Multimodal coronary functional assessment system based on ofr, qfr and acr techniques

Abstract

This invention discloses a multimodal coronary artery function assessment system based on OFR, QFR, and ACR technologies. It is characterized by comprising an angiography image data acquisition unit; a three-dimensional reconstruction unit; a maximum congestion velocity calculation unit; an ACR registration unit; a candidate lumen area correction unit; and an FFR calculation unit. Compared with existing technologies, this invention has the following advantages: by combining QFR, OFR, and ACR, it achieves accurate calculation of the lumen area of ​​the entire blood vessel. Since the lumen area is the most important parameter in the FFR calculation formula, the technical solution disclosed in this invention can provide accurate predictions of FFR; by using bifurcation classification and quantification, it corrects the lumen area at the bifurcation location, thereby making the technical solution disclosed in this invention applicable to blood vessels with bifurcation lesions, improving the universality of this invention.

Description

Technical Field

[0001] This invention relates to a multimodal coronary artery function assessment system. Background Technology

[0002] The fractional flow reserve (FFR) is an indicator of the functional severity of coronary stenosis. It requires inducing maximal hyperemia in the cardiovascular system using adenosine, followed by pressure wire measurement of the pressures distal to and proximal to the lesion. The ratio of the distal to proximal pressure is the FFR. However, this method of FFR measurement is time-consuming, requires the use of vasodilators such as adenosine triphosphate to induce maximal hyperemia, and is relatively expensive, thus limiting its application.

[0003] Therefore, how to combine imaging data and fluid dynamics principles to design an imaging-based FFR calculation technology that can avoid using drugs to cause maximum vascular congestion, shorten calculation time, and ultimately achieve accurate FFR calculation is a problem that urgently needs to be solved.

[0004] Currently, there are two main types of imaging-based FFR measurement techniques:

[0005] 1) QFR: Calculates FFR using coronary angiography data. This method mainly involves collecting angiography data from two angles, performing three-dimensional reconstruction, determining the lumen area, calculating blood flow velocity using a frame-by-frame method, and then calculating FFR using the hemodynamic equations. The main drawback of QFR is that the lumen area calculated using the three-dimensional reconstructed model has a certain error compared to the actual vessel area, thus reducing the accuracy of FFR.

[0006] 2) OFR: Using OCT pull-back scan images, the lumen area is calculated, and the blood flow velocity is calculated based on the assumption of baseline blood flow velocity. The disadvantage of OFR is that the end position of OCT pull-back is not necessarily at the proximal end of the vessel. Therefore, the calculated FFR value is not the pressure ratio from the distal end to the proximal end of the vessel. Furthermore, based on the assumed baseline blood flow velocity, it cannot accurately fit the blood flow velocity of different patients, resulting in a large error and thus affecting the calculation of the FFR value. Summary of the Invention

[0007] The technical problem to be solved by this invention is that current imaging-based FFR measurement technology suffers from low accuracy.

[0008] To address the aforementioned technical problems, the present invention provides a multimodal coronary artery function assessment system based on OFR, QFR, and ACR technologies, characterized by comprising:

[0009] The contrast imaging data acquisition unit is used to acquire contrast imaging data;

[0010] The three-dimensional reconstruction unit performs three-dimensional reconstruction of blood vessels based on the angiographic image data from the two angles before OCT pullback, obtains a three-dimensional model of blood vessels, and further calculates the area of ​​the entire blood vessel from the distal end to the proximal end, defining this area as the candidate lumen area.

[0011] Maximum congestion velocity calculation unit, used to calculate maximum congestion velocity V;

[0012] The ACR registration unit is used to register OCT-contrast images based on OCT lens markers and contrast data acquired during OCT pullback, thereby obtaining the motion trajectory of the OCT lens markers on the contrast images.

[0013] The candidate lumen area correction unit is used to correct the candidate lumen area to obtain an accurate calculation result of the lumen area of ​​the entire blood vessel, including the following steps:

[0014] Calculate the OCT lumen area for each OCT-contrast image;

[0015] Image registration is performed between any angiography image data before OCT retraction and the registered OCT-angiography image during OCT retraction to obtain the position of the angiography vessel segment during OCT retraction in the 3D model of the vessel. This position corresponds one-to-one with the OCT-angiography image. A relationship model is established between the candidate lumen area of ​​the OCT retraction segment in the angiography image and the OCT lumen area obtained based on the OCT-angiography image. This relationship model is used to correct the candidate lumen area from the end position of OCT retraction to the proximal end to obtain the corrected candidate lumen area, thereby obtaining the accurate lumen area calculation result of the entire vessel.

[0016] The FFR calculation unit uses the following model to calculate FFR:

[0017]

[0018] ΔP=FV+SV 2

[0019]

[0020]

[0021] In the formula: Pa is the pressure at the proximal end of the blood vessel, ΔP is the pressure difference, V is the maximum congestion velocity, F is the viscous frictional resistance coefficient, and S is the detachment force coefficient. It is the lumen area of ​​an artery without stenosis, corrected using the bifurcation and fractal law. This is the lumen area at the narrowing point of the artery after correction using the bifurcation and fractal law, where μ is the blood viscosity, L is the lumen length, k represents the inlet / outlet pressure drop coefficient, and ρ represents the blood density. Let D be the initial lumen area obtained through the candidate lumen area correction unit. 1 The bifurcation area of ​​the blood vessel bifurcation point is D. 2 The corrected lumen area D is then determined using the bifurcation and fractal law. * Represented as: D * =0.678×( 1 + 2 ).

[0022] Preferably, the angle difference between the two angles of the angiographic image data at the two angles before the OCT pullback used by the three-dimensional reconstruction unit is at least 25 degrees.

[0023] Preferably, the maximum congestion flow velocity calculation unit first calculates the contrast agent flow velocity V1 using the frame-by-frame method, and then calculates the hemorrhage flow velocity V using the relationship between the contrast agent and the maximum congestion flow velocity, as shown in the following formula:

[0024] V = a0 + 1 × V1 + a2 × V1 2

[0025] In the formula, a0, a1, and a2 are constants.

[0026] Preferably, the frame-based method used by the maximum congestion flow velocity calculation unit includes the following steps:

[0027] For angiographic image data at any angle, the number of frames in which the contrast agent first reaches the proximal and distal ends of the blood vessel is selected. The time it takes for the contrast agent to travel from the proximal end to the distal end of the blood vessel is calculated based on the difference in the number of frames and the frame rate, and denoted as T.

[0028] Mark the proximal and distal positions of the contrast agent vessel on an angiography image, extract the centerline of the vessel using the vascular skeleton algorithm, calculate the pixel distance of the centerline, and then calculate the actual distance of the centerline based on the resolution of the angiography image, denoted as L.

[0029] The flow rate V1 of the contrast agent can be calculated using the formula V1 = / .

[0030] Preferably, the ACR registration unit uses the following steps to register the OCT-contrast images:

[0031] Step 1: Select an angiographic image in which the retracted blood vessel and the OCT lens markers are clearly visible as the first frame image;

[0032] Step 2: Mark the position of the OCT lens marker and the pullback path, then extract the skeleton lines of other angiography images, use the iterative nearest neighbor algorithm to learn the projection matrix from the first frame image to other angiography images, and use this projection matrix to project the pullback path onto other angiography images to obtain the pullback path of each angiography image.

[0033] Step 3: Based on the template matching method and combined with the motion characteristics of the OCT lens marker from far to near, project the OCT lens marker of the first frame image onto the next frame image. Then, find the region pixel point that is most similar to the current frame image and the OCT lens marker, and ensure that the region pixel point is near the projection point. In this way, the OCT lens marker recognition of all frames is realized, and the motion trajectory of the OCT lens marker on the imaging image is obtained, thereby completing the registration.

[0034] Preferably, in the candidate lumen area correction unit, an OCT lumen segmentation model is implemented using the u-net algorithm based on the OCT-contrast image, and the OCT lumen area of ​​each OCT-contrast image is calculated based on the segmentation result.

[0035] Preferably, in the candidate lumen area correction unit, a relationship model between the candidate lumen area of ​​the OCT pull-back segment in the contrast image and the OCT lumen area obtained based on the OCT-contrast image is established by fitting calculation, and the obtained relationship model is a quadratic fitting relationship model.

[0036] Compared with existing technical solutions, the present invention has the following beneficial effects:

[0037] 1) By combining QFR, OFR and ACR, the lumen area of ​​the entire blood vessel can be accurately calculated. Since the lumen area is the most important parameter in the FFR calculation formula, the technical solution disclosed in this invention can be used to accurately predict the FFR.

[0038] (2) By using bifurcation classification and quantitative analysis, the lumen area at the bifurcation position is corrected, thereby making the technical solution disclosed in this invention applicable to blood vessels with bifurcation lesions, thus improving the universality of this invention. Attached Figure Description

[0039] Figure 1 This is a flowchart of the present invention. Detailed Implementation

[0040] The present invention will be further illustrated below with reference to specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. Furthermore, it should be understood that after reading the teachings of this invention, those skilled in the art can make various alterations or modifications to the invention, and these equivalent forms also fall within the scope defined by the appended claims.

[0041] Combination Figure 1 This embodiment discloses a multimodal coronary function assessment system based on OFR, QFR, and ACR technologies, which includes:

[0042] The contrast imaging data acquisition unit is used to acquire contrast imaging data.

[0043] 3D Reconstruction Unit:

[0044] Based on the angiography image data acquired by the angiography image data acquisition unit at the two angles before OCT pullback (in this embodiment, the angle difference between the two angles is at least 25 degrees), the three-dimensional reconstruction of the blood vessel is performed to obtain the three-dimensional model of the blood vessel. The area of ​​the entire blood vessel from the distal end to the proximal end is further calculated, and the area calculated by the three-dimensional reconstruction unit is defined as the candidate lumen area.

[0045] In this embodiment, the three-dimensional reconstruction unit utilizes existing 3D QCA technology to perform three-dimensional reconstruction of blood vessels, specifically including the following steps:

[0046] Step 1: Extract the lumen from two angles using key point extraction technology;

[0047] Step 2: Use RANSA to match key points from two angles, and at the same time, remove interference points;

[0048] Step 3: Finally, using one of the angles as a reference coordinate system, obtain the three-dimensional spatial position of the key points on the image from the other angle in this coordinate system, thus obtaining 3D point cloud data;

[0049] Step 4: Reconstruct the 3D blood vessel structure based on 3D point cloud data, thereby obtaining the area of ​​the blood vessel at each location.

[0050] However, since there are only two images and there is a certain error in the matching process, the calculated lumen area is the candidate lumen area, which needs to be corrected using subsequent OCT lumen area measurements.

[0051] The flow rate calculation unit is used to calculate the contrast agent flow rate and blood flow rate using a frame-by-frame method. Its implementation includes the following steps:

[0052] Step 1: For angiographic image data obtained by the angiographic image data acquisition unit at any angle, manually select the number of frames when the contrast agent first reaches the proximal and distal ends of the blood vessel. Calculate the time it takes for the contrast agent to travel from the proximal end to the distal end of the blood vessel based on the frame difference and frame rate, and denot it as T.

[0053] Step 2: Manually mark the proximal and distal positions of the contrast agent vessels on an angiography image. Use the vascular skeleton algorithm to extract the centerline of the vessels, calculate the pixel distance of the centerline, and then calculate the actual distance of the centerline based on the resolution of the angiography image, denoted as L.

[0054] Step 3: Calculate the contrast agent flow rate V1 using the formula V1 = / ;

[0055] Step 4: Calculate the hemorrhage velocity V using the relationship between the contrast agent and the maximum hyperemia flow rate, as shown in the following formula:

[0056] V = a0 + 1 × V1 + a2 × V1 2

[0057] In the formula, a0, a1, and a2 are constants. In this embodiment, a0 = 0.10, a1 = -0.93, and a2 = -0.93.

[0058] The ACR registration unit is used to register OCT-contrast images (i.e., OCT pull-back images) based on OCT lens markings and contrast data acquired during OCT pullback (referred to as ACR registration). Its implementation includes the following steps:

[0059] Step 1: For the angiography image data acquired by the angiography image data acquisition unit, manually select an angiography image in which the pulled blood vessel and the OCT lens marker (hereinafter referred to as "marker") are clearly visible, and set the selected angiography image as the first frame image;

[0060] Step 2: Manually mark the position of the marker and the pullback path, then extract the skeleton lines of other contrast images, use the iterative nearest neighbor algorithm to learn the projection matrix from the first frame image to other contrast images, and use this projection matrix to project the pullback path onto other contrast images to obtain the pullback path of each contrast image.

[0061] Step 3: Based on the template matching method and combined with the motion characteristics of the marker from far to near, the marker of the first frame image is projected onto the next frame image. Then, the region pixel point that is most similar to the marker in the current frame image is found, and it is ensured that the region pixel point is near the projection point. In this way, the marker recognition of all frames is realized, and the motion trajectory of the marker on the angiographic image is obtained, thus completing the semi-automatic ACR registration algorithm and obtaining the motion trajectory of the OCT pull-back image on the angiographic image.

[0062] The candidate lumen area correction unit is used to correct the candidate lumen area based on the OCT lumen area and the ACR registration result obtained by the ACR registration unit, including the following steps:

[0063] Step 1: Based on OCT-enhanced images, an OCT lumen segmentation model is implemented using the classic u-net algorithm. Based on the segmentation results, the OCT lumen area of ​​each OCT-enhanced image can be calculated. Since this OCT lumen area is calculated directly based on the cross-section of the blood vessel, its accuracy is equal to the candidate lumen area calculated by the 3D reconstruction unit. However, the lumen area calculated using OCT-enhanced images is only the lumen area of ​​a portion of the blood vessel, i.e., the lumen area from the marker start point (i.e., the CT pullback start position) to the end point (i.e., the OCT pullback end position). It cannot calculate the lumen area from the marker end point to the proximal end of the blood vessel. Therefore, it is necessary to combine the candidate lumen area, the OCT lumen area, and the ACR registration results to calculate the lumen area from the OCT pullback end position to the proximal end of the blood vessel.

[0064] Step 2: For any angiography image data before OCT pullback and the OCT angiography image during OCT pullback, use the iterative nearest neighbor algorithm to perform image registration, obtain the position of the angiography vessel segment pulled back by OCT in any angiography image before OCT pullback, and thus obtain the position of the angiography vessel segment pulled back by OCT in the three-dimensional model of the blood vessel. This position and the OCT angiography image can correspond one-to-one. Therefore, a quadratic fitting relationship model between the candidate lumen area of ​​the OCT pullback segment in the angiography image and the OCT lumen area obtained based on the OCT angiography image can be calculated by fitting.

[0065] Step 3: Substitute the candidate lumen area from the end of the OCT pull-back position to the proximal end of the blood vessel into the quadratic fitting relationship model to update the corrected lumen area, thereby achieving accurate calculation of the entire blood vessel area.

[0066] The FFR calculation unit uses the bifurcation fractal law to correct the area of ​​the bifurcation angle, and finally uses a fluid dynamics model to calculate the FFR.

[0067] According to the definition of fractional flow reserve (FFR), FFR can be expressed as the ratio of distal coronary artery pressure (Pd) to coronary artery orifice pressure (Pa), i.e., the pressure difference ΔP = Pa - 1 / 2, which gives the FFR value. Based on fluid dynamics principles, the main causes of pressure loss are related to the blood flow and vessel wall morphology of the target vessel. Increased friction between blood flow and the vessel wall when passing through a narrowed segment, as well as turbulent blood flow in the narrowed segment, can all cause pressure loss. Therefore, the pressure difference ΔP can be represented by the following model:

[0068] ΔP=FV+SV 2

[0069] In the formula, V is the blood flow velocity at maximum congestion, F is the viscous friction resistance coefficient, and S is the detachment force coefficient.

[0070] The viscous friction resistance coefficient F can be expressed as:

[0071]

[0072] The detachment force coefficient S can be expressed as:

[0073]

[0074] Where μ is the viscosity of blood, which is usually set to 4.0 × 10⁻⁶. -3 Pa·S; L represents the length of the blood vessel lumen; A n and A s These represent the lumen area of ​​the artery without stenosis and the lumen area at the stenosis, respectively; k represents the influence coefficient of the inlet and outlet on the pressure drop, usually taken as 1; ρ represents the blood density, taken as ρ = 1050 kg / m³. 3 Therefore, the key to calculating the pressure difference ΔP lies in calculating A. n and A s And blood flow velocity V.

[0075] A n and A s The lumen area can be calculated based on the identified lumen contour, but neglecting the influence of bifurcation on the calculation of the main vessel's lumen area will cause prediction errors in the FFR value. Therefore, this invention adopts the fractal bifurcation law, using the bifurcation area at the bifurcation point to correct the lumen area obtained by the candidate lumen area correction unit. Let's assume the initial lumen area obtained by the candidate lumen area correction unit is D. 1 The bifurcation area is D 2 The corrected lumen area D * It can be represented as:

[0076] D * =0.678×(D) 1 +D 2 )

[0077] Therefore, the pressure loss model constructed in this invention is as follows:

[0078] ΔP=FV+SV 2

[0079]

[0080]

[0081] in, and It is the lumen area after correction using the bifurcation and fractal law.

[0082] The final FFR value is calculated using the following formula:

[0083]

[0084] In the formula, Pa represents the pressure at the proximal end of the blood vessel. In this embodiment, a fixed value is used: Pa = 100 mmHg.

Claims

1. A multi-modal coronary function assessment system based on OFR, QFR and ACR techniques, characterized in that, include: The contrast imaging data acquisition unit is used to acquire contrast imaging data; The three-dimensional reconstruction unit performs three-dimensional reconstruction of blood vessels based on the angiographic image data from the two angles before OCT pullback, obtains a three-dimensional model of blood vessels, and further calculates the area of ​​the entire blood vessel from the distal end to the proximal end, defining this area as the candidate lumen area. maximum hyperemic flow velocity calculating unit for calculating a maximum hyperemic flow velocity ; The ACR registration unit is used to register OCT-contrast images based on OCT lens markers and contrast data acquired during OCT pullback, thereby obtaining the motion trajectory of the OCT lens markers on the contrast images. The candidate lumen area correction unit is used to correct the candidate lumen area to obtain an accurate calculation result of the lumen area of ​​the entire blood vessel, including the following steps: Calculate the OCT lumen area for each OCT-contrast image; Image registration is performed between any angiography image data before OCT retraction and the registered OCT-angiography image during OCT retraction to obtain the position of the angiography vessel segment during OCT retraction in the 3D model of the vessel. This position corresponds one-to-one with the OCT-angiography image. A relationship model is established between the candidate lumen area of ​​the OCT retraction segment in the angiography image and the OCT lumen area obtained based on the OCT-angiography image. This relationship model is used to correct the candidate lumen area from the end position of OCT retraction to the proximal end to obtain the corrected candidate lumen area, thereby obtaining the accurate lumen area calculation result of the entire vessel. The FFR calculation unit uses the following model to calculate FFR: In the formula: For the pressure in the proximal part of the blood vessel, For pressure difference, To achieve maximum blood flow rate, The coefficient of viscous frictional resistance. This is the detachment force coefficient. It is the lumen area of ​​an artery without stenosis, corrected using the bifurcation and fractal law. It is the lumen area at the narrowing point of the artery, corrected using the bifurcation and fractal law. It's the viscosity of the blood. The length of the blood vessel lumen. This represents the coefficient of influence of import and export on pressure drop. Let the blood density be denoted as , where the initial lumen area obtained through the candidate lumen area correction unit is . The bifurcation area of ​​the blood vessel bifurcation point is The corrected lumen area is then determined using the bifurcation and fractal law. Represented as: .

2. A multi-modal coronary function assessment system based on OFR, QFR and ACR technologies as claimed in claim 1, wherein, The angle difference between the two angles of the angiographic image data at the two angles before the OCT pullback used by the three-dimensional reconstruction unit is at least 25 degrees.

3. The multi-modal coronary function assessment system based on OFR, QFR and ACR technologies as claimed in claim 1, wherein, The maximum hyperemic flow rate calculating unit first calculates the contrast medium flow rate using a number frame method and then calculates the blood flow rate using the relationship between the contrast medium and the maximum hyperemic flow rate as shown in the following equation: wherein , , is a constant.

4. The multimodal coronary artery function assessment system based on OFR, QFR, and ACR technologies as described in claim 3, characterized in that, The frame-by-frame method used by the maximum congestion velocity calculation unit includes the following steps: For angiographic image data at any angle, the frames at which the contrast agent first reaches the proximal and distal ends of the vessel are selected. The time it takes for the contrast agent to travel from the proximal to the distal end of the vessel is calculated based on the frame difference and frame rate, and denoted as [the time is missing from the original text]. ; On a contrast-enhanced image, the proximal and distal positions of the contrast agent vessels are marked. The centerline of the vessels is extracted using a vascular skeleton algorithm, and the pixel distance of the centerline is calculated. Then, based on the resolution of the contrast-enhanced image, the actual distance of the centerline is calculated and denoted as [missing information]. ; The flow rate of the contrast agent is calculated using the formula .​ 5. The multi-modal coronary function assessment system based on OFR, QFR and ACR techniques as claimed in claim 1, wherein, The ACR registration unit uses the following steps to register OCT-contrast images: Step 1: Select an angiographic image in which the retracted blood vessel and the OCT lens markers are clearly visible as the first frame image; Step 2: Mark the position of the OCT lens marker and the pullback path, then extract the skeleton lines of other angiography images, use the iterative nearest neighbor algorithm to learn the projection matrix from the first frame image to other angiography images, and use this projection matrix to project the pullback path onto other angiography images to obtain the pullback path of each angiography image. Step 3: Based on the template matching method and combined with the motion characteristics of the OCT lens marker from far to near, project the OCT lens marker of the first frame image onto the next frame image. Then, find the region pixel point that is most similar to the current frame image and the OCT lens marker, and ensure that the region pixel point is near the projection point. In this way, the OCT lens marker recognition of all frames is realized, and the motion trajectory of the OCT lens marker on the imaging image is obtained, thereby completing the registration.

6. The multi-modal coronary function assessment system based on OFR, QFR and ACR techniques as claimed in claim 1, wherein, In the candidate lumen area correction unit, an OCT lumen segmentation model is implemented using the u-net algorithm based on the OCT-contrast image, and the OCT lumen area of ​​each OCT-contrast image is calculated based on the segmentation results.

7. The multi-modal coronary function assessment system based on OFR, QFR and ACR techniques as claimed in claim 1, wherein, In the candidate lumen area correction unit, a relationship model between the candidate lumen area of ​​the OCT pull-back segment in the contrast image and the OCT lumen area obtained based on the OCT-contrast image is established by fitting calculation. The obtained relationship model is a quadratic fitting relationship model.

Images