A multi-beam optical coherence tomography and angiography system and method

Abstract

The application discloses a kind of multi-beam optical coherence tomography and angiography system and method, comprising: heterogeneous light beam generation module cuts initial light wave, generates center detection light beam and diffracted light beam with optical path difference;Irradiation and scattering collection module generates multiplexed backscattering light field by irradiating two light beams to the sample to be measured;Photoelectric conversion and acquisition module makes light field interference and implements conversion, generates aliasing radio frequency electrical signal;Frequency domain disintegration and separation module cuts the signal according to optical path difference, filters out shallow first, deep second baseband data;Deep structure analysis module is converted into dynamic space amplitude by second data parameter Suppressor item;Cancelling and blood flow solving module calculates speckle pattern according to first data and Suppressor item background;Rendering and reconstruction module projects three-dimensional coordinate system to dark surface compensation and generates reconstruction image.The application solves the problem that multi-beam concurrent scanning lacks frequency domain isolation and deep suppression, leading to low blood flow purity and false image.

Description

Technical Field

[0001] This invention belongs to the technical field of angiography, and relates to a multi-beam optical coherence tomography and angiography system and method. Background Technology

[0002] Optical coherence tomography (OCT) is a non-invasive, high-resolution imaging technique widely used in the medical field. This technology reconstructs the deep structural information of tissues by measuring the interference signal between scattered photons within the tissue and a reference beam. When the probe beam penetrates deep tissues, especially in tissues containing pigment barriers such as the retinal pigment epithelium and choroid, the incident light energy is attenuated due to scattering and absorption. This attenuation leads to a decrease in the signal-to-noise ratio of deep structural signals and obstruction, increasing the difficulty of acquiring images of deep tissue microvascular networks.

[0003] To address the limitations of deep imaging capabilities, existing technologies typically employ both hardware improvements and post-processing algorithm optimizations. For instance, Chinese patent CN114305320A discloses a multi-beam polarization OCT imaging device and method, which utilizes polarization beam splitting to generate multiple probe beams with different polarization states to acquire polarization-sensitive information about the tissue. At the hardware level, common methods include using longer wavelength light sources to reduce the scattering coefficient within the tissue, employing adaptive optics to compensate for wavefront distortion, and utilizing multi-beam concurrent detection technology. This concurrent detection technology extracts and segments signals in the spectral or polarization domains by segmenting the spectrum of a broadband light source and assigning independent path delays, or by using polarization beam splitting to generate multiple probe beams with different polarization states. At the post-processing algorithm level, the projection phenomenon of superficial blood vessels onto deep tissues is typically estimated and analyzed based on an established physical model, and then artifact cancellation is performed in the image.

[0004] However, the aforementioned technical solutions exhibit corresponding limitations when simultaneously handling shallow high-resolution blood flow detection and diffraction imaging tasks in deep obscured environments. Specifically, hardware improvements using long-wavelength light sources require a balance between enhancing penetration and maintaining axial resolution. Existing multi-beam techniques, with their single segmentation mechanisms in the spectral or polarization domains, lack flexibility when dealing with beam configurations of different physical properties, and their separation process carries the risk of signal crosstalk when dealing with heterogeneous scattered echo signals. On the other hand, post-processing algorithms often rely on the assumption of correlation between shallow and deep signals, and their separation accuracy is limited in complex optical obscuration scenarios, easily leading to structural shadows or residual artifacts in the extracted final images, affecting the continuity of deep angiography images. Summary of the Invention

[0005] In a first aspect, the present invention provides a multi-beam optical coherence tomography and angiography system, comprising the following modules: To address the aforementioned problems, this invention provides a multi-beam optical coherence tomography and angiography system.

[0006] A multi-beam optical coherence tomography and angiography system includes: The heterogeneous beam generation module is used to split the continuous initial light wave, generate the center probe beam, and the diffraction compensation beam with a fixed physical optical path difference. The illumination and scattering collection module is used to illuminate the sample under test with the central probe beam and the diffraction compensation beam together, and capture the scattered light flow to generate a multiplexed backscattered light field. The photoelectric conversion and acquisition module is used to perform photoelectric conversion by interfering and superimposing the multiplexed backscattered light field with the light field of the same source reference arm, thereby generating a mixed continuous radio frequency interference signal. The frequency domain decomposition and separation module is used to perform frequency domain transformation on the aliased continuous radio frequency interference electrical signal, and to perform positioning and cutting by utilizing the mapping characteristics of the fixed physical optical path difference in the frequency domain, and to filter out the first interference baseband data representing shallow structure information and the second interference baseband data representing deep structure information respectively. The deep structure analysis module is used to extract the continuous amplitude distribution parameters of the second interferometric baseband data and transform them into a dynamic spatial amplitude suppression denominator term based on the shadow mutation identification condition. The cancellation and blood flow calculation module is used to perform background removal operations based on the first interference baseband data and the dynamic spatial amplitude suppression denominator term to calculate the true blood flow micro-level speckle pattern. The rendering and reconstruction module is used to project the real blood flow micro-level speckle map onto the three-dimensional coordinate system, perform weight compensation and smooth rendering on the level speckle voxel points extracted from the physically occluded dark surface set, and generate a shadowless angiography reconstruction image.

[0007] In a further embodiment of the present invention, the heterogeneous beam generation module specifically includes: An optical fiber beam splitter is used to divide the acquired sample arm light wave into a first-source photon flow and a second-source optical field flow according to energy. A front-end spatial light modulator is used to assign large numerical aperture and short wavelength physical parameters to the photon flow of the first branch from the same source, generating a central probe beam with a Gaussian mode spatial distribution; A hardware spatial delay array is used to introduce the fixed physical optical path difference into the optical field flow of the second branch from the same source to form a preset path bias; A topology modulation disk is used to configure low numerical aperture, long wavelength and high-order Bessel topological charge characteristics for the optical field flow of the second branch of the same source with the fixed physical optical path difference, so as to generate a diffraction-compensated beam.

[0008] In a further embodiment of the present invention, the irradiation and scattering collection module is specifically used for: The probe beam from the center penetrates the superficial capillary network of the sample under test and extracts the change in the scattering cross-sectional area, thereby exciting the first backscattered light flow. By relying on the spatial self-healing diffraction that occurs when the diffraction compensation beam encounters a deep absorption / scattering physiological barrier, a second backscattered light flow carrying deep weak structural information is excited. The first and second backscattered light flows are captured simultaneously by a broadband microscope objective group, and the two are multiplexed and introduced into a single-mode transmission fiber to generate a multiplexed backscattered light field.

[0009] In a further embodiment of the present invention, the photoelectric conversion and acquisition module includes a balanced photodetector array and a high-speed analog-to-digital converter, and this module is specifically used for: A frequency-domain composite interference light field is formed and projected onto the photosensitive demodulation surface of the corresponding balanced photodetector array to excite and perform the photoelectric conversion physical operation of amplification, contrast, and common-mode removal; The radio frequency current value output by the balanced photodetector array is acquired by a high-speed analog-to-digital converter, generating a mixed continuous radio frequency interference signal that simultaneously carries a shallow modulation clutter spectrum and a deep diffraction continuous envelope.

[0010] In a further embodiment of the present invention, the frequency domain decomposition and separation module is specifically used for: The parallel acceleration engine is invoked to perform fast Fourier transform calculations on the aliased continuous radio frequency interference electrical signals at each depth point, generating a full-band complex spectrum analysis matrix that contains the strength characteristics of the scattered echoes at the full depth of the target location. The pseudo-depth offset distance feature value is extracted based on the prior value generated by the fixed physical optical path difference; Using this pseudo-depth offset distance feature value, the intersecting main peak frequency bands formed by the two beams in the frequency domain distribution are located in the full-band complex spectrum analytical matrix; A rectangular spatial window filter is mounted in a specific depth range of intersecting main wave peak frequency bands to perform physical clipping, thereby filtering out the first interferometric baseband data of the shallow local spatial range and the second interferometric baseband data of the deep steady-state tissue diffraction background distribution.

[0011] In a further embodiment of the present invention, the deep structure analysis module is specifically used for: Extract the deep steady-state diffraction envelope hidden in the second interferometric baseband data, quantize and assign values ​​to the strong structure interface, and generate continuous amplitude distribution parameters; Extract the first-order spatial partial derivative numerical matrix of the continuous amplitude distribution parameters along the longitudinal tangent direction; When the comparison confirms that the local region of the first-order spatial partial derivative numerical matrix exceeds the preset edge fluctuation threshold, it is determined to be a shadow abrupt fault caused by deep physical occlusion and the trigger condition judgment logic is activated. By using the activated trigger condition judgment logic, the continuous amplitude distribution parameters are transformed into a dynamic spatial amplitude suppression denominator term that serves as a dividing factor.

[0012] A further aspect of this invention involves using the cancellation and blood flow calculation module to perform a voxel-by-voxel background removal division normalization operation. The operation process is as follows: The first interferometric baseband data is reconstructed into a dynamic complex field amplitude flow matrix; Extract the target voxel unit to be processed within the dynamic complex field amplitude flow matrix, and feed the dynamic spatial amplitude suppression denominator as the suppression reference element for data stripping into the calculation channel; The complex amplitude of the target voxel element under each scan sequence is divided point by the static amplitude within the same spatial coordinate point in the denominator by the dynamic spatial amplitude suppression term, and then spatial cross-product is performed. The normalized standard deviation of the intensity is obtained by calculating the standard deviation of the time series after division. At the baseband signal level, erroneous frequency band envelopes originating from deep, strongly reflective static tissues are suppressed, and real blood flow micro-level speckle patterns that present the true displacement dimension are separated.

[0013] A further aspect of this invention involves the rendering and reconstruction module extracting the set of physically occluded dark surfaces, which includes: Obtain the effective high-contrast microlevel amplitude values ​​from the real blood flow microlevel speckle pattern after removing noise points with a preset noise floor threshold. The effective high-contrast microlevel amplitude value is then projected into a three-dimensional physical coordinate lattice system. Information voids caused by low signal-to-noise ratio in the scanned 3D physical coordinate lattice system are extracted and aggregated into a set of associated physical occlusion dark surfaces.

[0014] In a further embodiment of the present invention, the rendering operation of the rendering and reconstruction module specifically includes: Identify level speckle voxel points that fall within the set of physically occluded dark surfaces and are retained after cross-modal cancellation; Force a preset opacity weight value to level speckle voxels; According to the spatial shading instructions defined by the preset opacity weight value, voxel interpolation smoothing and stereoscopic connectivity rendering connection operations are performed to output the deep choroidal shadowless angiography reconstruction image.

[0015] Secondly, the present invention provides a method for multi-beam optical coherence tomography and angiography, comprising the following steps: S1. The continuous initial light wave is split to generate a central probe beam and a diffraction compensation beam with a fixed physical optical path difference. S2. Illuminate the sample to be tested together with the central probe beam and the diffraction compensation beam, and capture the scattered light flow to generate a multiplexed backscattered light field; S3. The multiplexed backscattered light field and the same source reference arm light field are interferometrically superimposed and photoelectric conversion is performed to generate a mixed continuous radio frequency interference signal. S4. Perform frequency domain transformation on the aliased continuous radio frequency interference signal, and use the mapping characteristics of the fixed physical optical path difference in the frequency domain to perform positioning and clipping, and filter out the first interference baseband data representing shallow structure information and the second interference baseband data representing deep structure information respectively. S5. Extract the continuous amplitude distribution parameters of the second interferometric baseband data, and transform them into a dynamic spatial amplitude suppression denominator term based on the shadow mutation identification condition. S6. Based on the first interference baseband data and the dynamic spatial amplitude suppression denominator term, background removal operation is performed to calculate the true blood flow micro-level speckle pattern. S7. Project the real blood flow micro-level speckle map onto the three-dimensional coordinate system, perform weight compensation and smooth rendering on the level speckle voxel points extracted from the physically occluded dark surface set, and generate a shadowless angiography reconstruction image.

[0016] In summary, the present invention has the following beneficial technical effects: 1. This invention splits a broadband light source and configures differentiated physical parameters for different branches, forming a Gaussian mode central detection beam and a diffraction compensation beam. This optical path configuration allows the central detection beam to acquire backscattered light reflecting changes in the blood flow cross-sectional area in the superficial capillary network within the same scanning cycle; simultaneously, the diffraction compensation beam, leveraging its spatial self-healing diffraction properties, penetrates deeper into the tissue to acquire deep structural information below the pigment barrier. Through concurrent detection at the physical level, the simultaneous extraction of heterogeneous information at different tissue depths is achieved, helping to overcome the time mismatch and motion artifact problems caused by time-division acquisition.

[0017] 2. This invention introduces a fixed physical optical path difference in the propagation path of the diffraction compensation beam, causing the scattering signals from tissues at different depths to be converted into aliased radio frequency interferometric signals carrying prior depth offsets after interference with the reference beam. By using a rectangular spatial window filter tailored to this fixed optical path difference for frequency domain clipping, shallow interferometric baseband data and deep diffraction interferometric baseband data can be separated within the same aliasing spectrum based on depth offset characteristic values. This scheme, based on hardware preset bias combined with frequency domain physical clipping, reduces separation errors easily introduced by post-processing algorithm estimation, maintains the independence of shallow blood flow data and deep structural data, and provides a reliable data foundation for subsequent cross-modal signal cancellation operations.

[0018] 3. This invention extracts continuous amplitude distribution parameters from deep diffraction interferometry baseband data and, by comparing the first-order spatial partial derivative with the edge fluctuation threshold, locates abrupt shadow faults caused by physical occlusion, thereby generating a dynamic spatial amplitude suppression denominator. This denominator contains the intensity distribution information of the deep static structure. During background removal operations, this suppression denominator is used as a normalization factor for spatial cross-product calculation and standard deviation calculation. This can specifically suppress projection artifact components in shallow complex signals that are associated with the intensity of deep static structures at the baseband signal level, thus better preserving the dynamic changes caused by the movement of real blood cells, improving the solution accuracy of real blood flow micro-level speckle patterns and the purity of deep blood flow signals.

[0019] 4. After generating a realistic blood flow micro-level speckle map and mapping it to a 3D mesh, this invention provides an information gain mechanism for occluded regions. The system identifies information voids and dark areas caused by deep physical occlusion in the 3D coordinate system. For valid speckle voxel points located within the dark area set but retained during cancellation processing, specific opacity weights are assigned. In subsequent voxel interpolation smoothing and stereoscopic connectivity rendering, this weight control bridges the visually broken signal points caused by occlusion, presenting a continuous network structure. This processing reduces visual interference from sweep trailing artifacts and projected occlusion shadows, contributing to a more complete and realistic deep vascular network morphology. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. The drawings are used to provide a further understanding of the present invention.

[0021] Figure 1 This discloses a schematic diagram of the framework in the embodiments of this application.

[0022] Figure 2 This discloses a flowchart of an embodiment of this application. Detailed Implementation

[0023] The following is in conjunction with the appendix Figure 1 - Figure 2 A preferred description of the present invention is provided below.

[0024] See attached document Figure 1 - Figure 2 This invention proposes a multi-beam optical coherence tomography and angiography system, comprising the following modules: The heterogeneous beam generation module is used to split the continuous initial light wave, generate the center probe beam, and the diffraction compensation beam with a fixed physical optical path difference. The illumination and scattering collection module is used to illuminate the sample under test with the central probe beam and the diffraction compensation beam together, and capture the scattered light flow to generate a multiplexed backscattered light field. The photoelectric conversion and acquisition module is used to perform photoelectric conversion by interfering and superimposing the multiplexed backscattered light field with the light field of the same source reference arm, thereby generating a mixed continuous radio frequency interference signal. The frequency domain decomposition and separation module is used to perform frequency domain transformation on the aliased continuous radio frequency interference electrical signal, and to perform positioning and cutting by utilizing the mapping characteristics of the fixed physical optical path difference in the frequency domain, and to filter out the first interference baseband data representing shallow structure information and the second interference baseband data representing deep structure information respectively. The deep structure analysis module is used to extract the continuous amplitude distribution parameters of the second interferometric baseband data and transform them into a dynamic spatial amplitude suppression denominator term based on the shadow mutation identification condition. The cancellation and blood flow calculation module is used to perform background removal operations based on the first interference baseband data and the dynamic spatial amplitude suppression denominator term to calculate the true blood flow micro-level speckle pattern. The rendering and reconstruction module is used to project the real blood flow micro-level speckle map onto the three-dimensional coordinate system, perform weight compensation and smooth rendering on the level speckle voxel points extracted from the physically occluded dark surface set, and generate a shadowless angiography reconstruction image.

[0025] In one embodiment of the present invention, the heterogeneous beam generation module is used to perform the following steps: An optical fiber beam splitter is used to divide the acquired sample arm light wave into a first-source photon flow and a second-source optical field flow according to energy. A front-end spatial light modulator is used to assign large numerical aperture and short wavelength physical parameters to the photon flow of the first branch from the same source, generating a central probe beam with a Gaussian mode spatial distribution; A hardware spatial delay array is used to introduce the fixed physical optical path difference into the optical field flow of the second branch from the same source to form a preset path bias; A topology modulation disk is used to configure low numerical aperture, long wavelength and high-order Bessel topological charge characteristics for the optical field flow of the second branch of the same source with the fixed physical optical path difference, so as to generate a diffraction-compensated beam.

[0026] Specifically, the optical system splits a broadband light source and emits multiple probe beams with heterogeneous physical properties. This optical system consists of a broadband swept-frequency laser source, fiber optic beam splitter, and modulation components. First, the system activates the broadband swept-frequency laser source, causing it to output a continuous initial light wave with a continuously scanning wavelength. The broadband swept-frequency laser source is a swept-frequency laser based on a semiconductor optical amplifier, with its center wavelength set around 1050 nm and a spectral scanning bandwidth covering the range of 100 nm to 150 nm. This continuous initial light wave is then injected... A single-mode fiber coupler is used, with the splitting ratio precisely set to 50:50. Through this fiber coupler, the energy of the continuous initial light wave is proportionally divided in the physical transmission space into a first-source photon flow and a second-source optical field flow that enter two independent optical fibers. These two optical flows maintain strict coherence because they originate from the same laser pulse.

[0027] The first branch of the photon beam originating from the same source exits from the end of its optical fiber, passes through a collimating lens group to form a parallel beam, and sequentially passes through the pre-amplified spatial light modulation module. This module integrates a short-pass filter and a high-magnification microscope objective. The short-pass filter selectively extracts shorter wavelengths from the broadband spectrum. The short-wavelength physical parameters are determined by optical filtering based on the output spectrum of the broadband swept-frequency laser source, selecting a shorter wavelength range, such as 980nm to 1080nm, for shallow-level detection. The high-magnification microscope objective focuses the filtered beam, thus endowing the first branch of the photon beam with a large numerical aperture and short-wavelength physical parameters.

[0028] It should be understood that large numerical aperture is a dimensionless parameter, and its value range is set according to the lateral resolution requirements of tissue imaging, usually between 0.1 and 0.3. A larger numerical aperture is beneficial for achieving higher lateral resolution in shallow tissues.

[0029] After modulation, the spatial energy distribution of the beam exhibits a Gaussian mode shape, ultimately generating the central probe beam.

[0030] Simultaneously, the second branch optical field flow from the same source also exits from its fiber end and is collimated. This collimated beam first enters a hardware spatial delay array, which, through a set of precisely controlled mirrors, introduces an additional physical propagation distance for the beam, thereby pre-setting a fixed hardware optical path difference.

[0031] It should be noted that the fixed physical optical path difference It is the physical path length increment introduced in the optical path of the second branch optical field flow from the same source. Its value is preset according to the longitudinal scanning range of the system, with a typical setting range of 1mm to 5mm. The purpose of setting this optical path difference is to effectively separate the interference signals of the two beams on the depth axis in subsequent frequency domain signal processing, so as to avoid signal crosstalk.

[0032] A fixed physical optical path difference introduces a fixed phase delay. Its size is related to the optical path difference. and center wavelength Related, specifically:

[0033] The beam carrying this optical path difference then passes through a topological modulation disk, which consists of a long-pass filter, a spiral phase plate, and a low-magnification telescope system connected in series.

[0034] Long-wavelength pass filters are used to extract the longer wavelength portion of the spectrum. The physical parameters of the long wavelength are selected longer wavelengths, such as 1080nm to 1180nm. Long-wavelength light has a lower scattering coefficient in biological tissues, thus having stronger penetrating power.

[0035] The spiral phase plate applies a specific spiral phase delay to the wavefront of the light field. This specific delay can be described by a phase mask function; this function defines the phase change at each point on a plane perpendicular to the beam propagation direction. For a beam with a specific topological charge, the mathematical expression for its phase mask function is:

[0036] in, In polar coordinates Complex transmittance of the lower phase mask; It is the radial distance from the center of the beam to that point. It is the azimuth angle; symbol Represents the imaginary unit, which satisfies ;symbol It is an integer representing the order of the higher-order Bessel topological charge. It is an integer that determines the magnitude of the hollow structure and orbital angular momentum of a Bessel beam; its order is... The value of is greater than 1, and is usually set in the range of 2 to 10. A higher order helps to enhance the spatial self-healing ability of the beam when it encounters obstacles.

[0037] The low-magnification telescope system adjusts the divergence angle of the beam. This series of operations configures the optical field flow of the second branch from the same source with a combination of physical properties including low numerical aperture, long wavelength, and high-order Bessel topological charge, thereby generating a diffraction-compensated beam. The value of the low numerical aperture is set in the range of 0.01 to 0.05. The lower numerical aperture can form a longer depth of focus, which helps to maintain the energy concentration of the beam in deep tissues.

[0038] Finally, the central probe beam and the diffraction compensation beam are precisely combined in space into two coaxial beams using a two-color beam combiner, ensuring that their exit paths are aligned and preparing for subsequent synchronous scanning illumination.

[0039] For example, in a specific application scenario of the present invention, the system employs a broadband swept-frequency laser source with a center wavelength of 1050nm and a scanning bandwidth of 120nm, i.e., 990nm to 1110nm. The continuous initial light wave output is split into a first-branch photon stream and a second-branch optical field stream of the same origin via a 50:50 fiber coupler. For the first-branch photon stream, a short-pass filter in the pre-amplified spatial light modulation module intercepts the 990nm to 1050nm band, forming a short-wavelength light with a center wavelength of 1020nm. This beam is focused by an objective lens with a numerical aperture of 0.15 to generate a central probe beam. For the second-branch optical field stream of the same origin, the mirror of the hardware spatial delay array is fixed at an optical path length of... The beam is positioned such that a fixed physical optical path difference is formed. Subsequently, the beam passes through a long-pass filter, allowing only wavelengths from 1050 nm to 1110 nm to pass through, forming long-wavelength light with a center wavelength of 1080 nm. This beam is then incident perpendicularly onto a topological modulation disk etched with a spiral phase structure, the topological charge of which is set to... Finally, a low numerical aperture lens system shapes the beam into a diffraction-compensated beam with a numerical aperture of 0.04. The two beams are then combined into a coaxial beam by a dichroic mirror that splits the beam at 1050 nm, ready to be emitted towards the target biological tissue.

[0040] In one embodiment of the present invention, the irradiation and scattering collection module is used to perform the following steps: The probe beam from the center penetrates the superficial capillary network of the sample under test and extracts the change in the scattering cross-sectional area, thereby exciting the first backscattered light flow. By relying on the spatial self-healing diffraction that occurs when the diffraction compensation beam encounters a deep absorption / scattering physiological barrier, a second backscattered light flow carrying deep weak structural information is excited. The first and second backscattered light flows are captured simultaneously by a broadband microscope objective group, and the two are multiplexed and introduced into a single-mode transmission fiber to generate a multiplexed backscattered light field.

[0041] After generating the center probe beam and the diffraction compensation beam, these two beams are converged and simultaneously irradiated onto the sample. First, the system controls a coaxial dual-color galvanometer, a system capable of coaxial, synchronous two-dimensional scanning of two beams with different center wavelengths. The mirror surface of this galvanometer is coated with a dielectric film that provides high reflectivity for both the center probe beam and the diffraction compensation beam in their respective wavelength bands. The center probe beam and the diffraction compensation beam are precisely guided coaxially to the center of the galvanometer. The coaxial dual-color galvanometer deflects at a preset scanning frequency and mode, synchronously injecting the combined beam into the sample within a single longitudinal scan hardware microsecond-level integration period. This single longitudinal scan hardware microsecond-level integration period refers to the time required for the system to complete one depth scan and acquire the corresponding data points. The duration of this period is directly related to the system's A-scan scanning rate and is typically set to 10 to 100 microseconds. Synchronous irradiation within this period ensures that the tissue information acquired by the two beams is highly correlated in time.

[0042] The sample to be tested has a layered structure, characterized by a superficial capillary network and a deep, highly absorbent pigment barrier beneath it. In ophthalmic imaging applications, the superficial capillary network specifically refers to the capillary plexus in the plexiform layer or ganglion cell layer of the retina. The deep, highly absorbent pigment barrier specifically refers to pigment clusters, drusen, or hemorrhages in the retinal pigment epithelium or choroidal layer, which are strong scattering or strong absorption bodies.

[0043] During irradiation, the central probe beam, due to its large numerical aperture and short wavelength, focuses primarily on the superficial regions of the tissue. When it penetrates the superficial capillary network, photons are scattered by flowing red blood cells. The system detects this change in the scattering cross-sectional area caused by red blood cell movement—that is, the physical quantity by which the intensity of backscattered light changes over time due to the displacement of scattering particles (mainly red blood cells) within the probe volume. This is the basis for extracting blood flow signals in optical coherence tomography (OCT) angiography, thereby exciting and forming the first backscattered light stream carrying information about superficial hemodynamics.

[0044] Meanwhile, the diffraction-compensated beam, with its low numerical aperture, long wavelength, and the inherent spatial self-healing diffraction properties of Bessel beams, can bypass small-sized obstructions and reconverge behind them when encountering deep, high-absorption pigment barriers. Spatial self-healing diffraction is an inherent characteristic of high-order Bessel beams. When its central portion is obstructed by an opaque obstacle, its peripheral ring-shaped energy can bypass the obstacle and undergo constructive interference at a certain distance behind it, reforming the central spot. This allows the beam to penetrate into normally obscured deep regions, interacting with deep, weak structures and thus exciting a second backscattered light flow carrying information about the deep tissue structure.

[0045] Subsequently, the broadband microscope objective group simultaneously captures the first and second backscattered light streams returning from the tissue along a path coaxial with the incident light path. This broadband microscope objective group is a chromatic aberration-corrected optical lens system, designed to ensure consistent focusing performance and high collection efficiency across the entire spectral range, including both short and long wavelengths. The objective group converges these two light streams, each containing information from scattering at different depths, and multiplexes them into the core of a single-mode transmission fiber. This results in a multiplexed backscattered light field transmitted in a single channel for subsequent interferometric measurement steps. The single-mode transmission fiber ensures good spatial coherence of the collected backscattered light, a prerequisite for effective interference with the reference light.

[0046] For example, continuing from the preceding steps, a coaxial dual-color galvanometer simultaneously scans the target human retinal tissue with a central probe beam (center wavelength 1020 nm) and a diffraction compensation beam (center wavelength 1080 nm). The system's A-scan scanning rate is set to 50 kHz, therefore the hardware microsecond-level integration period for a single longitudinal scan is 20 microseconds. During this period, the central probe beam is focused on the superficial capillary network approximately 150 μm below the retinal surface and interacts with the red blood cells therein to generate a first backscattered light flow. Simultaneously, a deep, highly absorbent pigment barrier with a diameter of 60 μm exists approximately 400 μm below the retina, i.e., a dense pigment deposit. When the diffraction compensation beam irradiates this pigment deposit, its energy is blocked by the central portion, but due to its topological charge... Utilizing the Bessel beam characteristics, the beam undergoes spatial self-healing diffraction at a distance of approximately 80 μm behind the pigment point, reforming the beam spot and continuing to probe the choroidal capillaries beneath it, thereby generating a second backscattered optical flow carrying information about the area behind the obscured region. A broadband microscope objective group designed and optimized for the 990 nm to 1110 nm wavelength range simultaneously collects these two orthogonally returning backscattered optical flows and couples them into a single-mode transmission fiber with a core diameter of 9 μm, forming a multiplexed backscattered optical field containing both superficial blood flow and deep structural information, which is then stably transmitted to the interferometer section of the system.

[0047] In one embodiment of the present invention, the photoelectric conversion and acquisition module is used to perform the following steps: A frequency-domain composite interference light field is formed and projected onto the photosensitive demodulation surface of the corresponding balanced photodetector array to excite and perform the photoelectric conversion physical operation of amplification, contrast, and common-mode removal; The radio frequency current value output by the balanced photodetector array is acquired by a high-speed analog-to-digital converter, generating a mixed continuous radio frequency interference signal that simultaneously carries a shallow modulation clutter spectrum and a deep diffraction continuous envelope.

[0048] The optical field is converted into an electrical signal via a closed-loop detection link. This step begins at the output end of the single-mode transmission fiber. At this output end, a beam splitter is precisely aligned, directly guiding the multiplexed backscattered light field emitted from the fiber core into the sample arm input port of the broadband Michelson interferometer physics component. In this embodiment, the broadband Michelson interferometer physics component is an all-fiber interferometer based on optical fibers, designed to operate within a wavelength range that matches the output spectral range of a broadband swept-frequency laser source.

[0049] Meanwhile, another portion of the optical field from the broadband swept-frequency laser source, serving as the optical field of the same-source reference arm as internal feedback within the link, is also introduced into the interference component after passing through an independent reference arm optical path. The optical field of the same-source reference arm refers to the reference light that branches off from the main light source and passes through a fixed-length optical fiber, whose optical path is precisely controlled to define the zero-depth reference plane of the system.

[0050] Within the 2×2 fiber coupler of the interference assembly, the multiplexed backscattered light field and the light field of the same-source reference arm undergo spatial beat frequency superposition. Since the two light fields originate from the same source and have a stable phase relationship, this superposition process generates a frequency-domain composite interference light field. The time-frequency characteristics of this interference light field not only encode the true positional information of scatterers at different depths within the tissue, but also show that the frequency of the intensity change of the frequency-domain composite interference light field over time is proportional to the depth of the scatterers. Simultaneously, the optical field also incorporates multiple spatial voxel depth anomalous offsets caused by a pre-set fixed physical optical path difference. Next, the frequency-domain composite interference optical field is precisely projected onto two independent photosensitive demodulation surfaces of a balanced photodetector array through the two complementary output ports of a 2×2 fiber coupler. The balanced photodetector array has a sufficiently high bandwidth to respond to the rapid oscillations of the interference signal; its bandwidth is typically set to several hundred MHz. The photosensitive demodulation surfaces are made of InGaAs material with high quantum efficiency. The electronic circuitry within the balanced photodetector array immediately performs a photoelectric conversion physical operation of amplification and contrast common-mode cancellation, i.e., the current generated by the two photosensitive surfaces is differentially amplified. This operation effectively eliminates common-mode noise such as light source intensity fluctuations and enhances the intensity of the interference signal.

[0051] Finally, the high-speed analog-to-digital converter continuously acquires the fluctuating radio frequency current values ​​output by the balanced photodetector array at a sampling rate much higher than the Nyquist frequency. To avoid signal aliasing, the sampling rate of the high-speed analog-to-digital converter needs to satisfy the Nyquist sampling theorem, that is, the sampling rate must be greater than twice the maximum frequency of the interference signal. Its sampling rate is usually set between 100 MS / s and 1 GS / s.

[0052] After sampling, the radio frequency current value is quantized into a series of discrete digital signals, thereby generating an aliased continuous radio frequency interference signal that simultaneously carries the shallow modulation clutter spectrum originating from the central probe beam and the deep diffraction continuous envelope originating from the diffraction compensation beam; that is, the interference signal spectrum generated by the interaction between the central probe beam and the shallow tissue and the interference signal spectrum generated by the interaction between the diffraction compensation beam and the deep tissue, wherein, due to the fixed physical optical path difference, the envelope is shifted to a pseudo-depth position in the spectrum.

[0053] It should be understood that the radio frequency current value output by the aforementioned balanced photodetector array This is a direct manifestation of the interference process. Ideally, this current can be approximated as the superposition of the interference terms of the optical fields of the sample arm and the reference arm, specifically expressed as:

[0054] in, It is over time Changing radio frequency current; It is a frequency-sweeping light source in time wavenumber, and wavelength The relationship is ; and These represent the depths of the central probe beam and the diffraction compensation beam, respectively. and The power of backscattered light generated by the scatterer; It is the power of the optical field of the same reference arm; and It is the actual optical depth of the scatterer within the tissue; It is a fixed physical optical path difference.

[0055] This formula shows that the final electrical signal is a linear superposition of two sets of interference fringes, where the second set of fringes is due to... The existence of is such that it is translated as a whole in the depth domain.

[0056] For example, continuing from the preceding steps, the backscattered light field is multiplexed from a single-mode transmission fiber to the sample arm port of a broadband Michelson interferometer physics component. Assume the central probe beam detects a location located in the superficial layer of the retina. A blood cell was detected, while the diffraction-compensated beam detected a blood cell located deep within the choroid. The organizational structure at that location. The power of the optical field of the same source reference arm was set to 5mW. The scanning rate of the light source caused the wavenumber to change over time. for According to the principle of interference, signals from superficial blood cells will generate frequencies of... The radio frequency signal. For deep tissue structures, the effective depth corresponding to the signal is the sum of its actual depth and the fixed physical optical path difference, i.e. Therefore, it will generate a frequency of The radio frequency signal is generated by a balanced photodetector array that superimposes these two frequency components, along with signals from other depths, to output a complex radio frequency current. A high-speed analog-to-digital converter with a sampling rate of 200 MS / s samples this current, generating a digitized data stream, namely an aliased continuous radio frequency interference signal. Two distinct peaks will appear in the spectrum of this signal, around 4.36 MHz and 40.3 MHz, corresponding to superficial blood cells and deep tissue structures, respectively.

[0057] In one embodiment of the present invention, the frequency domain decomposition and separation module is used to perform the following steps: The parallel acceleration engine is invoked to perform fast Fourier transform calculations on the aliased continuous radio frequency interference electrical signals at each depth point, generating a full-band complex spectrum analysis matrix that contains the strength characteristics of the scattered echoes at the full depth of the target location. The pseudo-depth offset distance feature value is extracted based on the prior value generated by the fixed physical optical path difference; Using this pseudo-depth offset distance feature value, the intersecting main peak frequency bands formed by the two beams in the frequency domain distribution are located in the full-band complex spectrum analytical matrix; A rectangular spatial window filter is mounted in a specific depth range of intersecting main wave peak frequency bands to perform physical clipping, thereby filtering out the first interferometric baseband data of the shallow local spatial range and the second interferometric baseband data of the deep steady-state tissue diffraction background distribution.

[0058] After photoelectric conversion and generation of aliased continuous radio frequency interference signals, mathematical mapping and decomposition of the signals and null-plane interception of the response are performed. The Discrete Fast Fourier Transform (DFT) acceleration engine integrated within the field-programmable gate array (FPGA) on the system motherboard is invoked, using each segment of A-scan data from the aliased continuous RF interference signal as input to perform depth-point-by-depth transform calculations. The DFT acceleration engine is a piece of hardware logic or a dedicated IP core embedded in the FPGA; FPGAs are semiconductor devices whose internal logic circuits can be reconfigured and are widely used for real-time signal processing acceleration due to their parallel processing capabilities.

[0059] The Discrete Fast Fourier Transform maps the interference signal in the time domain to the frequency domain. Since the signal frequency is proportional to the depth in swept-frequency optical coherence tomography, this transformation essentially converts the signal from the time intensity domain to the depth-reflectivity domain, thereby generating a full-band complex spectral analytical matrix that contains the strength characteristics of the scattered echoes across the entire depth of the target area.

[0060] It should be noted that the full-band complex spectrum analytical matrix is ​​a one-dimensional complex array, where each element corresponds to the complex reflectivity at a depth position, containing amplitude and phase information.

[0061] The preset prior value of the fixed physical optical path difference is retrieved from its parameter configuration library. This value directly corresponds to the predictable pseudo-depth offset distance feature value. It can be seen that this feature value is a direct manifestation of the fixed physical optical path difference in the depth domain, and its value is related to... The signals are equal. The system utilizes this pseudo-depth offset distance characteristic value to search and locate the interleaved main peak frequency bands naturally formed by the frequency domain distribution of the two probe beams and separated by this pseudo-depth offset distance within the full-band complex spectrum analysis matrix. The interleaved main peak frequency bands refer to the signal generated by the central probe beam being concentrated near zero depth in the full-band complex spectrum analysis matrix, while the signal generated by the diffraction compensation beam is concentrated near the pseudo-depth offset distance characteristic value; the two are effectively separated on the depth axis.

[0062] The software-implemented rectangular spatial window filter module is activated, closing its switch within a specific depth range corresponding to the intersecting main peak frequency bands to perform a physical clipping operation. For this module, its window width... The settings are determined based on the required axial resolution of the system and the imaging depth range of the actual tissue. Typically, the settings are configured to fully cover the target structure while minimizing noise. Furthermore, the clipping operation of this rectangular spatial window filter module can be performed using a window function. To express it mathematically.

[0063] Assuming the center depth of the target frequency band is Window width is Then the window function can be expressed as:

[0064] in, These are depth coordinates, representing the full-band complex spectrum analytical matrix. Multiplying by this window function yields the filtered interferometric baseband data. .

[0065] This operation removes any incoherent high-frequency clutter and DC components falling outside the target frequency band by setting all spectral components outside the window to zero. By attaching the first window near zero depth, the center of the window corresponding to the filtered data is... Approaching zero, the system precisely filters out the first interferometric baseband data specific to the shallow local spatial range, which contains structural and blood flow information of the shallow tissue. By attaching a second window near the depth indicated by the pseudo-depth offset distance feature value, the center of the window corresponding to the filtered data is determined. Approximately equal to the pseudo-depth offset distance feature value, i.e. The system also accurately filters out the second interferometric baseband data specific to the diffraction background distribution of deep steady-state tissues. This data mainly contains the static structural information of deep tissues. These two sets of data are stored in separate memory blocks for subsequent processing.

[0066] For example, continuing from the preceding steps, the aliased continuous radio frequency interference signal is fed into a field-programmable gate array (FPGA). A discrete fast Fourier transform (FFT) acceleration engine performs a 2048-point FFT on it, generating a full-band complex spectral analysis matrix. Calculations show that the signal peak from superficial blood cells occurs at 4.36 MHz, and the signal peak from deep tissue occurs at 40.3 MHz. Among the system parameters, the light source wavenumber variation rate... for Therefore, depth With frequency The relationship is Therefore, 4.36MHz corresponds to a depth of 40.3MHz corresponds to a depth of The system retrieves the pseudo-depth offset distance feature value from the configuration library. The system confirms the apparent depth of the deep signal. The true depth of shallow signals This matches the offset value, thus locating the point where... and The two intersecting main wave peak frequency bands are centered on this. Subsequently, the system sets the window width. The center of the first rectangular spatial window filter is set at... Its coverage area is to This filters out the first interferometric baseband data containing the 4.36MHz peak signal. The center of the second rectangular spatial window filter is set at... Its coverage area is to This allows the filtering out of the second interference baseband data, which contains a peak signal of 40.3 MHz.

[0067] In one embodiment of the present invention, the deep structure analysis module is used to perform the following steps: Extract the deep steady-state diffraction envelope hidden in the second interferometric baseband data, quantize and assign values ​​to the strong structure interface, and generate continuous amplitude distribution parameters; Extract the first-order spatial partial derivative numerical matrix of the continuous amplitude distribution parameters along the longitudinal tangent direction; When the comparison confirms that the local region of the first-order spatial partial derivative numerical matrix exceeds the preset edge fluctuation threshold, it is determined to be a shadow abrupt fault caused by deep physical occlusion and the trigger condition judgment logic is activated. By using the activated trigger condition judgment logic, the continuous amplitude distribution parameters are transformed into a dynamic spatial amplitude suppression denominator term that serves as a dividing factor.

[0068] Specifically, the central processing unit retrieves the second interferometric baseband data from memory, which is a matrix composed of a series of complex numbers. The system performs a modulo operation on each complex value in the matrix, i.e., calculates its absolute value, thereby extracting its amplitude information. This amplitude information collectively constitutes the deep steady-state diffraction envelope hidden in the low-frequency band within the second interferometric baseband data. This deep steady-state diffraction envelope is the set of complex amplitudes of the second interferometric baseband data, which mainly reflects the reflectivity distribution of the deep static structure detected by the diffraction compensation beam.

[0069] As the scan progresses, the envelopes of multiple A-scans are combined into a B-scan image, and multiple B-scan images are then stacked into a three-dimensional data volume. This process essentially involves quantizing and assigning continuous amplitude distribution parameters in three-dimensional space to the strong structural interface caused by interference from highly reflective media, generating a continuous amplitude distribution parameter matrix representing the reflectivity distribution of deep static tissue.

[0070] It should be noted that the continuous amplitude distribution parameter is a three-dimensional array, where the value of each element represents the signal intensity detected by the diffraction compensation beam at the corresponding spatial location.

[0071] The system extracts the first-order spatial partial derivative values ​​voxel by voxel along the depth axis direction of the continuous amplitude distribution parameter matrix. This calculation is implemented using a discrete gradient operator to quantify the rate of change of the amplitude signal along the depth direction, thereby obtaining the first-order spatial partial derivative numerical matrix.

[0072] The value of this matrix reflects the degree of variation in tissue reflectivity along depth. In this calculation, let the continuous amplitude distribution parameter matrix be... ,in It is the horizontal coordinate. If the coordinates are depth coordinates, then the numerical matrix of the first-order spatial partial derivatives in the depth direction is... It can be approximated by first-order backward difference:

[0073] in, This indicates the previous voxel along the depth axis.

[0074] The system compares the absolute value of each partial derivative in the matrix with a preset edge fluctuation threshold. This preset edge fluctuation threshold... The setting is based on statistical analysis of a large number of normal tissue and tissue samples containing occluded lesions. It is set to be higher than the gradient value of the normal tissue structure boundary, but lower than the gradient value caused by the sharp attenuation of signal due to strong absorption or strong scattering.

[0075] In a typical embodiment, if the signal amplitude is normalized to the range of 0 to 1, the threshold can be set between 0.3 and 0.6. During this comparison process, the criteria for determining abrupt tomography are:

[0076] When the comparison confirms that the partial derivative value of a certain local area exceeds the threshold, the system determines that the area is a shadow abrupt fault caused by deep physical occlusion. It should be understood that the shadow abrupt fault here is a set of voxels in 3D data whose gradient value exceeds the preset edge fluctuation threshold, which spatially delineates the edge of the shadow cast by the physical occluder.

[0077] When any of the shadow abrupt faults is determined to form, that is, when the above conditions are met at any point in the data volume... Once established, the system immediately activates the trigger condition judgment logic. This trigger condition judgment logic is a software flag bit; when its state changes from "inactive" to "active," it triggers the execution of specific algorithm modules in the subsequent processing flow.

[0078] Through the activated trigger condition judgment logic, the entire continuous amplitude distribution parameter matrix carrying the scattering delay characteristics of deep static anatomical anomalies is specified as a whole and transformed into a dynamic spatial amplitude suppression denominator term as a division factor, and its data pointer is passed to the next calculation step. After activation, the dynamic spatial amplitude suppression denominator term... Defined as:

[0079] This formula shows that the denominator of the dynamic spatial amplitude suppression is numerically identical to the continuous amplitude distribution parameter matrix. As a whole data block, it encapsulates the complete scattering information of deep static organization and is used to correct shallow signals in the next step.

[0080] For example, continuing from the preceding steps, the system processes a continuous amplitude distribution parameter matrix representing deep tissue, generated from the second interferometric baseband data. Assume that a portion of the amplitude data for this matrix at a certain B-scan cross section is as follows, where the columns represent lateral positions. Line represents depth (Units are arbitrary units of brightness): For The position, amplitude sequence is ;for At that location, due to a deep physical occlusion, the amplitude sequence is... The system calculates the first-order spatial partial derivatives of these two data columns. At that point, the partial derivative numerical sequence is Their absolute values ​​are all less than the preset edge fluctuation threshold. .exist At the occlusion edge, the partial derivative value is .because The system determines the presence of a shadowed abrupt fault at this point and immediately activates the trigger condition judgment logic. Once activated, this logic will process the entire B-scan cross-sectional matrix containing the aforementioned amplitude data, i.e., including data such as... The complete data block of the sequence is designated as the denominator of the dynamic spatial amplitude suppression, and its memory address is sent to the processing unit in step six, ready for subsequent artifact suppression calculations.

[0081] In one embodiment of the present invention, the cancellation and blood flow calculation module is used to perform the following operations: The first interferometric baseband data is reconstructed into a dynamic complex field amplitude flow matrix; Extract the target voxel unit to be processed within the dynamic complex field amplitude flow matrix, and feed the dynamic spatial amplitude suppression denominator as the suppression reference element for data stripping into the calculation channel; The complex amplitude of the target voxel element under each scan sequence is divided point by the static amplitude within the same spatial coordinate point in the denominator by the dynamic spatial amplitude suppression term, and then spatial cross-product is performed. The normalized standard deviation of the intensity is obtained by calculating the standard deviation of the time series after division. At the baseband signal level, erroneous frequency band envelopes originating from deep, strongly reflective static tissues are suppressed, and real blood flow micro-level speckle patterns that present the true displacement dimension are separated.

[0082] After separating two independent baseband data from the aliased signal, the true blood flow signal is calculated by fusing the first interferometric baseband data with the denominator term of the dynamic spatial amplitude suppression. First, the system reconstructs the first interferometric baseband data, which includes multiple consecutive B-scan scans, into a four-dimensional dynamic complex field amplitude flow matrix. It should be noted that this dynamic complex field amplitude flow matrix is ​​a four-dimensional data structure that stores all complex interferometric signals obtained from repeated scans (typically 2 to 8 times) at the same location. The dimensions of this matrix represent the transverse scan position, longitudinal scan position, depth, and scan time series, respectively.

[0083] Meanwhile, the denominator term of the dynamic spatial amplitude suppression, as a three-dimensional static matrix, is used as the sole suppression element prior to data stripping. It is fed into and loaded in real-time into the arithmetic register node pool of the heterogeneous collaborative computing unit for high-speed access during large-scale parallel computing. In this embodiment, the heterogeneous collaborative computing unit refers to a computing system composed of a central processing unit and one or more graphics processing units, utilizing the powerful parallel computing capabilities of the graphics processing units to accelerate image processing algorithms. The arithmetic register node pool is not a physical entity, but rather refers to a shared memory or texture memory region within the graphics processing unit that can be accessed at high speed by all computing cores.

[0084] Heterogeneous collaborative computing units, such as graphics processing units, are controlled to perform a spatial cross-integration and standard deviation de-cluttering operation on the target voxel units contained in the dynamic complex field amplitude flow matrix and the set of co-source spatial coordinate points contained in the dynamic spatial amplitude suppression denominator.

[0085] It should be understood that this background clutter removal operation is a specific algorithmic process that performs pixel-by-pixel division on two spatially aligned datasets and then performs statistical analysis on the time series of the results. It is not a general neural network model.

[0086] Specifically, for each spatial voxel coordinate, the computational unit divides the time series data corresponding to that coordinate in the dynamic complex field amplitude flow matrix point by point with the static amplitude corresponding to the same coordinate in the denominator of the dynamic spatial amplitude suppression, and then calculates the standard deviation of the normalized time series.

[0087] To more clearly describe this core calculation process, which aims to calculate a normalized standard deviation to represent blood flow intensity, the specific calculation formula is as follows: For any spatial voxel coordinate Its blood flow signal intensity Calculated using the following formula:

[0088] in, Operators for calculating standard deviation; It is a time series index, from 1 to... , representing the conduct Repeated scans; It is the dynamic complex field amplitude flow matrix in coordinates Place, No. Complex amplitude during each scan; The dynamic space amplitude suppression denominator terms are on the same coordinate. The amplitude at that point; It is a very small positive constant used to prevent the denominator from being zero, and it is a regularization parameter introduced to ensure computational stability. Its value is much smaller than the average level of system noise; for example, it can be set to 0. .

[0089] This formula normalizes the dynamic changes of shallow signals with the intensity of deep static signals, and then evaluates their dispersion, thereby suppressing projection artifacts.

[0090] The system performs the background clutter removal operation, generating a normalized standard deviation for each voxel, i.e., the standard deviation value. When the variation in the shallow signal originates from the projection of a deep static structure, due to... and The ratio is directly proportional to the signal intensity, and after normalization, it approaches a constant with a standard deviation close to zero, thus acting as a damping suppression to generate the final blood flow spectrum. This method forcibly clears or greatly suppresses the erroneous frequency band envelope in the first interferometric baseband data, which is positively correlated with the intensity of deep structures due to forward scattering from underlying strongly reflective static tissue, at the baseband signal level. This allows for the precise capture and extraction of the true blood flow micro-level speckle spectrum from the extremely low background noise, where the signal intensity is contributed solely by the actual red blood cell displacement. The true blood flow micro-level speckle spectrum is a three-dimensional matrix, where each element's value... This represents the intensity of blood flow activity within the corresponding voxel.

[0091] For example, continuing from the previous steps, suppose the system performed [something] at the same location. Second B-scan. Consider two shallow voxels, A and B. Voxel A is located within a real blood vessel with no strong reflective structures below it; voxel B has no real blood flow, but is directly below a detected strong reflective obstruction. For voxel A, the complex amplitude sequence of the first interferometric baseband data acquired in four scans is... Possibly This exhibits significant fluctuations caused by blood cell movement. The dynamic spatial amplitude at the corresponding location suppresses the value of the denominator. Because there is no strong reflection below, the value is low, at 20. For voxel B, due to the projection of the strong reflective structure below it, its amplitude sequence... Although it's still very strong, the changes haven't been significant; it might be... Its corresponding position The value is very high, at 85. (Settings) The normalized amplitude sequence for voxel A is calculated as follows: Its standard deviation The normalized amplitude sequence for voxel B is calculated as follows: Its standard deviation Ultimately, in the generated real blood flow microlevel speckle pattern, voxel A will appear as a bright spot with a value of 0.64, while voxel B will appear as a dark spot with a value of only 0.02, and projection artifacts will be successfully suppressed.

[0092] In one embodiment of the present invention, the rendering and reconstruction module is used to perform the following steps: Obtain the effective high-contrast microlevel amplitude values ​​from the real blood flow microlevel speckle pattern after removing noise points with a preset noise floor threshold. The effective high-contrast microlevel amplitude value is then projected into a three-dimensional physical coordinate lattice system. Scan the information void regions in the three-dimensional physical coordinate lattice system caused by low signal-to-noise ratio, and extract and aggregate them into a set of associated physical occlusion dark surfaces; The rendering operation specifically includes: Identify level speckle voxel points that fall within the set of physically occluded dark surfaces and are retained after cross-modal cancellation; Force a preset opacity weight value to level speckle voxels; According to the spatial shading instructions defined by the preset opacity weight value, voxel interpolation smoothing and stereoscopic connectivity rendering connection operations are performed to output the deep choroidal shadowless angiography reconstruction image.

[0093] After finally calculating the true blood flow microlevel speckle pattern, it is subjected to 3D visualization rendering to generate clinically readable angiography images. First, the system performs a thresholding process on the true blood flow microlevel speckle pattern to remove all false positive noise points below a preset noise floor, resulting in a clean dataset containing all valid high-contrast microlevel amplitude values. Here, valid high-contrast microlevel amplitude values ​​refer to voxel values ​​in the true blood flow microlevel speckle pattern whose intensity values ​​are higher than a certain statistical threshold. This threshold is typically set to 2 to 3 times the average intensity of all voxels in the pattern, used to distinguish true blood flow signals from background noise.

[0094] Next, each numerical group in this dataset, that is, each intensity value representing a blood flow signal, is precisely projected onto a relative spatial three-dimensional physical coordinate lattice system that describes the microsecond-level integration cycle of a single longitudinal scan. This relative spatial three-dimensional physical coordinate lattice system is a digitized three-dimensional mesh, in which the coordinates of each grid point are... These correspond to the transverse, longitudinal, and depth locations within the tissue, typically measured in millimeters. This dot matrix is ​​constructed based on the scanning parameters of the galvanometer and the system's axial calibration data, ensuring that each signal value corresponds one-to-one with its actual physical location within the target biological tissue.

[0095] The scanning algorithm is initiated, traversing the entire three-dimensional physical coordinate lattice system to identify and mark information voids or regions with abruptly reduced signal-to-noise ratios (SNR). Information voids or SNR-reduced regions refer to areas in the real blood flow micro-level speckle pattern where continuous, large-scale voxel values ​​are lower than the noise floor, indicating that signal acquisition in these areas is hampered by occlusion, resulting in information loss. These regions typically correspond to locations completely obscured by deep absorbing / scattering physiological structures. The scanning algorithm organizes these marked regions into a set of physically occluded dark surfaces, which constitutes the three-dimensional coordinate set of the aforementioned information void regions.

[0096] For level speckle voxel points that fall within the set of physically occluded dark surfaces and remain after cross-modal cancellation—that is, discrete points whose signal intensity is significantly higher than the surrounding noise level, or even close to or exceeds the normal blood flow signal intensity, although located within the set of dark surfaces—these points are considered to be deep effective structural signals acquired after the diffraction compensation beam successfully penetrates the occlusion. A preset opacity weight is forcibly assigned to these points. It should be noted that the weight here is a parameter used for 3D rendering, typically a floating-point number between 0 and 1. Assigning a full weight means setting its opacity to 0 or maximizing its brightness gain to visually emphasize its presence.

[0097] The system controls the graphics processing module to perform voxel interpolation smoothing and stereoscopic connected rendering operations according to the spatial shading instructions defined by the preset opacity weight.

[0098] In this embodiment, voxel interpolation smoothing is a three-dimensional image processing technique that fills the gaps between data points by calculating the weighted average value between adjacent voxels, so that discrete points can be connected into a smooth surface; stereo connectivity rendering connection operation is a rendering technique based on connectivity component analysis, which identifies spatially adjacent blood flow signal point clusters and renders them as a continuous whole, thereby forming a visual vascular network structure.

[0099] Using a standard volumetric 3D rendering pipeline, all blood flow signal points are smoothly connected and colored, with special emphasis on ensuring that points assigned high weights are displayed opaquely or highlighted in the final rendered image, thus visually bridging breaks caused by occlusion. The final output is a shadowless angiography reconstruction image of the deep choroid lining, free from shadowing artifacts caused by timing delays during the multi-step frequency sweep process.

[0100] For example, following the steps described above, the system obtains a true blood flow microlevel speckle pattern. Assuming the intensity values ​​of this pattern range from 0 to 1, the system sets a noise floor threshold of 0.1, and all voxels below this value are set to zero. After processing, a blood flow signal point with an intensity value of 0.64 is retained. This signal point is mapped to coordinates in a relative three-dimensional physical coordinate lattice system based on its index in the B-scan and A-scan. The system scan found [the location]. Near the depth, there exists a... A circular region with a radius of 0.1 mm, centered on the object, contains all voxels with zero blood flow signal values, constituting part of the physical occlusion dark surface set. However, within this dark surface set... At a certain point, there exists a discrete speckle voxel with an intensity value of 0.35. The system then assigns a preset opacity weight to this point, setting its alpha value to 1.0. Finally, the graphics processing module initiates the rendering process. It uses the maximum density projection algorithm to... The signal point at that location is rendered as a bright pixel. When rendering deeper, although the surrounding area is darkened due to occlusion, the pixel located at that location... The point at that location, due to its full weight, is also rendered as a bright pixel. By applying a smoothing kernel function during rendering, these two depth-separated points are visually connected in the final 2D projected image, forming a continuous vascular image, ultimately presenting a deep choroidal angiography reconstruction image without obvious shadow breaks.

[0101] See appendix Figure 2 This invention proposes a multi-beam optical coherence tomography and angiography method, comprising the following steps: S1. The continuous initial light wave is split to generate a central probe beam and a diffraction compensation beam with a fixed physical optical path difference. S2. Illuminate the sample to be tested together with the central probe beam and the diffraction compensation beam, and capture the scattered light flow to generate a multiplexed backscattered light field; S3. The multiplexed backscattered light field and the same source reference arm light field are interferometrically superimposed and photoelectric conversion is performed to generate a mixed continuous radio frequency interference signal. S4. Perform frequency domain transformation on the aliased continuous radio frequency interference signal, and use the mapping characteristics of the fixed physical optical path difference in the frequency domain to perform positioning and clipping, and filter out the first interference baseband data representing shallow structure information and the second interference baseband data representing deep structure information respectively. S5. Extract the continuous amplitude distribution parameters of the second interferometric baseband data, and transform them into a dynamic spatial amplitude suppression denominator term based on the shadow mutation identification condition. S6. Based on the first interference baseband data and the dynamic spatial amplitude suppression denominator term, background removal operation is performed to calculate the true blood flow micro-level speckle pattern. S7. Project the real blood flow micro-level speckle map onto the three-dimensional coordinate system, perform weight compensation and smooth rendering on the level speckle voxel points extracted from the physically occluded dark surface set, and generate a shadowless angiography reconstruction image.

[0102] Each of the modules can be implemented in whole or in part through software, hardware, or a combination thereof. It supports hardware embedded in or independent of the processor in the computer device, and also supports software stored in the memory of the computer device, so that the processor can call and execute the operations corresponding to each of the above modules.

[0103] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A multi-beam optical coherence tomography and angiography system, characterized in that, include: The heterogeneous beam generation module is used to split the continuous initial light wave, generate the center probe beam, and the diffraction compensation beam with a fixed physical optical path difference. The illumination and scattering collection module is used to illuminate the sample under test with the central probe beam and the diffraction compensation beam together, and capture the scattered light flow to generate a multiplexed backscattered light field. The photoelectric conversion and acquisition module is used to perform photoelectric conversion by interfering and superimposing the multiplexed backscattered light field with the light field of the same source reference arm, thereby generating a mixed continuous radio frequency interference signal. The frequency domain decomposition and separation module is used to perform frequency domain transformation on the aliased continuous radio frequency interference electrical signal, and to perform positioning and cutting by utilizing the mapping characteristics of the fixed physical optical path difference in the frequency domain, and to filter out the first interference baseband data representing shallow structure information and the second interference baseband data representing deep structure information respectively. The deep structure analysis module is used to extract the continuous amplitude distribution parameters of the second interferometric baseband data and transform them into a dynamic spatial amplitude suppression denominator term based on the shadow mutation identification condition. The cancellation and blood flow calculation module is used to perform background removal operations based on the first interference baseband data and the dynamic spatial amplitude suppression denominator term to calculate the true blood flow micro-level speckle pattern. The rendering and reconstruction module is used to project the real blood flow micro-level speckle map onto the three-dimensional coordinate system, perform weight compensation and smooth rendering on the level speckle voxel points extracted from the physically occluded dark surface set, and generate a shadowless angiography reconstruction image.

2. The multi-beam optical coherence tomography and angiography system according to claim 1, characterized in that, The heterogeneous beam generation module specifically includes: An optical fiber beam splitter is used to divide the acquired sample arm light wave into a first-source photon flow and a second-source optical field flow according to energy. A front-end spatial light modulator is used to assign large numerical aperture and short wavelength physical parameters to the photon flow of the first branch from the same source, generating a central probe beam with a Gaussian mode spatial distribution; A hardware spatial delay array is used to introduce the fixed physical optical path difference into the optical field flow of the second branch from the same source to form a preset path bias; A topology modulation disk is used to configure low numerical aperture, long wavelength and high-order Bessel topological charge characteristics for the optical field flow of the second branch of the same source with the fixed physical optical path difference, so as to generate a diffraction-compensated beam.

3. The multi-beam optical coherence tomography and angiography system according to claim 1, characterized in that, The illumination and scattering collection module is specifically used for: The probe beam from the center penetrates the superficial capillary network of the sample under test and extracts the change in the scattering cross-sectional area, thereby exciting the first backscattered light flow. By relying on the spatial self-healing diffraction that occurs when the diffraction compensation beam encounters a deep absorption / scattering physiological barrier, a second backscattered light flow carrying deep weak structural information is excited. The first and second backscattered light flows are captured simultaneously by a broadband microscope objective group, and the two are multiplexed and introduced into a single-mode transmission fiber to generate a multiplexed backscattered light field.

4. The multi-beam optical coherence tomography and angiography system according to claim 1, characterized in that, The photoelectric conversion and acquisition module includes a balanced photodetector array and a high-speed analog-to-digital converter. This module is specifically used for: A frequency-domain composite interference light field is formed and projected onto the photosensitive demodulation surface of the corresponding balanced photodetector array to excite and perform the photoelectric conversion physical operation of amplification, contrast, and common-mode removal; The radio frequency current value output by the balanced photodetector array is acquired by a high-speed analog-to-digital converter, generating a mixed continuous radio frequency interference signal that simultaneously carries a shallow modulation clutter spectrum and a deep diffraction continuous envelope.

5. The multi-beam optical coherence tomography and angiography system according to claim 1, characterized in that, The frequency domain decomposition and separation module is specifically used for: The parallel acceleration engine is invoked to perform fast Fourier transform calculations on the aliased continuous radio frequency interference electrical signals at each depth point, generating a full-band complex spectrum analysis matrix that contains the strength characteristics of the scattered echoes at the full depth of the target location. The pseudo-depth offset distance feature value is extracted based on the prior value generated by the fixed physical optical path difference; Using this pseudo-depth offset distance feature value, the intersecting main peak frequency bands formed by the two beams in the frequency domain distribution are located in the full-band complex spectrum analytical matrix; A rectangular spatial window filter is mounted in a specific depth range of intersecting main wave peak frequency bands to perform physical clipping, thereby filtering out the first interferometric baseband data of the shallow local spatial range and the second interferometric baseband data of the deep steady-state tissue diffraction background distribution.

6. The multi-beam optical coherence tomography and angiography system according to claim 1, characterized in that, The deep structure analysis module is specifically used for: Extract the deep steady-state diffraction envelope hidden in the second interferometric baseband data, quantize and assign values ​​to the strong structure interface, and generate continuous amplitude distribution parameters; Extract the first-order spatial partial derivative numerical matrix of the continuous amplitude distribution parameters along the longitudinal tangent direction; When the comparison confirms that the local region of the first-order spatial partial derivative numerical matrix exceeds the preset edge fluctuation threshold, it is determined to be a shadow abrupt fault caused by deep physical occlusion and the trigger condition judgment logic is activated. By using the activated trigger condition judgment logic, the continuous amplitude distribution parameters are transformed into a dynamic spatial amplitude suppression denominator term that serves as a dividing factor.

7. The multi-beam optical coherence tomography and angiography system according to claim 6, characterized in that, The cancellation and blood flow calculation module is used to perform voxel-by-voxel background removal division normalization operations. The operation process is as follows: The first interferometric baseband data is reconstructed into a dynamic complex field amplitude flow matrix; Extract the target voxel unit to be processed within the dynamic complex field amplitude flow matrix, and feed the dynamic spatial amplitude suppression denominator as the suppression reference element for data stripping into the calculation channel; The complex amplitude of the target voxel element under each scan sequence is divided point by the static amplitude within the same spatial coordinate point in the denominator by the dynamic spatial amplitude suppression term, and then spatial cross-product is performed. The normalized standard deviation of the intensity is obtained by calculating the standard deviation of the time series after division. At the baseband signal level, erroneous frequency band envelopes originating from deep, strongly reflective static tissues are suppressed, and real blood flow micro-level speckle patterns that present the true displacement dimension are separated.

8. The multi-beam optical coherence tomography and angiography system according to claim 1, characterized in that, The process of extracting the physically occluded dark surface set by the rendering and reconstruction module includes: Obtain the effective high-contrast microlevel amplitude values ​​from the real blood flow microlevel speckle pattern after removing noise points with a preset noise floor threshold. The effective high-contrast microlevel amplitude value is then projected into a three-dimensional physical coordinate lattice system. Information voids caused by low signal-to-noise ratio in the scanned 3D physical coordinate lattice system are extracted and aggregated into a set of associated physical occlusion dark surfaces.

9. The multi-beam optical coherence tomography and angiography system according to claim 1, characterized in that, The rendering operations of the rendering and reconstruction module specifically include: Identify level speckle voxel points that fall within the set of physically occluded dark surfaces and are retained after cross-modal cancellation; Force a preset opacity weight value to level speckle voxels; According to the spatial shading instructions defined by the preset opacity weight value, voxel interpolation smoothing and stereoscopic connectivity rendering connection operations are performed to output the deep choroidal shadowless angiography reconstruction image.

10. A method for multi-beam optical coherence tomography and angiography, characterized in that, Includes the following steps: S1. The continuous initial light wave is split to generate a central probe beam and a diffraction compensation beam with a fixed physical optical path difference. S2. Illuminate the sample to be tested together with the central probe beam and the diffraction compensation beam, and capture the scattered light flow to generate a multiplexed backscattered light field; S3. The multiplexed backscattered light field and the same source reference arm light field are interferometrically superimposed and photoelectric conversion is performed to generate a mixed continuous radio frequency interference signal. S4. Perform frequency domain transformation on the aliased continuous radio frequency interference signal, and use the mapping characteristics of the fixed physical optical path difference in the frequency domain to perform positioning and clipping, and filter out the first interference baseband data representing shallow structure information and the second interference baseband data representing deep structure information respectively. S5. Extract the continuous amplitude distribution parameters of the second interferometric baseband data, and transform them into a dynamic spatial amplitude suppression denominator term based on the shadow mutation identification condition. S6. Based on the first interference baseband data and the dynamic spatial amplitude suppression denominator term, background removal operation is performed to calculate the true blood flow micro-level speckle pattern. S7. Project the real blood flow micro-level speckle map onto the three-dimensional coordinate system, perform weight compensation and smooth rendering on the level speckle voxel points extracted from the physically occluded dark surface set, and generate a shadowless angiography reconstruction image.

Images