Ultra-deep three-dimensional imaging device and method based on swept frequency optical coherence tomography in turbid media

Abstract

The present invention discloses an ultra-deep three-dimensional imaging device and method based on swept frequency optical coherence tomography in turbid media. The present invention uses a three-dimensional scattered field rapid detection method that combines parallel detection in the depth direction of the sample with scanning detection of the lateral field distribution to obtain a depth-resolved reflection matrix and provide the depth range required for ultra-deep imaging. The present invention obtains and compensates for the inherent intra-frame phase fluctuations of scanning detection by introducing a quasi-common-path reference arm into the system, and uses a matrix method that is insensitive to residual inter-frame phase drift to achieve image reconstruction. It is proved that the remaining uncompensated inter-frame DC drift and the uniform phase distribution error have no effect on image reconstruction. The total measurement time of the ultra-deep three-dimensional imaging (125 microns × 125 microns × 4 mm) achieved by the present invention is about 5 minutes, which is a significant leap compared with the time domain matrix method.

Description

Technical Field

[0001] The present invention belongs to the field of ultra-depth imaging in strong scattering media, and in particular relates to an ultra-depth three-dimensional imaging device and method based on frequency-sweeping optical coherence tomography technology in turbid media. Background Art

[0002] Light scattering is not only an information source for optical imaging of biological tissues, but also a limiting factor for optical imaging of deep tissues. Biological tissues can be described as non-uniform strongly scattering media, in which light propagating will experience light scattering caused by spatial fluctuations in the refractive index. Unfortunately, as the depth of light propagation increases, the number of single scattered photons decays rapidly in an exponential trend, while the number of multiply scattered photons increases continuously with depth. The target information carried by the single scattered photons returning from the sample is difficult to recover, making the penetration depth achievable by traditional imaging techniques based on single scattered photons extremely shallow. The current conventional focusing and imaging methods under the Born approximation are only effective under transparent or weak scattering conditions and cannot be applied to focusing and imaging under strong scattering conditions. Therefore, the problem of light propagation and imaging in biological tissues, especially the problem of deep tissue optical imaging, is a basic problem and scientific problem that needs to be solved urgently.

[0003] In order to achieve deep high-resolution imaging of biological tissues, many scholars have proposed effective solutions, such as using long-wavelength imaging or tissue light transparency technology to reduce the scattering coefficient of biological tissues and reduce multiple scattering of light in biological tissues; or using time-reversal-based wavefront shaping technology to form an inverse diffusion wavefront to counteract the effects of multiple scattering and distortion on beam focusing and imaging. So far, the most mature method to eliminate multiple scattered light is optical coherence tomography (OCT). As a non-invasive, non-contact, label-free, low-cost real-time optical imaging method, OCT uses a combination of spatial gate and time gate gating technologies to effectively separate single scattered light from multiple scattering backgrounds and construct high-resolution images based on single scattered light. However, due to the influence of strong scattering in biological tissues, the imaging depth of OCT in biological tissues is still limited to about 1-2 mm. In order to further give play to the advantages of OCT, its imaging depth limit needs to be broken through.

[0004] Alexandre Aubry's team at the Langevin Institute in France first proposed a time-domain OCT system based on a reflection matrix, which is expected to achieve at least a 2-fold imaging depth expansion. Its basic mechanism for reducing multi-scattered photons and enhancing single-scattered photons includes wide-field detection of complex scattered fields and image reconstruction based on iterative time reversal. In the matrix OCT method, it is crucial to accurately and quickly measure the backscattered complex amplitude field. The phase shift method is a common method to obtain the complex amplitude field, but phase shifting by mechanical motion is time-consuming and unstable. In order to improve the measurement speed of the reflection matrix, Chen Zhongping's team proposed a wide-field heterodyne detection method based on an acousto-optic modulator and a phase-locked camera, but the complexity and cost of the system are increased. It is well known that swept-frequency OCT (SS-OCT) performs better than time-domain OCT. However, the matrix OCT schemes reported so far are all based on time-domain OCT systems, which can only measure the reflection matrix corresponding to one depth at a time and then use it for two-dimensional image reconstruction. Three-dimensional (3D) ultra-deep OCT imaging has not yet been confirmed. Summary of the invention

[0005] The present invention improves upon the deficiencies of the prior art and provides an ultra-deep three-dimensional imaging device and method based on swept frequency optical coherence tomography in turbid media. The present invention is achieved through the following technical solutions:

[0006] An ultra-deep three-dimensional imaging device based on frequency sweeping optical coherence tomography in a turbid medium comprises a high-speed frequency sweeping light source, a Mach-Zehnder interferometer clock box, a coupler A, an interferometer including a sample arm and a reference arm, a coupler B, a balanced detector and a computer for processing data, which are connected in sequence; the reference arm comprises a collimating lens A and a focusing lens A; the sample arm comprises a collimating lens B, a beam splitter, a focusing lens B, a sample, which are connected in sequence, and a focusing lens C and a reference reflector which are connected in sequence on one side of the beam splitter, and a biaxial galvanometer A and a collimating lens C which are connected in sequence on the other side; the input end of the sample arm is connected to the coupler A through the collimating lens B, and the output end is connected to the coupler B through the collimating lens C, and the sample arm also comprises a biaxial galvanometer B arranged between the collimating lens B and the beam splitter, or also comprises a biaxial displacement platform arranged on the bottom surface of the sample to support the sample.

[0007] As a further improvement, the two-axis displacement platform described in the present invention is used for sample movement to achieve focused light scanning, or the two-axis galvanometer B is arranged at the input end of the sample arm to scan the incident light.

[0008] As a further improvement, the reference reflector described in the present invention is placed at the defocused position of the focusing lens C at its front end, and ensures that the position depth of the reference reflector and the position depth of the upper surface of the sample form a similar but distinguishable quasi-common path optical path.

[0009] As a further improvement, the effective scanning range of the dual-axis galvanometer A described in the present invention needs to be consistent with the range of the sample imaging area.

[0010] The present invention also discloses an ultra-deep three-dimensional imaging method based on an ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography technology in a turbid medium, comprising:

[0011] Use a two-axis displacement platform to move the sample or use the two-axis galvanometer B at the input end of the sample arm to scan the incident light, so as to achieve traversal scanning of each position of the sample by the focused light;

[0012] At each focused light scanning position, the interference spectrum signal of the corresponding three-dimensional backscattering field is detected and recorded using a dual-axis galvanometer A scan;

[0013] Performing inverse Fourier transform on the detected spectral signal to obtain the complex amplitude signal of the three-dimensional backscattered field from the sample and the reference reflector;

[0014] According to the clock signal output by the clock box of the Mach-Zehnder interferometer, the phase error caused by the wavelength change between different axial scanning detections is calculated and compensated to the corresponding axial scanning signal in the three-dimensional backscattering field;

[0015] According to the complex amplitude field of the reference reflector obtained by multiple measurements (corresponding to the focused light scanning different positions), the inter-frame variation of its light field distribution is calculated, and further compensated to the corresponding three-dimensional backscattering field;

[0016] After the phase compensation operation is completed, there is still some uncompensated phase left, but it does not affect the subsequent image reconstruction;

[0017] Rearrange the two-dimensional fields at the same depth in all phase-compensated three-dimensional backscattered fields into column vectors, and arrange the column vectors in a reflection matrix according to a scanning trajectory, thereby forming a reflection matrix at a specified depth;

[0018] The above reconstruction process is performed on the two-dimensional fields at different depths in the three-dimensional backscattered complex field to construct a depth-resolved reflection matrix;

[0019] A digital pinhole is used to filter out most of the multi-scattering contribution from the reflectance matrix at each depth, followed by singular value decomposition and 2D image reconstruction;

[0020] In image reconstruction, in order to determine the required number of main singular values, correlation analysis is performed on image pairs consisting of images reconstructed with different numbers of singular values ​​and images reconstructed with corresponding remaining singular values ​​to determine the optimal number of main singular values ​​and achieve optimal reconstruction of two-dimensional images;

[0021] Finally, the best reconstructed two-dimensional images obtained at each depth are arranged according to depth to obtain a successfully restored three-dimensional target, thus achieving ultra-depth imaging in turbid media.

[0022] As a further improvement, the phase error described in the present invention includes: the phase error caused by various interference factors in the device, including wavelength offset between axial scans, mechanical jitter of the dual-axis galvanometer A, and optical path mismatch, etc. According to the different rates of change, the phase error can be divided into intra-frame phase fluctuation and inter-frame phase drift: Where x and y are the orthogonal coordinates of the backscattered field, z is the depth corresponding to the sample signal, Scanning location for sample;

[0023] Depending on the sample scanning method, when the device uses a dual-axis galvanometer B to scan the incident light, the incident light will experience spatially varying aberrations, thereby causing additional intra-frame phase fluctuations; when the device uses a displacement platform to move the sample, the incident light is fixed and will not experience spatially varying aberrations.

[0024] As a further improvement, the present invention calculates the phase error caused by wavelength change between different axial scanning detections based on the clock signal output by the Mach-Zehnder interferometer clock box, and compensates it to the corresponding axial scanning signal in the three-dimensional backscattering field, including:

[0025] Calculate the Hilbert transformed phase difference of the reference signal output by the Mach-Zehnder interferometer clock box and perform the first phase compensation operation:

[0026]

[0027] Among them, z MZI is the clock signal corresponding to the depth, is the Hilbert phase difference between the interferometric signals generated by the Mach-Zehnder interferometer clock boxes, and They are the phase distribution of the original sample signal and the phase distribution after the first phase compensation.

[0028] As a further improvement, the present invention calculates the inter-frame variation of the light field distribution of the reference reflector according to the complex amplitude field of the reference reflector obtained by multiple measurements (corresponding to the focused light scanning different positions), and further compensates to the corresponding three-dimensional backscattering field, including:

[0029] Calculate the inter-frame phase difference of the light field of the reference mirror The purpose of compensating the phase distribution of each frame of the sample signal is to unify the inconsistent scanning errors between all fields of the sample signal into the same phase distribution error. The phase compensation process is described as:

[0030]

[0031] Among them, z R is the depth corresponding to the reference mirror signal, E R By E R =E R0 / |E R0 |Normalized to get, E R0 is the backscattered light field from the reference reflector, It is the phase distribution of the sample signal after phase compensation is finally completed.

[0032] As a further improvement, after the phase compensation operation of the present invention is completed, there is still an uncompensated phase remaining, but it does not affect the subsequent image reconstruction, including: the residual phase error is expressed as: in The specific phase distribution error introduced (simplified as ), is the inter-frame phase drift (DC quantity independent of background distribution), at this time, the reflection matrix R e It can be expressed as:

[0033]

[0034] Where R is the ideal reflection matrix without phase error, D is the error matrix composed of the phase distribution error of the reference frame, and E is the error matrix composed of the phase fluctuation between frames. The decomposition of the reflection matrix and the reconstruction process of the image are as follows:

[0035]

[0036] are all time reversal operators; the singular value decomposition of the matrix R can be expressed as represents conjugate transpose; U and V are unitary matrices; Σ is a diagonal matrix consisting of singular values ​​(positive real numbers) σ i (i{i∈[1,N]}) are arranged in descending order; the matrix U d =DU means that the matrix U is weighted by a phase-only diagonal matrix D. Represents the matrix V as a phase-only diagonal matrix Weighting; and is the column vector U of the matrix U and V i 、V iThe reshaped two-dimensional wavefront; M (M < N) is the optimized number of singular values for reconstruction. Therefore, the background phase change independent of the distribution has no impact on image reconstruction, that is, the image reconstruction is insensitive to the inter-frame phase drift; the unified phase distribution error has no impact on image reconstruction; the post-processing of the reflection matrix composed of the compensated complex scattering field and the image reconstruction are not affected by the device scanning detection.

[0037] Based on the above technical solutions, the beneficial effects of the present invention compared with the prior art are as follows:

[0038] 1. Compared with the existing technology, the present invention uses a method that combines parallel detection in the depth direction of the sample and scanning detection of the transverse field distribution to achieve rapid detection of the three-dimensional scattering field, making the measurement of the three-dimensional reflection matrix and the three-dimensional ultra-depth imaging process more efficient and accurate. The present invention adopts a scanning detection mechanism, and a high-speed swept-frequency light source and a matching high-speed point detector are used in the system to provide a sufficient depth range for ultra-depth imaging.

[0039] 2. Compared with the existing reflection matrix measurement technology, the measurement system provided by the present invention has a simple setting and low cost: only the scanning detection for the transverse field distribution and the measurement of the reference signal for phase compensation need to be added on the basis of the traditional swept-frequency optical tomography system; the system can be directly switched to the traditional swept-frequency optical tomography system by fixing the position of the biaxial galvanometer at the detection end of the sample arm at the center.

[0040] 3. The method provided by the present invention comprehensively studies for the first time the influence of the phase interference caused by scanning detection during the measurement process of the backscattering field on the optical coherence tomography imaging method based on matrix measurement, including intra-frame and inter-frame phase fluctuations. The present invention introduces a Mach-Zehnder interferometer in the system to generate a clock signal, and a reference mirror is placed at the other end of the beam splitter in the sample arm (utilizing half of the incident light that was originally wasted) to generate a quasi-common-path reference signal to jointly compensate for the intra-frame phase fluctuations; after compensation, the phase errors with inconsistent distributions among all fields are unified into a same specific distribution error. This method proves that the matrix method is insensitive to the remaining inter-frame phase drift and the unified phase distribution error, and the image reconstruction is not affected by the system scanning detection.

[0041] 4. The present invention realizes three-dimensional ultra-depth optical coherence tomography imaging in turbid media for the first time. The total measurement time of the ultra-depth three-dimensional imaging (125 μm × 125 μm × 4 mm) system realized by the present invention is about 5 minutes, which is a significant leap compared with the time-domain matrix OCT, and the latter needs 6 minutes to complete the two-dimensional target imaging of a specified depth only. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 is a system diagram of an ultra-depth three-dimensional imaging device based on swept-frequency optical coherence tomography technology of the present invention;

[0043] Figure 2 It is a processing flow chart of an ultra-deep three-dimensional imaging method based on swept frequency optical coherence tomography in turbid media of the present invention;

[0044] Figure 3 It is a schematic diagram of the phase compensation result in the present invention, taking the reflector as a sample and continuously measuring its backscattering field; the first three columns in Figure a are the phase distributions of the scattered fields at three different sample positions, and the last two columns are the phase differences to indicate whether the phase error is fully compensated; the uncompensated phase in Figure b is the inter-frame phase, which increases with the increase of the measurement time;

[0045] Figure 4 It is a result display diagram of the present invention for ultra-depth imaging in scattering media; the second and third columns in Figure a are the reconstruction results obtained by traditional SS-OCT and matrix SS-OCT respectively; Figure b is an enlarged view of the three-dimensional structure of the last three microspheres (labeled as 8, 9, and 10) near the maximum depth shown by the dotted rectangle in Figure a; Figure c is a two-dimensional image of the three microspheres and a normalized intensity comparison diagram of the light intensity distribution at their center of mass.

[0046] Figure 1 In: 1. High-speed swept light source, 2. Mach-Zehnder interferometer clock box, 3. Coupler A, 4. Collimating lens A, 5. Focusing lens A, 6. Polarization controller, 7. Collimating lens B, 8. Beam splitter, 9. Focusing lens B, 10. Sample, 11. Focusing lens C; 12. Reference reflector, 13. Dual-axis galvanometer A, 14. Collimating lens C, 15. Two-axis displacement platform, 16. Coupler B, 17. Balanced detector, 18. Computer, 19. Dual-axis galvanometer B. DETAILED DESCRIPTION

[0047] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0048] Figure 1The present invention is a system diagram of an ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography technology; an ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography technology in a turbid medium, comprising: a high-speed swept frequency light source (1), a Mach-Zehnder interferometer clock box (2), a coupler A (3), an interferometer including a sample arm and a reference arm, a coupler B (16), a balanced detector (17) and a computer (18) for processing data, which are connected in sequence; the reference arm comprises a collimating lens A (4) and a focusing lens A (5); the sample arm comprises a collimating lens B (7), a beam splitter ( 8), focusing lens B (9), sample (10), and focusing lens C (11) and reference reflector (12) connected in sequence on one side of the beam splitter (8), and biaxial galvanometer A (13) and collimating lens C (14) connected in sequence on the other side; the input end of the sample arm is connected to coupler A (3) through collimating lens B (7), and the output end is connected to coupler B (16) through collimating lens C (14); the sample arm also includes biaxial galvanometer B (19) arranged between collimating lens B (7) and beam splitter (8), or also includes biaxial displacement platform (15) arranged on the bottom surface of the sample to support the sample.

[0049] In the sample arm, incident light is divided into a transmitted light beam and a reflected light beam after passing through a beam splitter (8), and the focused light is formed by a focusing lens B (9) and a focusing lens C (11) and focused onto a sample (15) and a reference reflector (12) respectively; the reference reflector (12) is placed at a defocused position of the focusing lens C (11) at its front end, and the position depth of the reference reflector (12) and the position depth of the upper surface of the sample are ensured to form a quasi-common path optical path that is similar but distinguishable; a two-axis displacement platform (15) is used for sample movement to realize focused light scanning, or the two-axis galvanometer B (19) is arranged at the input end of the sample arm to scan the incident light. In the detection optical path of the sample arm, a dual-axis galvanometer A (13) is used to scan and detect the backscattered field from the sample and the reference reflector, and its effective scanning range must be consistent with the sample imaging area range; the returned light from the sample arm and the reference arm interferes in the coupler B (16) and is input to the balanced detector (17); the computer synchronously collects the signal from the Mach-Zehnder interferometer clock box (2) and the interference spectrum data from the balanced detector (17), and finally displays the three-dimensional structure of the object at ultra-depth in the turbid medium after data processing.

[0050] Figure 2 The present invention is a processing flow chart of an ultra-deep three-dimensional imaging method based on swept frequency optical coherence tomography in a turbid medium; the present invention also discloses an ultra-deep three-dimensional imaging method of an ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in a turbid medium, and the specific steps are as follows:

[0051] 1. Use a two-axis displacement platform to move the sample or use the two-axis galvanometer B at the input end of the sample arm to scan the incident light, so as to achieve traversal scanning of each position of the sample by the focused light;

[0052] 2. At each focused light scanning position, the dual-axis galvanometer A-scan is used to detect and record the interference spectrum signal of the corresponding three-dimensional backscattering field;

[0053] 3. The device has measurement phase errors caused by various interference factors, including wavelength offset between axial scans, mechanical jitter of the dual-axis galvanometer A, and optical path mismatch. According to the different rates of change, the phase error can be divided into intra-frame phase fluctuation and inter-frame phase drift: Where x and y are the orthogonal coordinates of the backscattered field, z is the depth corresponding to the sample signal, is the sample scanning position; depending on the sample scanning method, when the device adopts the dual-axis galvanometer B to scan the incident light, the incident light will experience spatially varying aberrations, thereby causing additional intra-frame phase fluctuations; when the device adopts a displacement platform to move the sample, the incident light is fixed and will not experience spatially varying aberrations.

[0054] 4. Perform inverse Fourier transform on the detected spectral signal to obtain the complex amplitude signal of the three-dimensional backscattered field from the sample and the reference reflector;

[0055] 5. In order to compensate for the measurement phase error, firstly, the phase error caused by the wavelength change between different axial scanning detections is calculated according to the clock signal output by the Mach-Zehnder interferometer clock box, and then compensated to the corresponding axial scanning signal in the three-dimensional backscattering field:

[0056]

[0057] Among them, z MZI is the clock signal corresponding to the depth, is the Hilbert phase difference between the interferometric signals generated by the Mach-Zehnder interferometer clock boxes, and They are the phase distribution of the original sample signal and the phase distribution after the first phase compensation.

[0058] 6. Based on the complex amplitude field of the reference reflector obtained by multiple measurements (corresponding to the focused light scanning different positions), calculate the inter-frame changes of its light field distribution, and further compensate to the corresponding three-dimensional backscattering field, including: calculating the inter-frame phase difference of the light field of the reference reflector The purpose of compensating the phase distribution of each frame of the sample signal is to unify the inconsistent scanning errors between all fields of the sample signal into the same phase distribution error. The phase compensation process is described as:

[0059]

[0060] Among them, z R is the depth corresponding to the reference mirror signal, E R By E R =E R0 / |E R0 |Normalized to get, E R0 is the backscattered light field from the reference reflector, It is the phase distribution of the sample signal after phase compensation is finally completed.

[0061] 7. Figure 3 is a schematic diagram of the phase compensation result in the present invention, showing the distribution change of the backscattered field after the above two-step phase compensation operation: the intra-frame phase error has been successfully compensated, but the inter-frame phase drift and an introduced specific phase distribution error remain, such as Figure 3 As shown in b.

[0062] 8. The phase error remaining after the phase compensation operation is completed is expressed as: in The specific phase distribution error introduced (simplified as ), is the inter-frame phase drift (DC quantity independent of background distribution), at this time, the reflection matrix R e It can be expressed as:

[0063]

[0064] Where R is the ideal reflection matrix without phase error, D is the error matrix composed of the phase distribution error of the reference frame, and E is the error matrix composed of the phase fluctuation between frames. The decomposition of the reflection matrix and the reconstruction process of the image are as follows:

[0065]

[0066] and are all time reversal operators; the singular value decomposition of the matrix R can be expressed as represents conjugate transpose; U and V are unitary matrices; Σ is a diagonal matrix consisting of singular values ​​(positive real numbers) σ i (i{i∈[1,N]}) are arranged in descending order; the matrix U d =DU means that the matrix U is weighted by a phase-only diagonal matrix D. Represents the matrix V as a phase-only diagonal matrix Weighting; and is the column vector U of the matrix U and V i 、V iThe reshaped two-dimensional wavefront; M (M < N) is the optimized number of singular values for reconstruction. Therefore, the background phase change independent of the distribution has no effect on image reconstruction, that is, the image reconstruction is insensitive to the inter-frame phase drift; the uniform phase distribution error has no effect on image reconstruction; the post-processing of the reflection matrix and image reconstruction composed of the compensated complex scattered field are not affected by the device scanning detection.

[0067] 8. Rearrange the two-dimensional fields at the same depth in the three-dimensional backscattered field after all phase compensations into column vectors, and arrange the column vectors in the reflection matrix according to the scanning trajectory, thereby forming a reflection matrix at a specified depth;

[0068] 9. Perform the above reshaping process on the two-dimensional fields at different depths in the three-dimensional backscattered complex field to construct a depth-resolved reflection matrix;

[0069] 10. Use a digital pinhole to filter out most of the multiple scattering contributions for the reflection matrix at each depth, and then perform singular value decomposition and two-dimensional image reconstruction;

[0070] 11. In image reconstruction, to determine the required number of dominant singular values, perform correlation analysis on the image pair composed of the images reconstructed with different numbers of singular values and the images reconstructed with the corresponding remaining singular values to determine the optimal number of dominant singular values and achieve the optimal reconstruction of the two-dimensional image;

[0071] 12. Arrange the optimally reconstructed two-dimensional images obtained at each depth according to the depth to obtain the successfully restored three-dimensional target and achieve super-depth imaging in the turbid medium.

[0072] Figure 4 It is the result display diagram of the present invention for super-depth imaging in a scattering medium; it can be seen from the results in the figure that the traditional SS-OCT method can only reconstruct microspheres at a relatively shallow depth; when the imaging depth increases (exceeding the corresponding depth of the microsphere marked as 1), due to multiple scattering of light in the turbid medium, the three-dimensional distribution structure of the microspheres becomes blurred and unrecognizable. In contrast, the matrix SS-OCT method successfully restores the three-dimensional structure distribution of the microspheres and maximally suppresses the background noise. It is found by comparison that the matrix SS-OCT system expands the imaging depth by three times compared with the traditional SS-OCT system and achieves the reconstruction of the three-dimensional structure with super depth.

[0073] The above is only a preferred embodiment of the present invention. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Any technician familiar with the art can make many possible changes and modifications to the technical solution of the present invention by using the above disclosed methods and technical contents without departing from the scope of the technical solution of the present invention, or modify it into an equivalent embodiment of equivalent changes. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention still falls within the scope of protection of the technical solution of the present invention.

Claims

1. An ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media, characterized in that: It includes a high-speed frequency sweeping light source, a Mach-Zehnder interferometer clock box, a coupler A, an interferometer including a sample arm and a reference arm, a coupler B, a balanced detector and a computer for processing data, which are connected in sequence; the reference arm includes a collimating lens A and a focusing lens A; the sample arm includes a collimating lens B, a beam splitter, a focusing lens B, a sample, which are connected in sequence, and a focusing lens C and a reference reflector which are connected in sequence on one side of the beam splitter, and a biaxial galvanometer A and a collimating lens C which are connected in sequence on the other side; the input end of the sample arm is connected to the coupler A through the collimating lens B, and the output end is connected to the coupler B through the collimating lens C, and the sample arm also includes a biaxial galvanometer B arranged between the collimating lens B and the beam splitter, or also includes a biaxial displacement platform arranged on the bottom surface of the sample to support the sample; The ultra-deep three-dimensional imaging device based on the swept frequency optical coherence tomography technology in turbid media is used for the ultra-deep three-dimensional imaging method, comprising: Use a two-axis displacement platform to move the sample or use the two-axis galvanometer B at the input end of the sample arm to scan the incident light, so as to achieve traversal scanning of each position of the sample by the focused light; At each focused light scanning position, the interference spectrum signal of the corresponding three-dimensional backscattering field is detected and recorded using a dual-axis galvanometer A scan; Performing inverse Fourier transform on the detected spectral signal to obtain the complex amplitude signal of the three-dimensional backscattered field from the sample and the reference reflector; According to the clock signal output by the clock box of the Mach-Zehnder interferometer, the phase error caused by the wavelength change between different axial scanning detections is calculated and compensated to the corresponding axial scanning signal in the three-dimensional backscattering field; Based on multiple measurements, i.e., the complex amplitude field of the reference reflector obtained when the focused light scans different positions, the inter-frame variation of its light field distribution is calculated, and further compensated to the corresponding three-dimensional backscattering field; After the phase compensation operation is completed, there is still some uncompensated phase left, but it does not affect the subsequent image reconstruction; Rearrange the two-dimensional fields at the same depth in all phase-compensated three-dimensional backscattered fields into column vectors, and arrange the column vectors in a reflection matrix according to a scanning trajectory, thereby forming a reflection matrix at a specified depth; The two-dimensional fields at different depths in the three-dimensional backscattered complex amplitude field are reshaped in the previous step to construct a depth-resolved reflection matrix; A digital pinhole is used to filter out most of the multi-scattering contribution from the reflectance matrix at each depth, followed by singular value decomposition and 2D image reconstruction; In image reconstruction, in order to determine the required number of main singular values, correlation analysis is performed on image pairs consisting of images reconstructed with different numbers of singular values ​​and images reconstructed with corresponding remaining singular values ​​to determine the optimal number of main singular values ​​and achieve optimal reconstruction of two-dimensional images; Finally, the best reconstructed two-dimensional images obtained at each depth are arranged according to depth to obtain a successfully restored three-dimensional target, thus achieving ultra-depth imaging in turbid media.

2. The ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media according to claim 1, characterized in that: The two-axis displacement platform is used for sample movement to realize focused light scanning, or the two-axis galvanometer B is arranged at the input end of the sample arm to scan the incident light.

3. The ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media according to claim 1 or 2, characterized in that: The reference reflector is placed at the defocused position of the focusing lens C at its front end, and the position depth of the reference reflector and the position depth of the upper surface of the sample are ensured to form a quasi-common path optical path that is similar but distinguishable.

4. The ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media according to claim 3, characterized in that: The effective scanning range of the dual-axis galvanometer A must be consistent with the imaging area of ​​the sample.

5. The ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media according to claim 1, characterized in that: The phase error includes: The device has phase errors caused by various interference factors, including wavelength offset between axial scans, mechanical jitter of the dual-axis galvanometer A, and optical path mismatch. According to the different rates of change, the phase error can be divided into intra-frame phase fluctuation and inter-frame phase drift: Where x and y are the orthogonal coordinates of the backscattered field, z is the depth corresponding to the sample signal, Scanning location for sample; Depending on the sample scanning method, when the device uses a dual-axis galvanometer B to scan the incident light, the incident light will experience spatially varying aberrations, thereby causing additional intra-frame phase fluctuations; when the device uses a displacement platform to move the sample, the incident light is fixed and will not experience spatially varying aberrations.

6. The ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media according to claim 1, characterized in that: According to the clock signal output by the Mach-Zehnder interferometer clock box, the phase error caused by the wavelength change between different axial scanning detections is calculated, and compensated to the corresponding axial scanning signal in the three-dimensional backscattering field, including: calculating the Hilbert transformed phase difference of the reference signal output by the Mach-Zehnder interferometer clock box, and performing the first phase compensation operation: Among them, z MZI is the clock signal corresponding to the depth, is the Hilbert phase difference between the interferometric signals generated by the Mach-Zehnder interferometer clock boxes, and They are the phase distribution of the original sample signal and the phase distribution after the first phase compensation.

7. The ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media according to claim 1, 5 or 6, characterized in that: The method of calculating the inter-frame variation of the light field distribution of the reference reflector based on the complex amplitude field of the reference reflector obtained by multiple measurements, and further compensating to the corresponding three-dimensional backscattering field, includes: Calculate the inter-frame phase difference of the light field of the reference mirror The purpose of compensating the phase distribution of each frame of the sample signal is to unify the inconsistent scanning errors between all fields of the sample signal into the same phase distribution error. The phase compensation process is described as: Among them, z R is the depth corresponding to the reference mirror signal, E R By E R =E R0 / |E R0 |Normalized to get, E R0 is the backscattered light field from the reference reflector, It is the phase distribution of the sample signal after phase compensation is finally completed.

8. The ultra-deep three-dimensional imaging device based on swept frequency optical coherence tomography in turbid media according to claim 7, characterized in that: After the phase compensation operation is completed, an uncompensated phase still remains, but does not affect subsequent image reconstruction, including: The residual phase error is expressed as: in The specific phase distribution error introduced is simplified to is the inter-frame phase drift, which is a DC quantity independent of the background distribution. At this time, the reflection matrix R e It can be expressed as: Where R is the ideal reflection matrix without phase error, D is the error matrix composed of the phase distribution error of the reference frame, and E is the error matrix composed of the phase fluctuation between frames; the decomposition of the reflection matrix and the reconstruction process of the image are as follows: and are all time reversal operators; the singular value decomposition of the matrix R can be expressed as represents conjugate transpose; U and V are unitary matrices; Σ is a diagonal matrix consisting of positive real singular values ​​σ i Arranged in descending order, i∈[1,N]; matrix U d =DU means that the matrix U is weighted by a phase-only diagonal matrix D. Represents the matrix V as a phase-only diagonal matrix Weighting; and is the column vector U of the matrix U and V i 、V i The reshaped two-dimensional wavefront; M is the optimal number of singular values ​​used for reconstruction, where M <N。

Images