Identification of vascular bifurcations

Abstract

In part, this invention relates to an automated method for detecting branching vessels in intravascular imaging using intravascular modal imaging such as OCT, IVUS, or other imaging modalities. In one embodiment, representations of A-lines and frames generated by the intravascular imaging system are used to identify candidate branches of the vessel. One or more operators (such as filters) can be applied to remove false positives associated with other detections.

Description

Technical Field

[0001] The present invention generally relates to systems and methods applicable to the field of intravascular diagnosis and imaging, and more specifically to systems and methods supporting the identification of collateral vessels, intersections or other parts or features of blood vessels. Background Technology

[0002] Coronary artery disease is one of the leading causes of death worldwide. Better capabilities in diagnosing, monitoring, and treating coronary artery disease can be crucial for saving lives. Intravascular optical coherence tomography (OCT) is a catheter-based imaging modality that uses light to visualize the walls of coronary arteries and generate images for study. Utilizing coherent light, interferometry, and micro-optics techniques, OCT provides in vivo tomography with video rates at micrometer-level resolution within diseased vessels.

[0003] The use of fiber optic probes to observe subsurface structures at high resolution makes OCT particularly useful for minimally invasive imaging of internal tissues and organs. OCT enables this level of detail, allowing clinicians to diagnose and monitor the progression of coronary artery disease. OCT images provide high-resolution visualization of coronary artery morphology and can be used alone or in combination with other information, such as angiographic data and patient data from other sources, to aid in diagnosis and treatment.

[0004] OCT imaging of various parts of a patient's body provides doctors and others with a useful diagnostic tool. For example, coronary artery imaging via intravascular OCT can show the location of narrowing or stenosis that reduces blood flow and increases the risk of ischemia. This information helps cardiologists choose between invasive coronary artery bypass grafting and catheter-based minimally invasive procedures (such as angioplasty or stent delivery) to relieve stenosis and restore blood flow. The presence of collateral arteries in the narrowed area also affects blood flow through the artery and is therefore an important factor in designing a patient's treatment plan.

[0005] Quantitative assessment of vascular pathology and its progression involves the calculation of various quantitative measures, such as pressure drop, which can depend on the accurate identification of fluid volume and luminal geometry (including collateral geometry). Collaterals extending from the lumen are often difficult to identify in OCT images. This is partly because collaterals may be obscured by guidewires used in various OCT probes, or by stent struts, blood, and shadows.

[0006] Furthermore, shadows and other imaging artifacts are difficult to resolve and eliminate. As a result, important landmarks along the length of the artery (such as collateral branches) may be misidentified as tissue or not identified at all. Given that placing a stent on a collateral branch would be harmful, or should be done intentionally during the procedure, a reliable technique for identifying collateral branches is needed.

[0007] This invention addresses these and other challenges. Summary of the Invention

[0008] Partially, the present invention relates to a method for detecting one or more branches of a blood vessel. The method includes storing one or more intravascular image datasets of blood vessels, each intravascular dataset including a plurality of A-lines; detecting a luminal boundary in a first A-line image generated from a set of A-lines from the plurality of A-lines, wherein the first A-line image has an r-dimensional dimension and an A-line dimension; specifying a search distance T; defining a search region bounded by the detected luminal boundary and a boundary offset at a distance T from the luminal boundary; detecting edges of the search region; and identifying candidate branch regions in response to the detected edges.

[0009] In one embodiment, the method includes flattening an A-line image using a first image processing operator; applying median smoothing to the A-line image using a second image processing operator; and applying smoothing to the A-line image using a third image processing operator to generate a filtered image. In one embodiment, the method includes identifying a first minimum-maximum pair in the filtered image, wherein one or more distances between the first minimum-maximum pair define a first search window. In one embodiment, the method includes identifying a second minimum-maximum pair in the filtered image, wherein one or more distances between the second minimum-maximum pair define a second search window.

[0010] In one embodiment, the method includes searching along the r-dimensional dimension of a corresponding preprocessed input image within a first search window. In one embodiment, the method includes designating pixels located within the first search window below a noise floor threshold as corresponding to the candidate branch region. In one embodiment, the noise floor threshold is less than approximately 2 mm. In one embodiment, the method includes dividing the candidate branch region into three bands, wherein the sum of the widths of the three bands is equal to T. In one embodiment, the method includes, for each band, accumulating pixels corresponding to the candidate branch region along each A-line.

[0011] In one embodiment, the method includes marking the A-line in the band as corresponding to a branch if a particular A-line has more than about 10% to about 30% of pixels marked as candidate branches. In one embodiment, the method includes outputting a set of A-lines corresponding to candidate branches for each band.

[0012] In one embodiment, the method includes generating a branch matrix using a pulled-back frame, the frame including A-line and angle data. In one embodiment, the method includes isolating pixels corresponding to groups of all three bands and pixels corresponding to groups of the first two bands to select pixels corresponding to side branches. In one embodiment, the method includes removing guidewire regions from the branch matrix. In one embodiment, the method includes eliminating branches that appear only in one frame. In one embodiment, the method includes replicating the branch matrix to account for overlap across zeros.

[0013] In one embodiment, the first band ranges from 0 to T / 3, the second band ranges from T / 3 to 2 / 3T, and the third band ranges from 2 / 3T to T. In another embodiment, the first band ranges from 0 to T / 3, the second band ranges from T / 3 to 2 / 3T, and the third band ranges from 2 / 3T to T. In one embodiment, the method includes displaying one or more detected side branches in a user interface. In another embodiment, the method includes validating one or more candidate side branches using a branch matrix generated using pixels selected from two or more bands, wherein the sum of the bands is T.

[0014] Other features and advantages of the disclosed embodiments will become apparent from the following description and accompanying drawings. Attached Figure Description

[0015] This patent or application document contains at least one color drawing. A copy of this patent or application publication with one or more color drawings will be provided by the competent authority upon request and payment of the necessary fees.

[0016] The accompanying drawings are not necessarily drawn to scale, but rather generally focus on illustrating the principles. These drawings should be considered illustrative in all respects and are not intended to limit the invention; their scope is defined only by the claims.

[0017] Figure 1 This is a diagram of an intravascular imaging system, including an automated collateral detection module according to an illustrative embodiment of the present invention.

[0018] Figure 2A This is a polar coordinate A-line image of a region of interest in a blood vessel with collateral branches according to an illustrative embodiment of the present invention.

[0019] Figure 2B This is an illustrative embodiment of the present invention. Figure 2A The cross-sectional image of the blood vessel corresponding to the A-line image.

[0020] Figure 3 This is an illustrative embodiment of the present invention. Figure 2A The image shown is a flattened version.

[0021] Figure 4 This is an image illustrating the edge detection of a side branch according to an illustrative embodiment of the present invention.

[0022] Figure 5A , Figure 5B , Figure 5C and Figure 5D These are flattened image A-line images after image processing according to an illustrative embodiment of the invention, and various plots relative to a noise substrate for different r values, wherein the radial dimension r is perpendicular to the A-line dimension.

[0023] Figure 6 Images show various bands or zones selected for searching for collateral vessels, wherein the bands or zones are specified based on different depths from the lumen.

[0024] Figure 7 It is a cross-sectional OCT view, an L-mode or longitudinal OCT view, and a branch matrix with candidate branches and guidewires.

[0025] Figure 8 It is a branch matrix according to an illustrative embodiment of the present invention, which is parsed to identify individual branch candidates processed or operated according to the branch matrix. Detailed Implementation

[0026] In part, this invention relates to an automated method for detecting branches of blood vessels imaged using intravascular modalities such as OCT, IVUS, or other imaging modalities. The term branch refers to one or more branches of a blood vessel (such as collateral branches). In one embodiment, the invention relates to branch detection as an intermediate step in a pipeline of software modules, operators, and stages. Each stage transforms the intravascular data and performs feature detection (such as shadow and lumen detection) on it. Branch detection can be performed after OCT or IVUS retraction, and the resulting intravascular data can be processed using a lumen detection software module to extract lumen data (such as information related to lumen boundaries).

[0027] In part, the present invention relates to various methods for collecting and processing data, such as intravascular data frames. In one embodiment, the intravascular data frame or image data frame includes a cross-sectional image generated from multiple A-lines (scan lines) obtained using a rotatable intravascular probe. The collection of scan lines as the probe rotates forms a cross-sectional image of the blood vessel.

[0028] In one implementation, shadow detection is performed prior to branch detection to identify regions of interest from potential intravascular data. Shadows are of interest because they can correspond to different features such as blood pools, branches (such as collaterals), and guidewire segments. Guidewire segments are generated by guidewires used to position intravascular imaging probes within the artery. In one implementation, once a guidewire (or multiple guidewires) has been identified and validated, location markers or pixel markers of the guidewire generated on a given frame or scan line can be provided to other intravascular data processing modules. As an example, validated guidewire detection can be input to a collateral detection module. The process of detecting collaterals can also be input to other processing levels to generate information of interest regarding intravascular pullback data.

[0029] In part, this invention describes various methods and sub-methods related to branch detection and the evaluation of associated parameters. In one embodiment, the method is an automated approach that operates on intravascular data based on user interface input to detect collaterals or as part of other image processing that uses collateral detection as input.

[0030] As arteries gradually narrow, collateral branches in the coronary arteries can be used to simulate the normal diameter of the artery at each segment. The location and diameter of the collateral branches are important inputs for calculating and predicting flow along the arteries.

[0031] A novel software algorithm has been developed that automatically detects the location of lateral branches in OCT images and provides an estimate of their diameter. The algorithm identifies frames and scan lines within OCT frames that are part of the branch regions.

[0032] In one implementation, a software-based method works against scan lines in polar coordinate space and uses a combination of image processing filters and algorithms to detect the rising and falling intensity gradients of the sidewalls. In one implementation, the software generates a matrix from the scan line data, organized based on frames along the pullback. The matrix includes data from the pullback, which collects information on offsets beyond the lumen or other distances extending into the tissue or sidewalls. This matrix (branching matrix) is parsed to obtain information about possible branch locations, and the matrix is ​​used to measure the branch diameter.

[0033] like Figure 1As shown, a data collection system 30 for collecting intravascular data includes a data collection probe 7 for imaging the blood vessel. This system can be an OCT system, an IVUS system, or other systems based on intravascular imaging modalities. A guidewire is used to introduce the data collection probe 7 into the blood vessel. The data collection probe 7 can be introduced and pulled back along the blood vessel while data is collected. As the imaging probe retracts (pulls back) along the length of the blood vessel, multiple scan datasets or intravascular datasets (OCT data, IVUS data, or other data) are collected as the probe or a portion thereof rotates. In one embodiment, this is referred to as pull-back.

[0034] In one implementation, the collection of these datasets or image data frames can be used to identify regions of interest (such as stenosis or deployed stents). In one implementation, the data collection probe 7 is an OCT probe. Probe 7 may include a probe tip 17. When the OCT probe is used as probe 7, it is configured for use with a version of system 10 that includes an interferometer and a data processing system. Distance measurements collected using the data collection probe 7 can be processed to generate image data frames (such as cross-sectional views or longitudinal views (L-mode views) of blood vessels). For clarity, cross-sectional views may include, but are not limited to, longitudinal views. These images can be processed using one or more image data processing modules or stages.

[0035] A data collection probe 7 is shown before or after insertion into a blood vessel. The data collection probe 7 is in optical communication with the OCT system 10. The OCT system 10, connected to the data collection probe 7 via an optical fiber 15, may include: a light source (such as a laser), an interferometer with a sample arm and a reference arm, various optical paths, a clock generator, a photodiode, and other OCT system components. The probe 7 is positioned in an artery 8 having a branch B and a blood pool BP.

[0036] In one embodiment, optical receiver 31 (such as a system based on balanced photodiodes) can receive light emitted from data collection probe 7. Computing device 40 (such as a computer, processor, ASIC, or other device) can be part of OCT system 10 or can be included as a separate subsystem communicating electrically or optically with OCT system 10. Computing device 40 may include storage device 41(s), memory, bus, and other components suitable for processing data, as well as software components 44 (such as an image data processing level configured for stent visualization, stent adhesion defect detection, lumen detection, offset generation, search region 151 definition, lateral branch detection 45, guidewire detection, branch matrix generation, pullback data collection, etc.). Although branch detection module 45 is shown as a separate software module, it can also be one of the software components 44. Branch matrix generation software can be part of branch detection module 45 or a separate software module.

[0037] In various embodiments, the computing device 40 includes or accesses software modules or programs 44, such as a side branch detection module, a guidewire detection module, a lumen detection module, a stent detection module, a median mask removal module, a strength averaging module, a stent poor adhesion detection module, a bulge detection module, and other software modules. For example, the computing device 40 may access a side branch detection module 45 to detect side branches. In particular, the module is calibrated to use specific branch characteristics as signature features to improve branch accuracy.

[0038] In one implementation, the collateral detection module 45 generates or manipulates a two-dimensional branch matrix and uses the matrix or other methods as described herein to isolate candidate collaterals. In one implementation, branch characteristics may include the arrangement of features detected within the blood vessel, such as a noisy base and ascending or descending gradients. The software module or program 44 may include an image data processing pipeline or component modules thereof, and one or more graphical user interfaces (GUIs).

[0039] Exemplary image processing pipelines are used to convert collected intravascular data into two-dimensional and three-dimensional views of blood vessels and stents. The image data processing pipelines or any methods described herein are stored in memory and executed using one or more computing devices, such as processors, devices, or other integrated circuits.

[0040] In one embodiment, software module 44 further includes additional features related to blood flow detection, or includes features that replace collateral detection. In one embodiment, the diameter of one or more collaterals and predicted blood flow through these collaterals are included. Software module 44 may also include or communicate with a user interface software component to turn collateral blood flow views on and off and display and switch various user interface display modes, such as stent planning, flythrough, and other observation modes described herein.

[0041] like Figure 1 As shown, the display 46 may also be part of the intravascular data collection and processing system 10 for displaying information 47 (such as cross-sectional and longitudinal views of the blood vessel generated using the collected image data).

[0042] The data acquisition system 10 can be used to display and associate blood flow-related image data with collateral vessels of detected vessels. In one embodiment, one or more steps may be performed automatically or without user input other than initial user input to navigate about one or more images, input information, select an input (such as a controller) or user interface component, or interact with an input (such as a controller) or user interface component, or otherwise indicate the output of one or more systems. In one embodiment, the blood flow view is presented as an option to select a two-dimensional or three-dimensional view for easy viewing of a representation of the vessel and one or more collateral vessels. Switching between one or more viewing modes in response to user input can be performed with respect to the various steps described herein.

[0043] A representation of the stent and lumen boundaries, such as their OCT or IVUS images, can be displayed to the user via display 46. Lateral branch detection, shadow detection, and stent detection are performed before displaying these features and any coding or labeling with identification markings that may be included in the displayed images. One or more graphical user interfaces (GUIs) can be used to display OCT-based information 47. Images are examples of information 47 that can be displayed and interacted with using a GUI and various input devices.

[0044] Additionally, this information 47 may include, but is not limited to, cross-sectional scan data, longitudinal scans, diameter maps, image masks, shaded areas, stents, poorly adhered areas, lumen boundaries, vertical distances measured relative to automatically detected lumen boundaries, vertical distances extending from the lumen boundaries with distance T, and other images or representations or potential distance measurements of the blood vessels obtained using an OCT system and data collection probes.

[0045] The computing device 40 may also include software or program 44, which may be stored in one or more storage devices 41 and configured to identify stent struts and poor apposition levels (such as comparisons based on thresholds and measured distances) and other vascular features using text, arrows, color coding, highlights, outlines or other suitable human or machine-readable markings.

[0046] According to one embodiment, the display 46 depicts various views of the blood vessel. The display may include menus for showing or hiding various features, such as a menu for selecting which blood vessel features to display, and a menu for selecting the virtual camera angle of the display. The user can switch between multiple viewpoints on the user's display. Additionally, the user can switch between different collateral branches on the user's display, such as by selecting a specific collateral branch and / or by selecting a view associated with a specific collateral branch.

[0047] For example, a user can select an orifice view, which in one implementation could be a default view or a carina / carina view, allowing them to view the carina of one or more collateral branches. In one implementation, the image processing pipeline and associated software modules use data collected during pullback to detect luminal boundaries, guidewires, other shadows, stents, and collaterals in the imaged artery.

[0048] For example, the lumen boundary can be detected using distance measurements obtained from optical signals collected at the probe tip 17 using a lumen detection software component or module. Instead of optical fibers, an ultrasound transducer can be adapted to collect IVUS signals about the vessel wall and one or more stents.

[0049] Lumen detection software may include one or more steps. For example, in one embodiment, to perform lumen detection, a filter or other image processing device may be applied to a two-dimensional image to detect edges in the image that represent lumen boundaries. In another embodiment, a scan-line-based method is used. During one or more pull-backs, optical or ultrasound signals are collected as scan lines relating to the blood vessel and one or more stents arranged in the lumen of the blood vessel. In one embodiment, lumen detection software that enables a computing device to generate one or more images from the collection of scan lines using the computing device.

[0050] Furthermore, lumen detection may include generating a binary mask of a blood vessel image using a computing device, wherein the binary mask is generated using an intensity threshold. As another step, multiple scan lines are defined in the binary mask. Regarding each of the multiple scan lines, in one embodiment, a region on that scan line is identified as lumen boundary tissue. A contour segment of the boundary is identified based on the presence of the lumen boundary tissue region. In one embodiment, the method identifies adjacent contour segments. The lumen boundary detection method may also include interpolating missing contour data between adjacent contour segments. As a result, in one embodiment, adjacent contour segments and interpolated missing contour data define the lumen boundary.

[0051] Once intravascular data (such as frames and scan lines from the pullback) is acquired using a probe and stored in memory 41, this intravascular data can be processed to generate information 47 (such as cross-sectional views, longitudinal views, and / or three-dimensional views or subsets thereof of the vessel along the length of the pullback region). These views can be depicted as part of the user interface shown. The vascular images generated using distance measurements obtained from the intravascular data acquisition system provide information about the vessel and objects arranged within it.

[0052] Therefore, in part, the present invention relates to software-based methods, related systems, and devices suitable for evaluating and depicting information about blood vessels, stents, or other vessels of interest. Intravascular data can be used to generate 2-D views (such as cross-sectional and longitudinal views of the vessel) before or after initial stent deployment, or before or after procedures related to stent correction. Intravascular data obtained using data collection probes and various data processing software modules can be used to identify, characterize, and visualize stents and / or one or more properties associated with the stent and / or the lumen in which it is placed.

[0053] The location of the stent relative to and in relation to collateral openings within the vessel wall can be visualized, ensuring that the collateral openings are not obstructed by the stent. In one implementation, collaterals are identified and visualized to assist in treatment planning and stent placement.

[0054] Figure 2A This is a polar A-line OCT image of a region of interest in a blood vessel with collateral branch 102. The vessel lumen 100 is located at the top of the image. As shown, the lumen boundary 106 (i.e., the edge of the vessel wall) is demarcated by a dashed line, and the vessel wall tissue 120 fills most of the image. The guidewire shadow 104 is a vertical feature on the right side of the image. In the image, the lumen boundary 106 provides the brightest response, and the OCT signal attenuates to the depth of tissue penetration. Collateral branch 102 appears as a vertical shadow or a signal-free area in the vessel wall. In part, this invention relates to the detection of collateral branches such as branch 102.

[0055] In one implementation, guidewire detection is first performed to exclude shadows and guidewire segments from collateral shadows, thereby improving detection accuracy. In various implementations, the intravascular data acquisition system and related software modules detect branching characteristics in OCT image data within a predetermined scan depth region T. The T value can be used to define the search region 151. The T value can be entered via a graphical user interface.

[0056] In one embodiment, T is specified as approximately 660 μm and used as an offset from the lumen to define the offset lumen boundary demarcated by line 108. In one embodiment, the region of interest / search area for branch detection is defined by line 108 and lumen boundary 106, which is the lumen boundary 106 shifted by a distance T. A curved bar or band 151 of width T, defined by the dashed line (lumen boundary 106) and the shifted lumen boundary 108, specifies a subset of the intravascular image data for searching for collateral branches. The range of T can be from approximately 500 μm to approximately 800 μm. Figure 2B Is with Figure 2A The A-line image 85 corresponds to the cross-sectional image 130 of the blood vessel. In Figure 2BThe image shows the lumen boundary 106 in relation to probe 7 and side branch 102. A portion of the search region 151 adjacent to side branch 102 is shown. Region 102 is an example of a branch region that can be detected using one or more methods and systems described herein.

[0057] Branch detection implementation methods

[0058] The software-based branching method first scans between a descending response gradient 112 and an ascending response gradient 114 for a branch signature feature or pattern defined by a noise base (also referred to as NF) 110 or a no-signal region. In one implementation, the noise base is a region or area where the tissue intensity has decreased to the same value as in a clear lumen. The noise base 110 may correspond to a transition region between the ascending and descending intensity gradients in the search region or area. Next, any image frames that match the signature feature or pattern are marked as candidate branch regions. The candidate branch regions are then combined across all frames. In turn, optionally, the branch regions are resolved with respect to the branch matrix, and the diameter of each lateral branch is estimated. The software-based branching method is described in more detail herein.

[0059] First, after the pullback, the original A-line image is preprocessed to flatten the image, making it easier to identify the lateral branches. Figure 3 This is intravascular image 125. Figure 2A The image shown is a flattened version. Using the detected lumen boundary ( Figure 2A Image 106 is used as the top edge of the flattened image to flatten the A-line image. Median and Gaussian image smoothing can also be used to filter image 127, such as... Figure 4 As shown.

[0060] analyze Figure 3 The preprocessed image 125 is used to detect the noise floor (NF). The noise floor (NF) can be used as a threshold to identify pixels below the noise floor as candidate branch regions. The noise floor threshold is calculated based on image histograms from different samples or known OCT noise levels. In one embodiment, the noise floor is determined to be approximately 1.5 mm to approximately 2.5 mm from the lumen in the T-offset direction. In another embodiment, the noise floor is approximately 2 mm.

[0061] In one implementation method Figure 3 The flattened image 125, containing a portion of the candidate side branches, is then processed using an edge detection filter 133 or filter F1 or F2. To reduce data processing and improve efficiency, filtering is preferably performed on the data corresponding to a predetermined scan depth T. Figure 4In this method, the scanning depth T is 660 μm from the lumen boundary, but any depth matching the OCT can be used. Image 127 is filtered using filter 133 to generate a filtered image 135. Edge detection filters can be applied to identify the left edge 126 and right edge 128 of the side branches and guidewire shadows. In one embodiment, filter 133 is a Gaussian filter. In one embodiment, filter 133 is two filters, such as filters 133a and 133b, which are also referred to as first filter F1 and second filter F2. In one embodiment, filters 133a and 133b are one or more separable edge-finding filters. In one embodiment, the edge-finding filter used herein is a boxcar type filter.

[0062] Furthermore, in one embodiment, the filter is a smoothing filter. In one embodiment, filter 133 is an edge-finding filter. In one embodiment, filter 133 is a combination of a smoothing filter and an edge-finding filter (which may be filters F1 and F2).

[0063] As an additional step, along as... Figure 5A Image 147 shows various r-value analyses of the edge-filtered image (such as...). Figure 4 Image 135). In Figure 5A Then, image 147 is analyzed for all or a subset of all A-lines for each r-dimensional value to detect candidate branch regions. Next, image 147 is filtered along the value of r=20 for each r value or a subset thereof. In this way, Figure 5B The filter is shown in Figure 149. Figure 5B The local minimum-maximum peaks in the filtered image 149 for r=20. In one implementation, the process is repeated for all r values ​​or subsets of r values ​​to generate a set of local minimum-maximum peaks.

[0064] like Figure 5A As shown, the region of interest (ROI) 128 relative to the cavity 126 is illustrated. The ROI corresponds to the portion of the cavity 126 detected by the displacement T. Figure 5A and Figure 5C The vertical axis corresponds to dimension r, where a dashed line is shown at r = 20, passing through a pair of lumen signals 126 and a pair of ROI signals 128. Between the minimum-maximum pair indices, it is formed by... Figure 5B The filtered image in the figure and Figure 5D The circled maximum and minimum values ​​in the intensity map of the preprocessed input image are shown.

[0065] about Figure 5B The vertical axis of the diagram is formed by... Figure 5A The filter output is generated by filtering, and the horizontal axis is the A-line. Figure 5D yes Figure 5C The intensity map of the preprocessed image. Figure 5D In this context, the vertical axis represents the intensity of the input image. In one implementation, the system searches for all or a subset of the r values ​​across the A-line of the corresponding preprocessed input image along the "r" dimension.

[0066] Figure 3 The noise floor is also Figure 5D The intensity map is shown. A noise base threshold is set, and candidate lateral branches 138 below this threshold are identified. In one embodiment, the noise base threshold is based on an estimate of the noise base and is set such that samples near the noise base are smaller than the threshold, while ensuring that tissue samples are larger than the threshold. The system uses one or more software components to find pixels below the noise base threshold in the flattened image ( Figure 5C In the corresponding area of ​​) Figure 5D As specified in drawing 161. (See reference) Figure 5A The edge-filtered image 147 is filtered to identify local maxima and minima at "r = 20". In one implementation, the filtering and plotting of the filtered results are performed on all or a subset of the r values. The r values ​​along... Figure 5A and Figure 5C The vertical axis in. Figure 5B In the diagram, the x-axis corresponds to line A-number, and the y-axis corresponds to r. (Reference) Figure 5B As shown in the figure, the min-max method identifies the respective local min-max pairs 130a, 130b and 132a, 132b.

[0067] In addition, regarding Figure 5B The method searches within plot 149 in the range of values ​​134 between each minimum-maximum pair. Figure 5C The corresponding flattened image 160 contains a noise floor. Intravascular data from the flattened image are plotted as A-lines (x-axis) against intensity (y-axis) to identify scan lines containing the weakest OCT response. Regions in the flattened image falling below a predetermined threshold 136 are considered candidate branch regions 138. In one implementation, the threshold 136 is the noise floor (NF). Therefore, the descending response gradient ( Figure 2A Local minima 130a and 132a (the set of circled values ​​below) are found in the ascending response gradient (112). Figure 2A Find the local maxima 130b and 132b (the set of values ​​circled above) in 114).

[0068] In one implementation, the appearance of peaks and valleys corresponds to changes in the gradient intensity defining the search space. If a signal below the noise floor exists in the search space, the corresponding pixel corresponds to a candidate branch region. These candidate regions are then analyzed to determine whether they are valid side branches.

[0069] In one implementation, such as Figure 6 As shown, candidate regions are divided into search bands or zones. In this way, specifying the band / zone to search helps determine candidate branch regions. In one implementation, for each A-line, it is divided into band 1, band 2, or band 3. In one implementation, the band is part of T. These bands identify which set of A-lines is potentially associated with a lateral branch. Each A-line is viewed at three or more different depths, where the depth is shown as... Figure 6 The dividing zone. There are three equal depths of 220 μm, which is equivalent to 1 / 3 of T when T is 660 μm.

[0070] In one implementation, the region is divided or subdivided into three bands (band 1, band 2, and band 3) as part of the selection / specification of the region to be searched. In one implementation, each band is processed individually. These bands can also be processed in parallel or differently, with features of one band compared to features in one or more other bands. Although three bands are shown, one, two, or more bands can be specified for candidate branch search.

[0071] In one implementation, for each specified search band, the method accumulates marked pixels along each A-line. If a particular A-line has more than 10%-35% of pixels, then the A-line in that band is marked as a branch. This method indicates that at least 10%-35% of the pixels in the search area are at or below the noise floor.

[0072] In one implementation, a software module is used to parse the branch matrix to isolate candidate branch regions that are most likely branches. This step is performed based on one or more rules for all three branch regions. In some implementations, the guidewire is removed.

[0073] In one implementation, blood pooling and other sources of false positives are addressed. Thrombosis, blood pooling, and other artifacts are the main causes of false positives. Blood attenuates the signal, which can simulate branching regions. An intensity threshold is calculated for each frame to identify pixels with blood pooling.

[0074] The obstruction index is calculated based on the number of blood stasis pixels detected within the lumen for each detected branch. This index correlates with blood stasis and thrombus accumulation within the lumen and provides a score for each detected branch. The index is high when there is blood or thrombus within the lumen that causes significant signal attenuation. Branches with high obstruction indices are rejected as false positives. Those within the acceptable range are retained as true positives. Branches with moderate obstruction indices can be marked for further review.

[0075] Figure 7 This includes cross-sectional OCT views, L-mode views, or longitudinal OCT views, as well as a branch matrix with candidate branches and guidewires. Color-coded legends for various detected depths are also shown. Typically, the depth from the lumen is associated with a color and / or letter representing the depth range. About Figure 7 The resolved branch matrix shows that red, or R, ranges from approximately 0 μm to approximately 220 μm. Subsequently, green, or G, ranges from approximately 220 μm to approximately 440 μm. Blue, or B, ranges from approximately 440 μm to approximately 660 μm. Yellow, or Y, ranges from approximately 0 μm to approximately 440 μm. Finally, cyan, or C, ranges from approximately 200 μm to approximately 600 μm. White, or W, corresponds to all three bands (band 1, band 2, and band 3) and the total distance T from the lumen, such as 0–660 μm. These color-coded markings are applicable to… Figure 7 and Figure 8 In various implementations, other markings and symbols may be used to indicate band groupings in addition to colors.

[0076] Branch matrix 250 and branch matrix 300 (respectively) Figure 7 and Figure 8 This shows branch candidate information from all frames for all A-lines and all three zones (band 1, band 2, and band 3). The following legend regarding the relationships between the different bands is applicable to the band grouping and color scheme. Figure 7 Branch matrix 250 and Figure 8 The branch matrix is ​​300. The colors are also identified by line connectors and the first letter of the color, but there are no arrows to distinguish the use of the letter AE to show candidate side branches.

[0077] Branch matrix diagram (distance from lumen / band group, e.g., T = 660 μm):

[0078] • Band 1 (0-220μm): Red (R)

[0079] • Band 2 (220μm-440μm): Green (G)

[0080] • Band 3 (440μm-660μm): Blue (B)

[0081] • Bands 1, 2, and 3 (0-660μm): White (W)

[0082] • Bands 1-2 (0-440μm): Yellow

[0083] • Bands 2-3 (220μm-660μm): Cyan

[0084] Figure 8 These are branch matrix 300 and the processed matrix 305. Matrix 300 is operated on to remove guidewire data. After removing the guidewire data, matrix 305 shows candidate branches A, B, C, D, and E with the guidewire data removed. In images 300 and 305, these branches are identified by multiple arrows with arrowheads. In one implementation, one or more morphological operators are applied to further isolate and enhance matrix 305, making the branches more clearly defined, as shown in 305 in the figure. Data from branch matrix 300 can be compared with other cross-frame data to improve detection accuracy. In one implementation, a given branch matrix is ​​sequentially combined on a frame-by-frame basis.

[0085] about Figure 8 The matrix 300 can be processed and manipulated to generate a processed or filtered branch candidate matrix 305. These steps may include one or more of the following processing or filtering steps. A method for detecting branches using the branch matrix may include generating the branch matrix. This method may include isolating white W (all three bands) pixels and yellow Y (the first two bands) pixels adjacent to the white pixels. This method may include removing the guide wire region.

[0086] The method may also include eliminating branches that appear only in a single frame. In one implementation, assuming the angle spans 360 degrees, depending on the orientation of the matrix (which is based on a cylindrical arrangement) and the overlap of the matrix ends, portions of the matrix may be copied to cover the upper and lower horizontal axes of the matrix. In one implementation, the morphological operator may include applying a 1D image opening operation (7 pixels) along the A-line dimension to eliminate a-lines with negligible data coverage. Furthermore, a filter emphasizing connectivity may be applied to perform connectivity component analysis to identify individual components as individual branches. Cross-frame data may also be used to connect binary large objects (blobs) that are part of a branch within the connectivity matrix.

[0087] Partially, the present invention relates to an automated method for branch detection, comprising the step of detecting branch characteristics within a region having a scan depth T. In one embodiment, T is a distance measured from the lumen and can be used to define a boundary offset by the distance T from the lumen boundary. In one embodiment, T ranges from about 400 μm to about 700 μm. In one embodiment, T is about 660 μm. In one embodiment, T is an approximation of the vessel wall thickness or scan depth thickness, and T is selected to specify the search area for locating collaterals. In one embodiment, the branch characteristics include one or more of the following: a noisy base, or a signal-free region between a descending gradient and an ascending gradient. The use of the ascending and descending segments of the noisy base, relative to the detection signature feature, advantageously improves the detection accuracy for large branches.

[0088] In one implementation, the automated branch detection method includes the step of combining candidate branch regions across all frames, substantially all frames, or M frames, where M is a value of 2 or greater. In one implementation, the automated branch detection method includes the steps of: parsing branch regions to identify candidate branches, and estimating the branch diameter of these candidate branches. In one implementation, side branches with a diameter D greater than about 1 mm are tracked. In one implementation, a large branch is a branch with a diameter D greater than or equal to about 1 mm.

[0089] Larger branches contribute to increased flow and can therefore significantly affect FFR, VFR, and other blood flow-based measurements. Consequently, once detected, branches with diameters of interest (such as ≥1 mm) are tracked and evaluated to verify their characteristics as branches rather than false detections (such as shadows). Other diameters of interest may include branch diameters D ranging from ≥0.4 mm to ≤2 mm.

[0090] In one implementation, the automated branch detection method includes the step of generating a representative two-dimensional matrix of A-lines (also called scan lines) relative to the frame to define the branch matrix. In one implementation, the vertical axis is used to represent A-lines having units of angle (such as degrees ranging from 0 to 360 degrees), and the horizontal axis has units corresponding to the frame number. The angle range can be shown as greater than 360 degrees; however, the additional angle portions of the matrix typically overlap with earlier angle values ​​in the matrix.

[0091] Therefore, in one implementation, the frame numbering can start from 0 or 1 and continue to J, where J is the frame number in the pullback. The automated branch detection method includes the step of parsing the branch matrix to isolate branch candidates. In one implementation, guidewire detection and lumen detection are performed before branch detection. In some implementations, the guidewire visible in the frame is removed.

[0092] A system of one or more computers can be configured to perform specific operations or actions by means of software, firmware, hardware, or a combination thereof installed on the system, which, in operation, causes the system to perform actions. One or more computer programs can be configured to perform specific operations or actions by including instructions that, when executed by a data processing device, cause that device or computing device to perform actions.

[0093] One general aspect includes a method for automatically detecting one or more collateral vessels, comprising generating a branch matrix including scan line data and frame indicators. The method also includes storing one or more intravascular image datasets of vessels using an intravascular imaging system; each intravascular dataset includes multiple A-lines. The method further includes generating A-line images with the luminal boundaries of the detected vessels. An offset T can also be used to shift the representation of the detected lumen in a direction away from the imaging probe to a tissue region or branch region.

[0094] In one embodiment, the method further includes increasing edge intensity in the A-line image. The method also includes suppressing smooth regions in the A-line image. The method further includes specifying a search distance offset T relative to the lumen. The offset T may define a region or band for searching candidate collateral regions. The method further includes identifying local minima-maxima pairs in the filtered image. In one embodiment, the method further includes searching for the radial dimension r in the corresponding preprocessed input image.

[0095] In one embodiment, the method further includes marking pixels in the preprocessed input image below a noise floor threshold as candidate branch regions. Other embodiments of this aspect include corresponding computer systems, apparatuses, and computer programs recorded on one or more computer storage devices, each computer system, apparatus, and computer program configured to perform the actions of the method.

[0096] The implementation may include one or more of the following features. The method further includes: smoothing the A-line image using a first image processing operator. The method may also include applying median smoothing to the A-line image using a second image processing operator. The method may further include applying Gaussian smoothing to the A-line image using a third image processing operator.

[0097] In one embodiment, the method further includes dividing the candidate branch region into N bands, such as 1, 2, 3 or more bands, and processing each band separately. In one embodiment, the bands have the same thickness. In one embodiment, for the N bands, the thickness or width of the band is T / N. The method also includes accumulating labeled pixels along each A-line. Pixels can be labeled or otherwise tracked using software to identify a given pixel as corresponding to a shadow, guidewire pixel, branch pixel (such as a collateral pixel), lumen pixel, blood pixel, and other pixels corresponding to an imaged intravascular object or shadow or its reflection.

[0098] In one implementation, if a particular A-line has more than about 10% to about 30% of pixels marked as branches, the method marks the A-line in that band as a branch or an A-line containing a branch. The method also includes generating a branch matrix during frame-by-frame processing. The method further includes isolating white pixels (all three bands) and yellow pixels (the first two bands) adjacent to the white pixels. The method also includes removing guide wire regions. The method further includes eliminating branches that appear only in one frame. Therefore, branches that do not appear in multiple frames can be used to exclude candidate branches. The method also includes replicating the branch matrix to account for overlap across zeros. Implementations of the described techniques may include hardware, methods or processes on a computer-accessible medium, or computer software.

[0099] Although the present invention relates to different aspects and embodiments, it should be understood that the different aspects and embodiments disclosed herein may be integrated together, wholly or partially, where appropriate. Therefore, each embodiment disclosed herein may be incorporated to varying degrees into aspects suitable for a given implementation, and steps from various methods may be combined without limitation. Despite the foregoing and other disclosures herein, the embodiments disclosed herein may also be applied within the scope of bipolar-based systems and methods (if applicable).

[0100] Non-restrictive software features and implementation methods for implementing branch detection

[0101] The following description is intended to provide an overview of the device hardware and other operating components suitable for performing the methods of the invention described herein. This description is not intended to limit the scope or applicability of the invention. Similarly, hardware and other operating components may be suitable as part of the aforementioned apparatus. The invention can be implemented in other system configurations, including personal computers, multiprocessor systems, microprocessor-based or programmable electronic devices, network PCs, minicomputers, mainframe computers, etc.

[0102] Parts of a specific implementation are presented based on algorithms and symbolic representations used to manipulate data bits within computer memory. These algorithms can be described and represented by those skilled in the art of computers and software. In one implementation, the algorithm is here and generally conceived as a consistent sequence of operations that produces a desired result. The operations performed here as method steps, or otherwise described, are operations requiring physical manipulation of physical quantities. Typically, but not necessarily, these quantities take the form of electrical or magnetic signals that can be stored, transmitted, combined, converted, compared, and otherwise manipulated.

[0103] It should be understood that, unless otherwise specified, it will be apparent from the following discussion that, throughout this specification, the use of terms such as “processing” or “computing” or “determining an angle” or “selecting” or “switching” or “calculating” or “comparing” or “arc length measurement” or “detecting” or “tracking” or “masking” or “sampling” or “operating” or “generating” or “determining” or “displaying” refers to the actions and processing of a computer system or similar electronic computing device that manipulates and converts data represented by physical (electronic) quantities in the registers and memory of the computer system into similar other data represented by physical quantities in the memory or registers of the computer system or other such information storage, transmission or display devices.

[0104] In some embodiments, the invention also relates to means for performing the operations described herein. This means may be specifically constructed for a desired purpose, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer.

[0105] The algorithms and displays presented in this paper do not inherently involve any particular computer or other device. Based on the teachings of this paper, various general-purpose systems can be used with the program, or it can be demonstrated that it is convenient to construct more specialized devices to perform the required method steps. The following description will present the necessary structures for various such systems.

[0106] Embodiments of the present invention can be implemented in many different forms, including but not limited to: computer program logic used with a processor (e.g., a microprocessor, microcontroller, digital signal processor, or general-purpose computer), programmable logic used with a programmable logic device (e.g., a field-programmable gate array (FPGA) or other PLD), discrete components, integrated circuits (e.g., application-specific integrated circuits (ASICs)), or any other means including any combination thereof. In a typical embodiment of the invention, some or all of the processing of data collected using OCT probes, FFR probes, angiography systems, and other imaging and target monitoring devices and processor-based systems is implemented as a set of computer program instructions, which are converted into a computer-executable form, stored in a computer-readable medium, and executed by a microprocessor under the control of an operating system. Thus, user interface instructions and triggers based on pull-back or fusion requests are, for example, converted into processor-understandable instructions suitable for generating OCT data and performing image processing using the various and other features and embodiments described above.

[0107] The computer program logic that implements all or part of the functionality described earlier in this document can be embodied in various forms, including, but not limited to, source code, computer-executable form, and various intermediate forms (e.g., forms produced by an assembler, compiler, linker, or locator). Source code can include a set of computer program instructions implemented in any of a variety of programming languages ​​(e.g., object code, assembly language, or high-level languages ​​such as Fortran, C, C++, JAVA, or Hypertext Markup Language (HTML)) for use with various operating systems or operating environments. Source code can define and use various data structures and communication messages. Source code can be in computer-executable form (e.g., executed by an interpreter), or source code can be transformed (e.g., via a translator, assembler, or compiler) into computer-executable form.

[0108] Computer programs can be permanently or temporarily embedded in tangible storage media in any form (e.g., source code, computer-executable, or intermediate form), such as semiconductor storage devices (e.g., RAM, ROM, PROM, EEPROM, or flash memory-programmable RAM), magnetic storage devices (e.g., disks or hard disks), optical storage devices (e.g., CD-ROMs), PC cards (e.g., PCMCIA cards), or other storage devices. Computer programs can be embedded in signals that can be transmitted to a computer using any of a variety of communication technologies, including but not limited to analog, data, optical, wireless (e.g., Bluetooth), networking, and interconnection technologies. Computer programs can be distributed in any form as removable storage media with printed or electronic documentation (e.g., compressed software packages), pre-loaded onto a computer system (e.g., on a system ROM or hard disk), or distributed from a server or electronic bulletin board on a communication system (e.g., the Internet or the World Wide Web).

[0109] The hardware logic (including programmable logic used with programmable logic devices) that implements all or part of the functions previously described herein can be designed using conventional manual methods, or can be designed electronically, captured, simulated, or documented using a variety of tools such as computer-aided design (CAD), hardware description languages ​​(e.g., VHDL or AHDL), or PLD programming languages ​​(e.g., PALASM, ABEL, or CUPL).

[0110] Programmable logic can be permanently or temporarily embedded in a tangible storage medium, such as semiconductor storage devices (e.g., RAM, ROM, PROM, EEPROM, or flash memory-programmable RAM), magnetic storage devices (e.g., disks or hard disks), optical storage devices (e.g., CD-ROMs), or other storage devices. Programmable logic can be embedded in a signal that can be transmitted to a computer using any of a variety of communication technologies, including but not limited to analog, digital, optical, wireless (e.g., Bluetooth), networking, and interconnection technologies. Programmable logic can be distributed as a removable storage medium with printed or electronic documentation (e.g., compressed software packages), pre-loaded onto a computer system (e.g., on a system ROM or hard disk), or distributed from a server or electronic bulletin board on a communication system (e.g., the Internet or the World Wide Web).

[0111] The following discusses various examples of suitable processing modules in more detail. As used herein, a module refers to software, hardware, or firmware suitable for performing a specific data processing or data transfer task. In one implementation, a module refers to a software routine, program, or other memory-resident application that is suitable for receiving, transforming, routing, and processing instructions or various types of data (e.g., angiography data, OCT data, FFR data, IVUS data, fusion data, pixels, branch matrices and orientations and coordinates, user interface signals and various graphic display elements, as well as other information of interest as described herein).

[0112] The computers and computer systems described herein may include operatively associated computer-readable media, such as memory for storing software applications used in acquiring, processing, storing, and / or transmitting data. It will be understood that such memory may be internal, external, remote, or local relative to the computer or computer system to which it is operatively associated.

[0113] The memory may also include any component for storing software or other instructions, including, for example, but not limited to, hard disks, optical discs, floppy disks, DVDs (Digital Universal Optical Discs), CDs (Compressed Disks), Memory Sticks, flash memory, ROMs (Read-Only Memory), RAMs (Random Access Memory), DRAMs (Dynamic Random Access Memory), PROMs (Programmable ROMs), EEPROMs (Extended Erasable Programmable ROMs), and / or other similar computer-readable media.

[0114] Generally, computer-readable storage media used in conjunction with embodiments of the invention described herein may include any storage medium capable of storing instructions executable by a programmable device. Where applicable, the method steps described herein may be embodied or executed as instructions stored on a computer-readable storage medium or storage medium. According to embodiments of the invention, these instructions may be software embodied in various programming languages, such as C++, C, Java, and / or various other types of software programming languages ​​applicable to creating the instructions.

[0115] The aspects, embodiments, features, and examples of this invention are to be considered illustrative in all respects and are not intended to limit the invention, the scope of which is defined only by the claims. Other embodiments, modifications, and uses will be apparent to those skilled in the art without departing from the spirit and scope of the claimed invention.

[0116] The use of headings and portions in this application is not intended to limit the invention; each portion may be applied to any aspect, implementation or feature of the invention.

[0117] Throughout this application, where a composition is described as having, including, or containing specific components, or a process is described as having, including, or containing specific process steps, it is contemplated that the composition of this teaching is also substantially composed of or consisting of the listed components, and the process of this teaching is also substantially composed of or consisting of the listed process steps.

[0118] In this application, where an element or component is stated to be included in and / or selected from the list of listed elements or components, it should be understood that an element or component can be any of the listed elements or components, or a group consisting of two or more of the listed elements or components. Furthermore, it should be understood that the elements and / or features of the components, apparatus, or methods described herein can be combined in various ways without departing from the spirit and scope of this teaching, whether explicit or implicit.

[0119] The use of the terms “including” or “having” should generally be understood as open-ended and non-restrictive, unless otherwise expressly stated.

[0120] The use of the singular in this document includes the plural (and vice versa), unless otherwise expressly stated. Additionally, the singular forms “a” and “the” include the plural forms, unless the context clearly indicates otherwise. Furthermore, where the term “about” precedes a numerical value, this teaching also includes the specific numerical value itself, unless otherwise expressly stated. As used herein, the term “about” refers to a variation of ±10% from the nominal value.

[0121] It should be understood that the order of steps or the sequence of actions is not important as long as this instruction remains operational. Furthermore, two or more steps or actions can be performed simultaneously.

[0122] It should be understood that the claimed aspects of the invention relate to a subset and sub-steps of the techniques disclosed herein. Furthermore, the terms and expressions used herein are descriptive rather than restrictive, and in using such terms and expressions, no exclusion is intended of any equivalents of the features shown and described or portions thereof, but it is recognized that various modifications may be made within the scope of the claimed invention. Therefore, it is desired that protection be obtained by patent certificate for the disclosure defined and distinguished in the following claims, including all equivalents.

[0123] The term "machine-readable medium" includes any medium capable of storing, encoding, or carrying a set of instructions executable by a machine, and enabling the machine to perform any one or more methods of the present invention. Although a machine-readable medium is shown as a single medium in the exemplary embodiments, the term "machine-readable medium" should be understood to include a single medium or multiple media (e.g., a database, one or more centralized or distributed databases and / or associated caches and servers) storing a set or more sets of instructions.

[0124] It is understood that in certain aspects of the invention, a single component may be replaced by multiple components, and multiple components may be replaced by a single component, to provide an element or structure or to perform a given one or more functions. Such substitutions are considered to be within the scope of the invention, except where they are inoperable for certain embodiments of the practice.

[0125] The examples presented herein are intended to illustrate potential and specific implementations of the invention. It will be understood by those skilled in the art that these examples are primarily intended to illustrate the purposes of the invention. Variations may exist in the figures or operations described herein without departing from the spirit of the invention. For example, in some cases, method steps or operations may be performed or executed in a different order, or operations may be added, deleted, or modified.

[0126] Furthermore, although specific embodiments of the invention have been described herein for illustrative purposes and not for limiting purposes, those skilled in the art will understand that various modifications to the details, materials, and arrangements of elements, steps, structures, and / or parts can be made within the principles and scope of the invention without departing from the invention as described in the claims.

Claims

1. A method for detecting one or more branches of a blood vessel, comprising: Store one or more intravascular image datasets of the blood vessel; Detecting the luminal boundary in an image generated from the image dataset of the blood vessels, wherein the image has a first dimension and a second dimension; Specify the search distance T; Define a search region, which is bounded by the detected lumen boundary and a boundary offset from the lumen boundary equal to the search distance T; Detecting edges in the search area; and In response to the detected edge recognition candidate branch region.

2. The method of claim 1, further comprising: The image is flattened using a first image processing operator; A second image processing operator is used to apply median smoothing to the image; as well as A third image processing operator is used to apply smoothing to the image to generate a filtered image.

3. The method of claim 2, further comprising: Identify a first minimum-maximum pair in the filtered image, wherein one or more distances between the first minimum-maximum pair define a first search window.

4. The method of claim 3, further comprising: Identify a second minimum-maximum pair in the filtered image, wherein one or more distances between the second minimum-maximum pairs define a second search window.

5. The method of claim 3, further comprising: The search is performed along the first dimension of the corresponding preprocessed input image within the first search window.

6. The method of claim 5, further comprising: Pixels located in the first search window that are below the noise floor threshold are designated as corresponding to the candidate branch regions.

7. The method of claim 6, wherein, The noise floor threshold is less than 2 mm.

8. The method of claim 6, further comprising: The candidate branch region is divided into three bands, where the sum of the widths of the three bands is equal to T.

9. The method of claim 8, further comprising: For each band, pixels corresponding to the candidate branch region are accumulated along each segment.

10. The method of claim 9, wherein, If a particular portion has more than 10% to 30% of pixels that are marked as candidate branches, then that portion of the band is marked as the corresponding branch.

11. The method of claim 10, further comprising: Output the set of parts corresponding to the candidate branches for each band.

12. The method of claim 3, further comprising: A branch matrix is ​​generated using pulled-back frames, which include angle data.

13. The method of claim 8, further comprising: Isolate the pixels corresponding to the groups of all three bands and the pixels corresponding to the groups of the first two bands to select the pixels corresponding to the side branches.

14. The method of claim 12, further comprising: Remove the guide wire region from the branch matrix.

15. The method of claim 14, further comprising: Eliminate branches that appear in only one frame.

16. The method of claim 12, further comprising: Duplicate the branch matrix to account for overlap across zeros.

17. The method of claim 13, wherein, The first band ranges from 0 to T / 3, the second band ranges from T / 3 to 2 / 3T, and the third band ranges from 2 / 3T to T.

18. The method of claim 8, wherein, The first band ranges from 0 to T / 3, the second band ranges from T / 3 to 2 / 3T, and the third band ranges from 2 / 3T to T.

19. The method of claim 1, further comprising: Display one or more detected branches in the user interface.

20. The method of claim 1, further comprising: One or more candidate branches are validated using a branch matrix generated using pixels selected from two or more bands, where the sum of the bands is T.

Images