A four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering

Abstract

This invention provides a four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering, comprising: extracting the temporal dynamic data of each voxel from the three-dimensional dynamic data as the data to be processed; processing to obtain a three-dimensional enhanced image intensity image and a maximum time image; the enhanced image intensity image is obtained by normalizing the absolute value of the one-dimensional discrete Fourier transform of the dynamic data correction value of the voxel; the pixel value of the maximum time image is the time coordinate value when the dynamic data correction value of the voxel reaches its maximum value in the time dimension; updating the enhanced image intensity image according to the artifact time value; and obtaining the display information of the RGB image of each voxel using the maximum time image as the color value and the enhanced image intensity image as the display intensity value. The method of this invention achieves high signal-to-noise ratio, high small vessel visibility, automatic artifact suppression, low dose, and more intuitive vessel segmentation and enhancement in four-dimensional angiography data based on original multi-phase dynamic data.

Description

Technical Field

[0001] This invention belongs to the field of image post-processing enhancement technology in medical vascular imaging, specifically relating to a four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering. Background Technology

[0002] In current clinical medicine, three-dimensional vascular examination techniques based on radiographic imaging equipment utilize contrast agents to enhance the visualization of blood vessels and other soft tissues. For example, iodine contrast agents are used in computed tomography (CT) angiography to enhance blood vessels, while gadolinium contrast agents are used in magnetic resonance imaging (MR) angiography. Furthermore, by altering the contrast agent injection method (rate, dosage, etc.) and allowing the imaging equipment to acquire images multiple times within a given timeframe, various examination methods can be developed, such as tissue perfusion examination and multi-phase dynamic contrast-enhanced angiography. Multi-phase dynamic data can provide additional information, such as comparing image enhancement at different phases or extracting dynamic blood enhancement curves, for use in disease diagnosis.

[0003] This method of viewing images requires repeatedly reviewing image data from different phases, which is cumbersome and inconvenient. The image quality of multi-phase dynamic imaging is also affected by factors such as patient movement, lack of cooperation, examination parameter settings, contrast agent concentration and dosage, and metal artifacts. Simultaneously, the signal-to-noise ratio is also affected by radiation dose and instrument performance. Current multi-phase dynamic vascular imaging examinations suffer from problems such as high radiation dose and high contrast agent dose. Is it possible to achieve high signal-to-noise ratio, high visibility of small vessels, automatic artifact suppression, and even reduce radiation dose and contrast agent, as well as achieve more intuitive vessel segmentation and enhanced image data based on multi-phase dynamic image data through post-processing algorithms?

[0004] The patent document ZL 2018 1 0203889.1 discloses an X-ray angiography method and a dynamic image reading method based on X-ray angiography, which introduces a method for dynamic angiography imaging based on frequency domain filtering. This method can achieve enhanced vascular imaging of dynamic angiography data from two-dimensional projection images. However, this method cannot directly process multi-phase dynamic angiography data in clinical medicine. This is because two-dimensional projection data is a superposition of the imaged sample in the projection direction. The absorption of X-rays by background tissues other than blood vessels can be considered equivalent to the non-uniformity of the incident X-ray intensity, and the contrast agent's image fluctuations will change proportionally. Therefore, using the DC averaging data in the patent method can effectively normalize the two-dimensional projection data and improve the signal-to-noise ratio.

[0005] However, for 3D image data, image voxels of different substances or tissue regions have different CT values ​​according to the sample's inherent absorption coefficient for X-rays (e.g., different concentrations of iodine contrast agents have corresponding CT values). Different substances in the sample also have different imaging parameter ranges in magnetic resonance imaging. The numerical fluctuation of pixels at the blood vessel of interest (CT value of the contrast agent) may be on the same order of magnitude as the fluctuation of noise or other interference at pixels in other regions, but the DC-averaged data distribution varies greatly, even among blood vessel pixels in different tissues. Therefore, using the DC-averaged data in the aforementioned patented method cannot effectively normalize the voxel data, resulting in uneven enhancement of blood vessel images, high noise levels, and a poor signal-to-noise ratio. Furthermore, the processing procedures for 3D dynamic data and 2D dynamic data also differ.

[0006] Therefore, a new four-dimensional angiography imaging method is needed to achieve high signal-to-noise ratio, high visibility of small vessels, automatic artifact suppression, and even reduce radiation dose and contrast agent, as well as achieve more intuitive vessel segmentation and enhancement during four-dimensional imaging. Summary of the Invention

[0007] The purpose of this invention is to provide a four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering, so as to achieve high signal-to-noise ratio, high visibility of small vessels, automatic artifact suppression, low dose, and more intuitive vessel segmentation and enhanced four-dimensional angiography data based on original multi-phase dynamic data.

[0008] To achieve the above objectives, the present invention provides a four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering, comprising:

[0009] S1: Extract the temporal dynamic data of each voxel in the three-dimensional dynamic data one by one, as the data to be processed;

[0010] S2: Process the data to be processed to obtain the three-dimensional enhanced image intensity image I(x,y,z) and the maximum time image T(x,y,z), where (x,y,z) are the spatial coordinates of the voxels;

[0011] Among them, the enhanced image intensity image I(x,y,z) is obtained by taking the absolute value of the one-dimensional discrete Fourier transform G(x,y,z,k) of the dynamic data correction value g(x,y,z,t) of the voxel, and normalizing the sum of one or more absolute values ​​of the other one-dimensional discrete Fourier transform G(x,y,z,k) based on the absolute value of the one-dimensional discrete Fourier transform G(x,y,z,0) at 0Hz.

[0012] The pixel value of T(x,y,z) at the maximum moment is the time coordinate value of the voxel's dynamic data correction value g(x,y,z,t) when it reaches its maximum value in the time dimension;

[0013] S3: Observe and find the location of artifacts in the 3D enhanced image intensity image I(x,y,z). Obtain the artifact time value based on the pixel value of the corresponding pixel in the 3D maximum time image T(x,y,z). Then, set the pixel value of the pixel point in the 3D maximum time image T(x,y,z) that is equal to the artifact time value to zero in the 3D enhanced image intensity image I(x,y,z) to update the 3D enhanced image intensity image I(x,y,z).

[0014] S4: Using the maximum moment image T(x,y,z) as the color value of different pseudo-color codes and the enhanced image intensity image I(x,y,z) as the display intensity value of the pixel, the display information of the RGB image of each voxel is obtained.

[0015] In step S2, the three-dimensional enhanced image intensity image I(x,y,z) and the maximum time image T(x,y,z) satisfy the following formula:

[0016] g(x,y,z,t)=f(x,y,z,t)-a,

[0017]

[0018]

[0019]

[0020] T(x,y,z)=maxloc(g(x,y,z,t)),

[0021] Where f(x,y,z,t) represents the time-domain dynamic data of the voxel, a is the pixel value representative constant at the blood vessel location, g(x,y,z,t) is the dynamic data correction value of the voxel, b is the first intermediate constant, and k is the frequency value of the discrete Fourier transform. And are integers, where N is the total number of frames of the 3D dynamic data, A(x,y,z,k) q ) is dynamic data with a time frequency of k q The amplitude component, k q Let k be the frequency value of the q-th discrete Fourier transform, and n be the total number of frequency values ​​of the discrete Fourier transform when calculating the intensity of the enhanced image. G(x,y,z,k) is the one-dimensional discrete Fourier transform of the dynamic data correction value g(x,y,z,t) of the voxel. The function max loc(g(x,y,z,t)) is defined as the coordinate t corresponding to the maximum value of the dynamic data correction value g(x,y,z,t) of the voxel.

[0022] n is at most the maximum value of k and is a natural number, and k has different indices. q The values ​​are different.

[0023] Step S1 specifically includes: acquiring three-dimensional dynamic data F(x,y,z,t), dividing the three-dimensional dynamic data into combinations of data from different slices, and then combining the data from each slice at different times into a two-dimensional dynamic image data set, thereby obtaining multiple two-dimensional dynamic image data sets; subsequently, extracting the temporal dynamic data of each pixel in the two-dimensional dynamic image data set as the data to be processed.

[0024] Step S2 specifically includes:

[0025] S21: For the temporal dynamic data f of each pixel in the two-dimensional dynamic image dataset i (x,y,t), processed to obtain a two-dimensional enhanced image intensity image I i (x,y);

[0026] The dynamic data correction value g of the pixel of the i-th slice in the two-dimensional dynamic image dataset i (x,y,t) is:

[0027] g i (x,y,t)=f i (x,y,t)-a,

[0028] Where 'a' is a pixel value at the blood vessel, representing a constant;

[0029] Two-dimensional enhanced image intensity image I i (x,y) is:

[0030]

[0031]

[0032]

[0033] Where b is the first intermediate constant, and k is the frequency value of the discrete Fourier transform. And it is an integer, where N is the total number of frames of the three-dimensional dynamic data, and A i (x,y,k q ) is dynamic data with a time frequency of k q The amplitude component, k q k is the frequency value of the q-th discrete Fourier transform, where q is 1-n and n is the total number of frequency values ​​of the discrete Fourier transform when calculating the intensity of the enhanced image; G i (x,y,k) is g iOne-dimensional discrete Fourier transform of (x,y,t); G i (x,y,k) and g i In (x,y,t), i represents the i-th slice, and x and y represent the horizontal and vertical coordinates of the slice image.

[0034] S22: For the temporal dynamic data f of each pixel in the two-dimensional dynamic image dataset i (x,y,t), processed to obtain the two-dimensional maximum time image T i (x,y);

[0035] Maximum moment like T i The pixel value at (x, y) is:

[0036] T i (x,y)=maxloc(g i (x,y,t)),

[0037] Among them, the function max loc(g i (x,y,t) is defined as the dynamic data correction value g of the pixel in the i-th slice of the two-dimensional dynamic image dataset. i (x,y,t) is the coordinate t corresponding to the maximum value;

[0038] S23: Combine the two-dimensional enhanced image intensity image I i (x,y) and the two-dimensional maximum time image T i (x,y), to obtain the three-dimensional enhanced image intensity image I(x,y,z) and the maximum time image T(x,y,z).

[0039] The artifact time values ​​are all between 1 and N, where N is the total number of frames of the three-dimensional dynamic data.

[0040] The number of artifact time values ​​selected is 2.

[0041] Step S4 specifically includes:

[0042] S41: For each slice i, take the maximum time image T corresponding to that slice. i (x, y) represent the color values ​​of different pseudo-color codes, with the enhanced image intensity image I corresponding to this slice. i (x,y) represents the display intensity value of a pixel, and the display information of the RGB image of each pixel in the slice is obtained.

[0043] S42: Replace with the next slice i+1, and repeat step S41 until all slices have been processed.

[0044] S43: After obtaining the display information of the RGB image of each pixel in all slices, combine them to obtain the display information of the RGB image of each voxel. At this time, the three-dimensional augmented data EI(x,y,z) is obtained.

[0045] The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering of this invention enables high signal-to-noise ratio and high visibility of small vessels in the original multi-phase dynamic data through the steps of this invention. Because only the contrast agent signal is specifically enhanced, and other signals are not enhanced, automatic vessel segmentation can be achieved, removing the influence of motion artifacts, bone, and metal artifacts on vessel imaging. At the same time, the normalization operation in step S2 and the masking operation in the frequency domain filtering in step S3 can remove motion artifacts and other interferences. In addition, the color enhanced images of vascular tissue generated by this invention provide intuitive display of blood flow dynamic information, making it easier to read the film and obtain imaging information, which is beneficial for the qualitative analysis of blood vessels, such as distinguishing between arteries and veins, determining the direction of blood flow, blood flow velocity, and degree of vascular occlusion, providing more imaging information for clinical vascular examinations. Furthermore, based on the high sensitivity of this image enhancement, this invention is expected to further reduce the amount of contrast agent used and radiation dose in a single clinical vascular radiology examination or relax the performance requirements of the examination instruments, while obtaining images that meet the quality requirements of clinical applications.

[0046] In summary, the four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering of this invention addresses the problems of complex, inconvenient, and unintuitive viewing methods for existing multi-phase dynamic enhanced vascular examination data in clinical practice. Furthermore, image quality is affected by factors such as patient movement, lack of cooperation, examination parameter settings, contrast agent concentration and dosage, and metal artifacts. Simultaneously, the signal-to-noise ratio is also affected by radiation dose and instrument performance. Without affecting the implementation of existing diagnostic methods, this invention achieves high signal-to-noise ratio, high visibility of small vessels, automatic artifact suppression, low dose, and more intuitive vascular segmentation and enhancement of four-dimensional angiography data (three-dimensional vascular tissue pseudo-colorization achieves four-dimensional fusion of three-dimensional space and one-dimensional time) based on original multi-phase dynamic data, providing more imaging information for clinical vascular examinations. Attached Figure Description

[0047] Figure 1 This is a flowchart of a four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to a first embodiment of the present invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments disclosed in the present invention will be described in further detail below with reference to the accompanying drawings.

[0049] This invention discloses a four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering. It utilizes three-dimensional multi-phase dynamic angiography enhancement / perfusion data, processing the temporal dynamic data of each voxel individually to form the enhanced display values. For example... Figure 1 As shown, according to the first embodiment of the present invention, the four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering may specifically include:

[0050] Step S1: Extract the temporal dynamic data of each voxel in the three-dimensional dynamic data (e.g., the temporal dynamic data f(x,y,z,t) of a voxel) as the data to be processed.

[0051] The three-dimensional dynamic data refers to dynamic data that combines three dimensions with a time dimension, representing the dynamic data of the same imaged object over time. This three-dimensional dynamic data can be multi-phase dynamic angiography enhancement / perfusion data, obtained through reconstruction from multi-phase CT / MRI data. For example, if the object processed by this invention is the CT / MRI data required for multi-phase (20-phase) angiography enhancement / perfusion data, that is, 20 consecutive sets of CT / MRI data at different times with blood vessels as the object of interest. Each phase of CT / MRI data is reconstructed in the same way into slice data with the required spatial resolution and slice thickness (i.e., one phase of three-dimensional angiography enhancement / perfusion data). These slice data can be directly obtained from the corresponding angiography equipment and its associated software.

[0052] The spatial coordinates of each voxel in the 3D dynamic data can be represented by (x, y, z). In this embodiment, z represents different slices, and x and y represent the horizontal and vertical coordinates on each slice. Since it is 3D dynamic data, each voxel has 20 image values ​​at different times, denoted by t, where t is an integer ranging from 1 to 20. These 20 image values ​​constitute the temporal dynamic data of that voxel. Therefore, this invention can extract the temporal dynamic data of each voxel in the 3D dynamic data as the data to be processed.

[0053] Step S1 specifically includes: acquiring three-dimensional dynamic data F(x,y,z,t), dividing the three-dimensional dynamic data into combinations of data from different slices (such as dividing it into different slices according to a certain direction), and then combining the data of each slice at different times into a two-dimensional dynamic image data set, thereby obtaining multiple two-dimensional dynamic image data sets; subsequently, extracting the temporal dynamic data of each pixel in the two-dimensional dynamic image data set as the data to be processed.

[0054] For multiple two-dimensional dynamic image datasets, the extracted slice data (i.e., the temporal dynamic data of each pixel) are f1(x,y,t), ..., fi (x,y,t), ...,f M (x,y,t), where i represents the i-th slice and M is the total number of slices.

[0055] Specifically, extracting the temporal dynamic data of each pixel in the two-dimensional dynamic image data set is equivalent to extracting the temporal dynamic data of each voxel in the three-dimensional dynamic data set.

[0056] Step S2: Process the data to be processed to obtain the three-dimensional enhanced image intensity image I(x,y,z) and the maximum time image T(x,y,z) for subsequent processing, where (x,y,z) are the spatial coordinates of the voxels;

[0057] In this embodiment, step S2 specifically includes:

[0058] Step S21: For the temporal dynamic data f of each pixel in the two-dimensional dynamic image data set i (x,y,t), processed to obtain a two-dimensional enhanced image intensity image I i (x,y);

[0059] Wherein, g is the dynamic data correction value of the pixel in the i-th slice of the two-dimensional dynamic image dataset. i (x,y,t) is:

[0060] g i (x,y,t)=f i (x,y,t)-a,(1)

[0061] Where 'a' is a pixel value at the blood vessel location, representing a constant.

[0062] It should be noted that the selection of the pixel value representing the constant 'a' at the blood vessel location will affect the intensity of the enhanced image (image I). i The calculated value of (x,y). Since the imaging target is a blood vessel, when the value of the constant 'a' representing the pixel value at a certain blood vessel is selected, it will cause I... i In the calculation formula (2) for (x,y) at blood vessels, the denominator becomes smaller than that at other tissues, thereby enhancing the imaging effect of blood vessel tissue. The range of values ​​for the constant a representing the pixel value at blood vessels varies depending on the original data type. The approximate standard is the mean value of the dynamic signal of the pixels at the blood vessel. It can be the mean value of a single blood vessel of interest, or the mean value of several blood vessel locations.

[0063] Two-dimensional enhanced image intensity image I i (x,y) is:

[0064]

[0065] Where b is the first intermediate constant, and k is the frequency value of the discrete Fourier transform. And it is an integer, where N is the total number of frames of the three-dimensional dynamic data, and A i (x,y,k q ) is dynamic data with a time frequency of k q The amplitude component, k q k is the frequency value of the q-th discrete Fourier transform, where q is 1-n and n is the total number of frequency values ​​of the discrete Fourier transform when calculating the intensity of the enhanced image.

[0066] It should be noted that all pixels share the same constants a and b.

[0067] In other words, A i (x, y, k1) is the amplitude component of the dynamic data with a time frequency of k1, A i (x, y, k2) is the amplitude component of the dynamic data with a time frequency of k2, A i (x, y, k3) represents the amplitude component of dynamic data with a time frequency of k3. k1, k2, and k3 represent three different values ​​of k. The difference between k1, k2, and k3 is that their values ​​are different, but there is no requirement for their order of magnitude.

[0068] in,

[0069]

[0070] Where k is the frequency value of the discrete Fourier transform. And are integers, where N is the total number of frames of the 3D dynamic data, and G i (x,y,k) is g i One-dimensional discrete Fourier transform of (x,y,t). G i (x,y,k) and g i In (x,y,t), i represents the i-th slice, and x and y represent the horizontal and vertical coordinates of the slice image.

[0071] It should be noted that in this embodiment, for k in formula (2), from k1 to k... n With multiple distinct k, n is at most the maximum value of k and is a natural number. That is, the numerator in formula (2) is composed of the amplitude values ​​A of any one or more distinct k. i Additive combinations of (x, y, k). For example, enhancing image intensity like I. i (x,y) can be represented by only one discrete Fourier transform frequency value k, A i (x,y,k) constitutes the equation, i.e., formula (2) can be... Or, enhance image intensity like I i(x,y) can be represented by the frequency values ​​k of the two discrete Fourier transforms, A i (x,y,k) constitutes the equation, i.e., formula (2) can be... wait.

[0072] Step S22: For the temporal dynamic data f of each pixel in the two-dimensional dynamic image data set i (x,y,t), processed to obtain the two-dimensional maximum time image T i (x,y);

[0073] Among them, the maximum time like T i The pixel value at (x, y) is:

[0074] T i (x,y)=maxloc(g i (x,y,t)),

[0075] Among them, the function max loc(g i (x,y,t) is defined as the dynamic data correction value g of the pixel in the i-th slice of the two-dimensional dynamic image dataset. i (x, y, t) represents the coordinate t corresponding to the maximum value. In other words, the image at the maximum moment is T. i The pixel value at spatial coordinates (x, y) is the dynamic data correction value g of the pixel in the i-th slice of the two-dimensional moving image dataset. i The time value at which (x,y,t) reaches its maximum value in the time dimension.

[0076] Therefore, for different spatial coordinates (x, y), the corresponding T i The time values ​​of (x, y) are different, and the maximum time in two dimensions is like T. i The graph (x, y) shows the times when the values ​​are at their respective spatial coordinates (x, y).

[0077] Step S23: Combine the two-dimensional enhanced image intensity image I i (x,y) and the two-dimensional maximum time image T i (x,y), to obtain the three-dimensional enhanced image intensity image I(x,y,z) and the maximum time image T(x,y,z).

[0078] In other embodiments, slicing may not be considered. Therefore, in step S2, the three-dimensional enhanced image intensity image I(x,y,z) and the maximum time image T(x,y,z) satisfy the following formula:

[0079] g(x,y,z,t)=f(x,y,z,t)-a,

[0080]

[0081]

[0082]

[0083] T(x,y,z)=maxloc(g(x,y,z,t)),

[0084] f(x,y,z,t) represents the time-domain dynamic data of the voxel, a is the pixel value representative constant at the blood vessel location, g(x,y,z,t) is the dynamic data correction value of the voxel, b is the first intermediate constant, and k is the frequency value of the discrete Fourier transform. And are integers, where N is the total number of frames of the 3D dynamic data, A(x,y,z,k) q ) is dynamic data with a time frequency of k q The amplitude component, k q Let k be the frequency value of the q-th Discrete Fourier Transform, and n be the total number of frequency values ​​of the Discrete Fourier Transform used to calculate the enhanced image intensity image. G(x,y,z,k) is the one-dimensional Discrete Fourier Transform of the dynamic data correction value g(x,y,z,t) of the voxel. In other words, the enhanced image intensity image I(x,y,z) is obtained by taking the absolute value of the one-dimensional Discrete Fourier Transform G(x,y,z,k) of the dynamic data correction value g(x,y,z,t) of the voxel, and normalizing the sum of one or more absolute values ​​of the remaining one-dimensional Discrete Fourier Transform G(x,y,z,k) based on the absolute value of the one-dimensional Discrete Fourier Transform G(x,y,z,0) at 0Hz.

[0085] The function max loc(g(x,y,z,t)) is defined as the coordinate t corresponding to the maximum value of the voxel's dynamic data correction value g(x,y,z,t). In other words, the pixel value of T(x,y,z) is the time coordinate value of the voxel's dynamic data correction value g(x,y,z,t) when it reaches its maximum value in the time dimension.

[0086] Step S3: Observe and find the location of artifacts in the 3D enhanced image intensity image I(x,y,z). Based on the pixel values ​​of the corresponding pixels in the 3D maximum time image T(x,y,z) of these artifacts, obtain the artifact time values ​​t1, t2...t. n Subsequently, the pixel values ​​in the maximum time-major image T(x,y,z) of the three dimensions are set equal to the artifact time values ​​t1, t2, ..., tt. nThe pixel values ​​corresponding to the pixels in the three-dimensional enhanced image intensity image I(x,y,z) are set to zero (i.e., I(T(x,y,z)=t1)=0, I(T(x,y,z)=t2)=0, ..., I(T(x,y,z)=t3)=0) to update the three-dimensional enhanced image intensity image I(x,y,z).

[0087] The advantage of step S3 is that it can filter out dynamic data in certain special cases, such as data that may be subject to motion artifact interference.

[0088] Among them, the artifact time values ​​t1, t2...t n These are specific moments in time when the imaging target is blood vessels. In dynamic angiography data, the peak times of the contrast agent are quite regular. However, during data processing, some artifacts closely resemble the temporal characteristics of the contrast agent. The resulting 3D enhanced image intensity image I(x,y,z) contains these artifacts, but the pixel values ​​in the 3D maximum time image T(x,y,z) corresponding to these artifacts (i.e., the artifact time values ​​t1, t2...t) are not included. n This will also be significantly different from the pixel values ​​in the 3D maximum time image T(x,y,z) of the pixels at the blood vessel location. Therefore, by assigning the pixel values ​​in the 3D maximum time image T(x,y,z) to t1, t2, ..., t... n All pixels are set to zero to filter out motion artifacts.

[0089] The 3D enhanced image intensity image I(x,y,z) and the 3D maximum time image T(x,y,z) share the same 3D dynamic data voxel spatial coordinates (x,y,z). For a pixel at a given coordinate (x,y,z), it has two characteristics: first, its intensity value within the intensity image I; and second, its time value within the maximum time image T. The process involves finding pixels in the maximum time image T whose values ​​are equal to t1, t2, ..., t... n The pixels are identified and their coordinates are recorded. Then, the pixel values ​​at these coordinates are set to zero in the intensity image I, thus removing these artifacts.

[0090] Artifact time values ​​t1, t2...t n The values ​​of are all between 1 and N, and are integers, where N is the total number of frames in the 3D dynamic data. The artifact time values ​​t1, t2...t... n Corresponding to different times, and the artifact time values ​​t1, t2...t n The total number is uncertain. In this embodiment, two artifact time values ​​are selected, which is sufficient to filter out artifacts effectively. In other embodiments, the number of artifact time values ​​can be different.

[0091] Step S4: Using the maximum moment image T(x,y,z) as the color value of different pseudo-color codes, and the enhanced image intensity image I(x,y,z) as the display intensity value of the pixel, the display information of the RGB image of each voxel is obtained. Since the above processing is performed on all voxels, the color enhancement image EI(x,y,z) is obtained.

[0092] Step S4 specifically includes:

[0093] Step S41: For each slice i, take the maximum time image T corresponding to that slice. i (x, y) represents the color values ​​of different pseudo-color codes (e.g., T). i The lower the (x,y) value, the warmer the color encoding. i The higher the (x,y) value, the cooler the color encoding; the enhanced image intensity corresponding to this slice is I. i (x, y) represents the display intensity value (i.e., brightness value) of a pixel, yielding the display information of the RGB image for each pixel in the slice. Since the above processing is performed on all pixels in the slice, the color enhancement image EI is obtained. i (x,y).

[0094] Step S42: Replace with the next slice i+1, repeat step S41, and obtain the display information of the RGB image of each pixel in the slice until all slices have been processed.

[0095] Step S43: After obtaining the display information of the RGB image of each pixel in all slices, combine them to obtain the display information of the RGB image of each voxel, that is, obtain the three-dimensional augmented data EI(x,y,z).

[0096] The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering of this invention enables high signal-to-noise ratio and high visibility of small vessels in the original multi-phase dynamic data through the steps of this invention. Because only the contrast agent signal is specifically enhanced, and other signals are not enhanced, automatic vessel segmentation can be achieved, removing the influence of motion artifacts, bone, and metal artifacts on vessel imaging. At the same time, the normalization operation in step S2 and the masking operation in the frequency domain filtering in step S3 can remove motion artifacts and other interferences. In addition, the color enhanced images of vascular tissue generated by this invention provide intuitive display of blood flow dynamic information, making it easier to read the film and obtain imaging information, which is beneficial for the qualitative analysis of blood vessels, such as distinguishing between arteries and veins, determining the direction of blood flow, blood flow velocity, and degree of vascular occlusion, providing more imaging information for clinical vascular examinations. Furthermore, based on the high sensitivity of this image enhancement, this invention is expected to further reduce the amount of contrast agent used and radiation dose in a single clinical vascular radiology examination or relax the performance requirements of the examination instruments, while obtaining images that meet the quality requirements of clinical applications.

[0097] In summary, the four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering of this invention addresses the problems of complex, inconvenient, and unintuitive viewing methods for existing multi-phase dynamic enhanced vascular examination data in clinical practice. Furthermore, image quality is affected by factors such as patient movement, lack of cooperation, examination parameter settings, contrast agent concentration and dosage, and metal artifacts. Simultaneously, the signal-to-noise ratio is also affected by radiation dose and instrument performance. Without affecting the implementation of existing diagnostic methods, this invention achieves high signal-to-noise ratio, high visibility of small vessels, automatic artifact suppression, low dose, and more intuitive vascular segmentation and enhancement of four-dimensional angiography data (three-dimensional vascular tissue pseudo-colorization achieves four-dimensional fusion of three-dimensional space and one-dimensional time) based on original multi-phase dynamic data, providing more imaging information for clinical vascular examinations.

[0098] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of the invention. Various variations can be made to the above embodiments of the present invention. All simple and equivalent changes and modifications made in accordance with the claims and description of this application fall within the protection scope of the claims of this patent. All aspects not described in detail in this invention are conventional technical content.

Claims

1. A four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering, characterized in that, include: Step S1: Extract the temporal dynamic data of each voxel in the three-dimensional dynamic data one by one, as the data to be processed; Step S2: Process the data to be processed to obtain a three-dimensional enhanced image intensity image. And the maximum moment ,in, For the spatial coordinates of the voxel; Among them, enhanced image intensity image It is achieved by obtaining the one-dimensional discrete Fourier transform of the dynamic data correction values ​​g(x,y,z,t) of voxels. The absolute value of the one-dimensional discrete Fourier transform at 0 Hz The absolute value is used as a reference for the remaining one-dimensional discrete Fourier transforms. The sum of one or more absolute values ​​is normalized to obtain the result; Maximum moment image The pixel value is the time coordinate value when the dynamic data correction value g(x,y,z,t) of the voxel reaches its maximum value in the time dimension; Step S3: In the three-dimensional enhanced image intensity image The artifact time values ​​are obtained from the data; subsequently, the maximum time image in three dimensions is calculated. The pixel with a mid-pixel value equal to the artifact time value in the 3D enhanced image intensity image The corresponding pixel value is set to zero to update the 3D enhanced image intensity image. ; Step S4: Use the maximum time image Different pseudo-color encoded color values ​​to enhance image intensity. The display intensity value of each pixel is used to obtain the display information of the RGB image of each voxel; In step S2, the three-dimensional enhanced image intensity image And the maximum moment Satisfy the following formula: ， ， ， ， ， in, This represents the temporal dynamic data of voxels, where 'a' is the pixel value at the blood vessel, representing a constant. is the dynamic data correction value for the voxel; b is the first intermediate constant; k is the frequency value of the discrete Fourier transform. And are integers, where N is the total number of frames of the three-dimensional dynamic data, A(x,y,z,k) q ) is dynamic data with a time frequency of k q The amplitude component, k q k is the frequency value of the q-th Discrete Fourier Transform, and n is the total number of frequency values ​​of the Discrete Fourier Transform when calculating the intensity of the enhanced image. It is the one-dimensional discrete Fourier transform of the dynamic data correction value g(x,y,z,t) of a voxel. The function maxloc(g(x,y,z,t)) is defined as the coordinates corresponding to the maximum value of the dynamic data correction value g(x,y,z,t) of a voxel. .

2. The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to claim 1, characterized in that, n is at most the maximum value of k and is a natural number, and k has different indices. q The values ​​are different.

3. The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to claim 1, characterized in that, Step S1 specifically includes: acquiring three-dimensional dynamic data. The three-dimensional dynamic data is divided into combinations of different slices, and the data of each slice at different times are combined into a two-dimensional dynamic image data set, thus obtaining multiple two-dimensional dynamic image data sets; then, the temporal dynamic data of each pixel in the two-dimensional dynamic image data set is extracted as the data to be processed.

4. The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to claim 3, characterized in that, Step S2 specifically includes: Step S21: For the temporal dynamic data of each pixel in the two-dimensional dynamic image data set The process yields a two-dimensional enhanced image intensity image. ; The dynamic data correction value g of the pixel of the i-th slice in the two-dimensional dynamic image dataset i (x,y,t) is: ， Where 'a' is a pixel value at the blood vessel, representing a constant; Two-dimensional enhanced image intensity image for: ， ， ， Where b is the first intermediate constant, and k is the frequency value of the discrete Fourier transform. And it is an integer, where N is the total number of frames of the three-dimensional dynamic data, and A i (x,y,k q ) is dynamic data with a time frequency of k q The amplitude component, k q k is the frequency value of the q-th discrete Fourier transform, where q is 1-n and n is the total number of frequency values ​​of the discrete Fourier transform when calculating the intensity image of the enhanced image. It is g i One-dimensional discrete Fourier transform of (x,y,t); With g i In (x,y,t), i represents the i-th slice, and x and y represent the horizontal and vertical coordinates of the slice image. Step S22: For the temporal dynamic data of each pixel in the two-dimensional dynamic image dataset... Processing yields the two-dimensional maximum time image ; Maximum moment image The pixel value is: ， Among them, the function max loc(g i (x,y,t) is defined as the dynamic data correction value g of the pixel in the i-th slice of the two-dimensional dynamic image data set. i (x, y, t) are the coordinates corresponding to the maximum value. ; Step S23: Combine the two-dimensional enhanced image intensity images and the maximum moment image in two dimensions To obtain a three-dimensional enhanced image intensity image And the maximum moment .

5. The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to claim 1, characterized in that, The artifact time values ​​are all between 1 and N, where N is the total number of frames of the three-dimensional dynamic data.

6. The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to claim 1, characterized in that, The number of artifact time values ​​selected is 2.

7. The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to claim 3, characterized in that, Step S4 specifically includes: Step S41: For each slice i, take the image at the maximum time corresponding to that slice. For different pseudo-color encoded color values, the enhanced image intensity image corresponding to this slice is... Given the display intensity value of a pixel, we obtain the display information of the RGB image of each pixel in the slice; Step S42: Replace with the next slice i+1, and repeat step S41 until all slices have been processed; Step S43: After obtaining the display information of the RGB image of each pixel in all slices, combine them to obtain the display information of the RGB image of each voxel. At this point, the 3D augmentation data is obtained. .

8. The four-dimensional angiography imaging method based on three-dimensional dynamic data and frequency domain filtering according to claim 1, characterized in that, In step S3, the intensity image of the three-dimensional enhanced image is... The artifact time value is obtained from the image, specifically including: the artifact time value in the three-dimensional enhanced image intensity image. The location of artifacts was found by observation, and the maximum time of these artifacts in three dimensions was used to identify them. The corresponding pixel value is used to obtain the artifact time value.

Images