Estimation method and calibration system and method for included angle between ultrasonic beam and blood flow movement direction
Through the method of combining neural network model and ultrasonic catheter, the angle between the ultrasonic beam and the blood flow movement direction is automatically estimated, which solves the problem of large error in blood flow velocity measurement in the existing technology, and realizes efficient and accurate blood flow velocity measurement. In multi-layer blood vessels, the instrument system is simplified and the cost is reduced.
Patent Information
- Application Number
- CN202411184832.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-27
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-08-27
AI Technical Summary
The prior art lacks accuracy in estimating the angle between the ultrasonic beam and the blood flow movement direction, resulting in large errors in measuring blood flow velocity. The existing methods are complex or unsuitable for popularization, especially when operating experience is dependent on in vitro measurement, which is difficult to achieve. Automated and accurate calibration.
The ultrasonic blood vessel wall image is segmented by neural network model, and the angle between the ultrasonic beam and the blood flow movement direction is automatically estimated through time series analysis and geometric fitting, and the blood flow velocity is calculated based on the ultrasonic Doppler effect relationship. Intravascular ultrasound catheter synchronizes images and Doppler shift signals to reduce external interference.
It realizes automated, fast and accurate blood flow velocity measurement, reduces measurement errors and repeatability differences, simplifies the cost of hardware development of instrument systems, is suitable for measurement of shallow and deep blood vessels, and reduces the compression effect on patient tissues.
Smart Images

Figure CN118986413A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of ultrasonic imaging, and in particular to a technology of intravascular ultrasonic imaging, specifically a method for estimating the angle between an ultrasonic beam and the direction of blood flow movement, and a technology for calibrating blood flow velocity using the angle. Background Art
[0002] Monitoring the blood flow velocity in blood vessels can help us understand the changes in blood vessels and is one of the most important parameters of blood vessel status. However, the current technology for measuring blood flow velocity using ultrasound is mainly based on the ultrasound Doppler effect, that is, there is a Doppler frequency shift in the echo signal from blood cells received by the ultrasound probe, and its amplitude and direction are related to the relative movement of the probe and blood. There is an angle between the ultrasound beam emitted by the probe and the direction of blood flow (called the beam angle or Doppler angle in some literature), which is the main source of error in blood flow velocity measurement.
[0003] The effectiveness of measuring blood flow velocity using in vitro probes such as linear arrays and convex arrays often depends on the operator's experience in scanning sections during the measurement operation, which is not conducive to the quantitative measurement of blood flow velocity. Because ultrasound B-mode imaging only supports the observation of limited cross-sectional information of blood vessels, it is difficult to determine the overall direction and structure of blood vessels. Therefore, when measuring blood flow velocity in vitro, the θ angle is generally set to 15° to 60°. The greater the difference between the actual θ and the set value, the greater the error in blood flow velocity measurement.
[0004] Traditional intravascular ultrasound imaging technology uses an external motor to drive a flexible transmission shaft to allow the ultrasound catheter to complete rotational scanning in the blood vessel. However, this imaging method has an inevitable disadvantage, that is, the angle between the ultrasound catheter and the central axis of the blood vessel changes all the time, and it is difficult to determine the size of θ through geometric feature relationships.
[0005] In order to obtain the angle between the ultrasound beam and the direction of blood flow, there are currently three main methods, namely manual method, dual-beam and multi-beam Doppler method, and three-dimensional vascular reconstruction method.
[0006] The manual method is mainly because the commercial color Doppler ultrasound diagnostic system on the market cannot automatically determine the angle between the ultrasound beam and the direction of blood flow movement, so the operating physician is required to analyze the color Doppler image and constantly adjust the sampling volume size and the direction of the sampling line. In general examinations, the sampling volume is less than 1 / 3 of the lumen, and the sampling line should always remain parallel to the lumen. However, manual calibration is time-consuming and labor-intensive, and the calibrated measurement results have the disadvantages of poor repeatability and strong subjectivity, which may interfere with the doctor's accurate judgment of changes in vascular hemodynamic function.
[0007] In the dual-beam and multi-beam Doppler methods, the ultrasound probe is equipped with piezoelectric transducers arranged at different inclination angles to receive blood flow echo signals at different ultrasound incident angles. Then, through geometric relationships and equation calculations, the angle θ between the ultrasound beam and the blood flow direction and the blood flow velocity v are derived. Although this technology can accurately calculate θ in theory, there are some defects in practical conditions. For example, when a physician holds a probe or uses an ultrasound patch for examination, it will inevitably compress the patient's tissue. Therefore, the surface of the probe in contact with the skin is deformed, and the blood vessels are also squeezed to a certain extent. As a result, the incident angle of the ultrasound beam is no longer equal to the inclination angle of the piezoelectric transducer, that is, it does not meet the assumptions of the theory. In addition, the blood flow velocity is measured by the dual-beam Doppler method. This technology has very high requirements for the manufacturing process of the probe and requires a large number of complex calculations to solve θ. It is not easy to expand to commercial color Doppler ultrasound monitoring systems.
[0008] The optical coherence tomography blood flow imaging technology of three-dimensional vascular reconstruction is an emerging technology in recent years, and there is no need to inject contrast agents into blood vessels. For example, in the prior art with publication number CN105286779A, three-dimensional vascular reconstruction generally requires coronary angiography to collect cross-sectional images of blood vessels. Optical scattering signals are obtained by circular scanning or horizontal scanning, and the two-dimensional images are analyzed and processed by the decorrelation method to perform three-dimensional reconstruction of the vascular system of the retina and choroid. The three-dimensional vascular structure is analyzed, the position of the central axis of the blood vessel is determined, and the angle between the incident light and the central axis of the blood vessel is obtained. Although the spatial resolution of optical coherence tomography technology is high, the penetration depth is as low as a few millimeters. Therefore, it can only be used to measure the blood flow velocity of blood vessels such as the superficial layer of the skin or the retina, and does not have the universality of ultrasound technology. The three-dimensional vascular reconstruction method has very high requirements on hardware performance, which undoubtedly increases the cost of instrument system development and is not conducive to promotion in grassroots health institutions.
[0009] Therefore, how to provide a method that is easy to promote and can accurately estimate the angle between the ultrasound beam and the direction of blood flow movement is a technical problem that needs to be solved urgently. Summary of the invention
[0010] In order to solve the technical problem that the prior art for accurately estimating the angle between an ultrasound beam and the direction of blood flow movement is relatively limited or inaccurate, the present invention proposes an estimation method, calibration system and method for the angle between an ultrasound beam and the direction of blood flow movement.
[0011] The method for estimating the angle between the ultrasonic beam and the blood flow movement direction of the present invention comprises:
[0012] A neural network model is trained in advance based on an image data set consisting of ultrasound vascular wall images of different blood vessels at different deflection angles;
[0013] Calling the neural network model to segment the ultrasonic vascular wall image of the current blood vessel to be estimated to obtain a blood vessel cross-sectional graph;
[0014] Obtain the pixel points that constitute the boundary of the blood vessel lumen and calculate the lumen area;
[0015] Time series analysis was used to analyze the temporal changes in lumen area;
[0016] If the lumen area changes steadily over time, an elliptical geometric figure corresponding to the blood vessel cross section is fitted; otherwise, a convex hull geometric figure is fitted;
[0017] Using formula Estimate the angle between the ultrasound beam and the direction of blood flow movement, a is the major semi-axis of the geometric figure, and b is the minor semi-axis of the geometric figure.
[0018] Furthermore, the pixel points constituting the boundary of the blood vessel lumen are found by analyzing the binary image of the lumen area of the blood vessel cross-sectional image.
[0019] Furthermore, when the lumen area changes steadily over time, the zero-order moment M of the image grayscale is calculated by the grayscale value f(i, j) of the pixel point of the geometric image: 00 =∑ i ∑ j f(i,j) and first-order moment M 10 =∑ i ∑ j i·f(i,j),M 01 =∑ i ∑ j j·f(i,f), get the coordinates of the center of mass of the geometric figure (x0,y0),
[0020] According to the orientation angle, major semi-axis and minor semi-axis of the ellipse, the standard tilted ellipse equation is derived by using coordinate transformation to obtain the elliptical geometric figure corresponding to the blood vessel cross section.
[0021] Furthermore, when the temporal change of the lumen area is not smooth, the distance relationship between the opposite pixel points is analyzed by the rotating caliper algorithm, and the two pixel points with the farthest distance are connected as the long axis of the geometric figure. The product of the slope of the short axis and the slope of the long axis is always equal to -1, and a point-slope equation is established to solve the equation of the line representing the short axis. The coordinates of the two pixel points overlapping in the pixel point set of the straight line equation and the pixel point set of the geometric figure are used as the endpoint coordinates of the short axis, the coordinates of the intersection of the short axis and the long axis are used as the coordinates of the center of mass of the geometric figure, and all vertices in the pixel point set of the set image are connected to obtain a convex hull geometric figure corresponding to the cross section of the blood vessel.
[0022] Furthermore, when there are no two overlapping pixel points in the pixel point set of the straight line equation and the pixel point set of the geometric figure, the distances from all pixel points in the pixel point set of the geometric figure to the minor axis are calculated, and the coordinates of the two pixel points with the smallest distance to the minor axis are taken as the endpoint coordinates of the minor axis, and the coordinates of the intersection of the minor axis and the major axis are taken as the coordinates of the center of mass of the geometric figure.
[0023] The blood flow velocity calibration method proposed by the present invention comprises:
[0024] Acquire B-mode ultrasound vascular wall images and Doppler frequency shift signals of blood cells in blood vessels;
[0025] The angle between the ultrasound beam and the blood flow movement direction is obtained by using the method for estimating the angle between the ultrasound beam and the blood flow movement direction of the above technical solution;
[0026] Relationship based on ultrasound Doppler effect The blood flow velocity is calculated, v is the blood flow velocity, c is the blood sound velocity, f d It is the Doppler frequency shift of the first echo signal reflected by the red blood cells in the blood flow, f0 is the center frequency of the device measuring the first echo signal, and θ is the angle between the ultrasound beam and the direction of blood flow movement.
[0027] The blood flow velocity calibration system proposed by the present invention comprises:
[0028] An ultrasound catheter, used for synchronously acquiring B-mode ultrasound vascular wall images and Doppler frequency shift signals of blood cells in the blood vessel;
[0029] The calculation module uses the method for estimating the angle between the ultrasound beam and the direction of blood flow described in the above technical solution to obtain the angle between the ultrasound beam and the direction of blood flow, and based on the relationship of the ultrasound Doppler effect The blood flow velocity is calculated, v is the blood flow velocity, c is the blood sound velocity, f d It is the Doppler frequency shift of the first echo signal reflected by the red blood cells in the blood flow, f0 is the center frequency of the device measuring the first echo signal, and θ is the angle between the ultrasound beam and the direction of blood flow movement.
[0030] Further, the ultrasonic catheter comprises:
[0031] tube body;
[0032] A transducer fixing block is arranged at one end of the tube body;
[0033] A first ultrasonic transducer is disposed at a first end of the transducer fixing block close to the end of the tube body, and is used to emit a first sound beam toward the outside of the tube body and receive a first echo signal reflected by red blood cells in the blood flow;
[0034] A second ultrasonic transducer is disposed at a second end opposite to the transducer fixing block and is disposed opposite to the first ultrasonic transducer, and is used to emit a second sound beam perpendicular to the tube wall of the tube body and receive a second echo signal;
[0035] The acoustic reflector is arranged in the tube body opposite to the second ultrasonic transducer, and comprises a reflector and a driving component. The reflector reflects the emitted second acoustic beam to be perpendicular to the tube wall, and the driving component drives the reflector to rotate around its axis.
[0036] Furthermore, the reflector is made of a cylindrical magnet, and the first end facing the second ultrasonic transducer is an inclined surface at 45° to the axis of the tube body, and the driving assembly includes an inner tube for fixing the reflector in the tube body, a coil wound on the inner tube, and a ball arranged in the inner tube at the second end of the reflector.
[0037] Compared with the prior art, the present invention has the following advantages and beneficial effects.
[0038] (1) Compared with the manual method in the prior art, the present invention can automatically estimate and calibrate the blood flow velocity, shorten the time spent on each calibration, and improve the accuracy and repeatability of blood flow velocity measurement.
[0039] (2) Compared with the dual-beam and multi-beam Doppler methods based on extracorporeal ultrasound probes, the ultrasound catheter of the present invention is placed inside the blood vessel, avoiding compression of tissues and blood vessels and reducing external interference caused by the measurement principle, thus facilitating more accurate measurement of the true blood flow velocity. Compared with the dual-beam Doppler method based on intravascular ultrasound catheters, the ultrasound catheter of the present invention has only two transducers, so the structure is more concise and compact, and the preparation cost is very low.
[0040] (3) Compared with the prior art three-dimensional vascular reconstruction method, the main steps of the three-dimensional vascular reconstruction method of the present invention need to include: obtaining the volume metadata of the blood vessel, three-dimensional segmentation, three-dimensional vascular reconstruction, analyzing the trend of the vascular central axis, etc. However, the present invention uses deep learning technology to segment the lumen area of the two-dimensional B-mode image, and then fits the lumen boundary to estimate. Therefore, the hardware development cost and computing time of the instrument system are reduced, which is more conducive to application to mid- and low-end color Doppler ultrasound diagnostic systems.
[0041] (4) Compared with the prior art, the present invention takes into account the turbulence caused by the deformation of the blood vessels of cardiovascular patients. When the blood vessels are diseased and compressed, the lumen shape of the resulting B-mode image is irregular. The variation pattern of the lumen area is analyzed, and the lumen boundary is fitted into an inclined ellipse or convex hull according to the actual situation. The Doppler angle is solved and calibrated using the main axis parameters. This technical idea is more in line with real clinical problems.
[0042] (5) In order to ensure the relative accuracy of the measurement, the cutting surface of the external ultrasound probe in the prior art cannot be parallel to the skin surface or the inclination angle is too large, so it is generally only applicable to measuring blood flow velocity in the case of 15°-60°. The intravascular ultrasound catheter of the present invention can have a deflection angle range of 0°-60° in the blood vessel, and is applicable to the case of 0°-60°.
[0043] (6) The present invention can be applied to the measurement and calibration of blood flow velocity in superficial and deep blood vessels. The displacement of blood cells is analyzed based on the ultrasonic Doppler effect, and the blood flow velocity is directly measured. Because there is no need to consider the influence of physiological parameters such as the patient's weight, blood vessel diameter, and blood pressure, the system error caused by converting other physiological parameters into blood flow velocity in the indirect measurement method is eliminated. BRIEF DESCRIPTION OF THE DRAWINGS
[0044] The present invention is described in detail below with reference to the embodiments and accompanying drawings, wherein:
[0045] Figure 1 It is the main flow chart of the angle estimation and blood flow velocity calibration method of the present invention.
[0046] Figure 2 A three-dimensional structural diagram of the ultrasonic catheter of the present invention.
[0047] Figure 3 is a cross-sectional view of an ultrasonic catheter according to an embodiment of the present invention.
[0048] Figure 4 It is a flow chart of blood flow detection of the present invention.
[0049] Figure 5 It is a flow chart of intravascular ultrasound imaging of the present invention.
[0050] FIG. 6( a ) is a set of pixel points representing the lumen boundary according to an embodiment of the present invention.
[0051] FIG6( b ) is the convex hull fitting result of the lumen boundary in FIG6( a ).
[0052] FIG7( a ) is the blood flow velocity result before calibration.
[0053] FIG. 7( b ) is the blood flow velocity result after calibration using the present invention.
[0054] Figure 8 This is a schematic diagram of the phenomenon of eccentricity and torsion of the ultrasound catheter in the blood vessel.
[0055] Fig. 9 It is a basic theoretical analysis schematic diagram of the angle estimation method of the present invention.
[0056] Fig.10It is a schematic diagram of the present invention for demonstrating that the angle between the ultrasound beam and the blood flow direction is equal to the angle between the axis of the ultrasound catheter and the central axis of the blood vessel.
[0057] Fig.11 (a) is an intravascular ultrasound image of a blood vessel acquired according to an embodiment of the present invention.
[0058] Fig.11 (b) is the mask marking the lumen area.
[0059] Fig.11 (c) is the segmentation result of the lumen area.
[0060] Fig.11 (d) is a binary image representing the lumen area.
[0061] Fig.11 (e) is a binary image representing the lumen boundary.
[0062] Fig.11 (f) is the ellipse fitting result of the lumen boundary.
[0063] Description of reference numerals:
[0064] 1. Transducer fixing block; 2. Tube body; 3. Inner tube; 4. Coil; 5. Reflector; 6. Ball; 7. Piezoelectric chip of the first ultrasonic transducer; 8. Matching layer of the first ultrasonic transducer; 9. Backing layer of the first ultrasonic transducer; 10. Piezoelectric chip of the second ultrasonic transducer; 11. Matching layer of the second ultrasonic transducer; 12. Backing layer of the second ultrasonic transducer. DETAILED DESCRIPTION
[0065] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.
[0066] Thus, a feature indicated in this specification will be used to illustrate one of the features of an embodiment of the present invention, rather than implying that each embodiment of the present invention must have the described feature. In addition, it should be noted that this specification describes many features. Although some features can be combined together to illustrate possible system designs, these features can also be used in other combinations that are not explicitly described. Thus, unless otherwise stated, the described combinations are not intended to be limiting.
[0067] The method for estimating the angle between the ultrasonic beam and the blood flow movement direction of the present invention mainly includes the following steps.
[0068] Before estimation, it is necessary to prepare in advance a neural network model for segmenting the image. To prepare the corresponding neural network model, it is necessary to first obtain an image data set, which needs to contain ultrasonic vascular wall images of different blood vessels and different deflection angles. After acquiring ultrasonic vascular wall images of different blood vessels and different deflection angles, a neural network model is trained based on the image data set composed of ultrasonic vascular wall images of different blood vessels with different deflection angles. The input of the neural network model is the ultrasonic vascular wall image of the blood vessel whose angle is to be estimated, and the output is the blood vessel cross-sectional diagram of the blood vessel whose angle is to be estimated.
[0069] After the trained neural network model is obtained, when it is necessary to estimate the direction of the ultrasound beam and blood flow, the ultrasound vascular wall image of the blood vessel to be estimated is first obtained, and the trained neural network model is called to segment the ultrasound vascular wall image of the current blood vessel to be estimated to obtain a blood vessel cross-sectional diagram;
[0070] Based on the obtained cross-sectional image of the blood vessel, the pixel points constituting the boundary of the vascular lumen of the blood vessel are obtained to calculate the lumen area of the blood vessel. When the ultrasound catheter is assembled, a standard annular wire target phantom can be used for B-mode imaging to test the axial and lateral resolution. The wire diameter of the nylon or metal wire target, as well as the position and angle buried in the vascular phantom are generally known and can be found in the instruction manual. Observe the delay of the echo signal reflected by the wire target on the oscilloscope, and use the formula d=c*t / 2 to calculate the Euclidean distance of the wire target from the center line of the ultrasound catheter. The B-mode image is composed of a two-dimensional matrix containing grayscale brightness values, so by analyzing the length of the set of highlighted pixels representing the wire target from the center of the image, the number of pixels constituting the set, etc., the actual size represented by a pixel point, that is, the scale of the image, can be inferred. The scale can be obtained based on the analysis of multiple groups of the above experiments. As long as the number of pixels constituting the lumen area is obtained, the area of the lumen area can be estimated.
[0071] Time series analysis was used to analyze the temporal changes in lumen area;
[0072] If the lumen area changes steadily over time, an elliptical geometric figure corresponding to the blood vessel cross section is fitted; otherwise, a convex hull geometric figure is fitted;
[0073] Using formula Estimate the angle between the ultrasound beam and the direction of blood flow movement, a is the major semi-axis of the geometric figure, that is, half of the major axis of the geometric figure, and b is the minor semi-axis of the geometric figure, that is, half of the minor axis of the geometric figure.
[0074] The present invention analyzes the temporal changes of the lumen area by time series analysis, and fits an ellipse or a convex hull according to different situations, so that the geometric figure of the obtained blood vessel cross section is closer to the original shape of the blood vessel. Compared with the prior art method of using an ellipse to represent the blood vessel cross section, the present invention can obtain a more accurate angle between the blood flow direction and the ultrasound beam.
[0075] On the basis of the above technical solution, an embodiment of the present invention is to find the pixel points constituting the boundary of the blood vessel lumen by analyzing the binary image of the lumen area of the blood vessel cross-sectional figure.
[0076] In one embodiment, when the lumen area changes steadily over time, it indicates that there is no embolism in the blood vessel and the blood vessel is normal. The zero-order moment M of the image grayscale is calculated by the grayscale value f(i, j) of the pixel point of the geometric image. 00 =∑ i ∑ j f(i,j) and first-order moment M 10 =∑ i ∑ j i·f(i,j),M 01 =∑ i ∑ j j·f(i,j), get the coordinates of the centroid of the geometric figure (x0,y0),
[0077] Then, according to the orientation angle, major semi-axis and minor semi-axis of the ellipse, the standard tilted ellipse equation is derived by coordinate transformation to obtain the elliptical geometric figure corresponding to the blood vessel cross section. That is, after obtaining the centroid of the geometric figure, the present invention fits to obtain the general ellipse equation, then obtains the orientation angle, major semi-axis and minor semi-axis of the ellipse corresponding to the general ellipse equation, and then obtains the standard tilted ellipse equation.
[0078] In one embodiment, when the temporal variation of the lumen area is not stable, the distance relationship between the opposite pixel points is analyzed by the rotating caliper algorithm, and the two pixel points with the farthest distance are connected as the long axis of the geometric figure. The product of the slope of the short axis and the slope of the long axis is always equal to -1, and a point-slope equation is established to solve the equation of the line representing the short axis. The coordinates of the two pixel points overlapping in the pixel point set of the straight line equation and the pixel point set of the geometric figure are used as the endpoint coordinates of the short axis, and the coordinates of the intersection of the short axis and the long axis are used as the coordinates of the centroid of the geometric figure, and all vertices in the pixel point set of the set image are connected to obtain the convex hull-shaped geometric figure corresponding to the cross section of the blood vessel. The reason why the present invention calculates the coordinates of the centroid is that when fitting the convex hull, the long axis of the convex hull does not necessarily pass through the calculated centroid, but must be near the long axis. The present invention assumes that the short axis passes through the centroid, and thus establishes a point-slope equation to solve the equation of the line of the short axis.
[0079] For blood vessels, there may be thrombi in the blood vessels, so that the cross-section of the blood vessels is not a circle or an ellipse. For example, with the development of atherosclerotic plaques, the blood vessels become narrower and narrower. Due to the irregular shape of coronary and carotid plaques, when high-speed arterial blood flows through the plaques, the velocity gradient of each particle changes randomly, and turbulence often occurs. When imaging a narrow blood vessel segment, the lumen shape of the blood vessel is often irregular. When measuring the blood flow velocity in a narrow blood vessel segment, the blood flow distribution law is not laminar. Therefore, the lumen boundary should not be fitted into an ellipse, but rather a convex hull to analyze the major and minor axes of the lumen area. When imaging normal blood vessels, the lumen area also changes periodically with the pulsation of the blood vessel. When imaging a narrow blood vessel containing plaques, the lumen shapes of adjacent image sequences are not nearly the same, so the lumen area changes randomly. By segmenting the lumen area of the B-mode image under a linear time series and analyzing the randomness of the lumen area change, it can be determined whether the ultrasound catheter is currently measuring blood flow velocity at a narrow blood vessel segment.
[0080] Assume that there are small thrombi or plaques in the blood vessels, and the plaques are randomly distributed on the blood vessel wall. At this time, the narrow blood vessel segment is imaged, and the B-mode image of the blood vessel with arc-shaped and irregular curve-shaped coexistence can be clearly observed. As shown in Figure 6(a), after segmenting the lumen boundary, the binary image result is analyzed to find the pixel point set S representing the lumen boundary. The arc-shaped structure represents the normal inner wall of the blood vessel, and the irregular curve represents the plaque part. Obviously, the points on the irregular curve are closer to the inner wall of the blood vessel, while the points on the arc are farther away from the inner wall of the blood vessel. The distance relationship between the opposite pixel points is analyzed by the rotating calcaneal algorithm, and the line segment l formed by connecting the two pixel points with the farthest distance is the convex hull diameter (major axis). The slope of the major axis is solved by the two-point equation. Theoretically, the minor axis is perpendicular to the major axis and passes through the center of mass. Then the product of the slope of the minor axis and the slope of the major axis is always equal to -1, and then the point-slope equation is used to solve the infinitely long straight line t representing the minor axis. However, there may be a situation where there are no two points in S that are exactly on the line t. In this case, the distance from all pixels in S to t should be analyzed, and the two points with the smallest distance should be retrieved as the two endpoints of the short axis. As shown in Figure 6(b), the positions of all pixels in the set S are analyzed, the convex hull is fitted, and then the vertices of the convex hull are connected to obtain a convex polygon. "*" represents the centroid of the lumen area, the dotted line represents the major axis of the lumen area, and the dot-dash line represents the minor axis of the lumen area.
[0081] The present invention is based on the above-mentioned method for estimating the angle between the ultrasound beam and the direction of blood flow movement, and also protects the blood flow velocity calibration method, such as Figure 1 As shown, the blood flow velocity calibration method of the present invention mainly includes the following steps.
[0082] Acquire B-mode ultrasound vascular wall images and Doppler frequency shift signals of blood cells in blood vessels;
[0083] The angle between the ultrasound beam and the blood flow movement direction is obtained by using the method for estimating the angle between the ultrasound beam and the blood flow movement direction of the above technical solution;
[0084] Relationship based on ultrasound Doppler effect Calculate the blood flow velocity.
[0085] v is the blood flow velocity, c is the blood sound velocity, f d It is the Doppler frequency shift of the first echo signal reflected by the red blood cells in the blood flow, f0 is the center frequency of the device measuring the first echo signal, and θ is the angle between the ultrasound beam and the direction of blood flow movement.
[0086] The blood flow velocity calibration system of the present invention comprises an ultrasonic catheter and a calculation module.
[0087] The ultrasound catheter is used to synchronously acquire B-mode ultrasound vascular wall images and Doppler frequency shift signals of blood cells in the blood vessel.
[0088] The calculation module uses the estimation method of the angle between the ultrasound beam and the blood flow direction of the above technical solution to obtain the angle between the ultrasound beam and the blood flow direction, and based on the relationship of the ultrasound Doppler effect The blood flow velocity is calculated, v is the blood flow velocity, c is the blood sound velocity, f d It is the Doppler frequency shift of the first echo signal reflected by the red blood cells in the blood flow, f0 is the center frequency of the device measuring the first echo signal, and θ is the angle between the ultrasound beam and the direction of blood flow movement.
[0089] In one embodiment, a display module may also be included to display the calibrated blood flow velocity result, namely, pulse Doppler blood flow imaging.
[0090] Figure 2 , Figure 3 The specific structural diagram of the ultrasonic catheter of the present invention is shown, and the ultrasonic catheter comprises a tube body 2, a transducer fixing block 1, a first ultrasonic transducer, a second ultrasonic transducer and an acoustic reflector.
[0091] The tube body is the outer shell of the 2 ultrasound catheters.
[0092] The transducer fixing block 1 is arranged at one end of the tube body 2 .
[0093] The first ultrasonic transducer is arranged at the first end of the transducer fixing block near the end of the tube body, and is used to emit a first sound beam toward the outside of the tube body and receive a first echo signal reflected by red blood cells in the blood flow; in the ultrasonic Doppler effect relationship, f dMeasured by the first transducer, f0 is the center frequency of the first transducer. The first ultrasonic transducer includes a piezoelectric wafer 7 of the first ultrasonic transducer, a matching layer 8 of the first ultrasonic transducer, and a backing layer 9 of the first ultrasonic transducer.
[0094] The second ultrasonic transducer is arranged at the second end opposite to the transducer fixing block 1 and is arranged opposite to the first ultrasonic transducer, and is used to emit a second sound beam perpendicular to the pipe wall of the pipe body and receive a second echo signal. The second ultrasonic transducer includes a piezoelectric chip 10 of the second ultrasonic transducer, a matching layer 11 of the second ultrasonic transducer, and a backing layer 12 of the second ultrasonic transducer.
[0095] The acoustic reflector is used to allow the second ultrasonic transducer to emit a second acoustic beam perpendicular to the tube wall. The acoustic reflector and the second ultrasonic transducer are arranged in the tube body relative to each other. The acoustic reflector includes a reflector 5 and a drive assembly. The reflector 5 reflects the emitted second acoustic beam perpendicular to the tube wall, and the drive assembly can drive the reflector to rotate around its axis. The drive assembly includes an inner tube, a coil, a ball, etc.
[0096] The reflector 5 is made of a cylindrical magnet, and the first end facing the second ultrasonic transducer is an inclined surface at 45° to the axis of the tube body. The inclined surface is plated with a metal film layer that can reflect sound waves, so it can act as a reflector and can also rotate 360 degrees with the coil 4. The driving component includes an inner tube 3 that fixes the reflector 5 in the tube body 2, and the coil 4 is wound on the inner tube 3. When the coil 4 is connected to a three-phase alternating current, the reflector 5 can rotate 360 degrees. The ball 6 is arranged in the inner tube at the second end of the reflector, wherein the inner tube is a pure copper thin tube, and the ball is a steel ball.
[0097] Figure 4 The flowchart of blood flow detection using the first transducer is shown. The first transducer is stimulated by a pulse transceiver so that the first transducer emits ultrasonic waves into the blood. The echo signal received by the first transducer is collected by an A / D acquisition card. The one-dimensional radio frequency (RF) data is arranged according to the direction of the fast signal and the slow signal, and reorganized into a two-dimensional RF data matrix. The signal is Fourier transformed, its frequency domain characteristics are analyzed, and a suitable low-pass filter and wall filter are designed. The low-pass filter can filter out the high-frequency component signal, and the wall filter can filter out the Doppler frequency shift signal of the vascular pulsation. The fast signal is demodulated, and the slow signal at a certain sampling depth is analyzed. The blood flow velocity is estimated by the spectral Doppler method, the autocorrelation method, the autoregression method, etc., and finally the blood flow velocity result is displayed.
[0098] Figure 5The flowchart of intravascular ultrasound imaging using the second transducer is shown. The second transducer is stimulated by a pulse transceiver, so that the second transducer emits ultrasound to the acoustic reflector. As the micromotor rotates, the ultrasound is evenly reflected by the acoustic reflector into the vascular tissue, achieving a 360° circular scan of the blood vessel. The echo signal received by the second transducer is collected using an A / D acquisition card. The one-dimensional radio frequency (RF) data is arranged in the direction of the fast signal and the slow signal, and reorganized into a two-dimensional RF data matrix. The signal is Fourier transformed, its frequency domain characteristics are analyzed, and a suitable bandpass filter is designed to complete the demodulation of the fast signal. Due to the scanning characteristics of the second transducer, the two-dimensional RF data matrix is arranged in polar coordinate form. In order to conform to people's habit of reading pictures, the two-dimensional RF data matrix needs to be transformed from the polar coordinate system to the Cartesian coordinate system through digital coordinate conversion. The maximum and minimum brightness of the pixel are often very different. In order to adapt to the grayscale level of the display, logarithmic compression should be performed. Then, algorithms such as bilinear interpolation and nearest neighbor interpolation are used to fill the gaps in each sound line in the image matrix, and then the annular B-mode image results are displayed.
[0099] The present invention places an ultrasonic catheter on a deflection angle adjustment platform to simulate the catheter's different degrees of eccentricity and torsion in a blood vessel. Collect echo signals to obtain a series of intravascular ultrasound B-mode images. Professional physicians annotate these images to construct a data set. When there are fewer images in the data set, data enhancement processing is performed through elastic transformation, affine transformation, etc. to increase the sample size of the data set. Two-dimensional Gaussian filtering and adaptive denoising filtering are used to suppress white noise and speckle noise in the image, and histogram equalization is used to improve image contrast. Pytorch, TensorFlow, etc. are used to design variant network structures of U-Net and DeepLabV3+, train the network until convergence, and output the segmentation model (input is). Deploy the inference code and call the model to segment the lumen area of the B-mode image in real time. Analyze the binary image of the segmentation result, detect the lumen boundary, and calculate the lumen area size. The time series change of the lumen area is analyzed by the differential autoregressive moving average model. If the area size is a stationary time series, the lumen boundary is fitted into an inclined ellipse, otherwise the lumen boundary is fitted into a convex hull. Analyze the fitted geometric figures and solve the parameters such as centroid, major axis, minor axis, etc. To solve the angle between the ultrasound beam and the direction of blood flow, use The blood flow velocity is calibrated and finally the calibrated blood flow velocity result is displayed.
[0100] In order to conveniently and effectively obtain the angle between the ultrasound beam and the blood flow direction, the present invention adopts a method of synchronously obtaining the B-mode ultrasound vessel wall image in the blood vessel and the Doppler frequency shift signal of the blood cells. The reliability of the calculation of the angle between the ultrasound beam and the blood flow direction of the present invention is described below.
[0101] When a plane intersects a cylindrical surface, three situations may occur. The shape of the intersection line depends on the relative position and angle between the plane and the cylindrical axis. When the cutting plane is parallel to the cylindrical axis, the shape of the intersection line is a rectangle. When the cutting plane is perpendicular to the cylindrical axis, the intersection line is a circle. When the cutting plane is inclined to the cylindrical axis, the intersection line is an ellipse, and the minor semi-axis of the ellipse is equal to the radius of the bottom surface of the cylinder. The second ultrasonic transducer of the present invention can emit a first sound beam perpendicular to the wall of the blood vessel. The sound waves it emits are reflected 360 degrees by the reflector on the blood vessel and can be regarded as a scanning plane, while the wall of the blood vessel can be regarded as a cylindrical surface.
[0102] The present invention uses a peristaltic pump, a blood phantom, an isolated blood vessel, an ultrasonic catheter, etc. to build a hemodynamic circulation loop. Taking the spectral Doppler method as an example, Figure 7(a) shows the power spectrum result within 0.29 seconds, and the grayscale value of the pixel in the coordinate area represents the amplitude of the Doppler frequency shift signal. At 0.04004 seconds, the blood flow velocity is about 0.453125 meters per second. Obviously, the blood phantom flows regularly, with each cycle of about 0.1s. The power spectral density integral and three-line fitting method are used to analyze the spectrum envelope and depict it with a solid line. Substitute the solved θ into As shown in FIG7( b ), at 0.04004 seconds, the calibrated blood flow velocity is about 0.482206 meters per second.
[0103] When the intravascular ultrasound catheter is impacted by high-speed blood, deflection and torsion will occur. Assuming that the second ultrasound transducer and the acoustic reflector are aligned in the catheter, the extension line of the ultrasound beam emitted by the first ultrasound transducer should be the normal vector of the scanning plane of the second transducer. For a finite length of blood vessel, under the action of blood friction and blood pressure, the cross-sectional shape of the blood vessel lumen is approximately circular. Figure 8 As shown in the figure, it is assumed that the vessel wall is a cylinder and the blood flow in the vessel is a periodic pulsating laminar flow. The angle between the ultrasonic beam emitted by the first ultrasonic transducer and the direction of blood flow is the beam angle θ, and the angle between the axis of the ultrasonic catheter and the axis of the blood vessel is δ. Figure 2 The catheters and blood vessels in the image are projected onto the xoz plane, and then Fig. 9 Perform geometric analysis. The DN line segment is the projection of the scanning plane of the second transducer on the xoz plane, and JK is the projection of the lumen area in the annular B-mode image on the xoz plane. The rectangle PQP'Q' represents a blood vessel of finite length, TR is on the central axis of the blood vessel, and the straight line L is the extension line of the ultrasound catheter, L⊥DN. Fig.10 A detailed argument can lead to θ=δ:
[0104] That is, for normal blood vessels, the imaging result of the vascular lumen is approximately circular or elliptical. In the present invention, the circle can be regarded as a special ellipse with equal major and minor axes. Therefore, by fitting the lumen boundary into an ellipse, the equation of the inclined ellipse can be derived to analyze the major and minor axes of the lumen area. Traditional image segmentation methods include edge detection, region growing, adaptive threshold segmentation, and the like. Image segmentation methods based on machine learning include simple linear iterative clustering, naive Bayesian method, minimum error Bayesian decision method, and the like. Image segmentation networks based on deep learning include U-Net and its variants, DeepLab V3+, Faster R-CNN, Mask R-CNN, and the like.
[0105] Fig.11 It is the result of using deep learning technology to segment the lumen area and fit the lumen boundary. Fig.11 (a) shows the image acquired by the second ultrasonic transducer. First, the images in the data set are preprocessed, such as affine transformation, histogram equalization, two-dimensional median filtering, etc. Fig.11 As shown in (b), through image annotation software, use circles or polygons to mark the lumen boundary along the direction of the inner wall of the blood vessel, and output labels and masks. Then design the convolutional neural network structure, such as the selection of activation function, the setting of upsampling convolution kernel, the setting of downsampling pooling kernel, the selection of loss function, etc. Distinguish the ratio of training set and validation set in the data set. Use GPU to train the neural network, adjust the learning rate, iterate multiple times until the loss function converges, and save the segmentation model. Fig.11 As shown in (c), the segmentation model is called to segment the lumen area from the intravascular ultrasound image. Fig.11 (d)- Fig.11 As shown in (e), a binary image of the lumen area is output, the connected components in the image are analyzed, and all the pixels that make up the lumen boundary are found.
[0106] A rectangular coordinate system is established with the center of the image as the origin, the coordinates of the pixel points are analyzed, the lumen boundary is fitted into an ellipse through an algorithm, and the standard ellipse equation is solved. There are many methods for fitting an ellipse, such as the least squares method, the Hough transform method, and the random sampling consistency method. The centroid coordinates P0 (x0, y0) of the image are calculated by normalizing the first-order central moment. As shown in equations (2)-(3), the grayscale value of the pixel in the two-dimensional image is represented by f(i, j), and the sum of the image grayscale is the zero-order moment M 00 , the first moment is M 10 and M 01 As shown in formula (4), the grayscale center of the image is the centroid of the image, that is, the ratio of the first-order moment to the zero-order moment. As shown in formula (5), according to the orientation angle of the ellipse (the angle α between the major axis and the x-axis), the major semi-axis a, and the minor semi-axis b, the standard tilt ellipse equation is derived using coordinate transformation. Fig.11As shown in (f), the result of ellipse fitting is depicted by a solid line near the lumen boundary, where α∈(0,π / 2).
[0107]
[0108] M 10 =∑ i ∑ j i·f(i,j),M 01 =∑ i ∑ j j·f(i,j) (2)
[0109]
[0110] Through the above description, the present invention has the following features.
[0111] The ultrasonic catheter used in the blood vessel of the present invention has two transducers with different central frequencies, which can synchronously detect blood flow and intravascular ultrasonic imaging, and use the synchronous signal to trigger the pulse transceiver to achieve the excitation of the first and second ultrasonic transducers. The echo signals of the two are collected to simultaneously complete the blood flow velocity measurement and the blood vessel annular B-mode imaging.
[0112] The present invention obtains B-mode images at different deflection angles when training the neural network model, and uses deep learning technology to segment the lumen area. Variants of convolutional neural network structures such as U-Net and DeepLabV3+ are designed, and the model is called by inference code to complete real-time segmentation.
[0113] The present invention analyzes the connected areas and connected components of the binary image to find the lumen boundary. The present invention determines whether to measure the blood flow velocity in a normal blood vessel or a stenotic blood vessel by analyzing the change law of the lumen area size, and thus uses different methods to fit the lumen boundary. After obtaining the lumen boundary, the present invention obtains the coordinates of the particle of the image by the image center distance. The fitting result of the lumen boundary is analyzed to solve the long axis and short axis length of the lumen area, and finally, based on the long axis and short axis, the angle θ between the ultrasound beam and the blood flow movement direction is estimated, and then the blood flow velocity is calibrated.
[0114] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for estimating the angle between the ultrasound beam and the direction of blood flow, It is characterized in that include: A neural network model is trained in advance based on an image data set consisting of ultrasound vascular wall images of different blood vessels at different deflection angles; Calling the neural network model to segment the ultrasonic vascular wall image of the current blood vessel to be estimated to obtain a blood vessel cross-sectional graph; Obtain the pixel points that constitute the boundary of the blood vessel lumen and calculate the lumen area; Time series analysis was used to analyze the temporal changes in lumen area; If the lumen area changes steadily over time, an elliptical geometric figure corresponding to the blood vessel cross section is fitted; Otherwise, a convex hull-like geometry is fitted; Using formula Estimate the angle between the ultrasound beam and the direction of blood flow movement, a is the major semi-axis of the geometric figure, and b is the minor semi-axis of the geometric figure.
2. The method for estimating the angle between the ultrasound beam and the direction of blood flow as claimed in claim 1, It is characterized in that By analyzing the binary image of the lumen area of the blood vessel cross-sectional image, the pixel points constituting the lumen boundary of the blood vessel are found.
3. The method for estimating the angle between the ultrasound beam and the direction of blood flow as claimed in claim 2, It is characterized in that When the lumen area changes steadily over time, the zero-order moment M of the image grayscale is calculated by the grayscale value f(i, j) of the pixel point of the geometric image. 00 =∑ i ∑ j f(i,j) and first-order moment M 10 =∑ i ∑ j i·f(i,j),M 01 =∑ i ∑ j j·f(i,j), get the coordinates of the centroid of the geometric figure (x 0 ,y 0 ), According to the orientation angle, major semi-axis and minor semi-axis of the ellipse, the standard tilted ellipse equation is derived by using coordinate transformation to obtain the elliptical geometric figure corresponding to the blood vessel cross section.
4. The method for estimating the angle between the ultrasound beam and the direction of blood flow as claimed in claim 2, It is characterized in that When the temporal variation of the lumen area is not smooth, the distance relationship between the opposite pixel points is analyzed by the rotating caliper algorithm, and the two pixel points with the farthest distance are connected as the long axis of the geometric figure. The product of the slope of the short axis and the slope of the long axis is always equal to -1, and a point-slope equation is established to solve the equation of the line representing the short axis. The coordinates of the two pixel points overlapping in the pixel point set of the straight line equation and the pixel point set of the geometric figure are used as the endpoint coordinates of the short axis, the coordinates of the intersection of the short axis and the long axis are used as the coordinates of the center of mass of the geometric figure, and all vertices in the pixel point set of the collective image are connected to obtain a convex hull geometric figure corresponding to the cross section of the blood vessel.
5. The method for estimating the angle between the ultrasound beam and the direction of blood flow as claimed in claim 4, It is characterized in that When there are no two overlapping pixel points in the pixel point set of the straight line equation and the pixel point set of the geometric figure, the distances from all pixel points in the pixel point set of the geometric figure to the minor axis are calculated, and the coordinates of the two pixel points with the smallest distance to the minor axis are taken as the endpoint coordinates of the minor axis, and the coordinates of the intersection of the minor axis and the major axis are taken as the coordinates of the center of mass of the geometric figure.
6. A blood flow velocity calibration method, It is characterized in that include: Acquire B-mode ultrasound vascular wall images and Doppler frequency shift signals of blood cells in blood vessels; The angle between the ultrasound beam and the direction of blood flow movement is obtained by using the method for estimating the angle between the ultrasound beam and the direction of blood flow movement as described in any one of claims 1 to 5; Relationship based on ultrasound Doppler effect The blood flow velocity is calculated, v is the blood flow velocity, c is the blood sound velocity, f d is the Doppler frequency shift of the first echo signal reflected by red blood cells in the blood flow, f 0 is the center frequency of the device measuring the first echo signal, and θ is the angle between the ultrasound beam and the direction of blood flow movement.
7. A blood flow velocity calibration system, It is characterized in that include: An ultrasound catheter, used for synchronously acquiring B-mode ultrasound vascular wall images and Doppler frequency shift signals of blood cells in the blood vessel; A calculation module, which uses the method for estimating the angle between the ultrasound beam and the direction of blood flow movement as described in any one of claims 1 to 5 to obtain the angle between the ultrasound beam and the direction of blood flow movement, and based on the relationship of the ultrasound Doppler effect The blood flow velocity is calculated, v is the blood flow velocity, c is the blood sound velocity, f d is the Doppler frequency shift of the first echo signal reflected by red blood cells in the blood flow, f 0 is the center frequency of the device measuring the first echo signal, and θ is the angle between the ultrasound beam and the direction of blood flow movement.
8. The blood flow velocity calibration system according to claim 7, It is characterized in that The ultrasonic catheter comprises: tube body; A transducer fixing block is arranged at one end of the tube body; A first ultrasonic transducer is disposed at a first end of the transducer fixing block close to the end of the tube body, and is used to emit a first sound beam toward the outside of the tube body and receive a first echo signal reflected by red blood cells in the blood flow; A second ultrasonic transducer is disposed at a second end opposite to the transducer fixing block and is disposed opposite to the first ultrasonic transducer, and is used to emit a second sound beam perpendicular to the tube wall of the tube body and receive a second echo signal; The acoustic reflector is arranged in the tube body opposite to the second ultrasonic transducer, and comprises an acoustic reflector and a driving component. The reflector reflects the emitted second acoustic beam to be perpendicular to the tube wall, and the driving component drives the reflector to rotate around its axis.
9. The blood flow velocity calibration system according to claim 8, It is characterized in that The acoustic reflector is made of a cylindrical magnet, and the first end facing the second ultrasonic transducer is an inclined surface at 45° to the axis of the tube body. The driving assembly includes an inner tube for fixing the reflector in the tube body, a coil wound on the inner tube, and a ball arranged in the inner tube at the second end of the reflector.
Citation Information
Patent Citations
Method of imaging of in vivo retina haemodynamics and measuring of absolute flow velocity
CN105286779A
Blood vessel identification-based fully-automatic measurement method, device, storage medium and system
CN110051385A
Ultrasonic probe for monitoring hemodynamic parameters
CN111281428A
Spectrum blood flow detection method of color spectrogram based on ultrasonic Doppler
CN111325202A
Quantification of aortoiliac endoluminal irregularity
US20040171932A1
