Selection of collateral vessels in retrograde chronic total occlusion procedures

Abstract

A system for selecting a collateral vessel to perform a retrograde chronic total occlusion (CTO) treatment procedure is provided. The system includes one or more processors configured to: receive angiographic image data (120) representative of an occlusion in a target blood vessel; analyzing the angiography image data (120) to identify one or more candidate collateral vessels (1501... i) for approaching the occlusion in the target vessel in a retrograde direction; and assigning a suitability score to each candidate collateral vessel (1501... i), the suitability score representing a suitability of the vessel for performing the retrograde CTO treatment procedure on the occlusion.

Description

Technical Field

[0001] This disclosure generally relates to the procedure for performing retrograde chronic total occlusion (CTO). More specifically, the invention relates to selecting collateral vessels to perform the retrograde CTO procedure. A system, a computer-implemented method, and a computer program product are disclosed. Background Technology

[0002] Over time, blood vessels can become blocked (or, in other words, occluded). For example, a blood vessel may become blocked due to plaque buildup or due to clots in the vessel. Occlusion can occur in blood vessels in various parts of the anatomy (including in the heart, in the brain, and in peripheral areas such as the legs).

[0003] In some cases, the blockage is partial, and blood flow in the vessel is restricted, but not completely blocked. For example, in the heart, partial occlusion of a coronary artery can cause the heart to have to work harder to maintain blood flow. However, in more serious cases, blood flow in the vessel may be completely blocked. Chronic total occlusion, or "CTO," is defined as a vessel being completely blocked for a duration of 3 months or more.

[0004] If a blood vessel becomes blocked, a procedure may be required. An example of such a procedure is balloon angioplasty. A stent can be placed after balloon angioplasty. Balloon angioplasty opens the lumen of the blood vessel, and the stent holds the lumen open. This restores blood flow through the vessel. Various interventional devices, including guidewires and (balloon) catheters, can be used for such procedures.

[0005] The initial step in such a procedure is to pass the interventional device (e.g., a guidewire) across the occlusion. This is called “crossing the occlusion.” This step is typically performed by crossing the occlusion in the anterograde direction (i.e., by crossing the occlusion in the direction of normal blood flow). If anterograde crossing of the occlusion fails, retrograde crossing of the occlusion (i.e., crossing the occlusion in the opposite direction of normal blood flow) is often attempted as the next step. Retrograde approach has been reported in 20% to 50% of peripheral coronary interventions (PCIs) performed on CTO worldwide, as Megaly, M. et al. noted in their article “Retrograde Approach to Chronic Total Occlusion Percutaneous Coronary Intervention” (Circ. Cardiovasc. Interv., 2020, 13:e008900). The article also states that the success rate of retrograde CTO PCI is 90%, compared to approximately 70% for anterograde approach.

[0006] However, retrograde approach to occlusion carries a higher risk of complications compared to antegrade approach. Therefore, it is important to select the optimal approach for a given object. Here, the physician attempts to select the optimal collateral vessel (e.g., bypass graft, diaphragmatic collateral, or epicardial collateral, in cases of occlusion within a coronary artery) for retrograde approach to the occlusion. However, collateral vessels are often difficult to detect in angiographic images because the contrast agent used to generate such images tends not to accumulate near the CTO. The selection of the optimal collateral vessel is also influenced by factors such as vessel tortuosity, the risk of tamponade, the difficulty of guidewire placement in the retrograde direction, and the ability to dilate the collateral vessel.

[0007] Therefore, there is a need to improve the selection of collateral vessels used in retrograde CTO procedures. Summary of the Invention

[0008] According to a first aspect of this disclosure, a system is provided for selecting collateral vessels to perform a retrograde chronic total occlusion (CTO) treatment procedure. The system includes one or more processors configured to: Receive angiographic image data representing occlusion in the target blood vessel; Analyze the angiographic image data to identify one or more candidate collateral vessels for approaching the occlusion in the target vessel in a retrograde direction; and Each candidate collateral vessel is assigned a suitability score, which indicates the vessel's suitability for performing the retrograde CTO procedure on the occlusion.

[0009] In this system, the identification of candidate collateral vessels and the assignment of suitability scores to them facilitate the physician's selection of the optimal vessel for performing the procedure. This helps physicians overcome the challenge of mentally weighting the many factors that influence their selection.

[0010] In related aspects, the one or more processors are also configured to: Receive second angiographic image data representing the interventional device in the selected collateral vessels; The second angiographic image data is analyzed to provide an estimate of the occlusion time period, which represents the total time the interventional device has obstructed the lumen of the selected collateral vessel; and A warning is output in response to the estimated blocking time period exceeding a threshold time period; or The output indicates the amount of remaining time available to execute the process, which is calculated based on the difference between the threshold time period and the blocking time period.

[0011] As stated above, retrograde approach occlusion carries a higher risk of complications compared to antegrade approach. These complications include, for example, the risk of myocardial infarction and the risk of thrombosis in collateral vessels. Both risks increase when the collateral vessel is blocked for too long a period of time. In this regard, the system analyzes second angiographic image data to estimate the total time the interventional device has blocked the lumen of the selected collateral vessel. If the estimated total time exceeds a threshold time period, a warning is output. Alternatively, an indication of the amount of remaining time available to perform the procedure is output. The warning, along with the similar indication of the amount of remaining time, alerts the physician to the increased risk of complications during the retrograde CTO procedure.

[0012] According to a second aspect of this disclosure, a system for determining a threshold time period is provided. The threshold time period represents the maximum total time during which an interventional device can safely occlude the lumen of a collateral vessel in a retrograde chronic total occlusion (CTO) procedure. In this respect, the system includes one or more processors configured to: Receive angiography image data, the angiography image data representing an obstruction in a target blood vessel and one or more collateral vessels for approaching the obstruction in the target blood vessel in a retrograde direction; The angiography image data is input into a neural network, which is trained to predict the threshold time period for each of the one or more collateral vessels; and In response to the input, the threshold time period for each of the one or more collateral vessels is output; and The neural network is trained to predict the threshold time period for each of the one or more collateral vessels based on the input angiography image data.

[0013] According to a third aspect of this disclosure, an alternative system for determining a threshold time period is provided. The threshold time period represents the maximum total time during which an interventional device can safely occlude the lumen of a collateral vessel in a retrograde chronic total occlusion (CTO) procedure. In this regard, the system includes one or more processors configured to: Receive angiography image data, the angiography image data representing an obstruction in a target blood vessel and one or more collateral vessels for approaching the obstruction in the target blood vessel in a retrograde direction; Extract one or more vascular parameters from each of the one or more collateral vessels in the angiography image data; One or more extracted vascular parameters are input into a neural network, which is trained to predict the threshold time period for each of the one or more collateral vessels; and In response to the input, the threshold time period for each of the one or more collateral vessels is output; and The neural network is trained to predict the threshold time period for each of the one or more collateral vessels based on one or more input vascular parameters.

[0014] The latter two aspects of the system provide a threshold time period for each of one or more collateral vessels. This threshold time period represents the maximum total time the interventional device can safely occlude the lumen of a collateral vessel during a retrograde chronic total occlusion (CTO) procedure. This information is useful for physicians in weighting their selection of collateral vessels for performing a retrograde CTO procedure. For example, if the system indicates that a particular collateral vessel has a relatively short threshold time period compared to other collateral vessels, the physician can decide to select a different vessel with a relatively long threshold time period to safely complete the procedure within its expected duration. If none of the collateral vessels has a threshold time period long enough to accommodate the expected duration of the procedure, the physician may even decide to pursue an antegrade approach to the CTO.

[0015] Other aspects, features, and advantages of this disclosure will become apparent from the following description of examples illustrated with reference to the accompanying drawings. Attached Figure Description

[0016] Figure 1This is an illustration of a heart according to some aspects of this disclosure, the illustration including an example of an occlusion 130 in a target vessel 140 and a path P for approaching the occlusion 130 in the anterograde direction. a .

[0017] Figure 2 This is an illustration of a heart according to some aspects of this disclosure, the illustration including an example of an occlusion 130 in a target vessel 140 and a path P for approaching the occlusion 130 in a retrograde direction via a collateral vessel 1501. r .

[0018] Figure 3 This is a schematic diagram illustrating an example of a system 100 for selecting collateral vessels to perform a retrograde CTO procedure, according to some aspects of this disclosure.

[0019] Figure 4 This is a flowchart illustrating an example of a computer-implemented method for selecting collateral vessels to perform a retrograde CTO procedure, based on some aspects of this disclosure.

[0020] Figure 5 This is a schematic diagram illustrating an example of a neural network 170 according to some aspects of the present disclosure, the neural network 170 being trained to: identify one or more candidate collateral vessels 1501, 1502 to approach an occlusion in a target vessel in a retrograde direction, and assign a fitness score to each candidate collateral vessel.

[0021] Figure 6 The diagram illustrates some aspects of this disclosure for determining the threshold time period T. ObsMax A schematic diagram of an example of system 300.

[0022] Figure 7a The diagram illustrates the determination of the threshold time period T based on some aspects of this disclosure. ObsMax A flowchart of the first example of a computer-implemented method.

[0023] Figure 7b The diagram illustrates the determination of the threshold time period T based on some aspects of this disclosure. ObsMax A flowchart of a second example of a computer-implemented method.

[0024] Figure 8a This is a schematic diagram illustrating an example of a neural network 210 according to some aspects of the present disclosure, which is trained to predict a threshold time period T for a selected collateral vessel based on extracted vascular parameters 220. ObsMax .

[0025] Figure 8bThe illustration is based on an analysis of some aspects of this disclosure of second angiography image data 390 to provide information on the occlusion time period T. Obs The estimation results and the response to the blocking time period T Obs The time period exceeding the threshold T ObsMax An illustration of an example of generating a warning. Detailed Implementation

[0026] Examples of this disclosure are provided with reference to the following description and accompanying drawings. In this specification, for purposes of explanation, numerous specific details of certain examples are set forth. References to “example,” “implementation,” or similar language in the specification mean that a feature, structure, or characteristic described in connection with the example is included in at least that example. It should also be understood that a feature described with respect to one example may also be used in another example, and for the sake of brevity, not all features need to be repeated in each example. For example, features described with respect to a system may be implemented in a corresponding manner in a computer-implemented method and a computer program product.

[0027] In the following description, various systems that can be used to support physicians in selecting collateral vessels for performing retrograde CTO procedures are referenced. In some examples, retrograde CTO procedures performed on occlusions located in coronary arteries are cited. However, it should be noted that the systems disclosed herein can also be used to support physicians in selecting collateral vessels for performing retrograde CTO procedures where the occlusion is located in other parts of the anatomy. For example, the systems disclosed herein can be used to support physicians in selecting collateral vessels for performing retrograde CTO procedures where the occlusion is located in the leg or brain. In the brain, for example, the system disclosed herein can be used to support physicians in selecting collateral vessels within the brain's "collateral circulation," which is a network of blood vessels that maintains blood flow when the main vascular pathway is partially blocked. Furthermore, it should be understood that the system disclosed herein can be used to support physicians in selecting collateral vessels to perform retrograde CTO procedures, where the occlusion is located in a vessel type other than an artery. For example, the system disclosed herein can be used to support physicians in selecting collateral vessels to perform retrograde CTO procedures, where the occlusion is located in a vein. Therefore, the system disclosed herein can be used to support physicians in selecting collateral vessels to perform retrograde CTO procedures, where the occlusion is generally located within a blood vessel in the body.

[0028] Note that the computer-implemented methods disclosed herein can be provided as a non-transient computer-readable storage medium having computer-readable instructions stored thereon, which, when executed by at least one processor, cause the at least one processor to perform the method. In other words, the computer-implemented methods can be implemented in a computer program product. The computer program product can be provided by dedicated hardware or hardware capable of running software in association with appropriate software. When provided by a processor, the functionality of the method features can be provided by a single dedicated processor, or by a single shared processor, or by multiple individual processors (some of which can be shared). The functionality of one or more method features can be provided, for example, by a processor shared within a networked processing architecture (e.g., client / server architecture, peer-to-peer architecture, the Internet, or the cloud).

[0029] The explicit use of the terms "processor" or "controller" should not be construed as specifically referring to hardware capable of running software, but may implicitly include, but is not limited to, digital signal processor (DSP) hardware, read-only memory (ROM) for storing software, random access memory (RAM), non-volatile storage devices, etc. Furthermore, examples of this disclosure may take the form of a computer program product accessible from a computer-usable storage medium or a computer-readable storage medium, which provides program code for use by or in connection with a computer or any instruction execution system. For the purposes of this specification, a computer-usable storage medium or a computer-readable storage medium may be any means capable of including, storing, communicating, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device. The medium may be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system or device or a propagation medium. Examples of computer-readable media include semiconductor or solid-state memory, magnetic tape, removable computer disks, random access memory (RAM), read-only memory (ROM), hard disks, and optical disks. Current examples of optical discs include CD-ROM, CD-R / W, Blu-ray discs, and DVDs.

[0030] It should also be noted that some operations performed by one or more processors of the system disclosed herein can be implemented using artificial intelligence techniques. Suitable techniques may include machine learning techniques and deep learning techniques (e.g., neural networks). For example, one or more neural networks can be trained in a supervised manner or, in some cases, in an unsupervised manner to implement operations performed by one or more processors.

[0031] As mentioned above, there is a need to improve the selection of collateral vessels for use in retrograde CTO procedures.

[0032] To illustrate the challenges of selecting such blood vessels, refer to Figure 1 , Figure 1 This is an illustration of a heart according to some aspects of this disclosure, the illustration including an example of an occlusion 130 in a target vessel 140 and a path P for approaching the occlusion 130 in the anterograde direction. a . Figure 1 The diagram also illustrates the various arteries supplying blood to the heart (including the left coronary artery LCA and the right coronary artery RCA). The LCA extends from its arterial ostium until its branches enter the left circumflex artery (LCX) and the left anterior descending artery (LAD). Figure 1 In the example shown, occupant 130 is located in the RCA.

[0033] As described above, handling blockages (e.g., Figure 1 The initial step in the occlusion (130) shown is to pass an interventional device (e.g., a guidewire) across the occlusion. This is referred to as "crossing the occlusion" and is typically performed by crossing the obstruction in the anterograde direction (i.e., by crossing the occlusion in the normal blood flow direction). This is in Figure 1 Path P a The associated arrow diagram is shown. If anterograde crossing of the occlusion fails, retrograde crossing of the occlusion (i.e., crossing the occlusion in the opposite direction of normal blood flow) is often attempted as the next step. Figure 1 The illustration also shows 150 various collateral vessels that can be used for this purpose. 1··4 Collateral vessels interconnect coronary arteries and provide an alternative source of blood supply to the myocardium in cases of occlusive coronary artery disease. Figure 1 In the example shown, these collateral vessels include diaphragmatic collateral vessels 1501 and 1502, epicardial collateral vessel 1503, and bypass graft vessel 1504.

[0034] Figure 2 This is an illustration of a heart according to some aspects of this disclosure, the illustration including an example of an occlusion 130 in a target vessel 140 and a path P for approaching the occlusion 130 in a retrograde direction via a collateral vessel 1501. r . Figure 2 The heart shown corresponds to Figure 1 The heart is shown. For the sake of brevity, it will not be repeated here. Figure 2 and Figure 1 A description of the common features shown in the diagrams. (and) Figure 1 To create a contrast, Figure 2 The path P shown r Approaching block 130 in the reverse direction, as indicated by its associated arrow. Path P r The 130th occlusion was approached via the diaphragmatic collateral vessel 1501.

[0035] from Figure 2 The example shown is understandable, although the collateral vessels are 150 1··4 Provided for Figure 1 The path P shown is for approaching block 130. a Alternative pathways exist, but such pathways are often difficult to detect in angiographic images because the contrast agent used to generate such images tends not to accumulate near the CTO. Compared to the target vessel 140, collateral vessels typically also have relatively smaller luminal sizes, relatively longer lengths, and relatively higher tortuosity. Therefore, collateral vessels 150 1··4 It is generally considered a second option after anterograde approach to occlusion fails. Compared to anterograde approach, for example via collateral vessels 150... 1··4 Offering retrograde access to the occlusion also carries a higher risk of complications. Therefore, it is important to select the optimal access for a given subject. Here, the physician attempts to select the optimal collateral vessel for accessing the occlusion in the retrograde direction. This selection is influenced by factors such as the tortuosity of each collateral vessel, the risk of tamponade, the difficulty of guidewire placement in the retrograde direction, and the ability to dilate the collateral vessel.

[0036] Figure 3 This is a schematic diagram illustrating an example of a system 100 for selecting collateral vessels to perform a retrograde CTO procedure, according to some aspects of this disclosure. Figure 4 This is a flowchart illustrating an example of a computer-implemented method for selecting collateral vessels to perform a retrograde CTO procedure, based on some aspects of this disclosure. Figure 3 The system 100 shown includes one or more processors 110. Note that operations described as being performed by one or more processors 110 can also be described as... Figure 4 The method shown is used to perform the operation. Furthermore, it should be noted that the description is as... Figure 4 The operations performed by the method shown can also be performed by Figure 3 One or more processors are used to execute the command.

[0037] refer to Figure 3 The system 100 for selecting collateral vessels to perform retrograde chronic total occlusion (CTO) treatment procedures includes one or more processors 110, which are configured to: Receive S110 angiographic image data 120 representing the occlusion 130 in the target blood vessel 140; Analyze S120 angiographic image data 120 to identify one or more candidate collateral vessels 150 for use in approaching the occlusion 130 in the target vessel 140 in the retrograde direction. 1··i ;as well as 150 to each candidate collateral vessel 1··iAssign a suitability score to S130, the suitability score representing the suitability of the vessel for performing a retrograde CTO procedure on the occlusion 130.

[0038] In this system, the identification of candidate collateral vessels and the assignment of suitability scores to them facilitate physicians' selection of the optimal vessel for the procedure. This helps physicians overcome the challenge of mentally weighting the many factors that influence their selection.

[0039] refer to Figure 3 In the system shown, during operation S110, one or more processors 110 receive angiographic image data 120 representing an occlusion 130 in a target blood vessel 140.

[0040] The angiographic image data 120 received in operation S110 can be generated by various types of imaging systems. These imaging systems include those that generate angiographic image data that can be reconstructed into volumetric images (i.e., volumetric image data), and those that generate angiographic image data that can be reconstructed into two-dimensional images (i.e., projection image data). For example, this image data can be generated by a magnetic resonance imaging (MRI) system, a computed tomography (CT) system, or a projection X-ray imaging system. Figure 3 In the example shown, the angiography image data 120 is generated by a projection X-ray imaging system. In examples where the angiography image data is generated by a CT or projection X-ray imaging system, the imaging system can be a spectral X-ray imaging system. In other words, the angiography image data 120 can be generated by a spectral CT imaging system or a spectral X-ray projection imaging system. Spectral X-ray imaging systems generate X-ray attenuation data representing X-ray attenuation across multiple different energy ranges. X-ray attenuation data from spectral X-ray imaging systems can be processed to distinguish media that have similar X-ray attenuation values ​​when measured within a single energy range and would be indistinguishable in conventional X-ray attenuation data. Therefore, X-ray attenuation data from spectral X-ray imaging systems can be used to provide images with improved specificity for materials such as contrast agents, tissues, and bone.

[0041] Generally, the angiographic image data 120 received in operation S110 can represent a single static image, or it can represent a time series of images, i.e., fluorescence fluoroscopy images. In some examples, the angiographic image data 120 can be generated immediately before being received by one or more processors 110. For example, the angiographic image data 120 can represent a live image sequence. Alternatively, the angiographic image data may have been generated in advance for several seconds, minutes, hours, or even days or longer. In the latter case, the angiographic image data 120 can be generated before performing a medical procedure on the anatomical region. In this case, the angiographic image data 120 can be referred to as preoperative angiographic image data.

[0042] In some examples, the angiographic image data 120 received in operation S110 may be generated after the contrast agent has been injected into the vascular system of the object. For example, angiographic image data generated by a CT imaging system or a projection X-ray imaging system can be generated in this manner. The contrast agent highlights the blood in the angiographic image data. In some examples, the angiographic image data 120 may represent a digital subtraction angiography (DSA) image. A DSA image is generated by subtracting the image intensity value at the corresponding location in an X-ray image generated before the contrast agent was injected from the image intensity value of an X-ray image generated after the contrast agent has been injected. Therefore, the DSA image provides a visualization of blood in an anatomical structure free from obscuring image features such as bone, which are common in images both before and after contrast agent injection.

[0043] The angiography image data 120 received in operation S110 can be received from various sources, including medical imaging systems (e.g., the imaging systems described above). Alternatively, the angiography image data 120 can be received from another source (e.g., a database, computer-readable storage medium, the Internet, the cloud, etc.). The angiography image data 120 can be received via any form of data communication (including wired and wireless communication). By way of examples, when using wired communication, communication can occur via cable or fiber optic cable, and when using wireless communication, communication can occur, for example, via RF or infrared signals.

[0044] The angiographic image data 120 received in operation S110 represents an occlusion in the target vessel. The angiographic image data may represent a CTO in the right coronary artery 140, such as... Figure 1 The example shown is illustrated below. In other examples, angiographic image data can represent an occlusion in another blood vessel within an anatomical structure.

[0045] Return to Figure 3In the system shown, during operation S120, one or more processors 110 analyze angiographic image data 120 to identify one or more candidate collateral vessels 150 for approaching the occlusion 130 in the target vessel 140 in the retrograde direction. 1··i .

[0046] In one example, the angiographic image data 120 received in operation S110 includes cardiac angiographic image data. In this example, candidate collateral vessels identified in operation S120 may include: one or more septal collateral vessels 1501, 1502, and / or one or more epicardial collateral vessels 1503, and / or one or more bypass graft vessels 1504. Examples of these vessels are provided in... Figure 1 The example is illustrated in the figure. In other examples, the angiographic image data 120 received in operation S120 may represent different regions of anatomy, thus allowing the identification of different collateral vessels in operation S120.

[0047] In one example, operation S120 is performed on segmented angiographic image data. In this example, one or more processors 110 are configured to segment the angiographic image data 120 to identify target vessels 140 and occlusions 130, as well as one or more candidate collateral vessels 150. 1··i One or more processors analyze segmented angiographic image data to identify one or more candidate collateral vessels.150 1··i The segmentation performed according to this example improves the distinction between different anatomical regions in the anatomical image data. Therefore, performing operation S120 on the segmented angiographic image data provides information on the collateral vessels 150. 1··i More reliable identification. Various segmentation techniques can be used for this purpose, including region growing, thresholding, feature detection, model-based segmentation, and segmentation using trained neural networks (e.g., U-Net).

[0048] In the relevant examples, tags can be applied to the segmented target vessel 140, occlusion 130, and one or more candidate collateral vessels 150. 1··i It also outputs a graphical representation of segmented angiographic image data 120, including corresponding labels. This helps inform doctors of different options for reaching the occlusion.

[0049] In another example, the angiographic image data 120 received in operation S110 is generated by a spectral X-ray imaging system and includes X-ray attenuation data representing X-ray attenuation across multiple different energy ranges. In this example, one or more processors 110 are configured to analyze the angiographic image data 120 in operation S120 by applying a material decomposition algorithm to identify target vessels 140, occlusions 130, and one or more candidate collateral vessels 150. 1··i Various material decomposition algorithms are known for this purpose.

[0050] In one example, one or more processors 110 are configured to analyze angiographic image data 120 to identify one or more candidate collateral vessels 150. 1··i : Identify one or more vessels having connections to a location distal to the occlusion 130, and for each identified vessel: Establishing the existence of a path between the connection and the proximal location of the occlusion; and Based on the existence of the pathway, vessels are assigned as candidate collateral vessels 150 1··i .

[0051] These operations can be performed using various image processing techniques. For example, reference Figure 3 As illustrated in the example, the above operations can begin by identifying the coronary arteries LCA, LCX, LAD, and RCA. This can be performed using the segmentation techniques described above (e.g., model-based segmentation). Subsequently, the location of the occlusion 130 in the target vessel 140 can be identified by evaluating the intensity gradient along each artery in the artery. The location of the occlusion 130 can be identified as the location of an anomalous change in the intensity gradient. Starting from the distal location of the occlusion, one or more vessels with connections to the distal location of the occlusion 130 can then be identified based on the presence of branches in the target vessel. At each branch, it can be determined whether a path to the proximal location of the occlusion exists. This operation can be determined by evaluating a connectivity metric of the path (i.e., a value representing the probability that adjacent pixels along the path are connected). If such a path exists, the vessel can be considered a candidate collateral vessel.

[0052] In another example, a neural network 170 is used to perform analysis of angiographic image data 120 in the S120 to identify one or more candidate collateral vessels 150 for approaching the occlusion 130 in the target vessel 140 in the retrograde direction. 1··i The operation is as follows. See below for reference. Figure 5 Let's describe this example in more detail.

[0053] Return to Figure 3In the system shown, during operation S130, one or more processors 110 direct the signal to each candidate collateral vessel 150. 1··i Assign a suitability score. The suitability score indicates the suitability of the vessel for performing a retrograde CTO procedure on occlusion 130.

[0054] Each candidate collateral vessel can be evaluated based on various factors. 1··i The fitness score, which includes one or more of the following factors: The tortuosity of candidate collateral vessels; Risk values ​​for occlusion of candidate collateral vessels; The complexity of the interventional device traveling along the candidate collateral vessels; The ability to dilate candidate collateral vessels; Electronic medical record data related to occlusion; Blood flow data for candidate collateral vessels.

[0055] These factors can be assessed in various ways. For example, the following items can be evaluated based on angiographic image data 120 to assess the tortuosity of collateral vessels: the integral of the deviation from the straight path along the vessel path, or the maximum deviation from the straight path at one or more points along the vessel path, or the radius of curvature of the vessel along one or more segments of the vessel path, or the presence of a spiral along the vessel path.

[0056] Closure occurs when there is perforation of a collateral vessel. Therefore, the risk of clogging can be assessed based on factors such as the thickness of the collateral vessel wall (relatively thinner walls are associated with a relatively higher risk of clogging), the tortuosity of the collateral vessel (relatively higher tortuosity is associated with a relatively higher risk of clogging), the use of anticoagulants in the CTO treatment process, and the type of collateral vessel (e.g., bypass graft). The risk of collateral vessel perforation is also affected by the angle between the collateral vessel and the target vessel where the occlusion is located. Therefore, if the angle between the collateral vessel and the (target) vessel where the occlusion is located exceeds a specified angle (e.g., 90 degrees), a relatively high risk of vessel perforation and therefore occlusion may be associated with the collateral vessel. Thus, the risk of collateral vessel clogging can be assessed based on angiographic image data120 by quantifying such factors using image processing techniques.

[0057] A measure representing the complexity of an interventional device traveling along a collateral vessel's route can be evaluated based on angiographic image data 120 by assigning a complexity metric to segments of candidate collateral vessels using image processing techniques. For example, the complexity metric can be assigned along branches of candidate collateral vessels (including branches selected by entering and exiting the collateral vessel) based on factors such as the diameter of the collateral vessel, the angle of path deviation, and the distance from the distal end of the collateral vessel to the distal cap of the occlusion.

[0058] It may be possible to dilate some collateral vessels using methods such as balloons or intravenous drug injection, and this facilitates improved maneuverability of interventional devices along collateral vessels. Therefore, vessels capable of dilation can be considered more suitable for performing retrograde CTO procedures than those without this capability. The ability to dilate collateral vessels can be determined from databases.

[0059] Electronic health record (EHR) data related to occlusion can also be used to determine the suitability of a given collateral vessel for retrograde CTO procedures. EHR data may include elements such as body mass index, smoking history, age, and sex. EHR data may also include elements related to clinical investigations already performed on the subject, such as ECG data, data related to the characteristics of the collateral vessel (e.g., its tortuosity, lumen diameter), the presence of artifacts such as calcification, the presence of implanted devices such as stents, and outcomes of the subject's historical procedures (e.g., bleeding incidence). EHR data may include elements related to whether a retrograde CTO procedure has been attempted in a given collateral vessel before. If the vessel has been used previously, it may be less suitable for a retrograde CTO procedure than a vessel that has not been used before due to the increased risk of rupture associated with previously used vessels. EHR data may include elements related to blood flow to the collateral vessel. Blood flow data can be used to determine the suitability of a vessel for performing a retrograde CTO procedure on a collateral vessel. For example, blood flow data may include the vessel's transit time. EHR data represents the time it takes for the injected contrast agent bolus to travel along the vessel and provides an indication of blood flow velocity along the vessel. If a given vessel has a relatively long transit time compared to other vessels, this may indicate that the vessel is unsuitable for performing a retrograde CTO procedure on the occlusion, as this may indicate a relatively narrow vessel along which navigation of the interventional device may be difficult. Blood flow data can be evaluated based on angiographic image data 120 by using image processing techniques to determine the time taken to travel between the proximal and distal positions of the injected contrast agent bolus within the vessel. EHR data may include elements related to the intravascular imaging procedure already performed on the target vessel or collateral vessel. For example, results obtained from intravascular ultrasound (IVUS) or optical coherence tomography (OCT) imaging data can be used. The contributions of individual elements from the EHR data (e.g., those mentioned above) can be weighted to provide a measure of suitability for each collateral vessel, or alternatively, the data can be fed into a neural network, as referenced below. Figure 5 As described in more detail, various elements of the EHR data (e.g., those mentioned above) can be weighted based on the above considerations to provide a combined weighted suitability for a given collateral vessel to perform a retrograde CTO procedure.

[0060] In another example, a neural network 170 is used to perform the delivery of energy to each candidate collateral vessel 150. 1··i The operation of assigning S130 fitness scores is described below. (See below for reference.) Figure 5 Let's describe this example in more detail.

[0061] Return to Figure 3In the system shown, after one or more fitness scores have been assigned to one or more candidate collateral vessels in operation S130, one or more processors 110 can then output the assigned fitness scores. The fitness scores can be output in various ways. In some examples, the fitness scores are output graphically. For example, in one example, the fitness scores can be output as a ranking list, which includes the identifier of each collateral vessel and its corresponding score. In another example, a graphical representation of one or more collateral vessels can be generated, and the corresponding one or more collateral vessels can be labeled or color-coded using the fitness scores. A graphical representation of one or more collateral vessels can be generated based on angiographic image data 120. In one example, the graphical representation of one or more collateral vessels includes a composite image of the collateral vessels, and the fitness scores are output by color-coding the collateral vessels according to their fitness.

[0062] In another example, one or more processors 110 are configured to output an indication of a recommended orientation for the projection X-ray imaging system 160 for acquiring identified candidate collateral vessels 150 during the CTO treatment procedure. 1··i Projected X-ray images of one or more candidate collateral vessels. Projected X-ray imaging systems are often used in CTO (Chronic Osteoarthritis Toxic) procedures due to their relatively low X-ray dose compared to CT imaging systems. However, projection X-ray images lack depth information, necessitating an appropriate orientation of the projection X-ray imaging system relative to the anatomical structure to provide useful vascular images during the procedure. In this example, the output recommended orientation helps reduce the repositioning effort required for the projection X-ray imaging system to obtain useful collateral vessel images during the procedure.

[0063] In this example, various techniques can be used to determine the recommended orientation of the projection X-ray imaging system 160. In one technique, a representation of the heart and (one or more) collateral vessels 150 is used. 1··iA 3D model is used. In this technique, angiographic image data 120 includes projected image data generated by a projection X-ray imaging system. The projected image data is registered to the 3D model to determine the current orientation of the projection X-ray imaging system relative to (one or more) collateral vessels in the 3D model. A database storing recommended orientations for acquiring X-ray images of each collateral vessel in the 3D model is then consulted to determine adjustments to the current orientation of the projection X-ray imaging system required for acquiring projected X-ray images of (one or more) collateral vessels. Recommended orientations for collateral vessels in the 3D model can be provided by a specialist physician. The required adjustments are then output. For example, the required adjustments can be output by displaying them on a monitor or by outputting control signals for automatically adjusting the orientation of the projection X-ray imaging system.

[0064] In another technique, the angiographic image data 120 includes volumetric image data. For example, the volumetric image data can be generated by a CT imaging system or an MRI imaging system. In this technique, the method for acquiring each candidate collateral vessel 150 can be determined by the following steps. 1··i Recommended orientation for projected X-ray images: The volumetric image data is reconstructed into a volumetric image, and the following orientation of a virtual plane passing through the candidate collateral vessel is calculated, which minimizes the least-squares error difference between the plane and each of multiple locations along the vessel. The orientation of the vector oriented relative to the virtual plane normal is then calculated. This provides a recommended orientation for the projected X-ray imaging system 160 to acquire projected X-ray images of vessels during CTO treatment procedures.

[0065] In another example technique, angiographic image data 120 is input into a neural network, which is trained to predict one or more candidate collateral vessels 150 in the angiographic image data 120 for acquisition by the projection X-ray imaging system 160. 1··i The recommended orientation is provided for the projected X-ray images of each candidate collateral vessel. The training data used to train the neural network includes angiographic training images representing one or more collateral vessels for each of several different objects, and a corresponding recommended orientation for each training image. The recommended orientation in the training data used in this technique can be provided by an expert.

[0066] In another example, one or more processors 110 are configured to output to the identified candidate collateral vessels 150. 1··i The risk score indicates the risk of medical complications arising from the use of a vessel in a retrograde CTO procedure.

[0067] In this example, the output risk score can represent the risk of various medical complications, such as myocardial infarction, thrombosis, pericardiocentesis, contrast agent-induced nephropathy (in one or more candidate collateral vessels), etc. It can also output an indication of the risk of medical complications arising from performing anterograde CTO procedures on occlusions. This facilitates physicians' comparative assessment of the risks of performing CTO procedures using anterograde and retrograde approaches.

[0068] Various techniques can be used to assess risk scores for (one or more) collateral vessels and for anterograde proximity. In one technique, a database of historical CTO procedure outcome data and corresponding EHR data for historical procedures performed on multiple different subjects are used to assess risk scores. In this technique, EHR data related to occlusion 130 is used to consult the database to identify historical CTO procedure outcome data for one or more similar subjects. Risk scores are then assessed based on the outcome data of these similar subjects. One or more similar subjects can be identified by applying a similarity measure to the EHR data. Similarity measures (e.g., the Dice coefficient) can be used. For example, a similarity measure can be used to identify similar subjects based on elements of the EHR data (e.g., age, sex, weight, height, body mass index, and type of collateral vessel). Risk scores can be assessed by applying statistical analysis to the outcome data of similar subjects. If a single subject is identified using this technique, the outcome data for that single subject can be output.

[0069] In another example technique, angiographic image data 120 is fed into a neural network trained to predict a risk score for each of one or more vessels in the angiographic image data 120. Training data for the neural network may include angiographic training images representing one or more vessels from multiple different objects, and a corresponding risk score for each training image. The risk scores in the training data may be provided by experts or curated based on historical outcome data from procedures already performed on the vessels.

[0070] As described above, in some examples, a neural network 170 is used to perform analysis of the angiography image data 120 of S120 to identify one or more candidate collateral vessels 150. 1··i And 150 to each candidate collateral vessel 1··i The operation of assigning S130 fitness scores. Now refer to... Figure 5 To describe these examples, Figure 5This is a schematic diagram illustrating an example of a neural network 170 according to some aspects of the present disclosure, the neural network 170 being trained to: identify one or more candidate collateral vessels 1501, 1502 for approaching an occlusion in a target vessel in a retrograde direction, and assign a fitness score to each candidate collateral vessel. Reference Figure 5 and Figure 3 In these examples, one or more processors 110 are configured to analyze angiographic image data 120 to identify one or more candidate collateral vessels 150. 1··i And / or assign a fitness score to each candidate collateral vessel: Angiographic image data 120 is input into at least one neural network 170; and In response to the input, output one or more identified candidate collateral vessels 150 1··i and / or one or more fitness scores assigned; and In this embodiment, at least one neural network 170 is trained to identify one or more candidate collateral vessels for approaching an occlusion in a target vessel in a retrograde direction based on the input angiographic image data and / or to assign a fitness score to each candidate collateral vessel.

[0071] exist Figure 5 In the example shown, neural network 170 may include one or more architectures (e.g., a convolutional neural network "CNN", a recurrent neural network "RNN", a variational autoencoder "VAE", a transformer, etc.). Figure 5 In the example shown, the angiography image data 120 is input into the neural network 120, such as... Figure 5 As shown in the upper left corner. In the example shown, angiographic image data 120 represents a single static image. Angiographic image data 120 can alternatively represent a time series of images. Figure 5 As shown in the upper right corner, in response to the input, the neural network 170 outputs one or more identified candidate collateral vessels 150. 1··i And / or have been allocated to 150 collateral vessels 1··i One or more fitness scores are assigned to the candidate collateral vessels 1501 and 1502. In the example shown, candidate collateral vessels 1501 and 1502 have been identified and output as dashed lines on the graphical representation of the angiographic image data 120. In the example shown, fitness scores 3 and 6 have been assigned to the candidate collateral vessels 1501 and 1502, and these fitness scores 3 and 6 have been output by labeling the corresponding vessels in the graphical representation of the angiographic image data 120. In other examples, the neural network 170 may receive additional input data and may output additional data. Some examples of this data are shown via dashed lines on the graph. Figure 5The image below illustrates this. These examples are described in more detail below.

[0072] The training of at least one neural network 170 will now be described in detail. Generally, training a neural network involves: inputting a training dataset into the neural network; and iteratively tuning the parameters of the neural network until the trained neural network provides accurate output. A graphics processing unit (GPU) or a dedicated neural processor (e.g., a neural processing unit (NPU) or tensor processing unit (TPU)) is often used to perform the training. Training often employs a centralized approach, where cloud-based or mainframe-based neural processors are used to train the neural network. After training with the training dataset, the trained neural network can be deployed to devices for analyzing new input data during inference. The processing requirements during inference are significantly less than those required during training, allowing the neural network to be deployed on a variety of systems (e.g., laptops, tablets, mobile phones, etc.). Inference can be performed, for example, by a central processing unit (CPU), GPU, NPU, or TPU on a server or in the cloud.

[0073] Therefore, the process of training one or more neural networks 170 described above includes adjusting their parameters. Parameters (or more specifically, weights and biases) control the operation of the activation functions in the neural network. In supervised learning, the training process automatically adjusts the weights and biases so that, given input data, the neural network accurately provides the corresponding predicted output data. To do this, a loss function, or error value, is calculated based on the difference between the predicted output data and the expected output data. The loss function can be calculated using functions such as negative log-likelihood loss, mean squared error, Huber loss, or cross-entropy loss. During training, the value of the loss function is typically minimized, and training terminates when the value of the loss function meets a stopping criterion. Sometimes, training terminates when the value of the loss function meets one or more of a plurality of criteria.

[0074] Various methods are known for solving the loss minimization problem, such as gradient descent, quasi-Newton methods, etc. Various algorithms have been developed to implement these methods and their variants, including but not limited to stochastic gradient descent (SGD), batch gradient descent, mini-batch gradient descent, Gauss-Newton, Levenberg-Marquardt, Momentum, Adam, Nadam, Adagrad, Adadelta, RMSProp, and the Adamex "optimizer". These algorithms use a chain rule to compute the derivative of the loss function with respect to the model parameters. This process is called backpropagation because the derivative is computed starting from the last layer or output layer, moving towards the first layer or input layer. These derivatives tell the algorithm how the model parameters must be adjusted to minimize the error function. That is, the model parameters are adjusted starting from the output layer and working backward through the network until the input layer is reached. In the first training iteration, the initial weights and biases are often randomized. The neural network then predicts the output data (which is also randomized). Backpropagation is then used to adjust the weights and biases. The training process is performed iteratively by adjusting the weights and biases in each iteration. Training terminates when the error or difference between the predicted output data and the expected output data is within an acceptable range of the training data or some validation data. Subsequently, the neural network can be deployed, and the trained neural network uses the training values ​​of its parameters to predict new input data. If the training process is successful, the trained neural network accurately predicts the expected output data based on the new input data.

[0075] In one example, at least one neural network 170 is trained to identify one or more candidate collateral vessels and / or assign a fitness score to each candidate collateral vessel by: Receive training data, which includes: angiographic image data representing occlusion in a target blood vessel for each of a plurality of objects; Receive baseline ground truth data corresponding to the training data, the baseline ground truth data including: for each of a plurality of objects, the identification of one or more collateral vessels for approaching the occlusion in the target vessel in the retrograde direction and / or the suitability score of each collateral vessel, the suitability score representing the vessel's suitability for performing a retrograde CTO procedure on the occlusion; and For each of the multiple objects: The training data is fed into at least one neural network; The parameters of at least one neural network are adjusted based on the value of a loss function representing the difference between the fitness score assigned to each collateral vessel by at least one neural network and the corresponding ground truth data; and Repeat the input and adjustment until the stopping criterion is met.

[0076] The training data used in this example can be curated based on a historical process in which experts have identified baseline ground truth candidate collateral vessels and / or fitness scores. The baseline ground truth fitness score can be evaluated based on the factors described above. For example, the baseline ground truth fitness score can be evaluated based on one or more of the following: The tortuosity of candidate collateral vessels; Risk values ​​for occlusion of candidate collateral vessels; The complexity of the interventional device traveling along the candidate collateral vessels; The ability to dilate candidate collateral vessels; Electronic medical record data related to occlusion; Blood flow data for candidate collateral vessels.

[0077] Alternatively, baseline truth fitness scores can be assigned based on expert experience. For example, experts could provide fitness scores for blood vessels in the form of a ranked list, where the ranking corresponds to the fitness of the blood vessel.

[0078] In a relevant example, at least one neural network 170 is trained to also identify one or more candidate collateral vessels and / or assign a fitness score to each candidate collateral vessel based on electronic medical record data 180 associated with the occlusion 130. In this example, one or more processors 110 are also configured to: Receive electronic medical record data 180 related to occlusion 130; and Electronic medical record data is input into at least one neural network 170; This results in the output of one or more candidate collateral vessels and / or the assigned fitness scores being based on electronic medical record data.

[0079] Figure 5 The example is illustrated, and it shows the EHR data 180 being input into... Figure 5One or more neural networks 170 are located in the left central portion of the image. According to this example, various types of EHR data described above can be input into one or more neural networks 170. In one example, the EHR data includes electrocardiogram (ECG) data related to the occlusion. The ECG data can be input in the form of a graphical representation of the ECG signal. The EHR data improves the accuracy of the output of at least one neural network 170 because the output of at least one neural network 170 can be generated based on trends identified in the EHR data during training. In this example, at least one neural network 170 can be trained to identify one or more candidate collateral vessels and / or assign a fitness score to each candidate collateral vessel by including electronic medical record data 180 related to the occlusion 130 in the training data described above.

[0080] In another related example, at least one neural network 170 is trained to predict the recommended type of interventional device to be used in a CTO treatment procedure based on input angiographic image data 120. In this example, one or more processors 110 are configured to output a recommended type of interventional device to be used in a CTO treatment procedure in response to inputting angiographic image data 120 into at least one neural network 170.

[0081] Figure 5 The example is illustrated, and it shows one or more neural network 170 output device types, such as... Figure 5 The right side of the diagram is shown. According to this example, various types of interventional devices can be output, including, for example, types of (balloon) catheters, guidewires, thrombectomy devices, etc. The output device type can include, for example, the model number of the interventional device or specifications for thickness, rigidity, or tip shape. The output of device types according to this example provides additional guidance to the physician during the procedure. In this example, at least one neural network 170 is trained to predict the recommended type of interventional device used in a CTO treatment procedure by including baseline ground truth data representing the recommended types of interventional devices used in a CTO treatment procedure in the training data described above, and also evaluating a loss function based on the difference between the recommended types of interventional devices predicted by one or more neural networks and the recommended types of interventional devices used in a CTO treatment procedure from the baseline ground truth data.

[0082] As stated above, retrograde approach to occlusion carries a higher risk of complications compared to antegrade approach. These complications include, for example, the risk of myocardial infarction and the risk of collateral vessel thrombosis. Both risks increase because the collateral vessel is blocked for too long a period of time. In another example, the system analyzes second angiographic image data to estimate the total time the interventional device has blocked the lumen of the selected collateral vessel, and outputs a warning if the total time exceeds a threshold time period or an indication of the amount of time remaining available to perform the procedure. The warning, along with the similar indication of the amount of time remaining available to perform the procedure, alerts the physician to the increased risk of complications during the retrograde CTO procedure.

[0083] In this example, the above reference Figure 3 One or more processors 110 of the described system are configured to: Receive second angiography image data 190 representing the selected collateral vessel in the interventional device 200; Analyze second angiography image data 190 to provide information on the occlusion time period T. Obs The estimated result, wherein the occlusion time period represents the total time that the interventional device 200 has occluded the lumen of the selected collateral vessel; and In response to the estimated blocking time period exceeding the threshold time period T ObsMax And output a warning; or The output indicates the amount of remaining time available to execute the process, based on a threshold time period T. ObsMax With the blocking time period T Obs It is calculated based on the difference between them.

[0084] In this example, it can be done in various ways (including, for example, via a display device, for example, Figure 3 The monitor 240 shown provides warnings visually or audibly. Warnings, as well as similarly, indications of the remaining time for performing the procedure, alert the physician to an increased risk of complications during the retrograde CTO procedure.

[0085] In this example, the second angiography image data 190 can represent a time series of images. Various techniques can be used to analyze the second angiography image data 190 to provide data specific to the occlusion time period T. Obs The estimation results. In one technique, image processing techniques such as feature detection can be used to detect the location of the interventional device in the time series of images in the second angiography image data 190. In this technique, the occlusion time period T can be estimated by detecting the location of the interventional device relative to a selected collateral vessel in the images in the time series and accumulating the total time the interventional device 200 spends within the selected collateral vessel. ObsThe location of the interventional device can be facilitated by providing one or more radiopaque markers on the device. In another technique described in more detail below, a neural network can be used to estimate the occlusion period T. Obs .

[0086] In this example, the threshold time period T ObsMax This indicates the maximum total time required for the interventional device to safely occlude the lumen of a collateral vessel during a retrograde chronic total occlusion (CTO) procedure. In this example, the term "occlusion" should be interpreted to encompass a range of occlusion scenarios (including "at least partial occlusion" and "complete occlusion"). In this example, the term "safely occlude" should be interpreted to encompass a safety margin that provides the physician, for example, sufficient time to remove the interventional device from the lumen before the onset of medical complications. For example, it could represent 90% of the estimated time prior to the expected occurrence of a myocardial infarction.

[0087] Various techniques can be used to estimate the threshold time period T. ObsMax The value of . In one example described below, the threshold time period T is estimated based on the ratio of the cross-sectional area of ​​the interventional device used in the CTO procedure to the cross-sectional area of ​​the collateral vessel. ObsMax The value of [value missing]. As reported by Wustmann, K. et al. in the article “Is There Functional Collateral Flow During Vascular Occlusion in Angiographically Normal Coronary Arteries?” (Circulation, 2003, 107, pp. 2213-2220), in a study, the coronary arteries of 100 individuals were temporarily occluded for 1 minute by an angioplasty balloon. During the temporary occlusion, the flow provided by collateral circulation was approximately 18% of the baseline flow, which was measured by the collateral flow index (CFI). During the 1-minute coronary artery occlusion time, approximately 25% of the individuals in the study did not develop angina, and 20% of the individuals did not develop signs of ischemia on the coronary ECG. Based on this information, it can be concluded that, in general, if 100%–82% (i.e., 82%) of the baseline flow is occluded, approximately 20% of the individuals will not develop signs of ischemia after 1 minute. This can be used as an upper limit, since 75% of individuals do develop angina, and 80% do develop ECG signs of ischemia. Therefore, it is possible to calculate x% of the lumen of the collateral vessel that the interventional device can safely occlude during a retrograde CTO procedure using the following formula ( The threshold time period T) ObsMax :

[0088] In the above formula, the value This can be calculated as the ratio of the cross-sectional area of ​​the interventional device used in the CTO procedure to the cross-sectional area of ​​the collateral vessels. For , The value can be limited to 1 minute. In cases where additional collateral vessels supply blood to tissue downstream of the occlusion, the modified value... It can be calculated as the ratio of the cross-sectional area of ​​the interventional device used in the CTO treatment procedure to the total cross-sectional area of ​​all collateral vessels supplying blood to the tissue downstream of the occlusion.

[0089] Alternatively, other techniques can be used to estimate the threshold time period T. ObsMax The value of . For example, in the example above, assume T in Formula 1 ObsMax and The relationship is linear. However, for smaller occlusions, T ObsMax This timeframe can become longer, therefore the relationship is non-linear. Therefore, in another technique, the threshold time period T can be calculated using the following formula. ObsMax During this threshold time period, the interventional device can safely occlude the lumen of collateral vessels during retrograde CTO procedures. ( ):

[0090] And among them, Note that other formulas besides these examples can be used alternatively to evaluate the threshold time period T. ObsMax In other techniques, the threshold time period can also be evaluated based on other factors (e.g., vessel type, EHR data, ECG data, etc.). For example, elements of the EHR data and characteristics of the ECG data can be used to weight the values ​​calculated above. Alternatively, other formulas can be used to estimate the threshold time period T in other techniques. ObsMax The value of .

[0091] One or more processors 110 of system 100 can use various techniques to obtain the threshold time period T. ObsMax The value of the threshold time period T. For example, in one technique, the threshold time period can be extracted from a lookup table based on the type of collateral vessels. The threshold time period can also be extracted from the lookup table using other data related to collateral vessels. For example, the threshold time period can be extracted from the lookup table by using EHR data such as ECG data of blood vessels to identify threshold time periods from objects with similar EHR data. In another technique described in more detail below, a neural network can be used to predict the threshold time period T. ObsMax .

[0092] As mentioned above, neural networks can be used to predict the blocking time period T in the above example. Obs In this example, the above reference... Figure 3 One or more processors 110 described are configured to analyze second angiography image data 190 to provide data for the occlusion time period T. Obs The estimation results are as follows: The second angiography image data 190 is input into the neural network 170, which is trained to estimate the occlusion time period T. Obs ;as well as The estimated blocking time period T is output in response to the input. Obs ;and The neural network was trained to estimate the duration of blockage based on second angiography image data 190.

[0093] This example is in Figure 5 The figure in the middle shows the estimated congestion time T. Obs The diagram is shown as being in Figure 5 The output is located in the central right part.

[0094] In this example, neural network 170 can be trained by including multiple time series of angiographic images of an interventional device in the lumen representing collateral vessels in the training data, and for each time series representing the occlusion time period T. Obs The occlusion time T is estimated using the corresponding baseline ground truth data (i.e., the total time during which the interventional device 200 has blocked the lumen of the collateral vessel). Obs Then, the neural network 170 is trained by inputting each time series of angiography images and using the neural network 170 to estimate the occlusion duration T. Obs And also based on the blocking time period T estimated by the neural network. Obs Corresponding blocking time period T from the baseline true data Obs The difference between them is used to evaluate the loss function.

[0095] As mentioned above, neural networks can be used to predict the threshold time period T in the above example. ObsMax The following describes two techniques for this operation. In one technique, a threshold time period T is predicted based on angiographic image data. ObsMax In this technology, reference Figure 3 One or more processors 110 described are configured to: Angiographic image data 120 and / or second angiographic image data 190 are input into neural network 170, which is trained to predict a threshold time period T for a selected collateral vessel. ObsMax;as well as Output a threshold time period T in response to the input. ObsMax ;and The neural network 170 is trained to predict the threshold time period for the selected collateral vessel based on the input angiography image data 120 and / or the second angiography image data 190.

[0096] In another technique, a threshold time period T is predicted based on vascular parameters extracted from angiographic image data. ObsMax In this technology, reference Figure 3 One or more processors 110 described are configured to: Extract one or more vascular parameters 220 of the selected collateral vessels from the angiography image data 120; One or more extracted vascular parameters 220 are input into a neural network 210, which is trained to predict a threshold time period T for a selected collateral vessel. ObsMax ;as well as In response to the input, output a threshold time period T for the selected collateral vessel. ObsMax .

[0097] In the second technique, vascular parameters (e.g., the type of collateral vessels, the supplying vessels downstream of the collateral vessels, and the diameter of the collateral vessels) can be extracted from the angiography image data 120. The vascular parameters 220 can be extracted using image processing techniques or by using a neural network trained to extract values ​​for one or more vascular parameters 220.

[0098] In both techniques, neural network 170 can be trained to predict the threshold time period T by performing the following operations. ObsMax : Receive threshold time period training data, the threshold time period training data includes angiography image data for each of multiple objects, the angiography image data representing: collateral vessels used to perform retrograde CTO treatment procedures, or extracted vascular parameters 220 of collateral vessels used to perform retrograde CTO treatment procedures; Receive baseline ground truth data corresponding to the training data for the threshold time period, the baseline ground truth data including, for each of a plurality of objects, an indication of the total time for which the interventional device can safely occlude the lumen of collateral vessels before medical complications occur; and For each of the multiple objects: The training data for the threshold time period was input into neural networks 170 and 210; and The parameters of the neural network are adjusted based on the value of a loss function that represents the difference between the threshold time period predicted by the neural network and the corresponding total time from the baseline ground truth data; and Repeat the input and adjustment until the stopping criterion is met.

[0099] In this example, the training data used to train the neural network 170 can be curated based on a historical process in which experts have determined the baseline total time (i.e., T) during which the interventional device can safely occlude the lumen of collateral vessels before medical complications occur. ObsMax Alternatively, Formula 1 above can be used to calculate Ti based on the time series of angiographic images used in the training data. ObsMax The value of .

[0100] In relevant examples, ECG data for occlusion can also be used to predict the threshold time period T. ObsMax In this example, neural network 170 was trained to also predict the threshold time period T based on ECG data for the occlusion. ObsMax In this example, refer to Figure 3 One or more processors 110 described are configured to: Receive electrocardiogram data generated during the CTO processing procedure; Inputting electrocardiogram data into a neural network; and It also outputs a threshold time period T based on electrocardiogram data. ObsMax .

[0101] In this example, neural network 170 is trained to predict the threshold time period T by including the electrocardiogram data of the corresponding object in the training data for the aforementioned threshold time period. ObsMax Using ECG data, as illustrated in this example, can help improve the accuracy of neural network predictions.

[0102] In another example, neural network 170 was trained to predict threshold time period T based on the type of interventional device used in the CTO treatment process. ObsMax In this example, refer to Figure 3 One or more processors 110 described are configured to: Receive device data 230, which includes an identifier of the type of interventional device used in the CTO treatment process; The device data 230 is input into the neural network; and It also outputs the threshold time period T based on the type of interventional device. ObsMax .

[0103] In this example, neural network 170 is trained to predict the threshold time period T by including the type of interventional device used in the CTO treatment process for the corresponding object in the training data for the aforementioned threshold time period. ObsMax According to this example, using interventional devices can help improve the accuracy of neural network predictions.

[0104] In another example, neural network 170 was trained to predict a threshold time period T based on electronic medical record data 180 associated with occlusion 130. ObsMax Furthermore, one or more processors 110 are configured to: Receive electronic medical record data 180 related to occlusion 130; and Inputting electronic medical record data into a neural network; and It also outputs a threshold time period T based on electronic medical record data 180. ObsMax .

[0105] The type of interventional device used in this example helps improve the accuracy of neural network predictions.

[0106] Note that in the above example, system 100 may be provided with one or more additional items. For example, system 100 may also include one or more of the following: an angiography imaging system 160 for providing angiography image data 120, 190, for example, such as Figure 3 The system includes a projection X-ray imaging system; an interventional device 200 for performing a retrograde CTO procedure on the occlusion 130; and a monitor 240 for displaying data output by one or more processors 110, such as angiographic image data 120, 190, and identified collateral vessels 150. 1··i Graphical representation of its associated (one or more) fitness scores, etc.; 250 hospital beds; and user input devices (in... Figure 3 (not shown), for example, a keyboard, mouse, touchscreen, etc., and configured to receive user input in relation to operations performed by one or more processors 110.

[0107] In another example, a computer-implemented method is provided for selecting a collateral vessel for performing a retrograde chronic total occlusion (CTO) treatment procedure. The method includes: Receive S110 angiographic image data 120 representing the occlusion 130 in the target blood vessel 140; Analyze S120 angiographic image data 120 to identify one or more candidate collateral vessels 150 for use in approaching the occlusion 130 in the target vessel 140 in the retrograde direction. 1··i ;as well as 150 to each candidate collateral vessel1··i Assign a suitability score to S130, which indicates the suitability of the vessel for performing a retrograde CTO procedure on occlusion 130.

[0108] This method is in Figure 4 The image in the middle shows...

[0109] In another example, a computer program product including instructions that, when executed by one or more processors 110, cause the one or more processors to perform a method for selecting a collateral vessel for performing a retrograde chronic total occlusion (CTO) treatment procedure. The method includes: Receive S110 angiographic image data 120 representing the occlusion 130 in the target blood vessel 140; Analyze S120 angiographic image data 120 to identify one or more candidate collateral vessels 150 for use in approaching the occlusion 130 in the target vessel 140 in the retrograde direction. 1··i ;as well as 150 to each candidate collateral vessel 1··i Assign a suitability score to S130, which indicates the suitability of the vessel for performing a retrograde CTO procedure on occlusion 130.

[0110] As described above, the second aspect of this disclosure relates to a method for determining a threshold time period T. ObsMax System 300. The threshold time period T in this example. ObsMax Corresponding to the above threshold time period T ObsMax However, in this second aspect, the threshold time period T ObsMax Provided by an independent system.

[0111] Figure 6 The diagram illustrates some aspects of this disclosure for determining the threshold time period T. ObsMax A schematic diagram of an example of system 300. Figure 7a The diagram illustrates the determination of the threshold time period T based on some aspects of this disclosure. ObsMax A flowchart of the first example of a computer-implemented method. Figure 7b The diagram illustrates the determination of the threshold time period T based on some aspects of this disclosure. ObsMax A flowchart of a second example of a computer-implemented method. Figure 6 The system 300 shown includes one or more processors 310. Note that operations described as being performed by one or more processors 310 can also be described as... Figure 7a and Figure 7b The method shown is executed as a part of the method. Furthermore, it should be noted that what is described as... Figure 7a and Figure 7b The operations performed by the method shown can also be performed by Figure 6 The execution is performed by one or more processors as shown. It should also be noted that one or more processors 310 can be... Figure 3 The same one or more processors 110 shown are provided.

[0112] refer to Figure 6 Used to determine the threshold time period T ObsMax The system 300 includes one or more processors 310, the threshold time period representing the maximum total time that the interventional device 200 can safely occlude the lumen of collateral vessels during a retrograde chronic total occlusion (CTO) procedure, the one or more processors 310 being configured to: Receive S310 angiography image data 320, the angiography image data 320 representing an occlusion 130 in a target vessel 140 and one or more collateral vessels 150 for approaching the occlusion in the target vessel in a retrograde direction. 1··i ; The angiography image data 320 is input into the neural network 170 via S320a, which is trained to predict the effects of one or more collateral vessels 150. 1··i Threshold time period T for each collateral vessel ObsMax ;as well as In response to the input, output S330a a threshold time period T for each of one or more collateral vessels. ObsMax ;and The neural network 170 is trained to predict the threshold time period T for each of one or more collateral vessels based on the input angiography image data. ObsMax .

[0113] The above operations are in Figure 7a The flowchart is illustrated. The neural network 170 can be constructed from components such as those shown above. Figure 5 The neural network described is a neural network 170 or similar, wherein the angiographic image data 120 input to the neural network 170 is instead provided by the angiographic image data 320, and the neural network 170 provides a threshold time period T. ObsMax As output. In this example, other inputs to the neural network 170 (i.e., EHR data 180 and second angiography image data 190) can also be used, but this is not required. In this example, the identified candidate collateral vessels 150 can optionally be output. 1··i ,exist Figure 5 The image shows 150 candidate collateral vessels identified. 1··i The corresponding (one or more) suitability scores, and the device type, as described in the following examples.

[0114] In this example, Figure 5 The neural network 170 shown can be trained to predict the threshold time period T by performing the following operations. ObsMax : Receive threshold time period training data, which includes angiographic image data for each of a plurality of objects, the angiographic image data representing collateral vessels used to perform retrograde CTO treatment procedures; Receive baseline ground truth data corresponding to training data for the threshold time period, the baseline ground truth data including an indication of the total time for each of the plurality of objects, during which the interventional device is able to safely occlude the lumen of the collateral vessel before medical complications occur; and For each of the multiple objects: The training data for the threshold time period was input into neural networks 170 and 210; and The parameters of the neural network are adjusted based on the value of a loss function that represents the difference between the threshold time period predicted by the neural network and the corresponding total time from the baseline ground truth data; and Repeat the input and adjustment until the stopping criterion is met.

[0115] In this example, used for training Figure 5 The training data for the neural network 170 shown can be programmed based on a historical process in which experts have determined the baseline total time (i.e., T) during which the interventional device can safely occlude the lumen of collateral vessels before medical complications occur. ObsMax ).

[0116] In an alternative implementation, the threshold time period T is not predicted directly from the angiography image data 320. ObsMax Instead, it extracts vascular parameters from angiography image data 320 and inputs the extracted vascular parameters 220 into neural network 210, which is then trained to predict threshold time period T. ObsMax The vascular parameters 220 can be extracted using image processing techniques or by using a neural network trained to extract the values ​​of one or more vascular parameters 220.

[0117] In this embodiment, Figure 6 One or more processors 310 shown execute Figure 7b The operation shown is not Figure 7a The operation shown determines the threshold time period T. ObsMax In this embodiment, the threshold time period T is used to determine... ObsMaxThe system 300 includes one or more processors 310, the threshold time period representing the maximum total time that the interventional device 200 can safely occlude the lumen of collateral vessels during a retrograde chronic total occlusion (CTO) procedure, the one or more processors 310 being configured to: Extract one or more collateral vessels of S310b from angiography image data 320 150 1··i One or more vascular parameters 220 for each collateral vessel in the vessel; The extracted one or more vascular parameters 220 are input into S320b to a neural network 210, which is trained to predict a threshold time period for each collateral vessel in one or more collateral vessels; and In response to the input, output S330b is the threshold time period T for each of one or more collateral vessels. ObsMax ;and The neural network 210 is trained to predict a threshold time period for each of one or more collateral vessels based on one or more input vascular parameters 220.

[0118] This implementation method is also Figure 8a The figure is shown in the middle. Figure 8a This is a schematic diagram illustrating an example of a neural network 210 according to some aspects of the present disclosure, which is trained to predict a threshold time period T for a selected collateral vessel based on extracted vascular parameters 220. ObsMax .

[0119] In this example, Figure 5 The neural network 170 shown can be trained to predict the threshold time period T by performing the following operations. ObsMax : Receive threshold time period training data, which includes extracted vascular parameters 220 for collateral vessels used to perform the retrograde CTO procedure for each of multiple objects; Receive baseline ground truth data corresponding to training data for the threshold time period, the baseline ground truth data including an indication of the total time for each of the plurality of objects, during which the interventional device is able to safely occlude the lumen of the collateral vessel before medical complications occur; and For each of the multiple objects: The training data for the threshold time period was input into neural networks 170 and 210; and The parameters of the neural network are adjusted based on the value of a loss function that represents the difference between the threshold time period predicted by the neural network and the corresponding total time from the baseline ground truth data; and Repeat the input and adjustment until the stopping criterion is met.

[0120] In this example, used for training Figure 8a The training data for the neural network 210 shown may include vascular parameters that have been programmed based on angiographic images used in a historical procedure in which experts have determined the baseline total time (i.e., T) during which the interventional device can safely occlude the lumen of collateral vessels before medical complications occur. ObsMax ).

[0121] The threshold time period represents the maximum total time that the interventional device can safely occlude the lumen of a collateral vessel during a retrograde chronic total occlusion (CTO) procedure. This information is useful for physicians in weighting their selection of collateral vessels for performing a retrograde CTO procedure. For example, if the system indicates that a particular collateral vessel has a relatively short threshold time period compared to other collateral vessels, the physician can decide to select a different vessel with a relatively long threshold time period to safely complete the procedure within its expected duration. If none of the collateral vessels has a threshold time period long enough to accommodate the expected duration of the procedure, the physician may even decide to pursue an antegrade approach to the CTO.

[0122] like Figure 5 and Figure 8a As shown, additional data can also be input into the aforementioned neural networks 170 and 210, and used to predict the threshold time period T. ObsMax For example, EHR data 180 or an indication of interventional device type 230 can also be input into neural networks 170 and 210, thereby generating the predicted threshold time period T based on that data. ObsMax .

[0123] Return to Figure 6 The system is shown. In one example, a warning is generated if the total time the interventional device 200 has obstructed the lumen of a collateral vessel exceeds a threshold time period. This can be done in various ways (including, for example, via a display device, e.g., Figure 3 The monitor 240 shown provides warnings visually or audibly. In another example, the output indicates the amount of time remaining to perform the procedure. Warnings, along with similar indications of the amount of time remaining to perform the procedure, alert the physician to an increased risk of complications during the retrograde CTO procedure. In these examples, one or more processors 310 are configured to: The received data represents the estimated blocking time period T. Obs The input data, wherein the obstruction time period indicates that the interventional device 200 has blocked one or more collateral vessels 150 1··iThe total time of the selected collateral vessels' lumen; and In response to the estimated blocking time period exceeding the threshold time period T ObsMax And output a warning; or The output indicates the amount of time remaining to execute the process, based on a threshold time period T. ObsMax With the blocking time period T Obs It is calculated based on the difference between them.

[0124] In this example, the input data can be provided from various sources. For instance, in one example, the input data includes a track generated by an interventional device tracking system. The tracking data can be generated by an interventional device tracking system configured to track the position of the interventional device during a retrograde CTO procedure. The tracking data can be analyzed to assess the duration of obstruction by tracking the position of the distal end of the interventional device relative to a medical image representing one or more collateral vessels. An interventional device tracking system (e.g., an electromagnetic tracking system, a fiber-optic tracking system) can be used for this purpose. In another example, the input data includes user input data provided by the user. The user input data can be provided in the form of accumulated time from a clock started by the user when the interventional device enters the lumen. In another example, the input data is provided by a robotic system configured to control the interventional device. In yet another example, the input data includes a time series of images, and the time series of images is analyzed to estimate the duration of obstruction T. Obs In this example, one or more processors 310 are configured as follows: Receive second angiography image data 390, the second angiography image data 390 including representations of one or more collateral vessels 150 1··i Time series of images of the interventional device 200 in the selected collateral vessels; Analyze second angiography image data 390 to provide information on the occlusion time period T. Obs The estimated result, wherein the occlusion time period represents the total time that the interventional device 200 has occluded the lumen of the selected collateral vessel; and In response to the estimated blocking time T Obs The time period exceeding the threshold T ObsMax And output a warning; or The output indicates the amount of time remaining to be used to execute the process, based on a threshold time period T. ObsMax With the blocking time period T Obs It is calculated based on the difference between them.

[0125] This example is in Figure 8b The figure is shown in the middle. Figure 8bThe illustration is based on an analysis of some aspects of this disclosure of second angiography image data 390 to provide information on the occlusion time period T. Obs The estimation results and the response to the blocking time period T Obs The time period exceeding the threshold T ObsMax An illustration of an example of generating a warning. Figure 8b In the example shown, the threshold time period T ObsMax This is provided by neural network 210. However, in another example, the threshold time period T... ObsMax Alternatively, it can be made by the above-mentioned Figure 5 The neural network 170 shown is provided.

[0126] In this example, various techniques can be used to analyze the second angiography image data 390 to provide information on the occlusion time period T. Obs The estimation results. In one technique, image processing techniques such as feature detection can be used to detect the location of the interventional device in the time series of images in the second angiography image data 390. In this technique, the occlusion time period T is estimated by detecting the location of the interventional device relative to a selected collateral vessel in the images in the time series and accumulating the total time of the interventional device 200 within the selected collateral vessel. Obs In another technique, a neural network is used to estimate the blocking time T. Obs In this technology, the above reference... Figure 6 One or more processors 310 described are configured to analyze second angiography image data 390 to provide data for the occlusion time period T. Obs The estimation results are as follows: The second angiography image data 390 is input into the neural network 170, which is trained to estimate the occlusion time period T. Obs ;as well as The estimated blocking time period T is output in response to the input. Obs ;and The neural network was trained to estimate the duration of blockage based on second angiography image data 390.

[0127] This example is in Figure 5 The figure is shown in the diagram, and in it, the estimated blocking time T is... Obs As shown in Figure 5 The output is in the right part.

[0128] In this example, neural network 170 can be trained to estimate the blocking time period T by performing the following operations. Obs: Provides multiple time series of angiographic images of interventional devices representing the lumen of collateral vessels as training data, and provides a value representing the occlusion time period T for each time series. Obs (i.e., the total time the interventional device 200 has blocked the lumen of the collateral vessel) is the corresponding baseline ground truth data. Then, the neural network is trained by inputting each time series of angiographic images and using neural network 170 to estimate the occlusion time period T. Obs Furthermore, it is based on the blocking time period T estimated by a neural network. Obs Corresponding blocking time period T from the baseline true data Obs The difference between them is used to evaluate the loss function.

[0129] Note that in the above example, system 300 may be provided with one or more additional items. For example, system 300 may also include one or more of the following: an angiography imaging system 160 for providing angiography image data 320, 390, for example, such as Figure 6 The system shown is a projection X-ray imaging system; an interventional device 200 for performing a retrograde CTO procedure on the occlusion 130; and a monitor 240 for displaying data output by one or more processors 310, such as angiographic image data 320, 390, and identified collateral vessels 150. 1··i Graphical representation of its associated (one or more) fitness scores, etc.; 250 hospital beds; and user input devices (in... Figure 6 (not shown), for example, a keyboard, mouse, touchscreen, etc., and configured to receive user input related to operations performed by one or more processors 310.

[0130] In another example, a method for determining the threshold time period T is provided. ObsMax A computer-implemented method. The threshold time period represents the maximum total time that the interventional device 200 can safely occlude the lumen of a collateral vessel during a retrograde chronic total occlusion (CTO) procedure. The method includes: Receive S310a angiography image data 320, the angiography image data representing an occlusion 130 in a target vessel 140 and one or more collateral vessels 150 for approaching the occlusion in the target vessel in a retrograde direction. 1··i ; The angiography image data 320 is input into the neural network 170 via S320a, which is trained to predict the effects of one or more collateral vessels 150. 1··i Threshold time period T for each collateral vessel ObsMax ;as well as In response to the input, output S330a a threshold time period T for each of the one or more collateral vessels. ObsMax ;and The neural network 170 is trained to predict the threshold time period T for each of one or more collateral vessels based on the input angiography image data. ObsMax ; or: Extract one or more collateral vessels of S310b from angiography image data 320 150 1··i One or more vascular parameters 220 for each collateral vessel in the vessel; The extracted one or more vascular parameters 220 are input into S320b to a neural network 210, which is trained to predict a threshold time period for each collateral vessel in one or more collateral vessels; and In response to the input, output S330b is the threshold time period T for each of one or more collateral vessels. ObsMax ;and The neural network 210 is trained to predict a threshold time period for each of one or more collateral vessels based on one or more input vascular parameters 220.

[0131] In another example, a computer program product is provided. The computer program product includes instructions that, when executed by one or more processors, cause the one or more processors to perform a determination of a threshold time period T. ObsMax The method. The threshold time period represents the maximum total time during which the interventional device 200 can safely occlude the lumen of collateral vessels in a retrograde chronic total occlusion (CTO) procedure. The method includes: Receive S310a angiography image data 320, the angiography image data representing an occlusion 130 in a target vessel 140 and one or more collateral vessels 150 for approaching the occlusion in the target vessel in a retrograde direction. 1··i ; The angiography image data 320 is input into the neural network 170 via S320a, which is trained to predict the effects of one or more collateral vessels 150. 1··i Threshold time period T for each collateral vessel ObsMax ;as well as In response to the input, output S330a a threshold time period T for each of the one or more collateral vessels. ObsMax ;and The neural network 170 is trained to predict the threshold time period T for each of one or more collateral vessels based on the input angiography image data. ObsMax ; or: Extract one or more collateral vessels of S310b from angiography image data 320 150 1··i Each collateral vessel parameter is 220; The extracted one or more vascular parameters 220 are input into S320b to a neural network 210, which is trained to predict a threshold time period for each collateral vessel in one or more collateral vessels; and In response to the input, output S330b is the threshold time period T for each of one or more collateral vessels. ObsMax ;and The neural network 210 is trained to predict a threshold time period for each of one or more collateral vessels based on one or more input vascular parameters 220.

[0132] The examples above should be understood as illustrative of this disclosure, not as limiting it. Other examples are also contemplated. For example, examples describing a system may also be provided as part of a corresponding computer-implemented method, or a corresponding computer program product, or a corresponding computer-readable storage medium. It should be understood that features described with respect to any example may be used alone or in combination with other described features, and may be used in combination with one or more features of another example, or in combination with other examples. Furthermore, equivalents and modifications not described above may be employed without departing from the scope of the invention as defined in the appended claims. In the claims, the word "comprising" does not exclude other elements or operations, and the words "a" or "an" do not exclude a plurality. The mere fact that certain features are recited in dissimilar dependent claims does not indicate that combinations of these features cannot be advantageously used. No reference numerals in the claims should be construed as limiting their scope.

Claims

1. A system (100) for selecting a collateral vessel for performing a retrograde chronic total occlusion, CTO, treatment procedure, the system comprising one or more processors (110) configured to: receive (SI 10) angiographic image data (120) representing an occlusion (130) in a target vessel (140); analyzing (S120) the angiogram image data (120) to identify one or more candidate collateral blood vessels (150) for accessing the occlusion (130) in the target blood vessel (140) in a retrograde direction 1··i ); and assigning (S130) a suitability score to each candidate collateral blood vessel (150 1··i ), the suitability score representing a suitability of the blood vessel for performing the retrograde CTO treatment procedure on the occlusion (130).

2. The system of claim 1, wherein, The one or more processors (110) are further configured to segment the angiogram image data (120) to identify the target vessel (140), and the occlusion (130), and the one or more candidate collateral vessels (150 1··i ); and wherein the one or more processors are configured to analyze the segmented angiogram image data to identify the one or more candidate collateral blood vessels (150 1··i ).

3. The system of claim 1 or claim 2, wherein, the one or more processors (110) being configured to analyze the angiographic image data (120) by: identifying one or more vessels having a connection to a distal location with respect to the occlusion (130), and for each identified vessel: establishing an existence of a path between the connection and a proximal location with respect to the occlusion; and assigning the vessel as a candidate collateral vessel based on the presence of the path (150 1··i ).

4. The system of any one of claims 1-3, wherein, The suitability score of each candidate collateral vessel (150 1··i ) is assessed based on one or more of the following factors: a tortuosity of the candidate collateral vessel; a risk value for the candidate collateral vessel; a complexity for a route for an interventional device to travel along the candidate collateral vessel; an ability to dilate the candidate collateral vessel; electronic medical record data related to the occlusion; blood flow data for the candidate collateral vessel.

5. The system of any preceding claim, wherein, the angiographic image data (120) comprising cardiac angiographic image data, and wherein the candidate collateral vessels comprise: one or more septal collateral vessels (150i, 1502), and / or one or more epicardial collateral vessels (1503), and / or one or more bypass graft vessels (1504).

6. The system of any preceding claim, wherein, The one or more processors (110) are also configured to output an indication of a recommended orientation of a projection X-ray imaging system (160) for acquiring projection X-ray images of one or more of the identified candidate collateral blood vessels (150 1··i ) during the CTO treatment procedure.

7. The system of any preceding claim, wherein, The one or more processors (110) are also configured to analyze the angiogram image data (120) to identify one or more candidate collateral blood vessels (150 1··i ) and / or assign a suitability score to each candidate collateral blood vessel by: inputting the angiographic image data (120) into at least one neural network (170); and outputting the identified one or more candidate collateral vessels (150 1··i ) and / or the assigned one or more suitability scores in response to the input; and wherein the at least one neural network (170) is trained to, based on the inputted angiographic image data, identify the one or more candidate collateral vessels for accessing the occlusion in the target vessel in a retrograde direction and / or assign a suitability score to each candidate collateral vessel.

8. The system of claim 7, wherein, the at least one neural network (170) is trained to, based further on electronic medical record data (180) related to the occlusion (130), identify the one or more candidate collateral vessels and / or assign a suitability score to each candidate collateral vessel, and wherein the one or more processors (110) are further configured to: receive electronic medical record data (180) related to the occlusion (130); and input the electronic medical record data into the at least one neural network (170); such that the outputted one or more candidate collateral vessels and / or the assigned one or more suitability scores are further based on the electronic medical record data.

9. The system of claim 7 or claim 8, wherein, the at least one neural network (170) is trained to identify the one or more candidate collateral vessels and / or assign a suitability score to each candidate collateral vessel by: receiving training data comprising, for each of a plurality of subjects, angiographic image data representing an occlusion in a target vessel; receiving ground truth data corresponding to the training data, the ground truth data including, for each of the plurality of subjects, an identification of one or more collateral blood vessels proximal to the occlusion in the target blood vessel for use in a retrograde CTO treatment procedure and / or a suitability score for each collateral blood vessel, the suitability score representing a suitability of the blood vessel for use in performing a retrograde CTO treatment procedure on the occlusion; and for each of the plurality of subjects: inputting the training data into the at least one neural network; and adjusting parameters of the at least one neural network based on a value of a loss function representing a difference between the one or more collateral blood vessels identified by the at least one neural network and / or the suitability score assigned to each collateral blood vessel by the at least one neural network and corresponding ground truth data; and repeating the inputting and the adjusting until a stopping criterion is met.

10. The system of any one of claims 7-9, wherein, the at least one neural network (170) is further trained to predict, based on the inputted angiogram image data (120), a recommended type of interventional device for use in the CTO treatment procedure; and wherein the one or more processors (110) are further configured to output, in response to the input, the recommended type of interventional device for use in the CTO treatment procedure.

11. The system of any preceding claim, wherein, the one or more processors (110) are further configured to: receive second angiogram image data (190) representing an interventional device (200) in the selected collateral blood vessel; analyzing the second angiogram image data (190) to provide an estimate of an occlusion time period (T Obs ) representing a total time during which the interventional device (200) has occluded the lumen of the selected collateral vessel; and outputting a warning in response to the estimated period of blockage exceeding a threshold period of time (T ObsMax ) or outputting an indication of an amount of time remaining available to perform the procedure, the amount of time remaining being calculated based on a difference between the threshold time period (T ObsMax ) and the blocking time period (T Obs ).

12. The system of claim 11, wherein, the one or more processors (110) are further configured to: inputting the angiogram image data (120) and / or the second angiogram image data (190) into a neural network (170) trained to predict the threshold period of time (T ObsMax ) for the selected collateral vessel; and outputting the threshold period of time (T ObsMax ) in response to the input; and wherein the neural network (170) is trained to predict, based on the inputted angiogram image data (120) and / or the second angiogram image data (190), the threshold time period for the selected collateral blood vessel; or wherein the one or more processors (110) are further configured to: extract one or more vessel parameters (220) of the selected collateral blood vessel from the angiogram image data (120); inputting the extracted one or more vessel parameters (220) into a neural network (210) trained to predict the threshold period of time (T ObsMax ) for the selected collateral vessel; and outputting, in response to the input, the threshold period of time (T ObsMax ) for the selected collateral blood vessel; and wherein the neural network (210) is trained to predict the threshold time period (T ObsMax ) for the selected collateral vessel based on the input one or more vessel parameters (220).

13. The system of claim 12, wherein, The neural network (170) is trained to predict the threshold time period (T ObsMax ) also based on electrocardiogram data for the occlusion; and wherein the one or more processors (110) are further configured to: receive electrocardiogram data generated during the CTO treatment procedure; input the electrocardiogram data into the neural network; and The threshold time period (T ObsMax ) is also output based on the electrocardiogram data.

14. The system of claim 13, wherein, The neural network (170) is trained to predict the threshold time period (T ObsMax ) also based on a type of an interventional device used in the CTO treatment procedure; and wherein the one or more processors (110) are further configured to: receive device data (230) including an identification of a type of interventional device used in the CTO treatment procedure; input the device data (230) into the neural network; and outputting the threshold time period (T ObsMax ) based on the type of the interventional device.

15. The system of any of claims 12-14, wherein, The neural network (170) is trained to predict the threshold time period (T ObsMax ): receive threshold time period training data including, for each of a plurality of subjects, angiogram image data representing: a collateral blood vessel for use in performing a retrograde CTO treatment procedure, or an extracted vessel parameter (220) of a collateral blood vessel for use in performing a retrograde CTO treatment procedure; receive ground truth data corresponding to the threshold time period training data, the ground truth data including, for each of the plurality of subjects, an indication of a total time for which a lumen of the collateral blood vessel can be safely occluded by the interventional device before a medical complication occurs; and receive threshold time period training data including, for each of a plurality of subjects, angiogram image data representing: a collateral blood vessel for use in performing a retrograde CTO treatment procedure, or an extracted vessel parameter (220) of a collateral blood vessel for use in performing a retrograde CTO treatment procedure; For each object in the plurality of objects: inputting the threshold time period training data into the neural network (170, 210); and adjusting parameters of the neural network based on a value of a loss function representing a difference between the threshold time period predicted by the neural network and a corresponding total time from the ground truth data; and repeating the inputting and the adjusting until a stopping criterion is met.

Images