Medical image processing method and system for analyzing coronary stenosis

Abstract

Methods and systems for simulating stents within coronary artery luminal structures are disclosed. A method of simulating a stent within a coronary artery luminal structure, the method comprising: reconstructing a three-dimensional coronary arterial tree from a segmented coronary artery luminal profile; replacing a portion of the three-dimensional arterial tree with a candidate stent structure; and simulating a pressure distribution of the coronary arterial tree to determine a fractional flow reserve through the candidate stent structure.

Description

Technical Field

[0001] This invention relates to a medical image processing method for analyzing coronary artery stenosis, and more particularly to the simulation of stents. Background Technology

[0002] Coronary artery disease occurs when atherosclerotic plaques gradually accumulate in the coronary arteries, causing narrowing of the lumen and reducing blood supply to the heart muscle. When coronary artery narrowing leads to ischemia, revascularization, including percutaneous coronary intervention (PCI), may be necessary.

[0003] Although invasive fractional flow reserve (FFR) measurement is the recommended method, its usage is relatively low due to various factors associated with invasive FFR measurement, including cost, time, and potential adverse effects. Furthermore, this method is difficult to apply to bifurcation and tandem lesions. We employ a method called FFR... B A system integrating computed tomography coronary angiography (CTCA) and reduced-order computational fluid dynamics (CFD) techniques was developed for non-invasive FFR assessment. Based on this platform, we propose evaluating the functional status of coronary artery lesions after PCI by modeling virtual stent implantation. This allows operators to comprehensively assess and compare different implantation strategies and their impact on post-PCI coronary physiology. This could improve the appropriateness of patient selection in invasive treatment strategies and personalize stent selection and placement planning, potentially revolutionizing clinical decision-making and treatment planning in coronary artery disease.

[0004] FFR is the "gold standard" for selecting patients who may benefit from revascularization and for assessing functional improvement after revascularization. However, during invasive coronary angiography (ICA), FFR needs to be measured under conditions of congestion induced by adenosine, 5'-adenosine triphosphate, or papaverine. When measuring FFR, a guidewire with a pressure sensor at the tip is inserted into the target coronary artery through a catheter, and the pressures at both ends of the stenosis are recorded and the ratio of the two is used to quantify FFR. For patients with coronary artery disease with FFR > 0.8, drug therapy is recommended. For lesions with FFR ≤ 0.8, coronary revascularization is recommended. If FFR > 0.8 after revascularization, the procedure is considered successful. Although FFR-guided coronary revascularization has been shown to improve clinical outcomes, FFR measurement is not widely used in clinical practice due to high medical costs, additional operation time, and potential complications involved in invasive procedures [5]. Therefore, a non-invasive method is needed to identify ischemic lesions and assess the results of revascularization.

[0005] In terms of revascularization, PCI has proven effective in restoring cardiac blood flow to patients with coronary artery disease. The global coronary stent market was valued at $5.2 billion in 2017 and is projected to grow to $8.4 billion by 2025. Decisions regarding implantation location, stent selection (e.g., size / type / number), and stent implantation strategy remain complex and challenging, particularly for the treatment of bifurcation and tandem lesions, which are prone to in-stent restenosis (ISR), in-stent thrombosis, and related adverse clinical events. While FFR measurement is recommended for PCI in the treatment of bifurcation lesions, the pressure guidewire often cannot re-pass through the waveband of the stent previously implanted in the main vessel. For tandem lesions, due to physiological interactions, it is difficult to use FFR measurement to separately assess the severity of various stenosis conditions. Therefore, a tool is needed to model the functional status of coronary artery lesions after PCI with different implantation strategies to aid in PCI treatment planning.

[0006] CTCA (Computed Tomography Angiography) is an important non-invasive detection technique for assessing coronary artery disease based on high-resolution anatomical images of lesion sites. However, while it has high sensitivity, its specificity is low in identifying ischemic coronary artery lesions. Currently, combining CTCA with CFD (Computed Tomography-Fractional Analysis) modeling of coronary hemodynamics has made it feasible to non-invasively assess the functional significance of coronary artery lesions. A non-invasive method, FFR (Free Front-End Reduction), employing reduced-order CFD modeling and novel exit boundary conditions, has been developed. B One example can be found in the following article: Zhang JM, Zhong L, Luo T, Lomarda AM, Huo Y, Yap J, Lim ST, Tan RS, Wong ASL, Tan JWC, Yeo KK, Fam JM, Keng FYJ, Wan M, Su B, Zhao X, Allen JC, Kassab GS, Chua TSJ, Tan SY, A simplified model of noninvasive fractional blood flow reserve based on CT images, PLOS ONE, 2016, Vol. 11, No. 5: e0153070 (DOI: 10.1371 / journal.pone.0153070). Summary of the Invention

[0007] According to a first aspect of the present invention, a method for simulating the structure of a coronary artery lumen and providing an intraluminal stent is provided. The method includes: reconstructing a three-dimensional coronary plexus based on a segmented coronary artery lumen profile; replacing a portion of the three-dimensional plexus with a candidate stent structure; and simulating the pressure distribution of the coronary plexus to determine the fractional noninvasive flow reserve through the candidate stent structure.

[0008] In one embodiment, the method further includes: identifying candidate locations of the candidate scaffold structure.

[0009] In one embodiment, identifying the candidate location of the candidate stent structure includes: determining the average lumen area as the vessel straightens; identifying the proximal end and distal end of the lesion based on the average lumen area as the vessel straightens; and determining the candidate stent location based on the proximal end and distal end of the lesion.

[0010] In one embodiment, identifying the proximal end and the distal end of the lesion based on the average lumen area varying with the straightened length of the vessel includes: determining the curvature of the average lumen area varying with the straightened length of the vessel; and identifying the proximal end and the distal end of the lesion as the maximum value of the absolute value of the curvature.

[0011] In one embodiment, identifying the proximal end and the distal end of the lesion based on the average lumen area varying with the straightened length of the vessel includes: determining the slope change of the average lumen area varying with the straightened length of the vessel; and identifying the proximal end and the distal end of the lesion as the maximum value of the slope change.

[0012] In one embodiment, identifying the proximal end and the distal end of the lesion based on the average lumen area that varies with the length of the straightened vessel includes: identifying the proximal end and the distal end of the lesion based on the location of the minimum average lumen area; and identifying the proximal end and the distal end of the lesion on each side of the minimum average lumen area.

[0013] In one embodiment, the method further includes determining the diameter of the candidate stent structure based on the average lumen diameter at the proximal end of the lesion and the distal end of the lesion.

[0014] In one embodiment, the method further includes determining the length of the candidate stent structure based on the location of the proximal end of the lesion and the location of the distal end of the lesion.

[0015] In one embodiment, the method further includes adding an extension allowance to the length of the candidate scaffold structure.

[0016] In one embodiment, the method further includes: segmenting the coronary artery lumen structure from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen contour.

[0017] In one embodiment, segmenting the coronary artery lumen structure from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen contour includes: designating a point at the aortic sinus as the origin of the coronary plexus; determining the vascular centerline of the arteries in the plexus; using the centerline, creating an extended multiplanar reformatted volume for each segment of the plexus; extracting longitudinal sections from the extended multiplanar reformatted volume; detecting the lumen boundary of the extracted longitudinal section; and using the detected lumen boundary, detecting the lumen boundary contour of each slice of the multiplanar reformatted volume.

[0018] In one embodiment, the longitudinal sections are extracted from the extended multiplane reformatted volume at 45-degree intervals.

[0019] In one implementation, the centerline of the blood vessel is determined using a Hessian filter.

[0020] According to a second aspect of the present invention, a computer-readable carrier medium carrying processor-executable instructions is provided, which, when executed by a processor, cause the processor to perform the above-described method.

[0021] According to a third aspect of the present invention, a medical image processing system for simulating stents within the coronary artery lumen structure is provided. The medical image processing system includes a processor and a data storage device storing executable computer program instructions for instructing the processor to: reconstruct a three-dimensional coronary plexus by segmenting the coronary artery lumen contour; replace a portion of the three-dimensional plexus with a candidate stent structure; and simulate the pressure distribution of the coronary plexus to determine the fractional noninvasive flow reserve through the candidate stent structure.

[0022] In one embodiment, the data storage device further stores executable computer program instructions for instructing the processor to: identify candidate locations of the candidate scaffold structures.

[0023] In one embodiment, the data storage device further stores executable computer program instructions for causing the processor to identify the candidate location of the candidate stent structure by: determining the average lumen area as the vessel straightening length varies; identifying the proximal end of the lesion and the distal end of the lesion based on the average lumen area as the vessel straightening length varies; and determining the candidate stent location based on the proximal end of the lesion and the distal end of the lesion.

[0024] In one embodiment, identifying the proximal end and the distal end of the lesion based on the average lumen area varying with the straightened length of the vessel includes: determining the curvature of the average lumen area varying with the straightened length of the vessel; and identifying the proximal end and the distal end of the lesion as the maximum value of the absolute value of the curvature.

[0025] In one embodiment, identifying the proximal end and the distal end of the lesion based on the average lumen area varying with the straightened length of the vessel includes: determining the slope change of the average lumen area varying with the straightened length of the vessel; and identifying the proximal end and the distal end of the lesion as the maximum value of the slope change.

[0026] In one embodiment, identifying the proximal end and the distal end of the lesion based on the average lumen area that varies with the straightened length of the blood vessel includes: identifying the location of the minimum average lumen area; and identifying the proximal end and the distal end of the lesion on each side of the minimum average lumen area.

[0027] In one embodiment, the data storage device further stores executable computer program instructions for: determining the diameter of the candidate stent structure based on the average lumen diameter at the proximal end of the lesion and the distal end of the lesion.

[0028] In one embodiment, the data storage device further stores executable computer program instructions for: determining the length of the candidate stent structure based on the location of the proximal end of the lesion and the location of the distal end of the lesion.

[0029] In one embodiment, the data storage device further stores executable computer program instructions for: adding an extension allowance to the length of the candidate support structure.

[0030] In one embodiment, the data storage device further stores executable computer program instructions for: segmenting the coronary artery lumen structure from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen outline.

[0031] In one embodiment, the data storage device further stores executable computer program instructions for segmenting the coronary artery lumen structure from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen contour by: designating a point at the aortic sinus as the origin of the coronary plexus; determining the vascular centerline of the arteries of the plexus; using the centerline, creating an extended multiplanar reformatted volume for each segment of the plexus; extracting longitudinal sections from the extended multiplanar reformatted volume; detecting the lumen boundary of the extracted longitudinal section; and using the detected lumen boundary, detecting the lumen boundary contour of each slice of the multiplanar reformatted volume.

[0032] In one embodiment, the longitudinal sections are extracted from the extended multiplane reformatted volume at 45-degree intervals.

[0033] In one implementation, the centerline of the blood vessel is determined using a Hessian filter. Attached Figure Description

[0034] In the following description, embodiments of the invention are illustrated by way of non-limiting examples, with reference to the accompanying drawings. In the drawings:

[0035] Figure 1 A flowchart illustrating a method for simulating stent implantation in a patient according to an embodiment of the present invention;

[0036] Figure 2 This is a block diagram of a medical image processing system according to an embodiment of the present invention;

[0037] Figure 3 A flowchart of a medical image processing method for simulating stent implantation according to an embodiment of the present invention;

[0038] Figures 4A to 4D The images shown are: a set of computed tomography coronary angiography (CTCA) images of the patient 410; multi-planar reformatted images obtained from CTCA images; the luminal contour depiction effect from both longitudinal and transverse perspectives; and a three-dimensional reconstruction model of the patient's coronary arteries;

[0039] Figures 5A to 5E The illustration shows an example of selecting stent parameters in a method according to an embodiment of the present invention.

[0040] Figures 6A to 6D The image shows a method for simulating the placement of the support in one embodiment of the present invention;

[0041] Figures 7A to 7D The figures shown are invasive FFR and non-invasive FFR. B Comparative assessment of narrow functional impairment;

[0042] Figure 8 The illustration shows a comparative example of three different stent implantation strategies when treating a patient with tandem lesions of the left anterior descending artery using the method according to an embodiment of the present invention.

[0043] Figure 9A The image shows the FFR and FFR before stent implantation. B The correlation between them Figure 9B The image shown is the corresponding Bland-Altman diagram; and

[0044] Figure 10A The image shows the FFR and FFR after stent implantation. B The correlation between them Figure 10B The image shown is the corresponding Bland-Altman diagram. Detailed Implementation

[0045] This invention relates to systems and methods for quantifying coronary artery stenosis, modeling percutaneous coronary intervention (PCI) procedures, and evaluating hemodynamics before and after virtual stent implantation. The entire process includes coronary artery segmentation, centerline extraction, centerline tracing, generation of arterial cross-sectional images, arterial lumen segmentation, stenosis detection / quantification, simulation of stent implantation (size, length, and number) through shape reconstruction, placement, and evaluation of the fractional flow reserve (FFR) before and after virtual stent implantation. B )Evaluate.

[0046] Figure 1 This is a flowchart illustrating a method for simulating stent implantation in a patient according to an embodiment of the present invention. Figure 1 Method 10 shown can be used to simulate the results of stent implantation surgery performed on a patient.

[0047] In step 12 of the CTCA imaging procedure, computed tomography (CT) images of the coronary angiography are acquired. Imaging is performed according to the guidelines of the Society of Cardiovascular Computed Tomography. CTCA imaging can be performed using existing multi-slice CT scanners with high image spatial resolution, such as the Toshiba Aquilion One 320 slice, Canon Aquilion ONE / Genesis 640, Siemens Somatom Force Dual Source, and Philips Brilliance iCT. For patients with a resting heart rate >65 bpm, medications such as beta-blockers for heart rate regulation, including ivabradine, may be administered, and nitroglycerin may be administered sublingually before each scan. CTCA scans can be performed using a prospective ECG-triggered scanning mode. All CTCA images can be saved in DICOM format.

[0048] In image segmentation step 14, CTCA image processing is performed to segment the coronary artery lumen structure. Specifically, the inner contour of each transverse slice is detected, guided by the intersection of the transverse and longitudinal contours. Furthermore, lumen contour delineation in both transverse and longitudinal planes is supported. The segmented lumen contours are saved for further processing.

[0049] In step 16 of the patient-specific 3D model reconstruction, all segmented coronary artery lumen contours are stitched together to generate a patient-specific coronary artery 3D reconstruction model.

[0050] In stent implantation simulation step 18, a portion of the patient's three-dimensional coronary plexus model is replaced with a simulated stent. In some embodiments, after a stenotic region of the coronary plexus is detected, the stenotic region is replaced with a simulated stent. In other embodiments, the user manually selects the stenotic region to be replaced by the simulated stent. More than one stenotic region may be replaced by a simulated stent.

[0051] Following stent implantation simulation step 18, a model cleaning / meshing step 20 is performed. In model cleaning / meshing step 20, the 3D model is cleaned to have inlet and outlet boundaries perpendicular to the corresponding vessel centerline, respectively. Subsequently, the computational domain of the cleaned model is discretized into tetrahedral, triangular prism, or hexahedral elements with boundary expansion layers.

[0052] Following this, a computational fluid dynamics (CFD) simulation is performed. Specifically, for the computational domain, the continuity and momentum conservation equations (also known as the Navier-Stokes equations) of blood flow within the coronary arteries are solved using the finite volume method, based on patient-specific boundary conditions. Boundary conditions refer to the physiological conditions existing at the boundaries of the simulated model. In this invention, the inlet conditions are set based on the patient's mean brachial artery blood pressure. The outlet boundary conditions are specified by a user-defined coronary microvascular system model function (parameterized with patient-specific information). No-slip boundary conditions are set on the walls.

[0053] Following CFD simulation 22, a post-processing step is performed. In post-processing step 24, the non-invasive FFR is calculated using the pressure distribution along the coronary plexus. B value.

[0054] Figure 1 Method 10 shown can be used to assist in patient selection during stent implantation. When implemented, this method eliminates the need for stent implantation simulation step 18 and subsequent FFR. B Calculate, and when any segment's FFR B A value ≤0.80 indicates the presence of ischemia, and coronary revascularization may be considered. FFR B A value >0.80 is considered indicative of the absence of ischemia, and the patient is treated with medication. These thresholds are based on existing medical knowledge regarding FFR applications, but these recommended values ​​may be adjusted as future clinical experience accumulates. For patients considering coronary revascularization, the outcome of a virtual stent implantation procedure can be estimated using method 10, which includes a stent implantation simulation step 18. When calculating the outcome of the virtual stent implantation procedure, modifications to the three-dimensional reconstruction model of the coronary arteries are required to insert one or more stents at the lesion site.

[0055] Figure 2 The diagram illustrates a medical image processing system according to an embodiment of the present invention. The medical image processing system 100 is a computer system with a memory storing computer program modules that implement the medical image processing method according to an embodiment of the present invention.

[0056] The medical image processing system 100 includes a processor 110, working memory 112, input interface 114, user interface 116, output interface 118, program memory 120, and data memory 140. The processor 110 may be implemented as one or more central processing unit (CPU) chips. The program memory 120 is a non-volatile storage device such as a hard disk drive that stores computer program modules. The computer program modules are loaded into the working memory 112 for execution by the processor 110. The input interface 114 is an interface that allows data such as patient computed tomography coronary angiography (CTCA) image data to be received by the medical image processing system 100. The input interface 114 may be a wireless network interface such as a Wi-Fi interface or a Bluetooth interface. Alternatively, it may be a wired interface. The user interface 116 allows the user of the medical image processing system 100 to input selection results and commands, and may be implemented as a graphical user interface. The output interface 118 outputs data and may be implemented as a display or data interface.

[0057] Program memory 120 stores segmentation module 122, 3D model reconstruction module 124, scaffold simulation module 126, and fluid dynamics simulation module 128. The computer program modules cause processor 110 to perform various medical image processing tasks, which are described in more detail below. In some contexts, program memory 120 may be referred to as a computer-readable storage medium and / or a non-transitory computer-readable medium. Figure 2 As shown, the computer program modules are different modules that respectively execute corresponding functions implemented by the medical image processing system 100. It should be understood that the boundaries between such modules are merely illustrative; in other embodiments, modules may be combined, or module functions may be divided in other ways. For example, the modules described herein may be divided into sub-modules to be executed by multiple computer processes, or optionally by multiple computers. Furthermore, in alternative embodiments, multiple instances of a module or sub-module may be combined with each other. It should also be understood that although this document describes software implementations of computer program modules, these software implementations may also be implemented as one or more hardware modules (such as field-programmable gate arrays or application-specific integrated circuits) that include circuit systems for implementing functions equivalent to those implemented in the software.

[0058] Figure 3 This is a flowchart of a medical image processing method for simulating a stent according to an embodiment of the present invention. Figure 3 The method 300 shown is by Figure 2 The medical image processing system 100 shown is running.

[0059] Method 300 processes medical image data, such as computed tomography coronary angiography (CTCA) image data of patients. Figure 4AThe image shown is a set of CTCA images of the patient (410). Figure 4A As shown, CTCA image 410 shows a slice passing through the patient's heart and surrounding area.

[0060] In step 302, the processor 110 of the medical image processing system 100 executes the segmentation module 122 to segment the patient's coronary artery structure from the patient's medical image data.

[0061] Figure 4B The image shows the result obtained from... Figure 4A The CTCA image 410 shown is a multiplanar reformatted image 420. Multiplanar reformatted image 420 shows the patient's coronary arteries 422.

[0062] Figure 4C This shows the effect of depicting the contour of the lumen from both longitudinal and transverse perspectives. The longitudinal view 432 shows the segmentation result of the lumen 432 along its length, and the transverse view 435 shows the segmentation result of the lumen 435 along its cross-section.

[0063] The coronary artery lumen structure is segmented using CTCA image processing. Points at the aortic coronary sinus are designated as the origins of the left and right coronary plexuses. Subsequently, the vessel centerline is obtained using a Hessian filter and vessel detection, or an artificial intelligence (AI) method, or a combination of both. Based on this centerline, an extended multiplanar reformatted (MPR) volume is created for each segment of interest. Four longitudinal sections are then extracted from the MPR volume at 45° intervals. The luminal boundaries in these four longitudinal images are then detected using model-guided minimum cost (MCA). Similarly, MCA with a circular lumen model or other lumen models can be used to detect the luminal boundary contours in the transverse slices of the MPR volume. Alternatively, AI methods, or a combination of AI methods, can be used to generate the aforementioned contours.

[0064] As mentioned above, artificial intelligence (AI) can be used as an alternative to Hesse filtering and vessel detection, and it can also be used as an alternative to, or in combination with, the MCA method used in contour drawing. AI methods for centerline extraction can include convolutional neural networks as described below: Wolterink JM, van Hamersvelt RW, Viergever MA, Leiner T, I, Extraction of coronary artery centerline from cardiac CT angiography using a CNN-based orientation classifier, Medical Image Analysis, 2019, Vol. 51: 46-60; and Ziqing Wan, Weimin Huang, Su Huang, Zhongkang Lu, Liang Zhong, Zhiping Lin, Extraction of coronary arteries from CT coronary angiography by augmentation of partially labeled data, 2021 IEEE International Conference on Medical and Biological Engineering.

[0065] Artificial intelligence methods for coronary artery lumen segmentation may include: deep learning frameworks using 3D U-Net: Weimin Huang, Lu Huang, Zhiping Lin, Su Huang, Yanling Chi, Jiayin Zhou, Junmei Zhang, Ru-San Tan, Liang Zhong, Coronary artery segmentation from computed tomography coronary angiography images using deep learning neural networks, 2018 IEEE International Conference on Medical and Biological Engineering: pp. 608-611; deep learning codecs such as Residual U-Net: Cheng Zhu, Xiaoyan Wang et al., Cascaded Residual U-Net for fully automated 3D carotid artery segmentation of high-resolution multi-contrast MR images, Medical Physics and Biophysics, 2021, Vol. 66, No. 4: 045003; or other methods and / or combinations thereof.

[0066] Refer again Figure 3 In step 304, the processor 110 of the medical image processing system 100 executes the 3D model reconstruction module 124 to reconstruct a 3D model of the patient's coronary artery plexus. This involves stitching together the segmented coronary artery lumen contours to generate a patient-specific 3D reconstructed coronary artery model. First, a surface mesh model of each coronary artery and its branches is constructed, represented in a physical space defined by computed tomography at the origin. The mesh is then mapped back to the image space defined at (0, 0, 0) or the first voxel of the image volume. Image voxels can then be created within the mesh using any voxelization method. Subsequently, images containing each coronary artery and its branches are added (with a resolution of 1 mm for branches, which is the current limit of resolution for branches, but this limit can be increased with advancements in imaging technology) to form a 3D image containing all coronary arteries. Using this image, new meshes can be created using meshing operations such as moving cubes, forming a surface mesh containing all coronary arteries. This process simplifies the mesh creation work that requires handling the intersections between different meshes at the mesh level for the entire coronary plexus.

[0067] Figure 4D The image shows a three-dimensional reconstruction model of the coronary arteries. Figure 4D As shown, the three-dimensional reconstruction model 440 of the coronary artery is the three-dimensional shape of the patient's coronary artery plexus.

[0068] In step 306, the processor 110 of the medical image processing system 100 executes the stent simulation module 124 to simulate stent implantation by replacing one or more portions of the three-dimensional coronary plexus with candidate stent structures. Key parameters for stent placement include the initiation and termination points of one or more stents, as well as each corresponding diameter and length.

[0069] In some implementations, step 306 includes: determining the location and size of a candidate stent structure; and subsequently simulating stent implantation by replacing one or more portions of a three-dimensional coronary plexus with one or more candidate stent structures.

[0070] In other embodiments, step 306 includes: receiving user input indicating one or more candidate stent structures and the location of the candidate stent structures; and subsequently simulating stent implantation by replacing one or more portions of the three-dimensional coronary plexus with one or more candidate stent structures.

[0071] Figures 5A to 5E The illustration shows an example of selecting stent parameters in a method according to an embodiment of the present invention.

[0072] Figure 5A This is a longitudinal view of the coronary artery lumen outline. From Figure 5A As can be seen, the lumen profile 510 includes a narrowing region 512 where the artery narrows.

[0073] Figure 5B This is a transverse view of the outline of the coronary artery lumen. (Example) Figure 5B As shown, the lumen profile illustrates the arterial cross-section at a point along the arterial length. Based on this, the lumen cross-sectional area along the lumen length can be determined.

[0074] Figure 5C For along Figure 5A The image shows the lumen area of ​​a narrowed coronary artery. The lengths in the image represent the straightened length of the artery from the proximal (left) to the distal (right) end, obtained by three-dimensional tortuous multiplanar reformatting of the coronary artery along an automatic centerline. The curves were smoothed using mid-pass or other low-pass filters to minimize noise. Figure 5CAs shown, the degree of coronary artery stenosis reaches its maximum at point "S," which corresponds to the minimum luminal area. The proximal endpoint "P" corresponds to the proximal end of the stenosis, and the distal endpoint "D" corresponds to the distal end. Points "P" and "D" can be identified by finding the maximum absolute value of the change in curvature or slope on both sides of point "S."

[0075] Figure 5D for Figure 5C The curvature plot of the curve shown is given. Curvature is defined as the reciprocal of the radius of the osculating circle and is calculated using the independent coordinate method for plane curves described below: Lewiner T, Gomes J, Lopes H, Craizer M, Tools for estimating bending and twisting based on parametric curve fitting, Computer and Graphics, 2005, Vol. 29: 641–655.

[0076] like Figure 5D As shown, points "P" and "D" can be identified as the points with the largest absolute values ​​of curvature on the left and right sides of point "S".

[0077] Figure 5E for Figure 5C The graph shows the slope variation of the curve. The slope variation is calculated as the difference in slope between two adjacent points on the curve.

[0078] like Figure 5E As shown, points "P" and "D" can be identified as the points with the largest absolute values ​​of slope changes on the left and right sides of point "S".

[0079] Whether using the maximum absolute value of curvature or the maximum absolute value of slope change, points "P" and "D" can be automatically determined by traversing left and right from "S".

[0080] The distance from point "P" to point "D" represents the minimum length of the narrow segment that needs to be replaced by a virtual support. User-defined extension margins (such as 2-3mm) can be added to both ends of "P" and "D" to ensure a sufficiently large anchoring area and to provide a degree of freedom when selecting the length of a commercially available support within a certain range.

[0081] from Figure 5C , Figure 5D and Figure 5E As can be seen, the straightening lengths corresponding to points "P" and "D" are 28.5mm and 39.2mm, respectively. Considering the anchoring area, the virtual support covers the region with straightening lengths from 27.6mm to 42.6mm, corresponding to 3.2mm (8.04mm) respectively. 2 ) and 2.8mm (6.16mm) 2 The average diameter (cross-sectional area) of the proximal and distal ends of the ).

[0082] In the 3D model, one can search for values ​​corresponding to 8.04mm. 2 and 6.16mm 2 The planes corresponding to these two areas are used to determine the starting and ending planes of the support segment (see...). Figure 6A Furthermore, the distance between the proximal and distal ends is 15 mm, and the average lumen diameter between these two points is 3.0 mm.

[0083] Existing commercially available stents have nominal diameters and recommended balloon dilation limits due to their perforated mesh design. The diameter of the chosen commercially available stent should span the average diameter at both proximal and distal locations. Many coronary stents employ perforated mesh designs that allow expansion beyond the nominal diameter. Depending on the stent type, the recommended maximum stent dilation limit may exceed the nominal stent diameter by 50%.

[0084] The following, combined with Figures 6A to 6D This describes a simulation of how the support is placed.

[0085] Figure 6A The image shows a patient-specific three-dimensional coronary artery model with a narrowed region. Figure 6A As shown, the coronary artery model 600 has a stenotic region 610 of vascular lesion in the left anterior descending artery.

[0086] Figure 6B The image shows the generation of a simulated or virtual scaffold for use in replacing narrow areas. For example... Figure 6B As shown, first identify the proximal position P and distal position D of the narrow region. These two points can be determined based on the above combination. Figures 5A to 5E The process involves identification. For example, the cross-sectional area is first examined along the centerline of the blood vessel to determine the relationship between the cross-sectional area and the cross-sectional area. Figure 5C Planes “P” and “D” with similar cross-sectional areas are shown. Then, the portion 620 between the proximal position P and the distal position D is removed from the model. A series of circular surfaces 622 perpendicular to the original lumen centerline 624 are generated. The distance between these circular surfaces 622 is 3 mm, and the diameter is 3 mm (the average diameter at the proximal and distal ends).

[0087] Figure 6C The image shows a simulated or virtual stent in a three-dimensional coronary artery model. (Example:) Figure 6C As shown, the circular surface 622 is interpolated using the B-spline interpolation method to form a virtual support segment 630.

[0088] Figure 6D The image shows a new model containing a virtual support segment. (See image for details.) Figure 6D As shown, the new model 650 has a portion 652 formed according to the above-described simulation or simulated support. The remaining portion of the new model 650 is... Figure 6A The patient's coronary artery model shown corresponds to 600.

[0089] Refer again Figure 3 In step 308, the processor 110 of the medical image processing system 100 executes the hydrodynamic simulation module 128 to simulate blood flow through the coronary plexus containing a virtual stent segment and determine the non-invasive FFR. B .

[0090] In the fluid dynamics simulation, the continuity and momentum conservation equations (also known as the Navier-Stokes equations) of blood flow in the coronary arteries are solved using the finite volume method, based on patient-specific boundary conditions, for the computational domain. Boundary conditions refer to the physiological conditions existing at the boundaries of the simulated model. In the method of this invention, the inlet conditions are set based on the patient's mean brachial artery blood pressure. The outlet boundary conditions are specified by a user-defined coronary microvascular system model function (parameterized with patient-specific information). No-slip boundary conditions are set on the walls.

[0091] Using the same technology, virtual stents with different corresponding stent diameters and lengths can be "placed" at multiple locations within a 3D coronary artery model. Subsequently, operators can interactively evaluate the comprehensive anatomical reconstruction effects of different stent implantation strategies, which may differ in stent number, location, length, and diameter. Once the model with placed stents is ready, the non-invasive FFR after virtual stent implantation can be calculated. B In this way, operators can comprehensively evaluate the physiological effects of the planned revascularization strategies.

[0092] As described above, during the execution of the above method, the proximal and distal endpoints, stent length, and stent diameter of each coronary artery lesion can be determined. User-defined margins (e.g., 2-3 mm, to ensure a sufficiently large stent anchorage area) at both ends of the virtual stent and user-defined stent diameter expansion margins (to simulate stent expansion beyond the nominal diameter in clinical practice) can be manually input by the user to generate one or more user-defined virtual stents. Furthermore, CFD modeling can be used to generate pressure maps and FFRs for one or more virtual stents for each coronary artery lesion. B picture.

[0093] In one embodiment of the clinical application of the above-described method proposed by the present invention, the user can select a commercially available stent that best matches one or more virtual stents with the desired coronary hemodynamic characteristics. Taking into account the length and diameter margins of the virtual stents defined by the user, commercially available stents that are close to but shorter and narrower than the selected virtual stents are first identified. Subsequently, the length and diameter of each selected commercially available stent can be input into the model to simulate the implantation of the commercially available stent and automatically generate a CFD model and FFR. BFigure 1. The stent can be automatically placed by centering it at the center of the lesion. Users can choose to manually adjust the placement of the virtual stent, while the CFD model and FFR... B The diagram is automatically generated. In addition, the length, nominal diameter, diameter expansion limit, and corresponding balloon expansion technical parameters of the commercially available stent selected by the user can be uploaded and updated to a database maintained by the business and / or user site of this invention.

[0094] Figures 7A to 7D The figures shown are invasive FFR and non-invasive FFR. B Narrow functional assessment comparison. This example demonstrates FFR. B Its application in the diagnosis of coronary artery disease before percutaneous coronary intervention (PCI) and in the prediction of PCI outcomes. For example... Figure 7A As shown, the patient has a 710 mm stenosis in the left anterior descending coronary artery. The invasive FFR measurement was 0.63.

[0095] Figure 7B The image shows the left anterior descending coronary artery after stent placement. The invasive FFR value measured after stent implantation was 0.81.

[0096] Figure 7C The image shows a coronary plexus model of a patient before stent implantation. Significant functional stenosis was diagnosed in the left anterior descending coronary artery, with FFR (free flow rate). B It was 0.59 (<0.80), which is close to the invasive FFR measurement of 0.63.

[0097] Figure 7D The image shows the predicted stent implantation surgery results according to the method of an embodiment of the present invention. Post-PCI FFR B The calculated value was 0.83, which is close to the invasive FFR measurement of 0.81.

[0098] Figure 8 The illustration shows an example of how an embodiment of the present invention is applied to compare different stent implantation strategies for tandem lesions of the left anterior descending artery.

[0099] like Figure 8 As shown, image 810 illustrates the first lesion 812 and the second lesion 814 in the left anterior descending artery. The invasive FFR measurement was 0.63.

[0100] For the patient, a three-dimensional arterial plexus model 820 is generated. This model includes a first lesion 812 and a second lesion 814. Based on this model, non-invasive FFR is performed. B The calculated value is 0.64.

[0101] Using the above method, three different stent implantation strategies were simulated. The first stent implantation strategy 830 includes: applying a first stent 822 to the region of the first lesion 812, and applying a second stent 824 to the region of the second lesion 814. For example... Figure 8 As shown, the predicted non-invasive FFR of the first stent implantation strategy 830 B The value is 0.88. The second stent implantation strategy 840 includes: applying the first stent 822 to the region of the first lesion 812, but not applying a stent to the region of the second lesion 814. For example... Figure 8 As shown, the predicted non-invasive FFR of the second stent implantation strategy 840 B The value is 0.87. The third stent implantation strategy 850 includes: applying a second stent 824 to the region of the second lesion 814, but not applying a stent to the region of the first lesion 812. For example... Figure 8 As shown, the predicted non-invasive FFR of the third stent implantation strategy 850 B It is 0.69.

[0102] Based on the above combination Figure 8 Based on the simulation results of the aforementioned stent implantation strategies, clinicians can choose the second strategy because it significantly improves the prediction of non-invasive FFR. B However, adding an additional stent to the second lesion 814 region only achieved a minor improvement.

[0103] To validate the above method, a proof-of-concept study was also conducted. This study recruited 18 patients with suspected or known coronary artery disease. These patients first underwent CTCA, followed by invasive coronary angiography and FFR measurement, and then PCI to treat hemodynamically significant lesions. A total of 21 vessels (22 lesions) underwent stent implantation, and FFR measurements were performed before and after implantation.

[0104] Figure 9A The image shows the FFR and FFR before stent implantation. B The correlation between them Figure 9B The image shown is the corresponding Bland-Altman diagram.

[0105] Figure 10A The image shows the FFR and FFR after stent implantation. B The correlation between them Figure 10B The image shown is the corresponding Bland-Altman diagram.

[0106] FFR B The calculation results showed that before stent implantation, it had an excellent correlation with FFR (R = 0.88, p < 0.001), while after stent implantation, it had a good correlation (R = 0.55, p < 0.001). Figure 9A and Figure 10A ). FFR B The mean difference between the FFR and the mean value before stent implantation was 0.038 (95% concordance limit: -0.09 to 0.17). Figure 9B The value after stent implantation was -0.036 (95% concordance limit: -0.22 to 0.15). Figure 10B Used to calculate residual ischemia after stent implantation (defined as FFR after stent implantation). B FFR ≤0.80 B The diagnostic accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were 81%, 80%, 81%, 57%, and 93%, respectively.

[0107] In summary, FFR has significant hemodynamic implications for diagnosing coronary artery stenosis and for calculating the hemodynamic outcomes of stent implantation procedures. B It is an indicator with good prospects.

[0108] Coronary artery disease (CAD) causes myocardial ischemia and accounts for 13% of global deaths (projected to rise to 15% by 2030 according to the World Health Organization's 2011 forecast). Percutaneous coronary intervention (PCI) with stent implantation is an effective revascularization therapy to reduce ischemia in CAD. Decisions regarding stent selection (e.g., size / type / number), implantation location, and implantation strategy are particularly challenging for complex bifurcation and tandem lesions prone to in-stent restenosis, late thrombosis, and related adverse clinical events. Therefore, tools for predicting the functional status of coronary artery disease before and after PCI can greatly assist in management decisions and surgical planning.

Claims

1. A method for simulating the structure of a stent within a coronary artery lumen, characterized in that, The method includes: Based on the segmented coronary artery lumen contour, a three-dimensional coronary plexus is reconstructed; Identifying the first candidate location of the first candidate stent structure includes: Determine the average lumen area as the vessel length changes with the straightened length; The maximum point of the lesion at the location of minimum average luminal area is identified, and the proximal and distal ends of the lesion on each side of the maximum point are identified by the curvature of the average luminal area as a function of the vessel straightening length; and The first candidate location is determined based on the proximal end point and the distal end point of the lesion; The first portion of the three-dimensional coronary plexus is replaced by the first candidate stent structure; and The pressure distribution of the coronary plexus is simulated to determine the fractional noninvasive blood flow reserve through the first candidate stent structure.

2. A method for simulating a stent within the lumen of a coronary artery, characterized in that, The method includes: Based on the segmented coronary artery lumen contour, a three-dimensional coronary plexus is reconstructed; Identifying the first candidate location of the first candidate stent structure includes: Determine the average lumen area as the vessel length changes with the straightened length; The maximum point of the lesion at the location of minimum average lumen area is identified, and the proximal and distal ends of the lesion on each side of the maximum point are identified by the slope change of the average lumen area as the vessel length changes with the straightening length of the vessel; and The first candidate location is determined based on the proximal end point and the distal end point of the lesion; The first portion of the three-dimensional coronary plexus is replaced by the first candidate stent structure; and The pressure distribution of the coronary plexus is simulated to determine the fractional noninvasive blood flow reserve through the first candidate stent structure.

3. The method according to claim 1, characterized in that, Identifying the proximal and distal ends of the lesion based on the curvature of the average lumen area as the vessel length changes with the straightened length includes: identifying the proximal and distal ends of the lesion as the maximum value of the absolute value of the curvature.

4. The method according to claim 2, characterized in that, Identifying the proximal and distal ends of the lesion based on the slope change of the average lumen area as the vessel length changes includes: identifying the proximal and distal ends of the lesion as the maximum values ​​of the slope change.

5. The method according to claim 1 or 2, characterized in that, Also includes: The diameter of the first candidate stent structure is determined based on the average lumen diameter at the proximal end of the lesion and the distal end of the lesion.

6. The method according to claim 1 or 2, characterized in that, Also includes: The length of the first candidate stent structure is determined based on the location of the proximal end of the lesion and the location of the distal end of the lesion.

7. The method according to claim 6, characterized in that, Also includes: An extension allowance is added to the length of the candidate scaffold structure.

8. The method according to claim 1 or 2, characterized in that, Also includes: The coronary artery lumen structure is segmented from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen contour.

9. The method according to claim 8, characterized in that, The coronary artery lumen structure is segmented from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen contour, including: The point at the aortic sinus is designated as the origin of the coronary plexus; Determine the centerline of the arteries in the arterial plexus; Using the centerline, create extended multiplanar reformatted volumes for each segment of the arterial plexus; Extract the longitudinal section from the extended multiplane reformatted volume; The lumen boundary of the extracted longitudinal section was detected; and Using the detected lumen boundaries, the lumen boundary contours of each slice of the multi-planar reformatted volume are detected.

10. The method according to claim 9, characterized in that, The longitudinal sections are extracted from the extended multiplane reformatted volume at 45-degree intervals.

11. The method according to claim 9 or 10, characterized in that, The centerline of the blood vessel was determined using a Hesse filter.

12. A computer-readable carrier medium carrying processor-executable instructions, characterized in that, When the processor-executable instructions are executed by the processor, the processor causes the processor to perform the method according to claim 1 or 2.

13. A medical image processing system for simulating the structure of a coronary artery stent, the medical image processing system comprising a processor and a data storage device storing executable computer program instructions, characterized in that, The computer program instructions are used to cause the processor to: Reconstruct the three-dimensional coronary plexus based on the segmented coronary artery lumen contour; Identifying the first candidate location of the first candidate stent structure includes: Determine the average lumen area as the vessel length changes with the straightened length; The maximum point of the lesion at the location of minimum average luminal area is identified, and the proximal and distal ends of the lesion on each side of the maximum point are identified by the curvature of the average luminal area as a function of the vessel straightening length; and The first candidate location is determined based on the proximal end point and the distal end point of the lesion; The first portion of the three-dimensional coronary plexus is replaced by the first candidate stent structure; and The pressure distribution of the coronary plexus is simulated to determine the fractional noninvasive blood flow reserve through the first candidate stent structure.

14. A medical image processing system for simulating the structure of a coronary artery stent, the medical image processing system comprising a processor and a data storage device storing executable computer program instructions, characterized in that, The computer program instructions are used to cause the processor to: Reconstruct the three-dimensional coronary plexus based on the segmented coronary artery lumen contour; Identifying the first candidate location of the first candidate stent structure includes: Determine the average lumen area as the vessel length changes with the straightened length; The maximum point of the lesion at the location of minimum average lumen area is identified, and the proximal and distal ends of the lesion on each side of the maximum point are identified by the slope change of the average lumen area as the vessel length changes with the straightening length of the vessel; and The first candidate location is determined based on the proximal end point and the distal end point of the lesion; The first portion of the three-dimensional coronary plexus is replaced by the first candidate stent structure; and The pressure distribution of the coronary plexus is simulated to determine the fractional noninvasive blood flow reserve through the first candidate stent structure.

15. The medical image processing system according to claim 13, characterized in that, Identifying the proximal and distal ends of the lesion based on the curvature of the average lumen area as the vessel length changes with the straightened length includes: identifying the proximal and distal ends of the lesion as the maximum value of the absolute value of the curvature.

16. The medical image processing system according to claim 14, characterized in that, Identifying the proximal and distal ends of the lesion based on the slope change of the average lumen area as the vessel length changes includes: identifying the proximal and distal ends of the lesion as the maximum values ​​of the slope change.

17. The medical image processing system according to claim 13 or 14, characterized in that, The data storage device also stores executable computer program instructions for: determining the diameter of the first candidate stent structure based on the average lumen diameter at the proximal end of the lesion and the distal end of the lesion.

18. The medical image processing system according to claim 13 or 14, characterized in that, The data storage device also stores executable computer program instructions for: determining the length of the first candidate stent structure based on the location of the proximal end of the lesion and the location of the distal end of the lesion.

19. The medical image processing system according to claim 18, characterized in that, The data storage device also stores executable computer program instructions for: adding an extension allowance to the length of the first candidate support structure.

20. The medical image processing system according to claim 13 or 14, characterized in that, The data storage device also stores executable computer program instructions for: segmenting the coronary artery lumen structure from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen outline.

21. The medical image processing system according to claim 20, characterized in that, The data storage device also stores executable computer program instructions for segmenting the coronary artery lumen structure from a set of computed tomography coronary angiography images to obtain the segmented coronary artery lumen contour: The point at the aortic sinus is designated as the origin of the coronary plexus; Determine the centerline of the arteries in the arterial plexus; Using the centerline, create extended multiplanar reformatted volumes for each segment of the arterial plexus; Extract the longitudinal section from the extended multiplane reformatted volume; Detect the cavity boundary of the extracted longitudinal section; as well as Using the detected lumen boundaries, the lumen boundary contours of each slice of the multi-planar reformatted volume are detected.

22. The medical image processing system according to claim 21, characterized in that, The longitudinal sections are extracted from the extended multiplane reformatted volume at 45-degree intervals.

23. The medical image processing system according to claim 21 or 22, characterized in that, The centerline of the blood vessel was determined using a Hesse filter.

24. The method according to claim 1 or 2, characterized in that, Also includes: Identify the second candidate location of the second candidate scaffold structure; The second portion of the three-dimensional coronary plexus is replaced with the second candidate stent structure; Simulate the pressure distribution of the coronary plexus to determine the fractional noninvasive blood flow reserve through the second candidate stent structure; as well as Display: Indication of the first candidate stent structure; Indication of the second candidate stent structure; Indication of the non-invasive fractional flow reserve of the first candidate stent structure; and Indication of the non-invasive fractional flow reserve of the second candidate stent structure.

25. The medical image processing system according to claim 13 or 14, characterized in that, The computer program instructions are further configured to cause the processor to: Identify the second candidate location of the second candidate scaffold structure; The second portion of the three-dimensional coronary plexus is replaced with the second candidate stent structure; Simulate the pressure distribution of the coronary plexus to determine the fractional noninvasive blood flow reserve through the second candidate stent structure; as well as Display: Indication of the first candidate stent structure; Indication of the second candidate stent structure; Indication of the non-invasive fractional flow reserve of the first candidate stent structure; and Indication of the non-invasive fractional flow reserve of the second candidate stent structure.

Images