

[0059] In some embodiments, each output of fiber splitter 220 may be coupled to an optical circulator. For example, the first reference arm may include optical circulator 210-2 and the second reference arm may include optical circulator 210-3.

[0060] In some embodiments, each optical circulator (e.g., 210-2 and 210-3) can direct light to reference reflectors 214-1 and 214-2, respectively (e.g., via optics 216-2 and 216-3, respectively). and 210-3 can receive light reflected by the corresponding reference reflectors 214-1 and 214-2 and direct the reflected light to the corresponding fiber coupler. For example, in some embodiments, the optical circulator 210-2 can direct the light reflected by the reference reflector 214-1 to a first fiber coupler 222, which is also configured to receive a portion of the light reflected from the sample 112 (e.g., from the fiber splitter 218). As another example, in some embodiments, the optical circulator 210-3 can direct the light reflected by the reference reflector 214-2 to a second fiber coupler 224, which is also configured to receive a portion of the light reflected from the sample 112 (e.g., from the fiber splitter 218). As shown in FIG. 2A, in some embodiments, a polarization control 204 can be disposed between the optical circulators of the reference arm to control the polarization of the light directed to the corresponding fiber coupler. In some embodiments, fiber couplers 222 and 224 may be implemented using any suitable fiber coupler. For example, fiber coupler 222 (and / or fiber coupler 224) may be a 99 / 1 fiber coupler configured to combine approximately half of the light reflected from sample 112 with the light reflected from reference reflector 214-1 (or from reference reflector 214-2).

[0061] In some embodiments, the optical path length of one or more reference arms can be controlled by positioning the reference reflectors 214-1 and / or 214-2 in any suitable manner. For example, in some embodiments, the reference reflectors 214-1 and / or 214-2 can be associated with actuators that can controllably position the reference reflectors. In more specific embodiments, one reference reflector (e.g., reference reflector 214-1) can be positioned to provide a path length that corresponds to a path length along from the sample arm to the surface of the sample 112, and the other reference reflector (e.g., reference reflector 214-2) can be positioned to provide a path length that corresponds to a path length along from the sample arm to a point within the sample 112 (e.g., more than half the maximum imaging depth range that can be realized using the other reference reflector).

[0062] In some embodiments, the combined light from each fiber coupler 222 and 224 can be directed to a separate port of optical switch 230. In some embodiments, the path from fiber coupler 222 to a first port of optical switch 230 can be a first length, and the path from fiber coupler 224 to a second port of optical switch 230 can be a second length. For example, as shown in FIG. 2A , the path from fiber coupler 224 can include a length of single mode optical waveguide 226 that can delay the arrival of light from fiber coupler 224 to optical switch 230.

[0063] In some embodiments, the length of the optical waveguide 226 is set to a length that is sufficient to guide the light from the fiber coupler 224 to the sensor of the spectrometer 124 for it to collect a single line of data. The optical switch 230 may be configured to delay the light from the fiber coupler 222 and the fiber coupler 224 by an amount of time that corresponds to the time it takes to collect one line of data. In such an embodiment, the system 200 may control the optical switch 230 to alternately provide light from the fiber coupler 222 and the fiber coupler 224 to the spectrometer 124 at a frequency based on the time it takes to collect one line of data. The delay provided by the optical waveguide 226 allows the light provided from the fiber coupler 224 to represent the state of the sample at substantially the same time as the state of the sample represented by the light provided from the fiber coupler 222. Thus, the light received sequentially by the spectrometer 124 from the fiber couplers 222 and 224 may substantially correspond to the state of the sample 112 within the same time window, thereby facilitating imaging at greater depth penetration while limiting the effects of sample motion on imaging that may cause artifacts in sequentially acquired images. Note that herein, the output of the fiber coupler 222 is sometimes referred to as channel 1 and the output of the fiber coupler 224 is sometimes referred to as channel 2.

[0064] In some embodiments, the image sensor of the spectrometer 124 can be operated in a line trigger mode that allows acquisition of a line of data on each rising edge of a trigger signal (e.g., a TTL signal) or each falling edge of a trigger signal. In such an embodiment, another signal can be used to control the optical switch 230 such that data is acquired alternately from channel 1 and channel 2. For example, the signal controlling the optical switch 230 can have a frequency that is synchronized with the trigger signal, but half lower. In such an example, the optical switch control signal can rise or fall in synchronization with the trigger signal such that the optical switch signal changes at the same time that the trigger signal changes. This synchronized operation ensures that sequential acquisition of data from channel 1 and channel 2 captures information that corresponds in time.

[0065] FIG. 2B illustrates another example of a system 250 for multi-reference arm spectral domain optical coherence tomography according to some embodiments of the disclosed subject matter. As shown in FIG. 2B, the system 250 can include a reference arm embodied with a variable delay line 252 instead of a reference reflector to create a path length that substantially corresponds to the path length to the sample 112 along the sample arm. In such an embodiment, multiple optical components can be omitted, such as the optical circulators 210-2 and 210-3, the optical systems 216-2 and 216-3, and the reference reflectors 214-1 and 214-2. In some embodiments, one of the reference arms can include a variable component 254 to facilitate control of the path length of the reference arm. For example, the variable component 254 can be an air gap with an adjustable length, a stretchable section of fiber with an adjustable length, or the like.

[0066] FIG. 3 shows an example of an image generated using multiple reference arms of a multi-reference arm spectral domain optical coherence tomography system embodied in some embodiments of the disclosed subject matter. As described in more detail below, the acquired two-dimensional data matrix can include data from both channel 1 and channel 2 in alternating columns (or rows), and an OCT image can be generated from the acquired data by deinterleaving the data from each channel, removing background components from each channel, remapping the wavelength data to wavenumber data, performing a fast Fourier transform on the wavenumber data, and blending the images generated from each channel. For example, the odd lines of the data matrix can include a spectrum from channel 1, and the even lines of the data matrix include a spectrum from channel 2 delayed with an offset (e.g., about 3 millimeters (mm)). In such an example, the odd and even lines of the data matrix can be separated to create separate images corresponding to channel 1 and channel 2, which can be processed separately. In some embodiments, the average of all spectra (e.g., in each channel) can be subtracted from each spectrum of the corresponding data set for background removal. For example, the spectra from channel 1 can be averaged and this average value subtracted from each value in the data matrix that corresponds to channel 1. Note that averaging can be done using combined data, but losses in the two channels can be different, resulting in differences in background values.

[0067] In some embodiments, after background subtraction, each spectrum can be remapped from a linear wavelength domain to a linear wavenumber domain using a mapping function calibrated for the spectrometer, and in some embodiments, the remapped spectra can be fast Fourier transformed to generate a reflectance profile in the depth domain.

[0068] 3, image 302 is a cross-sectional OCT image of the sample corresponding to channel 1, and image 304 is a cross-sectional OCT image of the sample corresponding to channel 2. As noted above, due to the longer path length of the reference arm corresponding to channel 2, image 304 represents a portion of the sample deeper within the tissue than the portion represented in image 302 (although there is significant overlap). In addition, the delay in channel 2 after the light reflected from the sample is combined with the light reflected from the reference reflector allows the data in image 304 to represent substantially the same time window as that represented in image 302.

[0069] In general, SD-OCT systems have a larger sensitivity roll-off than swept source OCT (SS-OCT) systems, and signal intensity decays rapidly with optical delay in SD-OCT systems. This relatively rapid signal intensity decay can be seen in images 302 and 304. For example, in image 302, the signal intensity for tissue surfaces closer to the imaging probe surface (indicated by solid arrows pointing "up") is stronger than that of tissue surfaces at greater distances from the probe surface (indicated by solid arrows pointing "down"). However, in image 304, which has an additional optical delay offset, tissue surfaces located at greater distances from the imaging probe surface (indicated by asterisks) are relatively stronger than those at the same points in image 302. Note that when capturing the data used to generate image 302, the zero optical delay is set just before the inner surface of the imaging probe wall, which results in a separation in the mirror image due to the ambiguity between positive and negative delays in the SD-OCT data. Conversely, zero optical delay is set in the imaging probe surface when capturing the data used to generate image 304, and thus mirror images are overlapped in image 304. Thus, in some embodiments, information from image 302 can be used to remove mirror image artifacts from image 304 before the images are combined to generate image 306. Image 308 is a Cartesian version of image 302, and image 310 is a Cartesian version of image 306. As shown in FIG. 3, image 310 contains more information at greater depths.

[0070] FIG. 4 illustrates an example process 400 for generating spectral domain optical coherence tomography image data using multiple reference arms according to some embodiments of the disclosed subject matter. As shown in FIG. 4, in step 402, the process 400 can include setting a first reference arm zero optical delay to be near the surface of a sample (e.g., an in vivo tissue sample) to be imaged. For example, the zero optical delay can be set to be just outside the surface of the sample to prevent the formation of mirror artifacts in the OCT image generated using the first reference arm. In some embodiments, the path length of the first reference arm can be set using any suitable method or combination of methods. For example, in some embodiments, the path length of the first reference arm can be determined based on an expected distance between the probe (e.g., including the optical system 216-1) and the tissue surface. As another example, the path length of the first reference arm can be adjusted manually. In a more specific example, the path length can be set to an initial value, and an operator can adjust the path length until a relatively small separation between the mirror images is observed between the positive and negative delay images generated based on the first reference arm. As yet another example, the path length of the first reference arm can be automatically adjusted to be just before the surface of the sample using any suitable method or combination of methods, such as an active or passive range finding operation and analysis of data acquired using the first reference arm.

[0071] In step 404, the process 400 can include setting a second reference arm zero optical delay to be within the sample being imaged. For example, the zero optical delay can be set to be about half of the maximum imaging depth achievable with the first reference arm. Since the first reference arm can be expected to have zero optical delay at the tissue surface, the data will typically be taken to be about half of the maximum imaging depth achievable if the zero optical delay was set within the tissue. Only half the depth is collected (although this will result in an image with mirror image overlap). In some embodiments, the path length of the second reference arm can be set using any suitable method or combination of methods. For example, in some embodiments, the path length of the second reference arm can be determined based on the expected imaging depth of the first reference arm (e.g., the path length can be determined to be about 3 mm longer than the path length of the first reference arm). As another example, the path length of the second reference arm can be adjusted manually. In a more specific example, the path length can be set to an initial value, and an operator can adjust the path length until the combined image (and / or an image generated using only information generated using only the second reference arm) includes data from tissue depths that overlap a sufficient amount with the first reference arm, as well as data from depths that significantly exceed the maximum imaging depth of the first reference arm. As yet another example, the path length of the second reference arm can be automatically adjusted to provide the maximum imaging depth.

[0072] In step 406, process 400 can provide light to a sample arm, a first reference arm, and a second reference arm. For example, as described above with respect to FIG. 2A, light can be provided to the sample arm via a first fiber splitter (e.g., fiber splitter 208) and light can be provided to the first reference arm and the second reference arm via two fiber splitters (e.g., fiber splitter 208 and fiber splitter 220).

[0073] In step 408, the process 400 can activate an optical switch to mix the light reflected from the sample arm with the light reflected by the first reference arm.

[0074] In step 410, the process 400 determines a time t n and recording image data corresponding to a near-surface region of the sample. For example, as described above in connection with FIGURE 2A, in step 410, process 400 can control an image sensor (e.g., a linear CCD or linear CMOS sensor) to capture data indicative of the depth to which light is reflected by the sample based on interference between light in the sample arm and light in a first reference arm.

[0075] In step 412, the process 400 can activate an optical switch to mix the light reflected from the sample arm with the light reflected by the second reference arm. In some embodiments, the process 400 can control the optical switch to account for the time taken to acquire image data from the first channel and the delay in the second channel. For example, as described below in connection with FIG. 5, the image sensor can be activated with a trigger signal having a frequency that has approximately the same period as the time it takes a linear sensor to acquire one line of data.

[0076] In step 414, the process 400 determines a time t n Image data corresponding to an area within the sample can be recorded using For example, as described above in connection with FIG. 2A, in step 414, process 400 can control an image sensor (e.g., a linear CCD or linear CMOS sensor) to capture data indicative of the depth to which light is reflected by the sample based on interference between light in the sample arm and light in a second reference arm. In some embodiments, process 400 returns to step 408 to switch back to the first channel and in step 410, at time t n+1 Obtain the data corresponding to

[0077] 5 illustrates an example timing diagram 500 illustrating the collection of image data according to some embodiments of the disclosed subject matter. As shown in FIG. 5, the trigger signal used to activate the linear sensor of the spectrometer can have a first frequency. In the example shown in FIG. 5, the linear sensor , configured to capture data in response to detection of a rising edge of the trigger signal. The optical switch control signal may have a frequency that is half the frequency of the trigger signal to control the optical switch to output light from channel 1 and channel 2 during continuous data acquisition by the linear sensor. Note that in some cases, the acquisition time of the sensor may be a large fraction of the period of the trigger signal. For example, as shown in FIG. 5, if a readout is triggered by a rising edge, acquisition by the sensor may not be completed until after the corresponding falling edge of the trigger signal.

[0078] FIG. 6 illustrates an example process 600 for combining image data generated with different reference arms according to some embodiments of the disclosed subject matter. In step 602, the process 600 can receive interleaved image data corresponding to data received sequentially from channel 1 and channel 2 of an imaging system. In some embodiments, each column or row of the interleaved image data can correspond to an A-scan of the sample, with a pair of consecutive columns corresponding to A-scans of the same lateral position captured at the same time. As discussed above in relation to FIG. 2A, in some embodiments, one of the channels can be associated with a delay such that consecutive spectra represent the state of the sample within the same time window but have different sensitivities at different depths (e.g., the reference arm with the delay can be longer, which can facilitate improved sensitivity for larger depth values). In some embodiments, the interleaved image data can be formatted into a wavelength domain, in which each column or row corresponds to a depth value. Alternatively, in some embodiments, the interleaved image data can be received in another format, such as a wavenumber domain.

[0079] In some embodiments, the interleaved image data can be described as a data matrix representing M interference spectra (e.g., representing uniform spacing of spectra corresponding to a range of depths), each of the M interference spectra associated with N samples. For example, the interleaved image data can be represented as I(λ,t) M × N It can be expressed as:

[0080] In step 604, the process 600 can separate the odd and even rows (or columns) to generate separate data matrices corresponding to the information received over channel 1 and channel 2, respectively. For example, the matrix I(λ,t) M × N is a set of two matrices, I Ch1 (λ,t ) M / 2 × N and I Ch2 (λ,t) M / 2 × N Each of these can be separated into different references represents an image of samples corresponding to a signal.

[0081] In step 606, the process 600 calculates the matrix I Ch1 (λ,t) M / 2 × N For each value in Calculate the average (i.e., mean) value and uniformly subtract the mean from the data. For example, the process 600 can be used to remove background components from the channel 1 image by Ch1 (λ,t) M / 2 × N The adjusted row corresponding to Column I' Ch1 (λ,t) M / 2 × N can be generated, which can be expressed as do.

[0082]

number

[0083] In the formula, meanI Ch1 (λ,t)1× N is a column vector of dimension N, and each element is a matrix I Ch1 (λ,t) M / 2 × N is the average value of all elements in the row of , and rep{} is the average vector (meanI Ch1 (λ,t)1× N ) is repeatedly transformed into a matrix consisting of a sequence of mean vectors.

[0084] In step 608, the process 600 calculates the adjusted matrix I' Ch1 (λ,t) M / 2 × N Represented by The wavelengths obtained can be mapped to the wavenumber domain (also called the k-domain). For example, the central wavelength λ corresponding to each row can be calculated by N With wave number k N (Generally, the wave number corresponds to the inverse of the frequency, i.e., k N =1 / λ N (Note that the mapping is I' Ch1 (λ,t) M / 2 × N →I Ch1 (k,t) M / 2 × N In some embodiments, the mapping function can be calibrated for the spectrometer using linear interpolation.

[0085] In step 610, process 600 computes a discrete Fourier transform (DFT) of the k-domain data matrix corresponding to channel 1 to generate a spatial domain representation of the channel 1 image. For example, process 600 may compute a discrete Fourier transform (DFT) of the k-domain data matrix corresponding to channel 1 for various depth values ​​Z N An empty string representing a sample in Inter-domain data matrix I Ch1 (z,t) M / 2 × N of the k-domain data matrix to generate In a more specific example, the process 600 may perform a fast Fourier transform of the k-domain data matrix, which may be expressed as:

[0086]

number

[0087] At step 612, process 600 may partition the channel 1 image (e.g., a spatial domain data matrix) to separate the mirror image in the channel 1 image due to the ambiguity between positive and negative delays. For example, process 600 may further partition the channel 1 image using the first N / 2 columns (or rows) to represent one version of the mirror image and the next N / 2 columns (or rows) to represent another version of the mirror image. In a more specific example, process 600 may partition the channel 1 image using a data matrix I' Ch1 (z,t) M / 2 × N / 2 and I Ch1 (z,t) M / 2 × N / 2 The data matrix I' can be generated. Ch1 (z,t) M / 2 × N / 2 I Ch1 (z,t) M / 2 × N Based on the 1st to (N / 2)th columns of Then, the data matrix I Ch1 (z,t) M / 2 × N / 2 I Ch1 (z,t) M / 2 × N In some embodiments, process 600 can omit generating one of the mirror images if only one of the mirror images is used in subsequent processing.

[0088] In step 614, the process 600 calculates the matrix I Ch2 (λ,t) M / 2 × N The average of each value ( The background component can be removed from the channel 2 image by calculating the I (i.e., arithmetic mean) value and uniformly subtracting the mean from the data. For example, the process 600 can be used to obtain an I image with the background component removed. Ch2 (λ,t) M / 2 × N The adjusted matrix I' corresponding to Ch2 (λ,t) M / 2 × N can be generated, which can be expressed as follows:

[0089]

number

[0090] In the formula, meanI Ch2 (λ,t)1× N is a column vector of dimension N, and each element is a matrix I Ch2 (λ,t) M / 2 × N is the average value of all elements in the row of , and rep{} is the average vector (meanI Ch2 (λ,t)1× N ) is repeatedly transformed into a matrix consisting of a sequence of mean vectors.

[0091] In step 616, the process 600 calculates the adjusted matrix I' Ch2 (λ,t) M / 2 × N Represented by The wavelengths obtained can be mapped to the wavenumber domain (also called the k-domain). For example, the central wavelength λ corresponding to each row can be calculated by N With wave number k N (Generally, the wave number corresponds to the inverse of the frequency, i.e., k N =1 / λ N (Note that the mapping is I' Ch2 (λ,t) M / 2 × N →I Ch2 (k,t) M / 2 × NIn some embodiments, the mapping function can be calibrated for the spectrometer using linear interpolation.

[0092] In step 618, the process 600 computes a discrete Fourier transform (DFT) of the k-domain data matrix corresponding to channel 2 to generate a spatial domain representation of the channel 2 image. For example, the process 600 may compute a discrete Fourier transform (DFT) of the k-domain data matrix corresponding to channel 2 for various depth values ​​Z N An empty string representing a sample in Inter-domain data matrix I Ch2 (z,t) M / 2 × N of the k-domain data matrix to generate In a more specific example, the process 600 may perform a fast Fourier transform of the k-domain data matrix, which may be expressed as:

[0093]

number

[0094] In step 620, the process 600 uses one of the segmented images generated in step 612 and the image generated in step 618 to determine a depth offset Z between the two channels, which corresponds to the path length difference between the two reference arms. shift can be determined In some embodiments, the process 600 may include detecting the Z between two channels using any suitable method or combination of methods. shift For example, the process 600 can determine , a cross-correlation function can be used to determine which columns in the mirror image from channel 1 correspond to columns in the image from channel 2. In some embodiments, the spatial offset only needs to be calculated once since the offset typically does not change. However, in some embodiments, the offset may be recalculated if the path length of one of the reference arms changes and / or after a predetermined period of time and / or after a predetermined number of images have been captured (e.g., because path length changes over time due to environmental factors such as temperature).

[0095] In step 622, the process 600 calculates the spatial offset Z shift Zero pad the mirror image extracted in step 612 based on This allows the depth of the image extracted in step 612 to approximately match the maximum depth of the image generated from channel 2. In some embodiments, the image I generated from the channel 2 image based on an offset (e.g., as described below with respect to step 624) can be used to extract the image from the channel 2 image. Ch2 (z,t) M / 2 × N' To match the dimensions of I Ch1 (z,t) M / 2 × N / 2 →I Ch1 (z,t) M / 2 × N' It can be expressed as:

[0096] In step 624, process 600 may crop the image generated in step 618 based on the spatial offset to remove portions of the mirror image that are not represented in the image from channel 1. For example, process 600 may generate the following image:

[0097]

number

[0098] In step 626, the process 600 extracts the binarized portion I′ of the cropped image generated in step 624 that corresponds to the portion that overlaps with the zero-padded image from channel 1 generated in step 622. Ch2 (k,t) M / 2 × N / 2 Generate a mask B that represents In some embodiments, mask B can be generated based on a comparison of the mirror image of channel 1 and the main (e.g., positive delay) image from channel 2. For example, the value of each element in the mirror image of channel 1 can be compared to a threshold. If the value of the element is above the threshold, binary mask B can contain a 0 (zero) at that location, and if the value of the element is below (or equal to or less than) the threshold, binary mask B can contain a 1 at that location. In a more specific example, the threshold can be based on noise values ​​(e.g., mean and standard deviation) from areas with no signal. For example, at 706, the threshold can be determined using the portion of the matrix within zero delay and within the imaging probe surface that has no reflected signal.

[0099] In step 628, the process 600 may calculate combining coefficients that can be used to combine overlapping data from the data matrices representing channel 1 and channel 2. In some embodiments, the process 600 may calculate a weighted combining coefficient matrix W=Cm*B, where Cm is a combining coefficient. For example, Cm is a coefficient that can be expressed as Cm=(tanh(x)+1) / 2, where x ranges from -2π to 2π, and can be used to generate a vector of values ​​for multiplication with the mask B. Note, however, that this is merely an example and other functions, such as linear variation functions or step functions, can be used to generate the weighting coefficients.

[0100] In step 630, the process 600 uses the zero-padded mirror image (corresponding to channel 1) generated in step 622 and the cropped image generated in step 624 to derive a combined image Im(z,x) using the combination coefficients (e.g., based on the matrix W). M / 2 × N' In some embodiments, the process 600 may generate any suitable Any method or combination of methods can be used to combine the images from channel 1 and channel 2. For example, to generate a portion that is represented in both the channel 1 and channel 2 images, Im(z,x) can be calculated using the following formula: M / 2 × N' can be generated.

[0101]

number

[0102] In some embodiments, after generating the combined image, process 600 may cause the image to be stored (e.g., in a memory, cache, long-term storage, etc.) and / or displayed (e.g., using a display).

[0103] 7A shows example images illustrating various operations that may be performed during the combination of image data generated using different reference arms according to some embodiments of the disclosed subject matter. As shown in FIG. 7A, image 702 represents the odd columns (or rows) of the 2D data matrix output by the spectrometer, and image 704 represents the even columns (or rows) of the 2D data matrix output by the spectrometer, corresponding to channel 1 and channel 2 after background subtraction. However, this is merely an example and channel 2 may correspond to the even columns (or rows). The spatial difference between the channel 1 and channel 2 images can be seen by visually comparing images 702 and 704.

[0104] In Figure 7A, image 706 represents a mirror image of the tissue (e.g., rather than a real image of the tissue), which can be determined because the zero delay was set just outside the imaging probe wall, and therefore the negative delay space is mostly occupied by empty space (i.e., reflection from the negative delay side of the zero delay is minimal), whereas in Figure 7A, image 708 represents an image of the tissue, which can be determined because the zero delay was set just outside the imaging probe wall, and therefore the positive delay space has been disambiguated from the negative delay space.

[0105] Image 710 is a cropped version of image 704 that can be used to remove the overlapping portion of image 710 with image 706, while image 712 is a version of image 708 with multiple (blank) lines added to the bottom of the image corresponding to the spatial shift.

[0106] 7B shows further example images illustrating various operations that may be performed during the combination of image data generated with different reference arms according to some embodiments of the disclosed subject matter. In FIG. 7B, image 712 and image 710 are shown again for reference. As shown in FIG. 7B, a binary mask can be generated for the overlapping portion of image 710 with zero padded image 712 based on a mirror image of the tissue in image 706.

[0107] As shown in FIG. 7B, a combined image 724 generated by combining images 710 and 712 using a binary mask 722 and weighting factor Cm is more sensitive than the channel 1 image alone (corresponding to an image that can be generated using conventional SD-OCT methods) and is free of the mirror artifacts that appear when zero optical delay is set in the sample.

[0108] FIG. 7C illustrates an example of weighted combining coefficients that may be used to combine image data generated using different reference arms, according to some embodiments of the disclosed subject matter.

[0109] FIG. 8 shows examples of sensitivity roll-off observed for the first reference arm, the second reference arm, and the combination of the first and second reference arms of a spectral domain optical coherence tomography system embodied in some embodiments of the disclosed subject matter. As shown in FIG. 8, the measured maximum sensitivity of the exemplary system corresponding to each of the two reference delays was about (~) 105 dB. More specifically, FIG. 8 shows the depth-dependent sensitivity for the two reference arms individually (Ch.1 and Ch.2 curves) and for the combined data (combined-linear and combined-tanh, respectively). The latter two curves show the resulting sensitivity roll-off when using linear and (tanh+1) / 2 functions, respectively, to generate the combination coefficients. As shown, the sensitivity for the first arm is about 105 dB near the zero delay line, but attenuated by about 20 dB at the end of the imaging depth range (i.e., about 6 mm from zero optical delay for the exemplary system described below). The delay of the other reference arm was set to be about half the maximum imaging depth range (i.e., about 3 mm away from the zero delay line for the exemplary system described below), resulting in a sensitivity of about 98 dB for a scan at about 6 mm, with a peak sensitivity of about 105 dB at about 3 mm. The offset between the two reference arms was selected such that reference 1 provided better sensitivity (105 dB to 98 dB) for imaging depths of 0 to 3 mm, and reference 2 provided better sensitivity (105 dB to 98 dB) for imaging depths of 3 to 6 mm. Thus, in some embodiments, the mechanisms described herein can be used to provide sensitivity of 105 dB to 98 dB across imaging depths. It is possible to generate SD-OCT images in which the sensitivity is maintained at B. This results in an improvement in sensitivity of over 10 dB compared to conventional SD-OCT systems, where the sensitivity is attenuated from 105 dB to 85 dB (e.g., shown as Ch. 1 in Figure 8).

[0110] An exemplary system according to Example 200 was constructed to obtain exemplary sensitivity. The exemplary system includes a 3 milliwatt (mW) superluminescent diode (SLD) with a 3 dB bandwidth of about 100 nanometers (nm) and a central wavelength of about 1310 nm. A booster optical amplifier (BOA, 87 nm FWHM) was used to amplify the light from the SLD. Since the optical amplification by the BOA used is polarization sensitive, a polarization controller was used between the SLD and the BOA to optimize the optical amplification and its spectral profile. A 75 / 25 fiber splitter was used to split the light from the BOA into a sample arm and a reference arm. In the sample arm, an optical circulator was used to direct the light to the sample and to collect the backreflected light from the sample. The backreflected light from the sample was split using a 50 / 50 fiber splitter to direct half of the backreflected light to a first reference arm and the other half to a second reference arm. Another 50 / 50 fiber splitter was used to split the light in the reference arm, where the length of one reference arm was set to be at the zero delay point relative to the sample arm (i.e., the path length was matched to the sample arm). A linear translation stage and collimation assembly was used to set the path length of the second reference arm to a predetermined path length offset from the first reference arm by a path length of 3 mm (which is approximately half the width depth provided by the 2048 image line scan camera used in the spectrometer used in the exemplary system).

[0111] A 99 / 1 fiber coupler was used to combine the light returned from the evenly split sample and reference arms. 99% of the output from these couplers was sent to the spectrometer via a 2×1 optical switch such that the output of one 99 / 1 coupler was connected directly to input port 1 of the optical switch, and the output of the other 99 / 1 coupler was sent through a 1.5 km fiber delay line (FDL) before being connected to input port 2 of the optical switch. The output port of the optical switch was connected to the spectrometer (Wasatch Photonics Cobra super SWIR1310). The length of the FDL, 1.5 km, was chosen to introduce a time delay equal to the acquisition time of one line of data by a 140 kHz line scan camera (7.14 microseconds (μs)) (Sensors Unlimited GL2048R) used with the spectrometer. Based on the input signal to the optical switch, the optical switch passes light from channel 1 (Ch-1) while blocking light returning from channel 2 (Ch-2) and vice versa (e.g., as described above with respect to FIG. 2A). The isolation efficiency of the optical switch is about 20 dB, which can result in crosstalk interference detected in the OCT images. However, the FDL eliminates the detection of crosstalk interference because it makes the optical path length in one channel much longer than the optical path length at a certain depth of the system. Two composite TTL signals at 140 kHz and 70 kHz are used to initiate the detection of the linear array data (2048 images / line scan) and to drive the optical switch. This configuration allows the camera to capture alternating interference spectra resulting from the interference between the samples (within the same time window) and the two different reference arms. The spectra associated with the interference between the sample and each reference arm are in turn digitized and processed separately (see, e.g., the description in relation to FIG. 6 and FIG. 7 above). Processing steps include background subtraction, resampling, fast Fourier transformation and algorithmic combination of A-lines to generate a single OCT image with reduced sensitivity roll-off.

[0112] The final combined image is generated by combining Image 1 (the image corresponding to Ref-1 and Ch-1) with Image 2 (the image corresponding to Ref-2 and Ch-2). However, due to the ambiguity of positive and negative distances resulting from the Fourier transform of the real data, the imaging process Tissue within 3 mm of the lobe surface can produce a mirror image artifact in Image 2. The mirror artifact is mitigated in the combined image, and its removal when combining the two images produces a final image with increased sensitivity to greater depths and increases the imaging range of the system over conventional SD-OCT systems.

[0113] To demonstrate the ability of the mechanism described herein to obtain high quality image data at depth, excised porcine colon tissue was used. A tethered OCT capsule with a diameter of 11 mm was used as the scanning probe to generate the image data. The imaging capsule probe was placed within the lumen of an intact porcine colon and circumferential images of the intestinal wall were obtained. For comparison, images obtained with one reference arm and images obtained with two reference arms were also obtained.

[0114] Images 702 and 704 depict OCT images acquired by a capsule probe from excised porcine colon tissue along Ch1 and Ch2. Images 706 and 708 are images segmented from image 702 to separate real and mirror images. Image 722 is a binary mask obtained from image 706, which is used to remove the overlapping mirror artifact from image 704. The optical delay offset between images 702 and 704 was calculated using a cross-correlation scheme. Image 710 depicts a cropped version of image 704 showing only positive optical delays relative to Ch1 image 702. Image 712 shows a zero-padded version of image 708 to match the dimensions of image 710 before combining. FIG. 7C shows the weighted combining coefficients calculated from the binary mask 722 and the combining function (here, a (tanh+1) / 2 combining function was used to gradually change the weights for the combining coefficients from image 1 to image 2). Image 724 shows the final combined image. As shown in Figures 7A and 7B, the OCT image intensity in image 702 generated using only data from Ch-1 is stronger for tissues close to the surface of the imaging capsule probe, while the OCT image intensity from tissues far from the surface of the capsule is lower due to sensitivity roll-off. As expected, deeper regions in image 702 show very weak image intensity. In contrast, image 704 shows better signal intensity for tissues located at greater distances from the tissue surface due to the additional delay in Ch2. Combining these features using the mechanisms described herein to generate combined image 724, which shows a relatively strong overall sensitivity, facilitates improved visualization of the lumen wall at all depths within the maximum depth of the SD-OCT system. The improvement provided by the dual-reference SD-OCT system can also be clearly seen by comparing the scan-converted single-reference and dual-reference images (e.g., images 308 and 310).

[0115] FIG. 9 illustrates example hardware 900 that may be used to implement imaging and / or computing devices that may be used with some embodiments of the multi-reference arm spectral domain optical coherence tomography mechanism embodied by some embodiments of the disclosed subject matter. For example, the hardware illustrated in FIG. 9 may be used to implement at least a portion of the spectrometer 124. As illustrated in FIG. 9, in some embodiments, the imaging system 910 may include a hardware processor 912, a user interface and / or display 914, one or more communication systems 918, memory 920, one or more light sources 922, one or more electromagnetic radiation detectors 926, and / or one or more optical connectors 926. In some embodiments, the hardware processor 912 may be any suitable hardware processor or combination of processors, such as a central processing unit (CPU), a graphics processing unit (GPU), a microcontroller (MCU), a field programmable gate array (FPGA), a dedicated image processor, or the like. In some embodiments, the input(s) and / or display 914 may be a computer monitor, a touch screen, a touch screen, a touch panel, a touch screen ... It may include any suitable display device(s), such as a screen, television, transparent or translucent display, head mounted display, etc., and / or input devices and / or sensors capable of receiving user input, such as a keyboard, mouse, touch screen, microphone, eye tracking system, motion sensor, etc.

[0116] In some embodiments, communications system 918 may include any suitable hardware, firmware, and / or software for communicating information over communications network 902 and / or any other suitable communications network. For example, communications system 918 may include one or more transceivers, one or more communications chips and / or sets of chips, etc. In more specific examples, communications system 918 may include hardware, firmware, and / or software that may be used to establish a Wi-Fi® connection, a Bluetooth® connection, a cellular connection, an Ethernet® connection, an optical connection, etc.

[0117] In some embodiments, the communication network 902 may be any suitable communication network or combination of communication networks. For example, the communication network 902 may include a Wi-Fi network (which may include one or more wireless routers, one or more switches, etc.), a peer-to-peer network (e.g., a Bluetooth network), a cellular network (e.g., a 3G network, a 4G network, etc., conforming to any suitable standard, such as CDMA, GSM, LTE, LTE-Advanced, WiMAX, etc.), a wired network, etc. In some embodiments, the communication network 902 may be a local area network, a wide area network, a public network (e.g., the Internet), a private or semi-private network (e.g., a corporate or university intranet), any suitable type of network, or any suitable combination of networks. The communication links shown in FIG. 9 may each be any suitable communication link or combination of communication links, such as a wired link, an optical fiber link, a Wi-Fi link, a Bluetooth link, a cellular communication link, etc.

[0118] In some embodiments, the memory 920 may include any suitable storage device or devices usable to store instructions, values, etc. that may be used by the hardware processor 912 to process image data generated by the one or more photodetectors, for example, to communicate with the computing device 930 via the communication system(s) 918, to present content using the input / display(s) 914, etc. The memory 920 may include any suitable volatile memory, non-volatile memory, storage, any other suitable type of storage medium, or any suitable combination thereof. For example, the memory 920 may include RAM, ROM, EEPROM, one or more flash drives, one or more hard disks, one or more solid state drives, one or more optical drives, etc. In some embodiments, the memory 920 may have encoded thereon a computer program for controlling the operation of the imaging system 910. In some such embodiments, the hardware processor 912 may execute at least a portion of a computer program to control one or more light sources and / or detectors (e.g., to capture OCT data as described above with respect to FIG. 4), generate images and / or calculate values ​​(e.g., OCT images), send information to or receive information from the computing device 930, combine OCT images from different channels to generate a combined OCT image with improved sensitivity roll-off (e.g., as described above with respect to FIG. 6), etc.

[0119] In some embodiments, the imaging system 910 may include a broadband or narrowband light source. The optical fiber 922 may include one or more light sources 922, such as a coherent or incoherent light source (e.g., a light emitting diode or combination of light emitting diodes, a white light source, etc.). For example, the bandwidth of the light source may be selected to provide a wavelength range that facilitates detection at depths spanning the maximum imaging range of the SD-OCT system. Additionally, in some embodiments, the light source 922 may be associated with one or more fibers.

[0120] In some embodiments, the imaging system 910 can include one or more photodetectors 924, such as one or more photodiodes and / or one or more image sensors (e.g., a CCD image sensor or a CMOS image sensor, either of which may be a linear or two-dimensional array). For example, in some embodiments, the detector 924 can include one or more detectors configured to detect particular wavelengths of light (e.g., using filters, or using optics to direct light of different wavelengths to different portions of the detector(s), etc.).

[0121] In some embodiments, the imaging system 910 can include one or more optical connectors 926. For example, such optical connectors can be fiber optic connectors configured to form an optical connection between an optical fiber and the light source(s) 922 and / or detector 924 (e.g., as part of a fiber optic cable).

[0122] In some embodiments, the computing device 930 may include a hardware processor 932, a display 934, one or more inputs 936, one or more communication systems 938, and / or memory 940. In some embodiments, the hardware processor 932 may be any suitable hardware processor or combination of processors, such as a CPU, a GPU, an MCU, an FPGA, a dedicated image processor, etc. In some embodiments, the display 934 may include any suitable display device, such as a computer monitor, a touch screen, a television, a transparent or translucent display, a head mounted display, etc. In some embodiments, the input 936 may include any suitable input device and / or sensor that may be used to receive user input, such as a keyboard, a mouse, a touch screen, a microphone, an eye tracking system, a motion sensor, etc.

[0123] In some embodiments, communications system 938 may include any suitable hardware, firmware, and / or software for communicating information over communications network 902 and / or any other suitable communications network. For example, communications system 938 may include one or more transceivers, one or more communications chips and / or sets of chips, etc. In more specific examples, communications system 938 may include hardware, firmware, and / or software that may be used to establish a Wi-Fi® connection, a Bluetooth® connection, a cellular connection, an Ethernet® connection, etc.

[0124] In some embodiments, memory 940 may include any suitable storage device or devices usable to store instructions, values, etc. that may be used by hardware processor 932 to, for example, present content using display 934 to communicate with one or more imaging devices. Memory 940 may include any suitable volatile memory, non-volatile memory, storage, any other suitable type of storage medium, or any suitable combination thereof. For example, memory 940 may include RAM, ROM, EEPROM, one or more flash drives, one or more hard disks, one or more solid state drives, one or more optical drives, etc. In some embodiments, memory 940 may be encoded with a computer program for controlling the operation of computing device 930. In some such embodiments, hardware processor 932 may be configured to execute at least a portion of the computer program. may be performed to receive content (e.g., content including images) from one or more imaging devices (e.g., imaging device 910), combine OCT images from different channels to generate a combined OCT image with improved sensitivity roll-off (e.g., as described above with respect to FIG. 6), present the content (e.g., images and / or values), transmit the content to one or more other computing devices and / or imaging systems, etc.

[0125] In some embodiments, computing device 930 may be any suitable computing device, such as a general purpose computer or a specialized computer. For example, in some embodiments, computing device 930 may be a smartphone, a wearable computer, a tablet computer, a laptop computer, a personal computer, a server, etc. As another example, in some embodiments, computing device 930 may be a medical device, a system controller, etc.

[0126] In some embodiments, any suitable computer-readable medium may be used to store instructions for performing the functions and / or processes described herein. For example, in some embodiments, the computer-readable medium may be a transitory or non-transitory medium. For example, a non-transitory computer-readable medium may include media such as magnetic media (hard disks, floppy disks, etc.), optical media (compact disks, digital video disks, Blu-ray® disks, etc.), semiconductor media (RAM, flash memory, electrically programmable read only memory (EPROM), electrically erasable programmable read only memory (EEPROM), etc.), any other suitable media that is not ephemeral in transmission or lacks a permanent appearance, and / or any suitable tangible media. As another example, a transitory computer-readable medium may include media such as signals on a network, in a communication line, in a conductor, in an optical fiber, in a circuit, any other suitable media that is not ephemeral in transmission or lacks a permanent appearance, and / or any suitable intangible media.

[0127] Although the disclosed subject matter has been described in connection with particular embodiments and examples, those skilled in the art will recognize that the invention is not necessarily limited thereto, and that many other embodiments, examples, uses, modifications, and departures from the embodiments, examples, and uses are within the scope of the appended claims. Each patent document and publication cited in this specification is incorporated herein by reference as if each was individually incorporated herein by reference.

[0128] Various features and advantages of the invention are set forth in the following claims.

Claims

1. 1. A method for spectral domain optical coherence tomography, comprising: a first vector of data representative of a state of the sample in a first time window that includes a first time, the first vector being generated using a first reference arm having a first path length; a second vector of data representative of the state of the sample in the first time window generated using a second reference arm having a second path length; a third vector of data representative of states of the samples in a second time window not including the first time, generated using the first reference arm; a fourth vector of data representative of the state of the sample in the second time window generated using the second reference arm; and generating, by the processor, first image data based on the first vector of data and the third vector of data; generating, by the processor, second image data based on the second vector of data and the fourth vector of data; calculating, by the processor, a spatial offset between the first image data and the second image data based on a comparison of a portion of the first image data and a portion of the second image data; appending, by the processor, a first plurality of vectors to the portion of the first image data in a number based on the spatial offset to generate a zero padded image; removing, by the processor, a second plurality of vectors from the second image data in a number based on the spatial offset to generate a cropped image; combining, by the processor, the zero padded image and the cropped image to generate an image of the sample having reduced sensitivity roll-off.

2. 2. The method of claim 1, wherein the first vector is a first row of the matrix and the second vector is a second row of the matrix.

3. the first vector includes at least N elements; Each of the N elements corresponds to a pixel of the image sensor; The method of claim 2 , wherein the matrix includes at least N columns.

4. The method of claim 3 , wherein the spatial offset corresponds to a particular number of the N columns.

5. each entry in the first vector corresponds to a wavelength; The method of claim 1 , further comprising mapping the wavelength for each element to a wave number k.

6. The method of claim 5 further comprising performing a discrete Fourier transform to convert each element of the first vector into a depth value.

7. 2. The method of claim 1, further comprising the step of generating a real image of the sample based on the first image data by further dividing the matrix into two sub-matrices by selecting vectors on the positive side of zero delay for inclusion in a first sub-matrix and selecting vectors on the negative side of the zero delay for inclusion in a second sub-matrix.

8. The method of claim 7 , wherein the portion of the first image data corresponds to the first sub-matrix.

9. 8. The method of claim 7, further comprising generating a binary mask based on a comparison of the second sub-matrix and a shifted sub-matrix including values ​​generated using the second reference arm.

10. The method of claim 9 , further comprising: generating a weighting matrix W comprising a plurality of weighting coefficients C m based on the binary mask and the combination function.

11. 11. The method of claim 10, wherein the combination function is Cm=(tanh(x)+1) / 2, where x is a value in the range of -2π to 2π.

12. the first vector and the second vector correspond to a first lateral position on a surface of the sample; The method of claim 1 , wherein the third vector and the fourth vector correspond to a second lateral direction on the surface of the sample.

Images