Modeling regions of interest in anatomical structures
Through non-transitory computer systems and processors, the three-dimensional medical image data is processed using the region of interest modeling engine, and the anatomical region of interest model is identified and generated, which solves the problem of region of interest identification of anatomical structures in the prior art, achieving high Accurate surgical planning and navigation.
Patent Information
- Application Number
- CN202080021987.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2019-04-04
- Filing Date
- 2020-04-03
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2040-04-03
AI Technical Summary
Existing imaging techniques are difficult to effectively identify and model areas of interest in human anatomy, especially in surgical and medical planning, lack of high-precision regional positioning and navigation tools.
Receive and process three-dimensional medical image data using a region of interest modeling engine through non-transitory computer systems and processors, identify and generate regional of interest models, including calcification, mineralization, and plaque areas, using Hounsfield Unit (HU) thresholds Evaluate and travel cube algorithms to generate surface models that characterize anatomical structures.
High-precision identification and modeling of the anatomical structures is achieved, effective surgical planning and navigation tools are provided, the accuracy of intraoperative positioning and guidance is improved, and the registration of images of different imaging modalities is promoted.
Smart Images

Figure CN113573640B_ABST
Abstract
Description
[0001] CROSS-REFERENCE TO RELATED APPLICATIONS
[0002] This application claims the benefit of U.S. Provisional Patent Application No. 62 / 829,378, filed on April 4, 2019, entitled “MODELING REGION OF INTEREST OF AN ANATOMIC STRUCTURE,” which is incorporated herein by reference in its entirety. Technical Field
[0003] The present disclosure relates to modeling one or more regions of interest of an anatomical structure. Background Art
[0004] A variety of imaging techniques can be used to digitally visualize human anatomy. Magnetic resonance imaging (MRI), computed tomography (CT), and positron emission tomography (PET) are just some examples of imaging techniques used to image a patient's anatomy. Because such imaging data can represent three-dimensional anatomy, a computer can be used to generate or render a three-dimensional (3D) image. A 3D image can be rendered based on imaging data received from an imaging device used to generate the imaging data. Summary of the invention
[0005] The present disclosure relates to modeling one or more regions of interest of an anatomical structure.
[0006] In an example, a system may include a non-transitory computer-readable medium that may be configured to store data and instructions and a processor that may be configured to access the non-transitory computer-readable medium and execute instructions. The instructions may include a region of interest method code and a region of interest model code. The region of interest method code may be programmed to receive segmented image volume data for at least one anatomical structure. The segmented image volume data may include a three-dimensional (3-D) image volume that may include at least one anatomical structure. The region of interest method code may also be programmed to search for at least one anatomical structure outward from a calculated centerline to detect the surface of one or more regions of interest of at least one anatomical structure by evaluating a value associated with a voxel or pixel representing a 3-D image volume that may include at least one anatomical structure relative to a threshold. The region of interest method code may be programmed to generate a region of interest model based on one or more identified regions of interest.
[0007] In another example, a computer-implemented method may include receiving segmented image volume data for at least one anatomical structure. The segmented image volume data may include a 3-D image volume that includes the at least one anatomical structure. The computer-implemented method may include searching the at least one anatomical structure outward from a calculated centerline to detect a surface of one or more regions of interest of the at least one anatomical structure by evaluating values associated with voxels or pixels representing the 3-D image volume that may include the at least one anatomical structure relative to a threshold. The computer-implemented method may also include isolating the one or more regions of interest of the at least one anatomical structure in response to the search and generating a region of interest model based on the isolated one or more regions of interest.
[0008] In a further example, a system may include a non-transitory computer-readable medium that may be configured to store data and instructions, the data may include segmented image volume data for at least one anatomical structure. The segmented image volume data may include a 3-D image volume that may include at least one anatomical structure. The system may also include a processor that may be configured to access the non-transitory computer-readable medium and execute instructions. The instructions may include a region of interest modeling engine. The region of interest modeling engine may be programmed to search for at least one anatomical structure outward from a calculated centerline to detect the surface of one or more regions of interest of the at least one anatomical structure by evaluating HU values associated with voxels or pixels representing a 3-D image volume that may include at least one anatomical structure relative to a Hounsfield unit (HU) threshold. The region of interest modeling engine may also be programmed to isolate one or more regions of interest of the at least one anatomical structure in response to the search by excluding the 3-D image volume for the identified one or more regions of interest from the segmented image volume data for the at least one anatomical structure. The region of interest modeling engine may also be programmed to generate a region of interest model by applying a marching cubes algorithm to a voxel data set for the isolated one or more regions of interest of the at least one anatomical structure. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] Figure 1 is an example of a system that may be implemented to model one or more regions of interest of an anatomical structure.
[0010] Figure 2 is an example of a computer system programmed to provide a model of one or more regions of interest of an anatomical structure.
[0011] Figure 3 is a flow chart depicting an example of a method for generating a region of interest model.
[0012] Figure 4An example of a segmented image including bones and anatomical structures is depicted.
[0013] Figure 5 An example of a graphical representation of an anatomical structure is depicted.
[0014] Figure 6 An example of a graphical representation of a calcification model is depicted.
[0015] Figure 7 An example of a graphical user interface displaying a calcification model overlaid on an anatomical model is depicted.
[0016] Figure 8 An example of another graphical user interface displaying a calcification model overlaid on an anatomical model is depicted. DETAILED DESCRIPTION
[0017] The present disclosure relates to modeling one or more regions of interest of anatomical structures. In some examples, a region of interest modeling engine (e.g., a processor executable software code) can be programmed to generate a region of interest model. The region of interest model can characterize (e.g., identify) abnormal or otherwise distinguishable blocks of anatomical structures, such as calcification, mineralization, and plaque areas of the patient's vascular system. The region of interest model can be used in a medical environment to help plan corresponding procedures (e.g., surgical procedures). In some examples, the region of interest model can be displayed relative to an anatomical model representing the anatomical structure, such as during a corresponding procedure (e.g., on an output device). The region of interest model can be overlaid relative to the anatomical structure to facilitate intraoperative positioning and guidance of the object relative to the patient. Therefore, by providing a region of interest model, such as a model of a calcified region of an anatomical structure, effective procedure planning and navigation can be allowed during the procedure. In some examples, the region of interest model can be utilized to facilitate the registration of images acquired according to different imaging modalities.
[0018] For example, to generate a region of interest model, the region of interest modeling engine can be programmed to receive image volume data, which can include one or more three-dimensional (3-D) medical image volumes that can be acquired preoperatively for a given patient by a corresponding imaging modality. The image volume data can include image data representing voxels (e.g., 3-D image elements) corresponding to the patient's anatomical structure. In some examples, the value can be provided in Hounsfield units (HU) and associated with the corresponding voxel in the image data. The region of interest modeling engine can be programmed to segment the image volume data to isolate the anatomical structure from one or more bones and / or other anatomical structures (e.g., organs). The region of interest modeling engine can be programmed to generate a centerline model that can characterize the geometry of the anatomical structure (e.g., each branch of a blood vessel) and a surface model for the anatomical structure (e.g., vascular structure) based on the segmented image volume data.
[0019] The region of interest modeling engine may be programmed to search outward (e.g., radially) from the centerline model to detect a surface characterized by a surface model of one or more regions of interest of the anatomical structure. The search may include evaluating a HU value associated with a voxel representing a 3-D image volume including at least one anatomical structure relative to a HU threshold. The region of interest modeling engine may be programmed to isolate one or more regions of interest of at least one anatomical structure in response to the search. For example, the region of interest modeling engine may be programmed to exclude a 3-D image volume for the identified one or more regions of interest from the segmented image volume data for at least one anatomical structure. The region of interest modeling engine may be programmed to generate a region of interest model based on the one or more identified regions of interest. In some examples, the region of interest modeling engine may be programmed to generate a region of interest model by applying a marching cubes algorithm to a voxel data set for the isolated one or more regions of interest of at least one anatomical structure. The region of interest model may provide a graphical rendering (e.g., representation) of one or more regions of interest. The one or more regions of interest may include a calcified region, a mineralized region, a plaque region, or a stent region.
[0020] Figure 1An example of a system 100 is depicted that can be implemented to model one or more regions of interest of an anatomical structure. The system 100 can be implemented in a stand-alone computer, a workstation, a dedicated machine, or in a network environment where one or more modules or data can reside locally or remotely relative to where a user interacts with the system 100. The system 100 may include a processor 102. The processor 102 may be configured to access a memory 104 and execute machine-readable instructions stored in the memory 104 to implement one or more functions, such as modeling of one or more regions of interest of an anatomical structure. The machine-readable instructions may include a region of interest modeling engine 106.
[0021] The region of interest modeling engine 106 may be programmed to receive the modeling data 108 and process the modeling data 108 to model (e.g., generate) a region of interest model 110 to provide a graphical rendering (e.g., representation) of one or more regions of interest of the anatomical structure. In some examples, each region of interest may be individually modeled by the region of interest modeling engine 106 to provide one or more region of interest models 110. In other examples, the region of interest modeling engine 106 may be programmed to provide an integrated region of interest model 110 to provide a compressed model of one or more regions of interest. In some examples, a user interface ( Figure 1 ) to allow a user to provide user input for initiating the modeling process of the region of interest modeling engine 106.
[0022] The user interface may also be used to set up and establish data paths and variables used by the region of interest modeling engine 106. As another example, the user interface may also be used to configure calculations performed by the region of interest modeling engine 106 and / or one or more output devices 112 that may be provided. For example, the user interface may be configured based on the modeling data 108 to form the type of methods and parameters for analysis and processing (e.g., modeling). In some examples, the processor 102 may be programmed to provide the region of interest model 110 to the output device 112 to provide a visualization of one or more regions of interest. The rendered visualization of the region of interest model 110 may be used to facilitate or assist in the delivery of a medical protocol controlled therapy.
[0023] For example, to model one or more regions of interest, the region of interest modeling engine 106 may be programmed to retrieve image volume data 114. The image volume data 114 may include one or more 3-D medical image volumes that may be acquired for a given patient preoperatively or intraoperatively by a corresponding imaging modality. For example, the image volume data 114 may correspond to a preoperative anatomical (e.g., arterial) computed tomography (CT) scan of a region of the patient. The image volume data 114 may include image data representing voxels (e.g., 3-D image elements) corresponding to the patient's anatomical structure. In some examples, the values may be provided in HU and associated with corresponding voxels in the image data. Each HU value associated with the corresponding voxel is proportional to the X-ray attenuation right fork of the material. Therefore, each HU may represent tissue density. The region of interest modeling engine 106 may be programmed to detect the surface of one or more regions of interest of the patient's anatomical structure based on the HU values associated with the voxels, as described herein. The detected surface of each region of interest may be analyzed by the region of interest modeling engine 106 to determine the density of each region of interest.
[0024] The region of interest modeling engine 106 can be programmed to segment the image volume data 114 to isolate the anatomical structure from one or more bones and / or other anatomical structures (e.g., organs). In some examples, the isolated anatomical structure can correspond to soft connective tissue (such as a major blood vessel and one or more branches that can extend from such a blood vessel). Therefore, the region of interest modeling engine 106 can be programmed to isolate tissue from the image volume data 114 to provide segmented image volume data.
[0025] In some examples, the region of interest modeling engine 106 can be programmed to generate a centerline model that can characterize the geometry of the anatomical structure (e.g., each branch of a blood vessel) and a surface model for the anatomical structure (e.g., a vascular structure) based on the segmented image volume data. The surface model can describe the geometry of the internal and external surfaces of the anatomical structure. The region of interest modeling engine 106 can be programmed to extract the centerline model as a pixel or voxel thickness extending longitudinally along the central axis of the anatomical structure. In some examples, the surface model can characterize (e.g., provide) a point cloud representation (e.g., x, y, z) to describe the shape of the anatomical structure (or a portion thereof) based on the segmented image volume data. Each point in the point cloud can have a corresponding intensity, density (e.g., HU value), position and / or direction. The region of interest modeling engine 106 can be programmed to transform the surface point cloud data based on a grid algorithm (such as a marching cube algorithm) to provide a 3-D image of the surface of the anatomical structure.
[0026] The region of interest modeling engine 106 may be programmed to search outward (e.g., radially) from the centerline model to detect the surface of one or more regions of interest of the anatomical structure. In some examples, the region of interest modeling engine 106 may be programmed to initially evaluate the anatomical structure to determine the thickness of the anatomical structure. The region of interest modeling engine 106 may be programmed to identify one or more portions of the anatomical structure that may have a thickness equal to or greater than a thickness threshold (e.g., a value or range). The region of interest modeling engine 106 may be programmed to mark one or more portions for further evaluation, such as a more detailed analysis (e.g., granularity evaluation) of one or more portions of the one or more regions of interest.
[0027] In some examples, the region of interest modeling engine 106 can be programmed to find each edge corresponding to a surface of one or more regions of interest. For example, the region of interest modeling engine 106 is programmed to calculate a gradient vector for each voxel (or pixel) based on the segmented image volume data. The region of interest modeling engine 106 can be programmed to evaluate each calculated gradient vector to identify a transition between each surface of the anatomical structure and the one or more regions of interest based on a gradient threshold (e.g., a value or a range of values). The region of interest modeling engine 106 can be programmed to identify a transition by determining whether the difference between the calculated gradient vectors (e.g., the calculated gradient magnitudes) is greater than or equal to the gradient threshold.
[0028] In some examples, the region of interest modeling engine 106 is programmed to evaluate one or more portions to identify one or more regions of interest based on a thickness threshold in some examples. One or more regions of interest may have a different type of material (e.g., density) than other surrounding regions of the anatomical structure. For example, the region of interest modeling engine 106 is programmed to detect the surface of one or more regions of interest based on HU values associated with voxels of the segmented image volume data representing the anatomical structure. As such, voxels of the segmented image volume data representing an abnormal portion of the anatomical structure may have HU values that are different from those expected for one or more normal portions of the anatomical structure. The region of interest modeling engine 106 may be programmed to compare each HU value with a HU threshold to identify voxels (or pixels) representing one or more regions of interest (e.g., identifying one or more abnormal or other distinguishable portions of the anatomical structure). In some examples, the HU threshold may be set based on the type of imaging modality used to provide the image volume data 114.
[0029] The region of interest modeling engine 106 may be programmed to isolate one or more identified regions of interest from the anatomical structure. For example, the region of interest modeling engine 106 may extract voxels representing the image volume of one or more isolated regions of interest from the segmented image volume data of the anatomical structure, which extracted voxels may be stored in the memory 104 and marked as corresponding to the region of interest based on its evaluation. The region of interest modeling engine 106 may be programmed to generate a region of interest model 110 based on the removed voxels of the image volume representing one or more isolated regions of interest. For example, one or more regions of interest may correspond to calcified regions, mineralized regions, plaque regions, or stent regions. In some examples, the region of interest modeling engine 106 may be programmed to generate a region of interest model 110 in response to user input via a user interface to enable overlay (e.g., overlay on an anatomical model) of one or more isolated regions of interest or a portion of one or more isolated regions of interest.
[0030] In some examples, the region of interest modeling engine 106 is programmed to generate the region of interest model 110 as a 3-D model of points (e.g., pixels or voxels) corresponding to one or more isolated regions of interest. For example, the region of interest modeling engine 106 is programmed to provide a point cloud representation of the anatomical structure and the one or more isolated regions, respectively, and identify transitions (e.g., boundaries) between the one or more regions of interest and other regions of the anatomical structure based on HU values associated with the cloud points representing the anatomical structure.
[0031] Thus, the region of interest modeling engine 106 can be programmed to generate a region of interest model 110 that can characterize (e.g., identify) abnormal regions of the anatomical structure, such as calcification, mineralization, and plaque areas. The region of interest model 110 can be used to help plan a corresponding procedure (e.g., a surgical procedure). In some examples, the region of interest model 110 can be displayed relative to an anatomical model representing the anatomical structure during the corresponding procedure (e.g., on an output device 112), for example, while using a tracking system to facilitate intraoperative positioning and guidance of the object relative to the patient. In some examples, the tracking system can correspond to the tracking system described in U.S. Patent Application No. 14 / 209,300 (hereinafter referred to as the "'300 Patent Application") entitled "Method and System to Facilitate Intraoperative Positioning and Guidance", which is incorporated herein by reference in its entirety. In other examples, different tracking systems can be used.
[0032] Figure 2is an example of a computer system 200 programmed to provide a model 228 of one or more regions of interest of an anatomical structure. In some examples, the model 228 corresponds to the region of interest model 110, such as Figure 1 . The computer system 200 may include one or more processors 202 and a memory 204. The memory 204 may include one or more non-transitory computer-readable media having instructions and data stored thereon. The memory 204 may include a local storage device about the computer system 200 (e.g., physically attached to the processor 202 or in the same housing as the processor 202) and / or it may be remote (e.g., a network storage device or in a cloud computing infrastructure). The computer system 200 may also include one or more input / output (I / O) devices, such as a user input device 206 and an output device 208. For example, the user input device 206 may include a keyboard, a mouse, a touch screen interface, a gesture interface, etc., which may be configured to provide instructions and commands that may be stored in the memory 204 as user input data 210. The output device 208 may include one or more displays, such as a monitor, a head-up display, or a virtual or augmented reality headset or goggles. In some examples, the user input device 206 and the output device 208 may be integrated (e.g., a touch screen or other display) to provide both input and output for the computer system 200.
[0033] As an example, the user input device 206 may provide user input data 210 to specify a position (e.g., spatial coordinates) and / or movement of a pointer (e.g., a cursor) relative to a graphical representation of image volume data 212 that may be displayed on the output device 208. Additionally or alternatively, the user input device 206 may be configured to provide user input data 210 to select a particular voxel or pixel to be displayed as part of an image on the output device 208. As another example, the user input device 206 may be used to activate a function presented at a given location (e.g., a region of the display 208), such as the functions described herein.
[0034] In some examples, memory 204 may include image volume data 212 (e.g., as a Digital Imaging and Communications in Medicine (DICOM) or other type of file). Image volume data 212 may be associated with, for example, Figure 1. The image volume data 212 may include one or more 3-D medical image volumes that may be acquired for a given patient preoperatively by a corresponding imaging modality. In some examples, a 3-D medical image volume may be acquired intraoperatively. As an example, the image volume data 212 may correspond to a preoperative anatomical (e.g., arterial) CT scan of a region of the patient. In some examples, the CT scan may be acquired on the same day or even days, weeks, or months before a corresponding procedure (e.g., surgical intervention). Other medical imaging modalities may be used to provide the image volume data 212, such as magnetic resonance imaging (MRI), ultrasound, positron emission tomography (PET), etc. Such scans may be a common part of preoperative planning in a surgical workflow to help make a diagnosis, plan a surgical procedure, or facilitate other interventions. The image volume data 212 may include image data representing voxels (3-D image elements) corresponding to the patient's anatomical structure. In some examples, the values may be provided in HU and associated with corresponding voxels in the image data. In a CT scan, HU is proportional to the X-ray attenuation right-handed tree of the material. As such, HU values may be used to assess tissue density.As disclosed herein, HU values associated with voxels may be used to detect the surface of one or more regions of interest of a patient's anatomy.
[0035] As mentioned above, memory 204 includes instructions that can be executed by processor 202 to perform the various functions disclosed herein. Figure 2 In an example of , the instructions may include an output generator 214 that may be programmed to generate an interactive output visualization that may be rendered as a graphical output based on the image volume data 212 and displayed on the output device 208. The output generator 214 may include one or more rendering methods and / or controls to control what the image looks like in the output visualization displayed by the output device 208 (e.g., size, graphic overlay, viewing angle, etc.). The instructions stored in the memory 204 may also include a user interface 216 to enable interaction with the displayed image and a graphical user interface (GUI) (e.g., such as Figure 7-8 ) and various functions, including segmentation method 218. In some examples, segmentation method 218 can form part of region of interest modeling engine 106, such as Figure 1 as shown in .
[0036] The segmentation method 218 can be programmed to segment the image volume data 214, such as to isolate anatomical structures (e.g., one or more blood vessels). In other examples, the segmentation method 218 can also be programmed to segment the image volume data 212 to isolate one or more bones and / or other anatomical structures (e.g., organs). In some examples, the segmentation method 218 can include a segmentation method as described in U.S. Patent Application No. 16 / 265,732, entitled “Segmentation of Anatomic Structures” (hereinafter “the '732 Patent Application”), which is incorporated herein by reference in its entirety. In other examples, a different segmentation method can be used.
[0037] As an example, the segmentation method 218 can be programmed to isolate tissue from the image volume data 212 (such as corresponding to an anatomical structure). As disclosed herein, the segmentation method 218 can provide image volume data for segmentation of an anatomical structure. In some examples, the isolated anatomical structure can correspond to soft connective tissue (such as a major blood vessel and one or more branches that can extend from such a blood vessel). As an example, the blood vessels can correspond to the patient's descending aorta and associated renal arteries and their other branches. In other examples, the anatomical structure can correspond to the intestine, a portion of the respiratory tract, the patient's digestive tract, or other anatomical structures (e.g., kidneys) that can be monitored or manipulated for various diagnostic and surgical purposes.
[0038] The system 200 may include an anatomical model generator 220 that may be programmed to generate an anatomical model representing anatomical model data 222 based on the segmented image volume data for the anatomical structure. In some examples, the anatomical model generator 220 may correspond to an anatomical model generator as described in the '732 patent application. The anatomical model data 222 may include a centerline model that may describe the geometry of the anatomical structure (e.g., each branch of a blood vessel) and a surface model for the anatomical structure (e.g., a vascular structure). The surface model may describe the geometry of the interior and exterior surfaces of the anatomical structure (e.g., the inner wall of a blood vessel and the wall of a blood vessel). Thus, in some examples, the anatomical model generator 220 may form part of the region of interest modeling engine 106, as described in the '732 patent application. Figure 1 as shown in .
[0039] The anatomical model generator 220 can be programmed to calculate a corresponding centerline for an anatomical structure (e.g., a blood vessel, such as the aorta) based on segmented image volume data for the anatomical structure. As an example, the centerline can be extracted from the segmented image volume data as a pixel or voxel thickness extending longitudinally along a central axis of the anatomical structure. In one example, the centerline can be extracted according to the method disclosed in U.S. Patent Publication No. 2011 / 0026793, entitled "Automated Centerline Extraction Method and Generation of Corresponding Analytical Expression and Use Thereof," which is incorporated herein by reference. In other examples, other methods can be used to identify and model the centerline.
[0040] The corresponding surface boundaries of the anatomical structure may be calculated by the anatomical model generator 220. The surface of the anatomical structure may have a corresponding functional relationship with respect to the centerline along the length of the anatomical structure, as calculated by the anatomical model generator 220. The anatomical model generator 220 may be programmed to provide a centerline model by calculating model parameters from segmented image volume data describing the geometry of the centerline, and may also be programmed to provide a surface model by calculating model parameters from segmented vessel image volumes describing the surface of the anatomical structure.
[0041] In some examples, the anatomical model generator 220 can be programmed to provide a point cloud representation (e.g., x, y, z) to describe the shape of the anatomical structure (or a portion thereof) based on the segmented image volume data. Each point in the point cloud can have a corresponding intensity, density (e.g., HU value), position and / or orientation. The anatomical model generator 220 can be programmed to transform the surface point cloud data based on a grid algorithm (such as a marching cube algorithm) to provide a 3-D image of the surface of the anatomical structure. Other similar grid algorithms can also be used.
[0042] As a further example, the memory 204 employs a region of interest method 224. In some examples, such as Figure 1As shown in , the region of interest modeling engine 106 can be programmed to implement a region of interest method 224. The region of interest method 224 can be programmed to search outward from the centerline (e.g., radially) to detect the surface of one or more regions of interest of the anatomical structure (e.g., based on an evaluation of the image data according to one or more search criteria). In some examples, the region of interest method 224 can be programmed to search until a region of interest is identified. For example, the region of interest method 224 can be programmed to initially evaluate the anatomical structure to determine the thickness of the anatomical structure. The region of interest method 224 can be programmed to identify one or more portions of the anatomical structure that may have a thickness greater than a certain thickness threshold (e.g., a value or a range of values). The region of interest method 224 can be programmed to mark one or more portions for further evaluation, such as a more detailed analysis (e.g., granularity evaluation) of one or more portions of the one or more regions of interest.
[0043] Additionally or alternatively, the user input device 206 can be configured to provide user input data 210 in response to a user marking (or labeling) a particular voxel (or pixel) displayed as a portion of an image of an anatomical structure on the output device 208, which voxel (or pixel) can correspond to one or more portions based on a user experience (e.g., a user's knowledge). The region of interest method 224 can be programmed to evaluate one or more portions as described herein to identify one or more regions of interest. The one or more regions of interest can have a different type of material (e.g., density) than other surrounding regions of the anatomical structure. In some examples, the region of interest method 224 can be programmed to identify one or more regions of interest by evaluating speckle noise present in the image volume data 212.
[0044] As another example, the region of interest method 224 is programmed to determine speckle noise characteristics of each of the anatomical structure, one or more regions of interest, and, in some examples, blood flowing through the anatomical structure. The region of interest method 224 can be programmed to evaluate the speckle noise characteristics to identify transitions of the anatomical structure between any spatially adjacent regions. For example, the region of interest method 224 can be programmed to evaluate the speckle noise value associated with each voxel (or pixel) of the image volume data 212 relative to neighboring values to identify transitions. The transitions can correspond to surfaces (e.g., boundaries) of one or more regions of interest.
[0045] In some examples, machine learning techniques may be used to detect the surface of one or more regions. For example, the region of interest method 224 may be programmed to employ an image classifier that has been trained based on image training data. The image classifier may be programmed to identify changes (e.g., transitions) between adjacent voxels (or pixels) in the image volume data 212. Example image classifiers (e.g., classification algorithms) may include decision tree classifiers, naive Bayes classifiers, artificial neural network classifiers (e.g., convolutional neural networks), and k-nearest neighbor classifiers. In other examples, different image classifiers may be employed. The image classifier may be programmed to receive the image volume data 212 and classify each object (e.g., one or more regions, anatomical structures, bones, organs, etc.) to identify one or more regions of interest. In some examples, the region of interest method 224 may be programmed to label each voxel (e.g., pixel) of the image volume data 212 (e.g., classified by a classifier code) and grouped (e.g., transformed by a transformation code) to a corresponding class. For example, a given class may include voxels (or pixels) from the image volume data 212 that correspond to (e.g., define) the surface of one or more regions of interest (e.g., plaques). Each class of labeled voxels (or pixels) may characterize a corresponding object in the image volume data 212. The region of interest method 224 may be programmed to evaluate the voxels (or pixels) of each class relative to each other to identify transitions of anatomical structures to one or more regions corresponding to the surface of one or more regions of interest.
[0046] In some examples, the region of interest method 224 can be programmed to detect the surface of one or more regions of interest based on the HU values associated with the voxels of the segmented image volume data representing the anatomical structure. In some examples, one or more regions of interest of the anatomical structure can correspond to an abnormal portion of the anatomical structure (e.g., an abnormal tissue block), which can have a specific HU (or HU range) that is different from the expected HU of one or more normal portions of the anatomical structure (e.g., a healthy tissue block). Thus, in some examples, one or more regions of interest can correspond to a calcified region, a mineralized region, a plaque region, or a stent region. The abnormal portion can have a different density than the surrounding area of the anatomical structure. As such, the voxels of the segmented image volume data representing the abnormal portion of the anatomical structure can have a different HU value than expected for one or more normal portions of the anatomical structure.
[0047] The region of interest method 224 may be programmed to compare each HU value a to a HU threshold value to identify voxels (or pixels) representing one or more regions of interest (e.g., identifying one or more abnormal portions of an anatomical structure). The HU threshold value may correspond to one of a HU threshold value (e.g., a value of 130) or a HU threshold range (e.g., a range of 130 to 400). In some examples, the HU threshold value may be user definable based on user input data 210 provided at the user input device 206. In an example, the HU threshold value may be set based on a given HU value associated with blood that has been mixed with a contrast agent administered to a patient (e.g., prior to imaging). In some examples, multiple HU threshold ranges may be associated with multiple densities of different materials. The region of interest method 224 may be programmed to detect a surface of one or more regions of an anatomical structure based on multiple HU threshold ranges. As an example, the region of interest method 224 may be programmed to identify a stent deployed within an anatomical structure (e.g., an aorta) based on the density of the material of the stent.
[0048] In a further example, the region of interest method 224 can be programmed to evaluate HU values associated with voxels of the segmented image volume data representing the anatomical structure relative to multiple HU thresholds. The region of interest method 224 can be programmed to identify voxels representing the stent in the segmented image volume data based on the evaluation. Thus, the region of interest method 224 can be programmed to detect the surface of one or more regions of interest of the anatomical structure based on one or more HU thresholds that have been associated with materials having different densities. The HU value and / or other density parameter that can be determined from the evaluation can be stored (e.g., as density metadata) in the memory 204. This HU and / or density parameter can be assigned to an individual voxel or a collection of voxels, such as depending on the granularity of the evaluation.
[0049] As another example, the region of interest method 224 can be programmed to isolate one or more identified regions of interest from the anatomical structure. For example, the region of interest method 224 can be programmed to extract voxels of the image volume representing the one or more isolated regions of interest from the segmented image volume data of the anatomical structure, and the extracted voxels can be stored in the memory 104 and marked as corresponding to the region of interest based on its evaluation.
[0050] The memory 204 may include a region of interest model generator 226. In some examples, the region of interest model generator 226 may form a Figure 12. The region of interest model generator 226 may be programmed to generate a region of interest model 228 based on the removed voxels of the image volume representing one or more isolated regions of interest. In some examples, the region of interest model generator 226 may be programmed to generate the region of interest model 228 to enable coverage of one or more isolated regions of interest or a portion of one or more isolated regions of interest in response to user input via the user interface 216.
[0051] In examples where one or more regions of interest will provide a further frame of reference that can be selectively added to or removed from the output visualization (e.g., via the user input device 206), the region of interest model generator 226 can be programmed to generate a region of interest model 228 as a 3-D model of points (e.g., pixels or voxels) corresponding to one or more isolated regions of interest provided by the region of interest method 224. For example, the region of interest model generator 226 can be programmed to generate the region of interest model 228 by taking a voxel data set corresponding to voxels representing the one or more isolated regions (e.g., the output of the region of interest method 224) and applying a marching cubes algorithm to generate a mesh of polygons (e.g., triangles). The region of interest model 228 can be stored in the memory 204.
[0052] In some examples, the region of interest model generator 226 can be programmed to provide a point cloud representation of the anatomical structure and correspondingly one or more isolated regions, and identify transitions (e.g., boundaries) between one or more regions of interest and other regions of the anatomical structure based on HU values associated with the cloud points representing the anatomical structure. Therefore, in some examples, the region of interest model generator 226 can be programmed to identify transitions between one or more regions of interest and other regions of the anatomical structure based on the type of material. The point cloud can identify boundaries between different material types based on density. In some examples, the HU value (or range) of blood (with contrast agent) can overlap with one or more regions (such as soft plaque), making it difficult to identify voxels (or pixels) representing one or more regions of interest. In the present disclosure, a second set of image data from another modality can be used to identify voxels that overlap with blood or other materials in density to facilitate the identification of one or more regions of interest. The image volume data 212 can include a first set of one or more 3-D medical image volumes and a second set of one or more 3-D medical image volumes acquired for a given patient preoperatively by an imaging modality (e.g., via a CT scan). A first set of 3-D medical image volumes can be acquired preoperatively for a patient before a contrast agent is administered to the patient. The contrast agent can be mixed with blood to enhance detectability of blood within a region of interest via an imaging modality. A second set of one or more 3-D medical image volumes can be acquired preoperatively after a contrast agent has been administered to the patient.
[0053] The segmentation method 218 may be programmed to segment a first set of one or more 3-D medical image volumes and a second set of one or more 3-D medical image volumes to isolate anatomical structures in each corresponding medical image volume according to the functional settings described herein. The segmentation method 218 may be programmed to provide first segmented image volume data of the anatomical structure and second segmented image volume data of the anatomical structure. The region of interest method 224 may be programmed to evaluate each segmented image volume data of the anatomical structure to detect contrast agent and thus detect blood within the anatomical structure. The region of interest method 224 may be programmed to subtract the first segmented image volume data from the second segmented image volume data to isolate (or identify) voxels of the image volume in the second segmented image representing contrast agent within the anatomical structure. In some examples, the region of interest method 224 may be programmed to compare the first segmented image volume data of the anatomical structure with the second segmented image volume data to identify differences in image volume between the first and second segmented image volume data (e.g., identifying voxels of the image volume in the second segmented image data representing contrast agent within the anatomical structure).
[0054] The region of interest method 224 can be programmed to search, for example, radially outward from a centerline calculated for the anatomical structure based on the first segmented image volume data to identify one or more regions of interest of the anatomical structure. The region of interest method 224 can be programmed to detect the surface of the one or more identified regions of interest of the anatomical structure based on the HU values associated with the voxels of the first segmented image volume data representing the anatomical structure and also based on the image volume isolated in the second segmented image of the anatomical structure representing the contrast agent. The region of interest method 224 can be programmed to remove the image volume in the first segmented image of the anatomical structure representing blood based on the image volume isolated in the second segmented image of the anatomical structure representing the contrast agent. The region of interest method 224 can be programmed to provide a filtered first segmented image of the anatomical structure that can be substantially free of voxels of the image volume representing the contrast agent.
[0055] The region of interest method 224 can be programmed to detect the surface of one or more regions of interest of the anatomical structure by evaluating the HU values associated with the voxels of the filtered segmented image volume data representing the anatomical structure relative to the HU threshold. The region of interest model generator 226 can be programmed to generate a region of interest model 228 based on one or more identified regions. In some examples, the anatomical model generator 220 can be programmed to generate an anatomical model based on the filtered segmented image volume data of the anatomical structure. In some examples, the region of interest model 228 can be used to help plan a corresponding procedure (e.g., a surgical procedure) and be displayed relative to the anatomical model during the corresponding procedure, for example, when using a tracking system to facilitate intraoperative positioning and guidance of an object relative to a patient. In some examples, the tracking system can correspond to the tracking system described in the '300 patent application incorporated above. In other examples, a different tracking system can be used.
[0056] The region of interest model 228 can be spatially registered with the anatomical model and employed during the corresponding procedure to help the clinician (e.g., doctor) visualize and navigate the anatomical structure by providing the clinician with real-time awareness of one or more regions of interest (e.g., calcified regions, etc.). In an example where the anatomical model is derived from the same image data (e.g., CT image data), the registration of the region of interest model 228 can be facilitated. Color coding, shading, or other markings can be used to distinguish the region of interest model from the anatomical model, as well as to specify the plaque type or other characteristics of the tissue in the region of interest model. In addition, a graphical rendering based on the region of interest model 228 can be added or removed relative to the anatomical model (or other underlying image) in response to user input at the user input device 206. For example, in the case where the region of interest corresponds to calcified tissue in or on the vessel wall, a user input can be provided to switch (e.g., turn on and / or off) and thereby control the output generator 214 to selectively render or not render the calcified tissue on the display provided by the output device 208.
[0057] In some examples, the region of interest model 228 for the region of interest can be combined with the anatomical structure model to provide an output visualization. For example, the output generator is configured to use the calcification model (e.g., the region of interest model 228) in conjunction with the anatomical structure modeling data to provide aggregated output visualization data for rendering a graphical output on a display. In some examples, the output generator can correspond to the output generator 302 described in the '300 patent application. In these examples, the output generator 302 can be configured to use the calcification model in conjunction with the model data 308 and 309 to provide corresponding output visualization data 304 for rendering on a display.
[0058] In view of the above structural and functional characteristics, reference Figure 3 The example method will be better understood. Although for the sake of simplicity of explanation, Figure 3 The example method is shown and described as being executed serially, but it should be understood and appreciated that the present example is not limited to the order shown, as some actions can occur in a different order, multiple times, and / or concurrently with what is shown and described herein in other examples.
[0059] Figure 3 Describes the process used to generate a model of a region of interest (e.g., Figure 1 In some examples, the method 300 can be performed by a modeling engine (e.g., Figure 1 The region of interest modeling engine 106 shown) or a computer system (e.g., Figure 2 The invention is implemented on the computer system 200 shown.
[0060] The method may begin at 302 by receiving (e.g., at the region of interest modeling engine 106 or the region of interest method 224) segmented image volume data of at least one anatomical structure. The segmented image volume data may include a 3-D image volume that includes at least one anatomical structure. At 304, the method includes searching outward from a calculated centerline of the at least one anatomical structure (e.g., by the region of interest modeling engine 106 or the region of interest method 224) to detect a surface of one or more regions of interest of the at least one anatomical structure. The surface of the one or more regions of interest may be detected by evaluating values (e.g., HU values) associated with voxels of the 3-D image volume representing the at least one anatomical structure radially located from the calculated centerline relative to a threshold. In some examples, the one or more regions of interest may correspond to one or more regions where at least calcium has been deposited and hardened. At 306, in response to the search, one or more regions of interest of the at least one anatomical structure are isolated. For example, the region of interest modeling engine 106 or the region of interest method 224 may be configured to isolate the one or more regions of interest of the at least one anatomical structure. At 308 , a region of interest model (eg, the region of interest model 110 or the region of interest model 228 ) is generated based on the one or more isolated regions of interest. The region of interest model may be generated by the region of interest modeling engine 106 or the region of interest method 224 .
[0061] Figure 4 An example of a segmented image 400 including a bone 402, an anatomical structure (eg, blood vessel) 404, and calcification 406 is depicted. The segmented image 400 may be generated by a computer system (eg, computer system 200, such as Figure 2 For example, the segmented image 400 may be obtained from the segmented blood vessels 404 (eg, generated by the segmentation method 218, such as Figure 2 ) and segmented skeleton 402 (e.g., generated by segmentation method 218, such as Figure 2 Volume rendering of the segmented image).
[0062] Figure 5 An example of a graphical representation 500 of an anatomical structure, such as a blood vessel 502, is depicted. The graphical representation 500 may be generated by a computer system, such as a computer system such as Figure 2 The computer system 200 shown in FIG. 1 provides and displays on a display (e.g., Figure 2 For example, Figure 5 As shown in , the blood vessel 502 is the aorta. The graphical representation 500 of the blood vessel 502 can be generated from an isolated segmented volume of the blood vessel (e.g., generated by the segmentation method 218, such as Figure 2 ) rendering.
[0063] Figure 6 An example of a graphical representation 600 of a calcification model (e.g., blood vessel model 502) is depicted. The graphical representation 600 can be generated by a computer system (e.g., Figure 2 The computer system 200 shown in FIG. 1 provides and displays on a display (e.g., Figure 2 In some examples, the calcification model can be rendered on an output device 208 as shown. Figure 2 The illustrated region of interest model 228 corresponds to. For example, the graphical representation 600 of the calcification model may be generated from a graph for one or more isolated regions (eg, generated by the region of interest method 224, such as Figure 2 Segmented volume rendering of ).
[0064] Figure 7 An example of a graphical user interface 700 showing a calcification model overlaid on an anatomical model is depicted. The graphical user interface 700 can be displayed by a computer system (such as computing system 200, such as Figure 2 The user input (such as at an input device (e.g., user input device 206, such as Figure 2 ) received at ) and selectively turning on and off visualization of the calcification model relative to the graphical user interface 700, as described herein.
[0065] Figure 8 An example of another graphical user interface 800 showing a calcification model overlaid on an anatomical model is depicted. The graphical user interface 800 can be displayed by a computer system (such as the computing system 200, such as Figure 2 The user input (such as at an input device (e.g., user input device 206, such as Figure 2 The visualization of the calcification model relative to the graphical user interface 800 is selectively turned on and off by the user interface 800 as received at the graphical user interface 800.
[0066] In view of the above structural and functional descriptions, it will be appreciated by those skilled in the art that parts of the systems and methods disclosed herein may be implemented as methods, data processing systems, or computer program products (such as non-transitory computer-readable media). Thus, these parts of the methods disclosed herein may take the form of a complete hardware embodiment, a complete software embodiment (e.g., in one or more non-transitory machine-readable media), or an embodiment combining software and hardware. In addition, parts of the systems and methods disclosed herein may be computer program products on a computer-usable storage medium having a computer-readable program code thereon. Any suitable computer-readable medium may be used, including but not limited to static and dynamic storage devices, hard disks, optical storage devices, and magnetic storage devices.
[0067] Certain embodiments have also been described herein with reference to block diagrams of methods, systems, and computer program products. It will be understood that the illustrated blocks and combinations of the blocks in the illustrated blocks may be implemented by computer executable instructions. These computer executable instructions may be provided to one or more processors of a general-purpose computer, a special-purpose computer, or other programmable data processing device (or a combination of a device and a circuit) to produce a machine, so that the instructions executed by the processor implement the functions specified in one or more blocks.
[0068] These computer executable instructions may also be stored in a computer readable memory, which may direct a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory result in the product including instructions for implementing the functions specified in one or more flowchart blocks. The computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more flowchart blocks.
[0069] What is described above is an example. Of course, it is impossible to describe every possible combination of components or methods, because further combinations and arrangements are possible. Therefore, the present invention is intended to include all such changes, modifications and variations that fall within the scope of the present application, including the appended claims. When "one", "first" or "another" element or its equivalent is stated in the present disclosure or claims, it should be interpreted as including one or more than one such element, neither requiring nor excluding two or more such elements. As used herein, the term "comprising" means including but not limited to, and the term "including" means including but not limited to. The term "based on" means at least partially based on.
Claims
1. A system for modeling a region of interest of an anatomical structure, comprising: a non-transitory computer-readable medium configured to store data and instructions; A processor configured to access a non-transitory computer readable medium and execute instructions, the instructions comprising: The region of interest method code is programmed as: receiving segmented image volume data for at least one anatomical structure, wherein the segmented image volume data comprises a first set of segmented image volume data and a second set of segmented image volume data, the first set of segmented image volume data comprising the at least one anatomical structure, and the second set of segmented image volume data comprising the at least one anatomical structure and a contrast agent within the at least one anatomical structure; subtracting the first set of segmented image volume data from the second set of segmented image volume data to provide an isolated image volume in the second set of segmented image volume data representing at least contrast agent within the at least one anatomical structure; Searching the at least one anatomical structure outward from the calculated centerline to identify one or more regions of interest of the at least one anatomical structure, wherein searching outward from the calculated centerline comprises: performing a first evaluation of the first set of segmented image volume data to identify one or more portions of the anatomical structure based on the one or more portions having a thickness equal to or greater than a thickness threshold and / or in response to a user input; and performing a second evaluation of the first set of segmented image volume data to detect surfaces of the one or more regions of interest of the at least one anatomical structure (i) based on the one or more portions of the anatomical structure identified by the first evaluation, (ii) based on evaluation values associated with voxels representing the first set of segmented image volume data, and (iii) based on the isolated image volume in the second set of segmented image volume data, wherein the one or more regions of interest are a portion of at least one of the one or more identified portions of the at least one anatomical structure; and A region of interest model is generated based on the one or more identified regions of interest.
2. The system of claim 1, wherein the region of interest method code is further programmed to apply a marching cubes algorithm to the voxel datasets of the one or more identified regions of interest of the at least one anatomical structure to generate a region of interest model.
3. The system of claim 1 , wherein the instructions further comprise segmentation method code programmed to receive image volume data comprising a first set of one or more 3-D image volumes of a patient region of a given patient and a second set of one or more 3-D image volumes of the patient region.
4. The system of claim 3, wherein the segmentation method code is further programmed to segment the first set of one or more 3-D image volumes and the second set of one or more 3-D image volumes to isolate the at least one anatomical structure in each set of the one or more 3-D image volumes.
5. The system of claim 4, wherein the code is programmed to isolate the at least one anatomical structure comprises providing a first set of the segmented image volume data of the at least one anatomical structure from a first set of the one or more 3-D image volumes and providing a second set of the segmented image volume data of the at least one anatomical structure from a second set of the one or more 3-D image volumes.
6. The system of claim 5, wherein the image volumes identified in the second set of image volume data comprise corresponding 3-D image volumes.
7. The system of claim 6, wherein the region of interest method code is further programmed to exclude 3-D image volumes in the first set of segmented image volume data for the at least one anatomical structure based on 3-D image volumes identified in the second set of segmented image volume data during a search to detect the surface of the one or more regions of interest.
8. The system of claim 6, wherein the region of interest method code is further programmed to exclude 3-D image volumes in the first set of segmented image volume data for the at least one anatomical structure based on the 3-D image volumes identified in the second set of segmented image volume data to provide filtered segmented image volume data.
9. The system of claim 8, wherein the surfaces of the one or more regions of interest are detected by evaluating values associated with voxels of the 3-D image volume representing the filtered segmented image volume data relative to corresponding thresholds.
10. The system of claim 1, wherein the region of interest model corresponds to a calcification model.
11. The system of claim 1, wherein the instructions further comprise an anatomical model generator programmed to generate model data for a geometry of the at least one anatomical structure based on the segmented image volume data.
12. The system of claim 11, wherein the instructions further comprise an output generator programmed to render on a display a model of the region of interest and an anatomical model, the anatomical model provided based on the model data.
13. A computer-implemented method for modeling a region of interest of an anatomical structure, comprising: receiving segmented image volume data for at least one anatomical structure, wherein the segmented image volume data comprises a first set of segmented image volume data and a second set of segmented image volume data, the first set of segmented image volume data comprising the at least one anatomical structure, and the second set of segmented image volume data comprising the at least one anatomical structure and a contrast agent within the at least one anatomical structure; subtracting the first set of segmented image volume data from the second set of segmented image volume data to provide an isolated image volume in the second set of segmented image volume data representing at least contrast agent within the at least one anatomical structure; Searching the at least one anatomical structure outward from the calculated centerline to identify one or more regions of interest of the at least one anatomical structure, wherein searching outward from the calculated centerline comprises: performing a first evaluation of the first set of segmented image volume data to identify one or more portions of an anatomical structure having a thickness equal to or greater than a thickness threshold; and performing a second evaluation of the first set of segmented image volume data to detect surfaces of the one or more regions of interest of the at least one anatomical structure (i) based on the one or more portions of the anatomical structure identified by the first evaluation, (ii) based on values associated with voxels representing the first set of segmented image volume data, and (iii) based on the isolated image volume in the second set of segmented image volume data, wherein the one or more regions of interest are a portion of at least one of the one or more identified portions of the at least one anatomical structure; and A region of interest model is generated based on the identified one or more regions of interest of the at least one anatomical structure.
14. The computer-implemented method of claim 13, wherein generating comprises applying a marching cubes algorithm to the voxel data set of the isolated one or more regions of interest of the at least one anatomical structure to generate a region of interest model.
15. A system for modeling a region of interest of an anatomical structure, comprising: a non-transitory computer readable medium configured to store data and instructions, the data comprising segmented image volume data for at least one anatomical structure, the segmented image volume data comprising a three-dimensional 3-D image volume including at least one anatomical structure; A processor configured to access a non-transitory computer readable medium and execute instructions, the instructions comprising a region of interest modeling engine, the region of interest modeling engine being programmed to: identifying an isolated image volume in a second set of segmented image volume data representing at least contrast agent within the at least one anatomical structure; Searching the at least one anatomical structure outward from the calculated centerline to identify one or more regions of interest of the at least one anatomical structure, wherein searching outward comprises: performing a first evaluation of the first set of segmented image volume data to identify one or more portions of the anatomical structure based on determining that the one or more portions have a thickness equal to or greater than a thickness threshold and / or in response to a user input instruction; and performing a second evaluation of the first set of segmented image volume data to detect a surface of the one or more regions of interest of the at least one anatomical structure based on at least two of (i) the one or more portions of the anatomical structure identified by the first evaluation, (ii) values of voxels or pixels representing the first set of segmented image volume data, and (iii) the isolated image volume in the second set of segmented image volume data, wherein the one or more regions of interest are a portion of at least one of the one or more identified portions of the at least one anatomical structure; and A region of interest model is generated for the one or more regions of interest of the at least one anatomical structure. The system of claim 15 , wherein the region of interest model corresponds to a calcification model.
17. The system of claim 15 or 16, wherein the instructions further comprise an anatomical model generator programmed to generate model data for the geometry of the at least one anatomical structure based on the segmented image volume data.
18. The system of claim 17, wherein the instructions further comprise an output generator programmed to render the region of interest model and the anatomical model on a display.
