An interframe interpolation method for optimizing a three-dimensional vessel model based on intravascular imaging

Abstract

The application discloses a kind of based on intravascular imaging optimization three-dimensional vascular model interframe interpolation method;Obtain the sequence image group collected by intravascular imaging technology;And each frame image in sequence image group is carried out denoising processing;Using segmentation algorithm extracts the lumen boundary in each frame image, obtains the two-dimensional profile sequence image group of vascular lumen with frame data;Interpolation times are calculated;According to interpolation times, difference image group is generated, and according to the frame data of each image in difference image group, it is inserted into two-dimensional profile sequence image group;According to the last obtained two-dimensional profile sequence image group, generate high-resolution three-dimensional vascular model.The application is obtained by the frame interpolation technology of image collected by intravascular imaging technology, and the frame distance of adjacent image is reduced, and the continuity and smoothness of three-dimensional vascular model are improved.

Description

Technical Field

[0001] This invention belongs to the field of diagnostic technology using sound waves, specifically an inter-frame interpolation method based on intravascular imaging (OCT / IVUS) to optimize a three-dimensional vascular model and subsequently calculate fractional flow reserve (FFR). Background Technology

[0002] Fractional flow reserve (FFR) is an important parameter for assessing the functional severity of coronary artery stenosis. It is defined as the ratio of blood pressure distal to proximal to the stenosis under maximal congestion. Traditional FFR data relies on guidewire measurements, which are invasive and costly.

[0003] However, when images acquired using less invasive intravascular imaging techniques are imported into fractional flow reserve for evaluation, problems arise where the reconstructed model is “discrete” and has a discontinuous surface due to the large frame interval.

[0004] In recent years, with the widespread application of intravascular imaging technologies such as OCT (Optical Coherence Tomography) and IVUS (Intravascular Ultrasound), vascular models can be reconstructed from images, and FFR values ​​can be estimated using one-dimensional or three-dimensional blood flow simulation methods. This approach offers advantages such as being non-invasive and highly efficient. However, when using intravascular imaging technologies with OCT and / or IVUS, the relatively large frame interval often leads to issues with the accuracy of vascular geometry modeling, especially in areas of drastic structural changes such as vascular stenosis. This can further amplify the FFR calculation error, and these problems are amplified as the requirements for surgical prediction results become increasingly stringent.

[0005] When images acquired by intravascular imaging technology are applied to one-dimensional FFR modeling, problems such as sparse centerline sampling and discontinuous changes in lumen area can occur, which can easily cause abnormal fluctuations in pressure distribution. When applied to three-dimensional CFD simulation, discrete model boundaries can lead to poor mesh quality and unstable fluid solutions.

[0006] Current technologies lack systematic optimization methods to address the aforementioned problems, particularly regarding how to use inter-frame image interpolation to improve the continuity of 3D vascular models, thereby enhancing the accuracy of FFR calculations. Summary of the Invention

[0007] To address the problems existing in the background technology, this invention provides an inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging (OCT / IVUS). The technical solution includes:

[0008] A sequence of images acquired by intravascular imaging technology is obtained; each frame in the sequence of images is denoised; the lumen boundary in each frame is extracted using a segmentation algorithm to obtain a two-dimensional contour sequence of the vascular lumen with frame data;

[0009] Calculate the interpolation order f;

[0010] A difference image group is generated based on the number of interpolations f, and the frame data of each image in the difference image group is inserted into a two-dimensional contour sequence image group.

[0011] A high-resolution three-dimensional blood vessel model is generated based on the final obtained two-dimensional contour image sequence.

[0012] The formula for calculating the interpolation degree f is as follows:

[0013] ,

[0014] Where Δz0 is the original frame spacing, Δz e The desired frame spacing is less than 0.1mm, ceil() is the function that rounds up, and n is the maximum number of interpolation iterations.

[0015] The three-dimensional vascular model is used to fit the curve of the vascular inner wall generated by the interpolation points, and then the fractional flow reserve is calculated based on the vascular geometry model.

[0016] The process of extracting the lumen boundary in each frame of the image using the segmentation algorithm is as follows: the lumen grayscale image of the interpolated frame is formed by interpolation using the grayscale guided weight method, and then the contour of the blood vessel lumen is reconstructed.

[0017] The maximum number of interpolation iterations, n, is equal to the effective number of calculations of the marginal net benefit;

[0018] Marginal net profit Δn=B g −λC n ,

[0019] Where λ is the cost weighting parameter, and g is 1 in the initial calculation; B g =B0r g−1 B0 is the unitized profit brought by this interpolation, and r is the profit decay factor, r∈[0,1];

[0020] C n =C02 g−1 C0 is the baseline computational cost after this frame interpolation;

[0021] When Δn > 0, g is incremented by 1, and the marginal net gain Δn is calculated again;

[0022] The calculation stops when Δn≤0, the number of calculations is invalid, and the effective number of marginal net profit calculations is assigned to n.

[0023] The process of generating a set of difference images based on the number of interpolations f, and inserting the frame data of each image into the set of two-dimensional contour sequence images, includes:

[0024] Step 41: Obtain the data sequence for which intermediate frames need to be inserted:

[0025] Based on the frame data carried in the two-dimensional contour sequence image group, generate the intermediate number between each frame data as the sequence of intermediate frame data to be inserted; for each intermediate frame data in the sequence of frame data to be inserted, execute step 42;

[0026] Step 42: For each input intermediate frame data, perform an inter-frame interpolation algorithm to obtain the blood vessel cross-sectional contour belonging to the corresponding frame data;

[0027] Step 43: Based on the frame data positions provided by the intermediate frame data, insert the blood vessel cross-sectional contours corresponding to each intermediate frame data back into the two-dimensional contour sequence image group in sequence to obtain a new two-dimensional contour sequence image group; thereby inserting several intermediate frames between any two frames to generate an enhanced image sequence.

[0028] Step 42 includes:

[0029] Step 421, Centerline Interpolation:

[0030] Read the vessel centerlines C1(s) and C2(s) from two adjacent frames of images, and construct the intermediate interpolated centerline C. t (s), the formula is:

[0031] ,

[0032] ,

[0033] Where t is the interpolation parameter, α is the weighting parameter, and the arc length parameter s is a variable describing how far the vessel centerline has traveled along the path;

[0034] Step 422: Image interpolation uses the gray-level guided weighting method.

[0035] Read the grayscale information from two adjacent frames of images, and perform weighted fusion on the interpolated image. The grayscale value of a certain pixel x is calculated as follows:

[0036] ,

[0037] , ,

[0038] ,

[0039] Where, 𝐼1(𝑥) and 𝐼2(𝑥) represent the gray levels of point x in two adjacent frames; 𝑤1(𝑥) and 𝑤2(𝑥) are the gray level guiding weights of point x in two adjacent frames, 𝑒 is a constant, 𝜎 is a preset constant; 𝜇 is the average gray level of the corresponding pixels in the two frames;

[0040] Step 423: Reconstruct the cross-sectional contour of the blood vessel:

[0041] Based on the interpolated centerline Local cross-sectional grayscale scanning is performed along the normal direction in its orthogonal direction to restore the outline of the blood vessel cross section, determine the edge position, and reconstruct the complete blood vessel cross section shape.

[0042] The process of calculating fractional flow reserve based on vascular geometry model includes: performing three-dimensional CFD simulation and performing one-dimensional FFR modeling.

[0043] The three-dimensional CFD simulation includes:

[0044] Import the 3D reconstruction model into CFD software;

[0045] The blood flow field and pressure distribution are calculated, and the ratio of pressure at the distal end of the stenosis to the aortic pressure is calculated to obtain the fractional flow reserve.

[0046] The one-dimensional FFR modeling includes:

[0047] Extract the blood vessel centerline and the curve showing the change in lumen area along the vessel;

[0048] The solution is based on a one-dimensional blood flow model;

[0049] Based on the calculated blood pressure distribution along the path, the ratio of pressure distal to the stenosis to the aortic pressure is calculated to obtain the fractional flow reserve.

[0050] The beneficial effects of this invention are as follows:

[0051] 1. This invention is the first to propose using a multiple interpolation algorithm to perform inter-frame interpolation on OCT / IVUS sequence images, systematically improving image density and the smoothness of the three-dimensional vascular model, and optimizing the quality of vascular geometric modeling. By using inter-frame interpolation technology on images acquired by intravascular imaging technology, the frame interval between adjacent images is reduced, improving the continuity and smoothness of the three-dimensional vascular model, enhancing the geometric accuracy of CFD simulation and one-dimensional models, thereby significantly improving the accuracy and stability of FFR calculation results. This is a structural enhancement to existing FFR and other workflows.

[0052] 2. The interpolation method of this invention directly calculates the number of interpolations based on the original frame interval, which allows for rapid implementation of the entire interpolation process and achieves optimal optimization results, avoiding blindly calculating the number of interpolations. It is also applicable to OCT, IVUS, or fused image sequences, and can be integrated into existing medical image processing platforms or embedded in edge computing devices, demonstrating strong applicability.

[0053] 3. Structural-fidelity interpolation of lumen area and centerline changes: An inter-frame interpolation method combining spline centerline interpolation and grayscale guided image fusion is proposed. The interpolation process does not only interpolate the image boundary positions; it provides dual guarantees of physical consistency and image grayscale consistency, preserving the true anatomical structure and area change trend of the vascular stenosis area. This maximizes the restoration of the original anatomical morphology in the interpolated image, providing a more realistic foundation for physical simulation and mathematical modeling. The interpolation algorithm has high computational efficiency and good engineering implementation value.

[0054] 4. Subsequently, FFR calculations are performed on the 3D vascular model. The inter-frame interpolation system is systematically introduced into the OCT / IVUS image FFR calculation process. Currently, most virtual FFR calculation methods use the original intravascular image sequence, ignoring the structural continuity between images; therefore, this invention is a key step in achieving efficient, non-invasive FFR calculation. The number of interpolations is automatically calculated using an interpolation algorithm, automating the FFR calculation process.

[0055] 5. A unified interpolation method that is compatible with both 3D CFD and 1D FFR models: Traditional CFD and 1D models use data with different precision and format requirements, requiring separate processing. The interpolation framework proposed in this invention enhances image continuity while being compatible with both 1D and 3D modeling processes, significantly improving both types of FFR models. It has high versatility and systems engineering value; moreover, the results are closer to the measured values ​​of the guidewire, significantly reducing computational errors.

[0056] 6. Optimize the accuracy of 3D CFD simulation: The interpolated blood vessel model surface is smoother, the CFD mesh quality is improved, the numerical solution is more stable, and the FFR error is significantly reduced.

[0057] 7. Improve the stability of the one-dimensional FFR model: The smooth transition of the blood vessel cross-sectional area curve reduces non-physical abrupt changes in pressure calculation, improves the discreteness and stability problems in one-dimensional FFR modeling, and enhances the computational stability and the reliability of the results. Attached Figure Description

[0058] Figure 1 This is a schematic flowchart of an embodiment of the inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to the present invention.

[0059] Figure 2This is a schematic diagram of the process for generating FFR calculation results in an embodiment of the present invention.

[0060] Figure 3 This is a comparison diagram of FFR obtained from embodiments and comparative examples of the present invention.

[0061] Figure 4 This is a comparison chart of the lumen area curves obtained by secondary interpolation in the embodiments and comparative examples of the present invention.

[0062] Figure 5 This is a comparative diagram of the lumen area curve and FFR curve obtained by secondary interpolation in an embodiment of the present invention.

[0063] Figure 6 This is a schematic diagram of the image generation process in each step of the present invention. Detailed Implementation

[0064] The present invention will be further described in detail below with reference to the accompanying drawings.

[0065] like Figure 1 , Figure 2 and Figure 6 The embodiment of the present invention shown includes:

[0066] Step 1: Acquire a sequence of images captured by intravascular imaging technology (OCT or IVUS) under the condition of maximum vascular congestion; and perform denoising processing on each frame of the sequence of images (such as median filtering, wavelet denoising, etc.).

[0067] Step 2: Use segmentation algorithms (such as active contour models, deep learning models, etc.) to extract the lumen boundary in each frame of the image; thereby obtaining a two-dimensional contour sequence image group of the blood vessel lumen with frame data;

[0068] Step 3: Calculate the interpolation degree f:

[0069] The formula for calculating the interpolation degree f is as follows:

[0070] ,

[0071] Where Δz0 is the original frame spacing, Δz e The desired frame spacing is less than 0.1mm; ceil() rounds up the function, and n is the maximum number of interpolation iterations, calculated to be 3 based on the principle of maximizing overall net benefit; in this embodiment, the original frame spacing Δz0 = 0.2mm for the sequence image group, and the desired frame spacing Δz... e =0.05mm;

[0072] Step 4: Generate a difference image group based on the number of interpolations f, and insert the frame data of each image in the difference image group into the two-dimensional contour sequence image group;

[0073] Step 5: Generate a high-resolution three-dimensional blood vessel model based on the final obtained two-dimensional contour image sequence.

[0074] Step 6: Use the 3D vascular model to perform curve fitting (such as spline curve, Bezier curve, etc.) on the vascular inner wall surface generated by the interpolation points, and output the vascular geometry model in readable formats such as STL, VTK or CFD.

[0075] Step 7: Calculate the fractional flow reserve (FFR, the ratio of pressure distal to the stenosis to aortic pressure) based on the vascular geometry model.

[0076] In step 2 of this embodiment, the process of extracting the lumen boundary in each frame image using a segmentation algorithm is as follows: the lumen grayscale image of the interpolated frame is formed by interpolation using the grayscale guided weight method, and then the contour of the blood vessel lumen is reconstructed.

[0077] In step 3 of this embodiment, the formula for calculating the interpolation number f is derived from the original frame spacing formula. The influence of the interpolation number f on the frame spacing is as follows:

[0078] ,

[0079] in This is the original frame spacing. is the interpolated frame spacing, and f is the number of interpolations.

[0080] In scenarios where FFR is calculated based on intravascular imaging (OCT / IVUS), a frame interval of 0.05 mm is desired. Interpolation to reduce the frame interval to less than 0.1 mm can significantly optimize the problem of "discrete" and discontinuous surfaces in the reconstructed model caused by the frame interval, thus affecting the accuracy of vascular geometry modeling. For example, the original frame interval of intravascular imaging (OCT / IVUS) is 0.2 mm. After one interpolation, the frame interval becomes 0.1 mm, doubling the number of images. After a second interpolation, the frame interval becomes 0.05 mm, quadrupling the number of images. After a third interpolation, the frame interval becomes 0.025 mm, significantly increasing the continuity of subsequent FFR calculations and greatly reducing errors. However, the computational cost of third interpolation is high, and such a small frame interval is unnecessary for FFR calculation; therefore, in this embodiment, n=2 is calculated.

[0081] The formula for calculating the interpolation order in step 3 applies to first-order, second-order, and third-order interpolation.

[0082] The principle of maximizing overall net benefit in step 3 of this embodiment takes into account that the computational cost doubles with each interpolation (inserting one frame between each pair of adjacent frames): the overall computational cost C for the nth interpolation... n =C02g−1 C0 is the baseline computation cost of performing a single frame interpolation on the entire sequence; the benefits of frame interpolation (improved geometric continuity / stability, and improved FFR accuracy, etc.) are represented by the geometric decay model: B g =B0r g−1 B0 is the "unitized gain" (relative improvement in average error) from a single interpolation, B g g is an intermediate variable, initially set to 1; r is the revenue decay factor, r∈[0,1]; marginal net benefit Δn=B g −λC n λ is the cost weighting parameter, used to convert the computational cost into a measurement system similar to that of revenue. The value range of the cost weighting parameter λ is 10. -3 ≤λ≤10 3 The algorithm can be dynamically adjusted based on computational resource prices, computation time, energy consumption, or hardware utilization. The decision is as follows: if and only if Δn > 0, the current frame interpolation is considered "worth doing," and after g+1, the marginal net benefit is recalculated; otherwise (Δn ≤ 0), the current calculation is invalid, and the valid calculation count of the marginal net benefit is assigned to n, and the process ends. The optimal stopping point is also given at the point where the total net benefit (cumulative) is maximized. In this example, B0 = 10, C0 = 1, r = 0.8, λ = 1, and the maximum number of interpolation counts, n = 3, is calculated based on the principle of maximizing total net benefit.

[0083] Step 4 in this embodiment specifically includes:

[0084] Step 41: Obtain the data sequence for which intermediate frames need to be inserted:

[0085] Based on the frame data carried in the two-dimensional contour sequence image group, generate the intermediate number between each frame data as the sequence of intermediate frame data to be inserted; for each intermediate frame data in the sequence of frame data to be inserted, execute step 42;

[0086] Step 42: For each input intermediate frame data, perform an inter-frame interpolation algorithm to obtain the blood vessel cross-sectional contour belonging to the corresponding frame data;

[0087] Step 43: Based on the frame data positions provided by the intermediate frame data, insert the blood vessel cross-sectional contours corresponding to each intermediate frame data back into the two-dimensional contour sequence image group to obtain a new two-dimensional contour sequence image group; thereby inserting several intermediate frames between any two frames to generate an enhanced image sequence, enhancing the continuity and consistency of the blood vessel structure, and significantly improving the stability and accuracy of subsequent three-dimensional / one-dimensional FFR calculations.

[0088] In step 42, an inter-frame interpolation algorithm is used, taking two adjacent frames as endpoints, and interpolating boundary points using inter-frame interpolation methods. During interpolation, the positional change trend and grayscale information evolution of the boundary points are considered to ensure the geometric and visual coherence of the interpolated image. Specifically, a spatial interpolation strategy (e.g., cubic spline interpolation) combined with image grayscale-guided weighted interpolation improves the consistency and structural clarity of blood vessel contours in the sequence images, thereby improving the stability and accuracy of subsequent one-dimensional and three-dimensional CFD calculations of FFR, including:

[0089] Step 421, Centerline Interpolation

[0090] Read the vessel centerlines C1(s) and C2(s) from two adjacent frames of images. Based on their arc length parameter s∈[0,1], construct the intermediate interpolated centerline C using cubic spline interpolation. t (s), which is defined as:

[0091] ,

[0092] ,

[0093] Where t is the interpolation parameter, t∈[0,1]; 𝛼 is the weight parameter, 𝛼∈[0,1]; C m (s) is the linear weighted average of the two centerlines at the same arc length parameter s. The arc length parameter s is a variable describing how far the vessel centerline has traveled along the path. It can be used to orderly, continuously, and smoothly identify and interpolate points on the centerline; C i (s)=(x i (s),y i (s) is used to represent the coordinates of the center point at position i with arc length s, where x i (s) is the x-coordinate, y i (s) represents the ordinate (i=1,2,m,t); its biggest advantage over linear or quadratic interpolation is that it is smoother and the transition is more natural, making it especially suitable for scenarios such as the centerline of blood vessels that require geometric consistency and physical continuity.

[0094] Step 422: Image interpolation uses the gray-level guided weighting method.

[0095] To enhance image structure consistency, grayscale information from two adjacent frames is read, and the interpolated image is weighted and fused. The grayscale value of a certain pixel x is calculated as follows:

[0096] ,

[0097] , ,

[0098] ,

[0099] Wherein, 𝐼1(𝑥) and 𝐼2(𝑥) represent the gray levels of point x in two adjacent frames; 𝑤1(𝑥) and 𝑤2(𝑥) are the gray level guiding weights of point x in two adjacent frames. This weighting strategy makes the interpolated image more inclined to regions with stable gray level changes and consistent structure, effectively preserving the edges of blood vessels; 𝑒 is a constant, 𝜎 is a preset constant with a value range of [10, 50]; 𝜇 is the average gray level of corresponding pixels in the two frames.

[0100] Step 423: Reconstruct the cross-sectional contour of the blood vessel

[0101] Based on the interpolated centerline C t (s) Local cross-sectional grayscale scanning is performed along the normal direction in its orthogonal direction to restore the outline of the blood vessel cross section; edge positions are determined by methods such as Gaussian fitting, guided filtering or gradient search, and the complete blood vessel cross section shape is reconstructed.

[0102] In step 7 of this embodiment, the process of calculating the fractional flow reserve based on the vascular geometry model includes: performing three-dimensional CFD simulation and performing one-dimensional FFR modeling; wherein, performing three-dimensional CFD simulation includes:

[0103] Step 711: Import the 3D reconstruction model into CFD software (such as ANSYS Fluent, OpenFOAM, etc.).

[0104] Step 712: Set boundary conditions (inlet velocity, outlet pressure, or impedance);

[0105] Step 713: Calculate the blood flow field and pressure distribution, calculate the ratio of pressure at the distal end of the stenosis to the aortic pressure, and obtain the fractional flow reserve (FFR).

[0106] One-dimensional FFR modeling includes:

[0107] Step 721: Extract the vessel centerline and the curve showing the change in lumen area along the vessel;

[0108] Step 722: Solve based on a one-dimensional blood flow model;

[0109] Step 723: Calculate the blood pressure distribution along the path and calculate the ratio of pressure distal to the stenosis to the aortic pressure to obtain the fractional flow reserve (FFR).

[0110] The fractional flow reserve (FFR) variation curves along the vessel length obtained by one-dimensional FFR modeling in this embodiment and the comparative example are shown below. Figure 3As shown in the figure. The vertical axis represents the FFR value, and the horizontal axis represents the axial distance in the opposite direction with the distal end of the blood vessel as the origin (0). The vascular geometry data is derived from a medical image sequence with a frame interval of 0.2 mm.

[0111] The original FFR value was 0.87, and the FFR value calculated after lumen interpolation was 0.79. The interpolated image set generated based on the results of the two lumen extractions optimized the reconstruction accuracy of the vascular geometry model. The more accurate geometry improved the accuracy of calculating frictional resistance and local resistance in blood flow simulation, thereby reducing the calculation error of FFR and making it closer to the actual measurement value.

[0112] The vessel lumen area curves obtained by one-dimensional FFR modeling in this embodiment and the comparative example are shown below. Figure 4 As shown in the figure, the vertical axis represents the lumen area of ​​the blood vessel, and the horizontal axis represents the length of the blood vessel. The point where the horizontal axis is 0 is the distal end of the blood vessel. It can be seen from the figure that the lumen area of ​​the blood vessel segment from 0 to 5 mm is small and varies drastically. After lumen interpolation, the continuity of the lumen area is better, and the area abrupt change can be well optimized through interpolation. In particular, when there is a sudden increase or decrease in area, lumen interpolation has a significant impact on the FFR calculation results.

[0113] In this embodiment, a secondary interpolation image was generated from the acquired frame interval of 0.2 mm, thereby reducing the frame interval from 0.2 mm to 0.05 mm. This effectively increases the information of the blood vessel cross-section, ultimately generating a high-resolution three-dimensional blood vessel model with the lumen area and FFR curve as shown below. Figure 5 As shown in the figure, the horizontal axis represents the same frame data of the two curves, and 0 represents the distal end of the blood vessel. As can be seen from the figure, after processing by the method of this embodiment, the problem of abrupt area change and the resulting large FFR calculation error can be optimized at the same time.

Claims

1. A frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging, characterized in that, include: A sequence of images acquired by intravascular imaging technology is obtained; each frame in the sequence of images is denoised; the lumen boundary in each frame is extracted using a segmentation algorithm to obtain a two-dimensional contour sequence of the vascular lumen with frame data; Calculate the interpolation order f; Generate a difference image group based on the number of interpolations f, and insert the frame data of each image in the difference image group into the two-dimensional contour sequence image group; A high-resolution three-dimensional blood vessel model is generated based on the final obtained two-dimensional contour image sequence. The formula for calculating the interpolation degree f is as follows: ， Where Δz0 is the original frame spacing, Δz e The desired frame spacing is less than 0.1mm, ceil() is the function that rounds up, and n is the maximum number of interpolation iterations; The maximum number of interpolation iterations, n, is equal to the effective number of calculations of the marginal net benefit, where: Marginal net profit Δn=B g −λC n , Where λ is the cost weighting parameter, and g is 1 in the initial calculation; B g =B0r g−1 B0 is the unitized profit brought by this interpolation, and r is the profit decay factor, r∈[0,1]; C n =C02 g−1 C0 is the baseline computational value after this interpolation; When Δn > 0, g is incremented by 1, and the marginal net gain Δn is calculated again; The calculation stops when Δn≤0, the number of calculations is invalid, and the effective number of marginal net profit calculations is assigned to n.

2. The inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to claim 1, characterized in that, The three-dimensional vascular model is used to fit the curve of the vascular inner wall generated by the interpolation points, and then the fractional flow reserve is calculated based on the vascular geometry model.

3. The inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to claim 1, characterized in that, The process of extracting the lumen boundary in each frame of the image using the segmentation algorithm is as follows: the lumen grayscale image of the interpolated frame is formed by interpolation using the grayscale guided weight method, and then the contour of the blood vessel lumen is reconstructed.

4. The inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to claim 1, characterized in that, The process of generating a difference image group based on the interpolation number f, and inserting the frame data of each image in the difference image group into the two-dimensional contour sequence image group includes: Step 41: Obtain the sequence of intermediate frame data to be inserted: Based on the frame data carried in the two-dimensional contour sequence image group, generate intermediate frame data between each frame data as the intermediate frame data sequence to be inserted; for each intermediate frame data in the intermediate frame data sequence to be inserted, execute step 42. Step 42: For each intermediate frame data, perform an inter-frame interpolation algorithm to obtain the blood vessel cross-sectional contour belonging to the corresponding frame data; Step 43: Based on the frame data positions provided by the intermediate frame data, insert the blood vessel cross-sectional contours corresponding to each intermediate frame data back into the two-dimensional contour sequence image group in sequence to obtain a new two-dimensional contour sequence image group; thereby inserting several intermediate frames between any two frames to generate an enhanced image sequence.

5. The inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to claim 4, characterized in that, Step 42 includes: Step 421, Centerline Interpolation: Read the vessel centerlines C1(s) and C2(s) from two adjacent frames of images, and construct the intermediate interpolated centerline C. t (s), the formula is: ， ， Where t is the interpolation parameter, α is the weighting parameter, and the arc length parameter s is a variable describing how far the vessel centerline has traveled along the path; Step 422: Image interpolation uses the gray-level guided weighting method. An interpolated image is generated by weighted fusion, and grayscale information from two adjacent frames is read to produce the interpolated image. The grayscale value of a certain pixel x is calculated as follows: ， ， ， ， Wherein, 𝐼1(𝑥) and 𝐼2(𝑥) represent the gray levels of point x in two adjacent frames; 𝑤1(𝑥) and 𝑤2(𝑥) are the gray level guiding weights of point x in two adjacent frames; where 𝐼1(𝑥) and 𝐼2(𝑥) represent the gray levels of point x in two adjacent frames; 𝑤1(𝑥) and 𝑤2(𝑥) are the gray level guiding weights of point x in two adjacent frames, 𝑒 is a constant, 𝜎 is a preset constant; 𝜇 is the average gray level of corresponding pixels in the two frames. Step 423: Reconstruct the cross-sectional contour of the blood vessel: Based on the interpolated centerline C t (s) Local cross-sectional grayscale scanning is performed along the normal direction in its orthogonal direction to restore the outline of the blood vessel cross section, determine the edge position, and reconstruct the complete blood vessel cross section shape.

6. The inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to claim 2, characterized in that, The process of calculating fractional flow reserve based on vascular geometry model includes: performing three-dimensional CFD simulation and performing one-dimensional FFR modeling.

7. The inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to claim 6, characterized in that, The three-dimensional CFD simulation includes: Import the 3D reconstruction model into CFD software; The blood flow field and pressure distribution are calculated, and the ratio of pressure at the distal end of the stenosis to the aortic pressure is calculated to obtain the fractional flow reserve.

8. The inter-frame interpolation method for optimizing a three-dimensional vascular model based on intravascular imaging according to claim 7, characterized in that, The one-dimensional FFR modeling includes: Extract the blood vessel centerline and the curve showing the change in lumen area along the vessel; The solution is based on a one-dimensional blood flow model; Based on the calculated blood pressure distribution along the path, the ratio of pressure distal to the stenosis to the aortic pressure is calculated to obtain the fractional flow reserve.

Images