Optical coherence tomography angiography method and apparatus, and electronic device and storage medium

Abstract

Provided in the present application are an optical coherence tomography angiography method and apparatus, and an electronic device and a storage medium. The method comprises: acquiring time-domain signals of a target area, wherein the time-domain signals are N interference spectrum signals obtained by performing N repeated A-scans on the target area; performing a scale transform on each of the N interference spectrum signals on the basis of k scales, so as to obtain N×k scale transform signals; performing a Fourier transform on each scale transform signal and taking a logarithm, so as to obtain axial frequency-domain signals in a logarithmic space, and denoising the axial frequency-domain signals; dividing the N×k denoised axial frequency-domain signals into k groups of single-scale signals on the basis of the k scales, and performing de-correlation calculation on each group of single-scale signals, so as to obtain a single-scale blood flow signal; obtaining a multi-scale blood flow signal on the basis of k groups of single-scale blood flow signals; and obtaining a blood flow image on the basis of the multi-scale blood flow signal. By means of the present application, the accuracy of identifying a blood flow signal is improved.

Description

Optical coherence tomography angiography method, device, electronic device and storage medium Technical Field

[0001] The present application relates to the field of optical coherence tomography, and in particular to an optical coherence tomography angiography method, device, electronic device and storage medium. Background Art

[0002] Optical coherence tomography (OCT) is a high-sensitivity, high-resolution, high-speed, non-invasive imaging method that has been widely used in the diagnosis of fundus diseases and is of great significance for the detection and treatment of ophthalmic diseases.

[0003] As a tomographic imaging method, OCT uses the coherence of light to scan the fundus. Each scan is called an A-scan, and multiple adjacent continuous scans combined together are called a B-scan. B-scan is also the OCT cross-sectional image commonly seen and is the most important OCT imaging method in medical diagnosis.

[0004] In the field of OCT, OCTA (Optical Coherence Tomography Angiography) is a new technology developed in recent years. Retinal OCTA images can intuitively display the shape and distribution of retinal blood flow, playing a vital role in the diagnosis of fundus diseases. In a static eye, the only moving substances in the fundus are the blood cells flowing in the blood vessels. Therefore, OCTA can use the red blood cells flowing in the blood vessels as a contrast agent to measure the OCT signal differences caused by the moving cells, providing blood flow information and rapidly and non-invasively reconstructing the three-dimensional structure of the retinal choroidal vessels, thereby visualizing the vascular network. Angiography requires repeated imaging of the same location, calculating the differences between the B-scans taken multiple times to obtain the blood flow signal at the current location.

[0005] However, OCTA's high axial resolution makes it very sensitive to pulsation and jitter noise, which can easily cause imaging errors. Therefore, reducing the axial resolution is generally adopted to reduce the impact of noise; however, reducing the axial resolution may lead to a decrease in the accuracy of blood flow signal identification.

[0006] Summary of the Invention

[0007] In order to improve the accuracy of blood flow signal recognition, the present application provides an optical coherence tomography angiography method, device, electronic device and storage medium.

[0008] In a first aspect, the present application provides an optical coherence tomography angiography method, which adopts the following technical solution:

[0009] An optical coherence tomography angiography method, comprising:

[0010] Acquire a time domain signal of a target area, wherein the time domain signal is N interference spectrum signals obtained by repeating N times A scanning of the target area;

[0011] Performing a scale transformation on each of the N interference spectrum signals based on k scales to obtain N×k scale-transformed signals;

[0012] The step of performing scale transformation on any of the interference spectrum signals based on the k scales includes: decomposing the any interference spectrum signal based on each of the K scales to obtain the k scale transformation signals corresponding to the any interference spectrum signal;

[0013] Performing Fourier transformation on each of the scale-transformed signals and taking the logarithm thereof to obtain an axial frequency domain signal in a logarithmic space, and denoising the axial frequency domain signal;

[0014] Dividing the N×k denoised axial frequency domain signals into k groups of single-scale signals based on k scales, and performing decorrelation calculation on each group of single-scale signals to obtain a single-scale blood flow signal;

[0015] obtaining a multi-scale blood flow signal based on the k groups of single-scale blood flow signals;

[0016] A blood flow image is obtained based on the multi-scale blood flow signal.

[0017] By adopting the above technical solution, the blood flow signal is a non-steady-state signal, and its changes are localized. An interference spectrum signal is decomposed into multiple segments after a scale transformation. During the subsequent decorrelation calculation, the axial frequency domain signals corresponding to each small segment of the interference spectrum signal corresponding to the same target area at different times can be compared. Through segmented comparison, the changing signals in the non-steady state can be captured more accurately, thereby achieving accurate extraction of the blood flow signal.

[0018] Furthermore, by transforming an interference spectrum signal at multiple scales, each scale corresponds to a different number of decomposition segments, meaning that the scales of segmented comparison vary. This allows for measuring changes in non-stationary signals at different scales. Therefore, simultaneously transforming and fusing an interference spectrum signal at multiple scales can reduce the error rate associated with a single scale, improving the accuracy of blood flow signal extraction and processing efficiency. Furthermore, by preserving all points in a single interference spectrum signal, more image detail can be restored, improving the signal-to-noise ratio.

[0019] In a possible implementation, the interference spectrum signal corresponds to a sampling time period, and any interference spectrum signal is decomposed based on each of the K scales to obtain k scale-transformed signals corresponding to the any interference spectrum signal, including:

[0020] Decomposing the any interference spectrum signal based on any scale among the k scales to obtain a scale-transformed signal of the any interference spectrum signal corresponding to the any scale, the operation includes:

[0021] Determine a decomposed signal corresponding to any one of the scales, where the decomposed signal is a signal within the sampling time period T, and the decomposed signal includes at least one time group, where the time group includes an adjacent first time period and a second time period, two adjacent first time periods, or two adjacent second time periods;

[0022] Multiplying any interference spectrum signal corresponding to the first time period by a forward signal to obtain a first signal;

[0023] The second signal is obtained by multiplying any interference spectrum signal corresponding to the second time period by an inverse signal; and a scale transformation signal of the any interference spectrum signal corresponding to the any scale is obtained based on at least one first signal and at least one second signal.

[0024] By adopting the above technical solution, the sampling time period T is divided based on at least one time group corresponding to the scale, and the interference spectrum signal is divided into at least one first signal and at least one second signal based on the forward signal and the reverse signal, thereby realizing segmented segmentation of the interference spectrum signal and obtaining a scale transformation signal corresponding to the scale.

[0025] In a possible implementation, denoising the axial frequency domain signal in the logarithmic space to obtain the denoised axial frequency domain signal includes:

[0026] Determine the segmentation threshold;

[0027] If the amplitude of the axial frequency domain signal is greater than or equal to the segmentation threshold, then the denoised axial frequency domain signal is equal to the axial frequency domain signal in the logarithmic space;

[0028] If the amplitude of the axial frequency domain signal is less than the segmentation threshold, the denoised axial frequency domain signal is equal to the segmentation threshold.

[0029] By adopting the above technical solution, the axial frequency domain signal is subjected to noise reduction division based on the segmentation threshold to improve the signal-to-noise ratio.

[0030] In a possible implementation, determining the segmentation threshold includes:

[0031] Extracting noise signals corresponding to all the axial frequency domain signals based on a clustering algorithm;

[0032] A segmentation threshold is determined based on the noise signal.

[0033] By adopting the above technical solution, the noise signal is extracted and the segmentation threshold is determined by unsupervised learning, and the threshold is automatically determined by the axial frequency domain signal, which improves the accuracy of determining the segmentation threshold and improves the signal-to-noise ratio.

[0034] In a possible implementation, performing decorrelation calculation on each group of the single-scale signals to obtain a single-scale blood flow signal includes:

[0035] Determining first-order statistics, second-order statistics, and third-order statistics of each group of single-scale signals;

[0036] The single-scale blood flow signal is determined by performing decorrelation calculation based on the first-order statistics, the second-order statistics, and the third-order statistics.

[0037] By adopting the above technical solution, high-order statistics can be used to extract more useful information from single-scale signals and perform decorrelation calculations, thereby improving the accuracy of single blood flow scale signal extraction.

[0038] In a possible implementation, after performing Fourier transform on each of the scale-transformed signals and taking the logarithm to obtain an axial frequency domain signal in a logarithmic space, the method further includes:

[0039] Performing an alignment operation on all the axial frequency domain signals based on a phase correlation algorithm;

[0040] The denoised axial frequency domain signal is a signal obtained by denoising the axial frequency domain signal after the alignment operation.

[0041] By adopting the above technical solution, the alignment operation can reduce the interference of jitter on the axial frequency domain signal, thereby achieving the effect of improving the signal-to-noise ratio.

[0042] In a possible implementation manner, before performing scale transformation on each of the N interference spectrum signals based on k scales, the method further includes: performing dispersion compensation on each of the interference spectrum signals in the time domain signal.

[0043] In a second aspect, the present application provides an optical coherence tomography angiography device, which adopts the following technical solution:

[0044] An optical coherence tomography angiography device, comprising:

[0045] An acquisition module is used to acquire a time domain signal of a target area, wherein the time domain signal is N interference spectrum signals obtained by repeating N times of A scanning on the target area;

[0046] a scale transformation module, configured to perform scale transformation on each of the N interference spectrum signals based on k scales to obtain N×k scale transformation signals;

[0047] The step of performing scale transformation on any of the interference spectrum signals based on the k scales includes: decomposing the any interference spectrum signal based on each of the K scales to obtain the k scale transformation signals corresponding to the any interference spectrum signal;

[0048] a frequency domain transformation module, configured to perform Fourier transformation on each of the scale-transformed signals and take the logarithm thereof to obtain an axial frequency domain signal in a logarithmic space, and to denoise the axial frequency domain signal;

[0049] a decorrelation module, configured to divide the N×k denoised axial frequency domain signals into k groups of single-scale signals based on k scales, and perform decorrelation calculation on each group of single-scale signals to obtain a single-scale blood flow signal;

[0050] a fusion module, configured to obtain a multi-scale blood flow signal based on the k groups of single-scale blood flow signals;

[0051] An imaging module is used to obtain a blood flow image based on the multi-scale blood flow signal.

[0052] By adopting the above technical solution, the blood flow signal is a non-steady-state signal, and its changes are localized. An interference spectrum signal is decomposed into multiple segments after a scale transformation. During the subsequent decorrelation calculation, the axial frequency domain signals corresponding to each small segment of the interference spectrum signal corresponding to the same target area at different times can be compared. Through segmented comparison, the changing signals in the non-steady state can be captured more accurately, thereby achieving accurate extraction of the blood flow signal.

[0053] Furthermore, by transforming an interference spectrum signal at multiple scales, each scale corresponds to a different number of decomposition segments, meaning that the scales of segmented comparison vary. This allows for measuring changes in non-stationary signals at different scales. Therefore, simultaneously transforming and fusing an interference spectrum signal at multiple scales can reduce the error rate associated with a single scale, improving the accuracy of blood flow signal extraction and processing efficiency. Furthermore, by preserving all points in a single interference spectrum signal, more image detail can be restored, improving the signal-to-noise ratio.

[0054] In a possible implementation, the scale conversion module, when decomposing any interference spectrum signal based on any scale to obtain the scale conversion signal, is specifically configured to:

[0055] The interference spectrum signal corresponds to a sampling time period T;

[0056] Determine a decomposed signal corresponding to any of the scales, where the decomposed signal is a signal within the sampling time period T, and the decomposed signal includes at least one time group, where the time group includes an adjacent first time period and a second time period, two adjacent first time periods, or two adjacent second time periods;

[0057] Multiplying the interference spectrum signal corresponding to the first time period by the forward signal to obtain a first signal;

[0058] The interference spectrum signal corresponding to the second time period is multiplied by an inverse signal to obtain a second signal; and the scale transformation signal is obtained based on at least one first signal and at least one second signal.

[0059] In a possible implementation, the apparatus further includes a denoising module, which denoises the axial frequency domain signal in the logarithmic space to obtain the denoised axial frequency domain signal, specifically configured to:

[0060] Determine the segmentation threshold;

[0061] If the amplitude of the axial frequency domain signal is greater than or equal to the segmentation threshold, then the denoised axial frequency domain signal is equal to the axial frequency domain signal in the logarithmic space;

[0062] If the amplitude of the axial frequency domain signal is less than the segmentation threshold, the denoised axial frequency domain signal is equal to the segmentation threshold.

[0063] In a possible implementation, when determining the segmentation threshold, the denoising module is specifically configured to:

[0064] Extracting noise signals corresponding to all the axial frequency domain signals based on a clustering algorithm;

[0065] A segmentation threshold is determined based on the noise signal.

[0066] In a possible implementation, when performing decorrelation calculation on each group of the single-scale signals to obtain the single-scale blood flow signal, the decorrelation module is specifically configured to:

[0067] Determining first-order statistics, second-order statistics, and third-order statistics of each group of single-scale signals;

[0068] The single-scale blood flow signal is determined by performing decorrelation calculation based on the first-order statistics, the second-order statistics, and the third-order statistics.

[0069] In a possible implementation, the device further includes an alignment module, which is specifically configured to: after obtaining the axial frequency domain signal in the logarithmic space,

[0070] Performing an alignment operation on all the axial frequency domain signals based on a phase correlation algorithm;

[0071] The denoised axial frequency domain signal is a signal obtained by denoising the axial frequency domain signal after the alignment operation.

[0072] In a possible implementation, the apparatus further includes a dispersion compensation module, where the dispersion compensation module is specifically configured to:

[0073] Before performing scale transformation on each of the N interference spectrum signals based on k scales, dispersion compensation is performed on each of the interference spectrum signals in the time domain signal.

[0074] In a third aspect, the present application provides an electronic device, which adopts the following technical solution:

[0075] An electronic device, comprising:

[0076] one or more processors;

[0077] Memory;

[0078] One or more computer programs are stored in the memory and configured to be executed by one or more processors, and the one or more computer programs are configured to: execute the above-mentioned optical coherence tomography angiography method.

[0079] In a fourth aspect, the present application provides a computer-readable storage medium, which adopts the following technical solution:

[0080] A computer-readable storage medium includes: a computer program stored therein that can be loaded by a processor and execute the above-mentioned optical coherence tomography angiography method.

[0081] In summary, this application has at least the following beneficial technical effects:

[0082] After undergoing a scale transformation, an interference spectrum signal is decomposed into multiple segments. During the subsequent decorrelation calculation, the axial frequency domain signals corresponding to each small segment of the interference spectrum signal at different times and within the same target region can be compared. This segmented comparison allows for more accurate capture of changing signals in the non-stationary state, enabling accurate extraction of blood flow signals. Furthermore, all points in a single interference spectrum signal are retained, restoring more image detail and improving the signal-to-noise ratio. Simultaneously, by transforming an interference spectrum signal at multiple scales, each scale corresponds to a different number of decomposition segments, i.e., the scales of the segmented comparisons are different, allowing the changes in the non-stationary signal to be measured at different scales. Therefore, simultaneously transforming an interference spectrum signal and fusing multiple scales can reduce the misjudgment rate at a single scale, improving the accuracy of blood flow signal extraction and processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] FIG1 is a schematic flow chart of an optical coherence tomography angiography method according to an embodiment of the present application;

[0084] FIG2 is a schematic diagram of performing multi-scale transformation on a time domain signal;

[0085] FIG3 is a diagram of (a) a glass membrane blood flow image provided by an embodiment of the present application; (b) a deep layer image; and (c) a superficial layer blood flow image.

[0086] FIG4 is a flow chart of steps S021 to S023 of an embodiment of the present application;

[0087] FIG5 is a flow chart of steps S031 to S033 of an embodiment of the present application;

[0088] FIG6 is a schematic diagram of decorrelation calculation based on a single-scale signal according to an embodiment of the present application;

[0089] FIG7 is a schematic diagram of an optical coherence tomography angiography apparatus according to an embodiment of the present application;

[0090] FIG8 is a schematic diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0091] The present application is further described in detail below with reference to Figures 1 to 8.

[0092] After reading this specification, those skilled in the art may make non-creative modifications to this embodiment as needed, but such modifications are protected by patent law as long as they fall within the scope of the claims of this application.

[0093] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0094] In this document, the term "and / or" simply describes a relationship between related objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document, unless otherwise specified, generally indicates an "or" relationship between the related objects.

[0095] In order to facilitate understanding of the technical solutions proposed in this application, several elements that will be introduced in the description of this application are first introduced here. It should be understood that the following introduction is only for the convenience of understanding these elements, so as to understand the content of the embodiments of this application, and does not necessarily cover all possible situations.

[0096] (1) A-scan: The depth reflection profile reflected from the sample in the depth direction. One A-scan obtains an interference spectrum signal. The image information of one axial depth of the sample is contained in the interference spectrum signal. Through its Fourier transform, one axial depth image of one sample can be obtained.

[0097] (2) B-scan: contains multiple adjacent, continuous, and different position interference spectrum signals (A-scan signals), for example: contains N1 interference spectrum signals, each interference spectrum signal contains X points, and these spectra form an N1×X data packet; specifically, time-domain optical coherence tomography technology requires depth scanning, which is not conducive to high-speed real-time imaging of OCT; while the frequency-domain optical coherence tomography system can obtain all the information in the depth direction (Z direction) (A-scan signal) by Fourier transforming the coherent spectra of the sample arm and the reference arm. It does not require mechanical scanning in the depth direction (A-scan), but only requires lateral scanning (X-direction scanning and Y-direction scanning). One X-direction scan forms a B-scan image, and a Y-direction scan forms an OCT three-dimensional signal.

[0098] (3) Axial resolution of the OCTA system: refers to the minimum distance that can be distinguished by axial depth tomography of the sample. The number of X points contained in each interference spectrum signal is determined by the axial resolution of the OCTA system. The lateral resolution of the OCTA image is independent of the axial resolution. The OCTA system can provide an axial resolution of approximately (1-15 μm) that is independent of the lateral resolution. The axial resolution mainly depends on the coherence length of the width light source.

[0099] (4) Signal-to-noise ratio: It is an indicator for image evaluation by calculating the ratio of signal to noise in the image. It is a global evaluation standard. For OCTA images, the larger the signal-to-noise ratio, the less speckle noise in the image, the better the image denoising effect, and the better the image quality.

[0100] The present application provides an optical coherence tomography angiography method, which is performed by an electronic device. Referring to FIG. 1 , the method includes:

[0101] Step S101: Acquire a time domain signal of a target area.

[0102] Among them, the time domain signal is N interference spectrum signals obtained by repeating A-scan on the target area N times. The expression of the time domain signal is I(i,x,t), i=1,2,...N, x is the scanning position (i.e., the target area), t is the scanning time point under the current A-scan, and the duration corresponding to each A-scan is the sampling time period T. Generally, the scanning rate of A-scan is usually configured to 7000 times per second, 10000 times per second, etc.

[0103] Among them, the interference spectrum signal collected by an A-scan is Fourier transformed to obtain the sample depth information. The more data points corresponding to an interference spectrum signal, the more points there are in the image after Fourier transformation, and the higher the resolution of the image. For angiography, high-resolution images can more accurately identify blood flow signals in different tissues and different layers.

[0104] The duration of a single A-scan (sampling time period T) is manually set. When configuring a single A-scan, factors such as incident light power, sample reflectivity, measurement depth, and the speed at which depth information is acquired must be considered. For example, in weakly scattering media, the sensitivity of the OCT system determines the minimum detectable light signal; in strongly scattering media, the sensitivity of the OCT system determines the maximum imaging depth. Extending the signal acquisition time of a single A-scan and reducing the speed at which depth information is acquired can improve OCT sensitivity.

[0105] When analyzing stationary signals, the most commonly used and main analysis method is Fourier transform. However, since Fourier analysis uses a global transform (corresponding to a sampling time period T), it cannot express the time-frequency local properties of the signal. The time-frequency local properties are precisely the most fundamental properties of non-stationary signals. The blood flow signal studied in the embodiment of this application is a non-stationary signal.

[0106] To extract changes in blood flow signals, the related split-spectrum amplitude-decorrelation angiography (SSADA-OCT) has proposed an improved algorithm. SSADA divides the entire OCT spectrum into several narrow bands, calculates the correlation between different B-scans in these bands, and then averages them. The specific steps are: during the data processing process, the spectrum is separated using bandpass filters at different positions (for example, four equally spaced bandpass filters). This step is called spectral splitting; then, M independent decorrelation images are obtained based on the relationship between each pair of B-scans. The obtained images have a lower axial resolution and the same lateral resolution.

[0107] However, although the above-mentioned spectrum splitting method can obtain more accurate changes in blood flow signals and improve the signal-to-noise ratio, it discards some data points on the interference spectrum signal, thereby reducing the axial resolution of the restored image.

[0108] In order to retain high axial resolution and accurately extract the changing blood flow signal, the embodiment of the present application retains all data points corresponding to an A-scan, decomposes the signal corresponding to an A-scan into multiple unit signals (i.e., a scale transformation) based on the nature of local changes in the blood flow signal, and analyzes the changes in the blood flow signal within each unit signal to accurately extract the blood flow signal.

[0109] Step S102 : performing scale transformation on each of the N interference spectrum signals based on k scales to obtain N×k scale-transformed signals.

[0110] The scaling of any interference spectrum signal based on k scales includes: decomposing any interference spectrum signal based on each scale to obtain k scale-transformed signals corresponding to any interference spectrum signal.

[0111] When performing a scale transformation on the interference spectrum signal, the interference spectrum signal is decomposed into multiple unit signals. When performing decorrelation calculation, the axial frequency domain signals corresponding to each unit signal can be compared. The segmented comparison can more accurately capture the non-steady-state change signal in a single interference spectrum signal, thereby realizing the accurate extraction of the blood flow signal.

[0112] The interference spectrum signal is transformed into multiple scales, and the number of decomposition segments corresponding to each scale is different, that is, the scale of segmented comparison changes, and the non-steady-state change signal can be measured at different scales.

[0113] Compared with the spectrum splitting method, there may be overlap between the frequency bands during spectrum splitting, and repeated calculations will lead to reduced calculation efficiency. In the embodiment of the present application, not only can all points in an interference spectrum signal be retained, but repeated calculation of data can also be avoided.

[0114] Step S103: Perform Fourier transform on each scale-transformed signal and take its logarithm to obtain an axial frequency domain signal in a logarithmic space.

[0115] Step S104 : dividing the N×k denoised axial frequency domain signals into k groups of single-scale signals based on k scales, and performing decorrelation calculation on each group of single-scale signals to obtain a single-scale blood flow signal.

[0116] For the embodiment of the present application, the principle of denoising the axial frequency domain signal is as follows: the quantification effect of the decorrelation algorithm on the motion contrast has a significant dependence on the noise level of the original A-scan signal. As the signal intensity decays (such as in deep tissue areas), random noise will gradually occupy the main component, and will also produce a larger decorrelation value, bringing decorrelation artifacts. For example: the motion contrast usually generated based on the decorrelation operation cannot distinguish between the randomness of the noise and the decorrelation caused by the movement of red blood cells. Therefore, areas with relatively weak signal-to-noise ratios are easily misjudged as blood flow signal areas, thereby seriously affecting the contrast of the blood flow image. After denoising, the denoised axial frequency domain signal can reduce or eliminate the error caused by the random noise signal in the subsequent decorrelation degree calculation.

[0117] Step S105: obtaining a multi-scale blood flow signal based on the k groups of single-scale blood flow signals.

[0118] Among them, a B-scan signal includes multiple A-scan signals at different positions. After processing through steps S101 to S104, the A-scan input signal at any position is the multi-scale blood flow signal corresponding to any position; by analyzing the scale blood flow signal at each position in a B-scan signal, the vascular signal of the entire structural section can be obtained.

[0119] According to the embodiment of the present application, a scale transformation and multiple scale fusion are simultaneously performed on an initially acquired interference spectrum signal, which can improve the extraction accuracy and processing efficiency of the blood flow signal.

[0120] Step S106: Obtain a blood flow image based on the multi-scale blood flow signal.

[0121] Specifically, visualization software was used to convert multi-scale blood flow signals into vascular images. Figure 3 shows the corresponding images of (a) the vitreous membrane blood flow image, (b) the deep layer image, and (c) the superficial layer blood flow image of the fused multi-scale blood flow signals. The fusion of multiple scales reduces the misjudgment rate of single-scale analysis and improves the signal-to-noise ratio, enabling accurate identification of blood flow signals corresponding to different tissues and layers.

[0122] A possible implementation of the embodiment of the present application, referring to FIG. 2 and FIG. 4 , in step S102 : decomposing any interference spectrum signal based on any scale to obtain a scale-transformed signal, includes:

[0123] Step S021: Determine the decomposition signal corresponding to any scale.

[0124] Each scale corresponds to a preset decomposition signal, and the form of the decomposition signal is as follows:

[0125] [1,1,-1,-1]

[0126] [1,1,-1,-1]

[0127] [1,-1,-1,1]

[0128] [1,-1,1,-1]

[0129] …

[0130] The decomposed signal is a signal within a sampling time period T, and the decomposed signal includes at least one time group, and the time group includes an adjacent first time period and a second time period, two adjacent first time periods, or two adjacent second time periods.

[0131] Take [1,1,-1,-1] as an example: the decomposed signal corresponds to two time groups, namely the first time group (T1,T1) and the second time group (T2,T2). The first time group is located in front of the second time group within the sampling time period T.

[0132] In addition, taking [1,-1,-1,1] as an example, the decomposed signal corresponds to two time groups, namely the first time group (T1, T2) and the second time group (T2, T1). The first time group is located in front of the second time group within the sampling time period T.

[0133] Furthermore, in order to conveniently represent the k decomposition signals, the k decomposition signals corresponding to the k scales can be arranged in rows to form a non-zero orthogonal signal S. The non-zero orthogonal signal S can be expressed as:

[0134]

[0135] Wherein, Walsh(k) represents the k-dimensional Walsh transformation matrix;

[0136] is the Kronecker product;

[0137] M is the number of current A-scan sampling points;

[0138] j=1,2,......k;

[0139] t is the time scale of the signal;

[0140] For example, when k=4,

[0141]

[0142] Step S022: multiply the interference spectrum signal corresponding to the first time period by the forward signal to obtain a first signal, and multiply the interference spectrum signal corresponding to the second time period by the reverse signal to obtain a second signal.

[0143] Among them, the first signal and the second signal are different. Referring to Figure 2, the decomposed signal illustrated in the embodiment of the present application is a rectangular wave signal. In the embodiment of the present application, the first signal and the second signal also include other forms of signals, for example: the first signal and the second signal can jump smoothly.

[0144] Continuing with the previous example, 1 is the forward signal, -1 is the reverse signal, T1 corresponds to the forward signal, and T2 corresponds to the reverse signal.

[0145] Step S023: Obtain a scale-transformed signal based on the at least one first signal and the at least one second signal.

[0146] The obtained scale transformation signal is shown in the following formula (2):

[0147] I(i,x,t)*[1,1,-1,-1] Formula (2)

[0148] Furthermore, each of the N interference spectrum signals is scale-transformed based on k scales to obtain N×k scale-transformed signals, which can be expressed as:

[0149] I Multi_Scalar (i,j,x,t)=I(i,x,t)*S(j,t) Formula (3)

[0150] Among them, I Multi_Scalar (i, j, x, t) contains N×k scale-transformed signals.

[0151] For the embodiments of the present application, different scales and decomposition signals corresponding to the scales can be set according to different application requirements. After scale transformation, each interference spectrum signal corresponds to k scale transformation signals, and N interference spectrum signals obtain a total of N×k scale transformation signals.

[0152] A possible implementation of the embodiment of the present application includes, in step S101, after acquiring the time domain signal of the target area, and before performing scale transformation on each of the N interference spectrum signals based on k scales, further comprising: performing dispersion compensation on each interference spectrum signal in the time domain signal.

[0153] The axial resolution of the OCT system is inversely proportional to the half-width of the interference spectrum signal. In the actual system optical path, the different lengths of the dispersive material between the reference arm and the sample arm will lead to dispersion mismatch and introduce phase changes in the interference spectrum signal, causing the spectrum of the interference spectrum signal to be broadened, which may cause a decrease in the axial resolution of the OCTA image and a reduction in the signal-to-noise ratio.

[0154] Dispersion mismatch is known to originate from system components such as collimators, scanning objectives, optical fibers, and the sample to be measured. For different depths of the sample, the length of the dispersive medium through which the probe light passes is different, resulting in different amounts of dispersion mismatch. This results in different axial resolutions at different depths of the sample. Furthermore, the k-nonlinear acquisition of spectral signals introduces a phase factor related to the optical path difference between the two arms and the wave vector k, which has the same impact as dispersion mismatch. Therefore, in order to keep the axial resolution at different depths within the imaging depth range close to the theoretical axial resolution, it is particularly important to compensate for dispersion as it changes with depth.

[0155] Currently, OCTA system dispersion compensation methods can be divided into two categories: physical compensation and numerical compensation. The physical compensation method compensates for the dispersion mismatch between the two arms by adding hardware devices to the reference arm. The physical compensation method requires the addition and modification of hardware devices in the optical path, which complicates the system, introduces additional noise, and increases costs. The numerical compensation method only requires subsequent processing of the data collected by OCTA to eliminate the dispersion mismatch. The embodiment of the present application uses the numerical compensation method to perform dispersion compensation on the time domain signal.

[0156] Specifically, there are several numerical compensation methods:

[0157] The first is the iterative compensation method, which uses the spatial image quality of the A-scan (e.g., sharpness, Rayleigh entropy, etc.) as the evaluation function to iteratively search for the dispersion compensation coefficient that achieves the best compensation effect. MarKs et al. used an iterative method based on the golden section method to compensate for dispersion and improve the resolution of the OCTA system.

[0158] The second is the deconvolution method, which calculates the dispersion parameter convolution kernel that varies with sample depth based on the sample structure and material, and eliminates the dispersion mismatch through deconvolution operation;

[0159] The third is the short-time Fourier transform method, which uses a window function to divide the spectrum and calculate the dispersion phase that should be compensated for light of different frequencies;

[0160] The fourth is the spatial domain extraction fitting method, which sets a window function according to a certain threshold in the spectrum of the interference spectrum signal, extracts the Gaussian peak signal at different depths of the sample, and then uses the fitting method to remove the second-order and third-order dispersion terms representing the dispersion mismatch in the corresponding spectrum, thereby compensating for the dispersion mismatch at different depths;

[0161] The fifth is the dispersion coefficient fitting method, which uses an iterative method to pre-calculate the dispersion coefficient of the sample at different depths, thereby obtaining the material information of the sample and then compensating for the weak signal at deeper positions of the sample.

[0162] In an embodiment of the present application, after dispersion compensation is performed on the time domain signal in step S101 to obtain an updated time domain signal, in step S102, scale transformation is performed on each interference spectrum signal in the dispersion compensated time domain signal based on k scales to obtain N×k scale-transformed signals.

[0163] OCTA data acquisition requires continuous acquisition of B-scans at the same location with a certain time interval. The microvascular image is reconstructed by comparing the decorrelated signals between the B-scans. During the data acquisition process, it is inevitably affected by sample motion, environmental noise, system vibration, etc., which makes the data discontinuous and manifests as distortion and fracture of the vascular image, generating motion artifacts in the image. The artifacts caused by sample motion will seriously reduce the imaging quality of OCTA and affect the quantitative study of tissue microvessels.

[0164] Among them, sample motion has multiple sources, such as breathing and heartbeat, involuntary shaking of the measured part, etc. Even very small movements, such as micron-level movements caused by heartbeats in the skin, will increase background noise and ultimately affect the vascular imaging effect. The longer the time interval between data acquisition, the more serious the motion artifact. Therefore, in the OCTA vascular extraction algorithm, eliminating motion artifacts is a key content, and the main method for artifact elimination is image registration.

[0165] Specifically, image registration includes methods such as transformation, interpolation, and displacement, the purpose of which is to correct the distortion and fracture caused by motion in the image to obtain an image without motion artifacts.

[0166] For the embodiment of the present application, the main influencing factor for reducing the signal-to-noise ratio is translation. Therefore, a possible implementation method of the embodiment of the present application, after step S103: performing Fourier transform on each scale-transformed signal and taking the logarithm to obtain the axial frequency domain signal in the logarithmic space, also includes: performing an alignment operation on all axial frequency domain signals based on a phase correlation algorithm; wherein the denoised axial frequency domain signal is a signal obtained by denoising the axial frequency domain signal after the alignment operation.

[0167] The phase correlation algorithm can handle image translation, and the Fourier-Mellin (FMT) algorithm combined with the logarithmic polar coordinate model is an extension of the Fourier phase correlation algorithm. FMT can directly process rotated and scaled images. This frequency domain-based method uses the entire content information of the image for alignment.

[0168] Specifically, the process of the phase correlation algorithm for translation processing is:

[0169] Assume f1(x,y) and f2(x,y) are only translated (x0, y0):

[0170] f2(x,y)=(x-x0,y-y0) Formula (4)

[0171] After Fourier transform:

[0172]

[0173] The cross power spectrum of the two images is:

[0174]

[0175] in is the complex conjugate of F2(ξ,η). Performing a two-dimensional inverse Fourier transform on formula (6) yields an impulse function that is approximately 1 at (x0, y0) and approximately 0 at other locations. This allows the translation parameter (x0, y0) to be determined.

[0176] Based on the above-mentioned phase correlation algorithm, all axial frequency domain signals in the logarithmic space in step S103 are aligned based on the above-mentioned phase correlation algorithm to obtain all aligned axial frequency domain signals, and then all aligned axial frequency domain signals are denoised to obtain denoised axial frequency domain signals.

[0177] In a possible implementation of the embodiment of the present application, referring to FIG. 5 , after step S103 , denoising the axial frequency domain signal in the logarithmic space to obtain a denoised axial frequency domain signal includes:

[0178] Step S031: Determine the segmentation threshold.

[0179] The denoising method for decorrelation artifacts is as follows: when counting the changes in A-scan signals, an intensity threshold is set to isolate the impact of noise signals on OCTA. By generating an intensity mask to remove all signals with low signal-to-noise ratios, the error caused by the random noise signal in the subsequent decorrelation degree calculation can be reduced or eliminated. The segmentation threshold in the embodiment of the present application is the intensity mask. Usually, the segmentation threshold is set based on human experience. In order to further improve the accuracy, in the embodiment of the present application, the segmentation threshold is determined, including: extracting the noise signals corresponding to all axial frequency domain signals based on a clustering algorithm; and determining the segmentation threshold based on the noise signal.

[0180] The clustering method is to find outliers in the signal by discovering clusters in the signal and then delete them to achieve the purpose of removing noise signals. In the embodiment of the present application, the signal falling outside the cluster is the noise signal.

[0181] Among them, the K-means algorithm is a classic distance-based clustering algorithm. It takes K as a parameter input, randomly selects K center points, and finally divides n objects into K clusters. In these K clusters, members in the same cluster have high similarity, and members in different clusters have high dissimilarity. Specifically, the cluster center in the K-means clustering algorithm is determined by calculating the mean of the attributes of all signal objects in a cluster. By performing K-means clustering analysis on the image signal intensity, the distribution of the noise signal class can be obtained.

[0182] Specifically, the process of extracting noise signals based on the K-means clustering algorithm includes step S11 (not shown in the figure), step S12 (not shown in the figure), step S13 (not shown in the figure), step S14 (not shown in the figure) and step S15 (not shown in the figure):

[0183] Step S11: Use the farthest-first strategy to select K cluster centers as current cluster centers, where K is a natural number;

[0184] Step S12: clustering all axial frequency domain signals according to the current cluster center, and clustering each axial frequency domain signal into the cluster represented by the cluster center closest to it;

[0185] Step S12: Calculate the mean of each current cluster as the new cluster center;

[0186] Step S13: determine whether the new cluster center is the same as the previous cluster center. If so, execute step S14; if not, use the new cluster center as the current cluster center and loop through steps S12 to S13.

[0187] Step S14: Calculate the distance between any two cluster centers among all new cluster centers;

[0188] Step S15: Determine whether the distance between any two cluster centers is greater than the set reference threshold. If so, extract the clusters whose distance between any two cluster centers is greater than the set reference threshold, and use the axial frequency domain signals corresponding to the filtered clusters as noise signals; if not, output information indicating that there is no noise signal.

[0189] After obtaining the noise signal, the noise signal is distributed and statistically analyzed to obtain the mean and variance of the noise signal, and then the segmentation threshold is determined based on the mean and variance. Specifically, the method for selecting the optimal threshold includes: the maximum inter-class variance method; the method for determining the threshold based on the mean and variance is known to those skilled in the art and will not be repeated here.

[0190] After determining the segmentation threshold, the axial frequency domain signal can be segmented based on the signal intensity corresponding to the segmentation threshold, separating the effective target from the noise to improve the signal-to-noise ratio of the image.

[0191] Step S032: If the amplitude of the axial frequency domain signal is greater than or equal to the segmentation threshold, the denoised axial frequency domain signal is equal to the axial frequency domain signal in the logarithmic space.

[0192] Step S033: If the amplitude of the axial frequency domain signal is less than the segmentation threshold, the denoised axial frequency domain signal is equal to the segmentation threshold.

[0193] Step S032 and step S033 are specifically as follows:

[0194]

[0195] in, is the axial frequency domain signal in the logarithmic space after denoising;

[0196] th is the segmentation threshold;

[0197] j is any scale among k scales;

[0198] F Multi_Scalar (i, j, x, z) is the axial frequency domain signal in the logarithmic space;

[0199] After obtaining the denoised axial frequency domain signal, based on the principle of the decorrelation algorithm, the changes in signal correlation caused by blood flow are used to perform differential analysis on adjacent B-scan images to extract the blood flow signal.

[0200] A possible implementation of the embodiment of the present application, referring to FIG6 , is as follows: in step S105, decorrelation calculation is performed on each group of single-scale signals to obtain a single-scale blood flow signal, including: determining the first-order statistics, second-order statistics, and third-order statistics of each group of single-scale signals; and performing decorrelation calculation based on the first-order statistics, second-order statistics, and third-order statistics to determine the single-scale blood flow signal.

[0201] in,

[0202] The first-order statistics are:

[0203] The second-order statistics are:

[0204] The third-order statistics are:

[0205] From a statistical perspective, the statistical characteristics of a normally distributed random variable (vector) can be fully expressed by first-order and second-order statistics. For example, for a Gaussian distributed random vector, if its mathematical expectation and covariance matrix are known, its joint probability density function can be known; for a Gaussian random process, if its mean and autocorrelation function (or autocovariance function) are known, its probability structure can be known, that is, its entire statistical characteristics can be known.

[0206] However, when the random variable or random process does not obey the Gaussian distribution, the first-order and second-order statistics cannot fully represent its statistical characteristics; in other words, not all the information is contained in the first- and second-order statistics, and higher-order statistics also contain a lot of useful information.

[0207] The main motivations for using higher-order statistics in signal processing include: 1) suppressing the effects of additive colored noise of unknown power spectrum; 2) identifying non-minimum phase systems or reconstructing non-minimum phase signals; 3) extracting various information due to Gaussian deviations; 4) detecting and characterizing nonlinearities in signals and even nonlinear systems.

[0208] Therefore, high-order statistical signal processing methods are a way to extract useful information from the higher-order statistics of non-Gaussian signals, especially information that cannot be extracted from first- and second-order statistics. From this perspective, high-order statistical methods are not only an important supplement to random signal processing methods based on correlation functions or power spectra, but also provide a means to solve many signal processing problems that second-order statistical methods cannot solve.

[0209] A single-scale blood flow signal is determined by decorrelation calculation based on first-order statistics, second-order statistics, and third-order statistics, including:

[0210] Flow0(j,x,z)=α1J3(j,x,z)+α2J2(j,x,z)*J1(j,x,z)+α3J1(j,x,z) 3 +α4J2(j,x,z)+α5J1(j,x,z) 2 Formula (11)

[0211] Considering the sum of the cubic error and 2 times the variance of the blood flow signal, the following relationship is expressed:

[0212]

[0213]

[0214] Therefore, [α1,α2,α3,α4,α5] can be taken as [1,-3,2,2,-2].

[0215] In a possible implementation of the embodiment of the present application, in step S106, a multi-scale blood flow signal is obtained based on k groups of single-scale blood flow signals, including: superimposing the k groups of single-scale blood flow signals to obtain the multi-scale blood flow signal.

[0216] Flow(x,z)=∑ j Flow0(j,x,z) Formula (13)

[0217] The above embodiment introduces an optical coherence tomography angiography method from the perspective of method flow. The following embodiment introduces an optical coherence tomography angiography device 100 from the perspective of a virtual module or virtual unit. Please refer to the following embodiment for details.

[0218] The present invention provides an optical coherence tomography angiography apparatus 100. Referring to FIG. 7 , the optical coherence tomography angiography apparatus 100 includes:

[0219] An acquisition module 1001 is configured to acquire a time domain signal of a target area, where the time domain signal is N interference spectrum signals obtained by performing N repeated A scans on the target area.

[0220] A scale transformation module 1002 is configured to perform scale transformation on each of the N interference spectrum signals based on k scales to obtain N×k scale transformation signals;

[0221] The step of performing scale transformation on any interference spectrum signal based on k scales includes: decomposing any interference spectrum signal based on each scale to obtain k scale transformation signals corresponding to any interference spectrum signal;

[0222] A frequency domain transformation module 1003 is used to perform Fourier transformation on each scale-transformed signal and take its logarithm to obtain an axial frequency domain signal in a logarithmic space;

[0223] A decorrelation module 1004 is configured to divide the N×k denoised axial frequency domain signals into k groups of single-scale signals based on k scales, and perform decorrelation calculation on each group of single-scale signals to obtain a single-scale blood flow signal;

[0224] The fusion module 1005 is configured to obtain a multi-scale blood flow signal based on k groups of single-scale blood flow signals.

[0225] The imaging module 1006 is configured to obtain a blood flow image based on the multi-scale blood flow signal.

[0226] By adopting the above technical solution, the blood flow signal is a non-steady-state signal, and its changes are localized. An interference spectrum signal is decomposed into multiple segments after a scale transformation. During the subsequent decorrelation calculation, the axial frequency domain signals corresponding to each small segment of the interference spectrum signal corresponding to the same target area at different times can be compared. Through segmented comparison, the changing signals in the non-steady state can be captured more accurately, thereby achieving accurate extraction of the blood flow signal.

[0227] Furthermore, by transforming an interference spectrum signal at multiple scales, each scale corresponds to a different number of decomposition segments, meaning that the scales of segmented comparison vary. This allows for measuring changes in non-stationary signals at different scales. Therefore, simultaneously transforming and fusing an interference spectrum signal at multiple scales can reduce the error rate associated with a single scale, improving the accuracy of blood flow signal extraction and processing efficiency. Furthermore, by preserving all points in a single interference spectrum signal, more image detail can be restored, improving the signal-to-noise ratio.

[0228] In one possible implementation, the scale conversion module 1002, when decomposing any interference spectrum signal based on any scale to obtain a scale conversion signal, is specifically configured to:

[0229] Determine a decomposed signal corresponding to any scale, where the decomposed signal is a signal within a sampling time period T, and the decomposed signal includes at least one time group, where the time group includes an adjacent first time period and a second time period, two adjacent first time periods, or two adjacent second time periods;

[0230] Multiplying the interference spectrum signal corresponding to the first time period by the forward signal to obtain a first signal;

[0231] The interference spectrum signal corresponding to the second time period is multiplied by the reverse signal to obtain a second signal; and a scale transformation signal is obtained based on the at least one first signal and the at least one second signal.

[0232] In one possible implementation, the device further includes a denoising module, which denoises the axial frequency domain signal in the logarithmic space to obtain the denoised axial frequency domain signal, specifically for:

[0233] Determine the segmentation threshold;

[0234] If the amplitude of the axial frequency domain signal is greater than or equal to the segmentation threshold, the denoised axial frequency domain signal is equal to the axial frequency domain signal in the logarithmic space;

[0235] If the amplitude of the axial frequency domain signal is less than the segmentation threshold, the denoised axial frequency domain signal is equal to the segmentation threshold.

[0236] In one possible implementation, when determining the segmentation threshold, the denoising module is specifically configured to:

[0237] Extract the noise signals corresponding to all axial frequency domain signals based on clustering algorithm;

[0238] A segmentation threshold is determined based on the noise signal.

[0239] In a possible implementation, when performing decorrelation calculation on each group of single-scale signals to obtain a single-scale blood flow signal, the decorrelation module 1004 is specifically configured to:

[0240] Determine the first-order statistics, second-order statistics, and third-order statistics of each set of single-scale signals;

[0241] A single-scale blood flow signal is determined by performing decorrelation calculation based on first-order statistics, second-order statistics, and third-order statistics.

[0242] In one possible implementation, the apparatus further includes an alignment module that performs a Fourier transform on each scale-transformed signal and takes a logarithm to obtain an axial frequency domain signal in a logarithmic space. The alignment module is specifically configured to:

[0243] Perform alignment operations on all axial frequency domain signals based on phase correlation algorithm;

[0244] The denoised axial frequency domain signal is a signal obtained by denoising the axial frequency domain signal after the alignment operation.

[0245] In a possible implementation, the apparatus further includes a dispersion compensation module, which is specifically configured to:

[0246] Before performing scale transformation on each interference spectrum signal in the N interference spectrum signals based on k scales, dispersion compensation is performed on each interference spectrum signal in the time domain signal.

[0247] An optical coherence tomography angiography device provided in an embodiment of the present application is applicable to the above method embodiment and will not be described in detail here.

[0248] An electronic device is provided in an embodiment of the present application, as shown in FIG8 . The electronic device 1000 shown in FIG8 includes: a processor 1001 and a memory 1003. The processor 1001 and the memory 1003 are connected, for example, via a bus 1002. Optionally, the electronic device 1000 may further include a transceiver 1004. It should be noted that in actual applications, the number of transceivers 1004 is not limited to one, and the structure of the electronic device 1000 does not constitute a limitation on the embodiments of the present application.

[0249] Processor 1001 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 1001 may also be a combination that implements computing functions, such as a combination of one or more microprocessors, a combination of a DSP and a microprocessor, and the like.

[0250] Bus 1002 may include a path for transmitting information between the aforementioned components. Bus 1002 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, for example. Bus 1002 may be divided into an address bus, a data bus, a control bus, and so on. For ease of illustration, FIG8 shows only one thick line, but this does not mean that there is only one bus or only one type of bus.

[0251] The memory 1003 may be a ROM (Read Only Memory) or other type of static storage device that can store static information and instructions, a RAM (Random Access Memory) or other type of dynamic storage device that can store information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), a CD-ROM (Compact Disc Read Only Memory) or other optical disk storage, optical disk storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto.

[0252] The memory 1003 is used to store application code for executing the solution of the present application, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the application code stored in the memory 1003 to implement the content shown in the above method embodiment.

[0253] Electronic devices include, but are not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. They may also include servers. The electronic device shown in FIG8 is merely an example and should not limit the functionality or scope of use of the embodiments of the present disclosure.

[0254] The embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon. When the computer-readable storage medium is run on a computer, the computer can execute the corresponding content of the aforementioned method embodiment. Compared with the prior art, in the embodiment of the present application, when a user makes an appointment for a certain service type in a certain time period, the actual occupied time period and the corresponding time period occupancy threshold are used to determine whether the source time period of other service types related to the actual occupied time period is occupied, rather than directly determining the source time period corresponding to other service types as occupied. In other words, if it is determined based on the actual occupied time period that the source time period of other service types related to it is not occupied, this time can still be reserved by the user, thereby reducing fragmented time, reducing waste of resources, and further reducing the waiting time of medical staff and patients, which can improve the efficiency of consultation.

[0255] It should be understood that although the steps in the flowcharts of the accompanying drawings are shown in sequence as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some of the steps in the flowcharts of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0256] The above are only some of the implementation methods of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. An optical coherence tomography angiography method, characterized in that, comprising: acquiring a time-domain signal of a target region, the time-domain signal being N interference spectral signals obtained by performing A-scans on the target region N times repeatedly; performing scale transformation on each of the N interference spectral signals based on k scales to obtain N×k scale-transformed signals; wherein performing scale transformation on any one of the interference spectral signals based on k scales includes: decomposing the any one interference spectral signal based on each of K scales to obtain k scale-transformed signals corresponding to the any one interference spectral signal; performing Fourier transform on each of the scale-transformed signals and taking the logarithm to obtain an axial frequency-domain signal in the logarithmic space, and denoising the axial frequency-domain signal; dividing the N×k denoised axial frequency-domain signals into k groups of single-scale signals based on k scales, and performing decorrelation calculation on each group of the single-scale signals to obtain single-scale blood flow signals; obtaining a multi-scale blood flow signal based on k groups of the single-scale blood flow signals; obtaining a blood flow image based on the multi-scale blood flow signal.

2. The method according to claim 1, characterized in that, the interference spectral signal corresponds to a sampling time period T, and decomposing the any one interference spectral signal based on each of K scales to obtain k scale-transformed signals corresponding to the any one interference spectral signal includes: decomposing the any one interference spectral signal based on any one of k scales to obtain a scale-transformed signal corresponding to the any one interference spectral signal for the any one scale, and this operation includes: determining a decomposed signal corresponding to the any one scale, the decomposed signal being a signal within the sampling time period T, the decomposed signal including at least one time group, the time group including adjacent first and second time periods, adjacent first time periods, or adjacent second time periods; multiplying the any one interference spectral signal corresponding to the first time period by a forward signal to obtain a first signal; multiplying the any one interference spectral signal corresponding to the second time period by a reverse signal to obtain a second signal; obtaining a scale-transformed signal corresponding to the any one interference spectral signal for the any one scale based on at least one of the first signals and at least one of the second signals.

3. The method according to claim 1, characterized in that, denoising the axial frequency-domain signal in the logarithmic space to obtain the denoised axial frequency-domain signal includes: determining a segmentation threshold; if the amplitude of the axial frequency-domain signal is greater than or equal to the segmentation threshold, the denoised axial frequency-domain signal is equal to the axial frequency-domain signal in the logarithmic space; if the amplitude of the axial frequency-domain signal is less than the segmentation threshold, the denoised axial frequency-domain signal is equal to the segmentation threshold.

4. The method according to claim 3, characterized in that, the determining of the segmentation threshold includes: extracting noise signals corresponding to all the axial frequency-domain signals based on a clustering algorithm; determining a segmentation threshold based on the noise signals.

5. The method according to claim 1, It is characterized in that the calculation of decorrelating each group of the single-scale signals to obtain single-scale blood flow signals includes: determining the first-order statistic, second-order statistic, and third-order statistic of each group of the single-scale signals; performing decorrelating calculation based on the first-order statistic, second-order statistic, and third-order statistic to determine the single-scale blood flow signals.

6. The method according to claim 1, it is characterized in that after performing Fourier transform on each of the scale-transformed signals and taking the logarithm to obtain axial frequency-domain signals in the logarithmic space, it further includes: performing alignment operation on all the axial frequency-domain signals based on the phase correlation algorithm.

7. The method according to claim 1, it is characterized in that before performing scale transformation on each of the N interference spectral signals based on k scales, it further includes: performing dispersion compensation on each of the interference spectral signals in the time-domain signal.

8. An optical coherence tomography angiography device, it is characterized in that it includes: an acquisition module, configured to acquire a time-domain signal of a target region, where the time-domain signal is N interference spectral signals obtained by performing A-scan on the target region N times repeatedly; a scale transformation module, configured to perform scale transformation on each of the N interference spectral signals based on k scales to obtain N×k scale-transformed signals; wherein, performing scale transformation on any one of the interference spectral signals based on k scales includes: decomposing any one of the interference spectral signals based on each of the K scales to obtain k scale-transformed signals corresponding to any one of the interference spectral signals; a frequency-domain transformation module, configured to perform Fourier transform on each of the scale-transformed signals and take the logarithm to obtain axial frequency-domain signals in the logarithmic space, and perform denoising on the axial frequency-domain signals; a decorrelation module, configured to divide the N×k denoised axial frequency-domain signals into k groups of single-scale signals based on k scales, and perform decorrelating calculation on each group of the single-scale signals to obtain single-scale blood flow signals; a fusion module, configured to obtain multi-scale blood flow signals based on k groups of the single-scale blood flow signals; an imaging module, configured to obtain a blood flow image based on the multi-scale blood flow signals.

9. An electronic device, it is characterized in that the electronic device includes: one or more processors; a memory; wherein, one or more computer programs are stored in the memory and configured to be executed by one or more processors, and the one or more computer programs are configured to: execute the optical coherence tomography angiography method according to any one of claims 1-7.

10. A computer-readable storage medium, it is characterized in that it includes: a computer program stored with the ability to be loaded and executed by a processor and perform the optical coherence tomography angiography method according to any one of claims 1-7.

Images