Detecting and representing anatomical features of anatomical structures

Abstract

An exemplary processing system accesses a three-dimensional (3D) model of a patient's anatomical structure and applies a detection process to the 3D model to detect single-layer anatomical features in the anatomical structure. The detection process includes generating a probability map of candidate points for the single-layer anatomical features from the 3D model and generating a single-layer mesh representing the single-layer anatomical features based on the probability map of the candidate points.

Description

[0001] Related applications

[0002] This application claims priority to U.S. Provisional Patent Application No. 62 / 860,196, filed on June 11, 2019, entitled “DETECTING AND REPRESENTING ANATOMICAL FEATURES OF AN ANATOMICAL STRUCTURE,” the contents of which are incorporated herein by reference in their entirety. Background Art

[0003] Preoperative images of patients are often used to plan surgical procedures. Preoperative images may include images captured using computed tomography (CT) scans, magnetic resonance imaging (MRI), or other such imaging modalities. The surgeon, other surgical team members, and / or the procedure planning computing system may use such preoperative images to determine planned incision points, instrument paths, and other aspects of the procedure. However, identifying certain anatomical features from preoperative images can be difficult or time-consuming. Summary of the Invention

[0004] The following description presents a simplified overview of one or more aspects of the systems and methods described herein. This overview is not an extensive overview of all contemplated aspects and is neither intended to identify key or critical elements of all aspects nor to delineate the scope of any or all aspects. Its sole purpose is to present one or more aspects of the systems and methods described herein as a prelude to the detailed description presented below.

[0005] An exemplary method includes a processor accessing a three-dimensional (3D) model of a patient's anatomical structure and applying a detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, the detection process including generating a probability map of candidate points for the single-layer anatomical features from the 3D model, and generating a single-layer mesh representing the single-layer anatomical features based on the probability map of the candidate points.

[0006] An exemplary system includes a memory storing instructions and a processor communicatively coupled to the memory, the processor configured to execute the instructions to access a three-dimensional (3D) model of a patient's anatomical structure and apply a detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, the detection process including generating a probability map of candidate points for the single-layer anatomical features from the 3D model, and generating a single-layer mesh representing the single-layer anatomical features based on the probability map of the candidate points.

[0007] An exemplary non-transitory computer-readable medium stores instructions that, when executed, direct at least one processor of a computing device to access a three-dimensional (3D) model of a patient's anatomical structure and apply a detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, the detection process comprising generating a probability map of candidate points for the single-layer anatomical features from the 3D model, and generating a single-layer mesh representing the single-layer anatomical features based on the probability map of the candidate points.

[0008] Another exemplary method includes a processor accessing a three-dimensional (3D) model of an anatomical structure of a patient, applying a first detection process to the 3D model to detect a first type of anatomical features in the anatomical structure, applying a second detection process different from the first detection process to the 3D model to detect a second type of anatomical features in the anatomical structure, and providing a visualization of the anatomical structure based on the 3D model, the first type of anatomical features detected in the anatomical structure, and the second type of anatomical features detected in the anatomical structure.

[0009] Another exemplary system includes a memory storing instructions and a processor communicatively coupled to the memory, the processor configured to execute the instructions to access a three-dimensional (3D) model of an anatomical structure of a patient, apply a first detection process to the 3D model to detect a first type of anatomical features in the anatomical structure, apply a second detection process different from the first detection process to the 3D model to detect a second type of anatomical features in the anatomical structure, and provide a visualization of the anatomical structure based on the 3D model, the first type of anatomical features detected in the anatomical structure, and the second type of anatomical features detected in the anatomical structure.

[0010] Another exemplary non-transitory computer-readable medium stores instructions that, when executed, direct at least one processor of a computing device to access a three-dimensional (3D) model of an anatomical structure of a patient, apply a first detection process to the 3D model to detect a first type of anatomical features in the anatomical structure, apply a second detection process different from the first detection process to the 3D model to detect a second type of anatomical features in the anatomical structure, and provide a visualization of the anatomical structure based on the 3D model, the first type of anatomical features detected in the anatomical structure, and the second type of anatomical features detected in the anatomical structure.

[0011] Another exemplary method includes applying a first detection process to a set of images to generate a three-dimensional (3D) model of an anatomical structure of a patient, applying a second detection process different from the first detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, and providing a visualization of the anatomical structure based on the 3D model and the detected single-layer anatomical features in the anatomical structure.

[0012] Another exemplary system includes a memory storing instructions and a processor communicatively coupled to the memory, the processor configured to execute the instructions to apply a first detection process to a set of images to generate a three-dimensional (3D) model of an anatomical structure of a patient, apply a second detection process different from the first detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, and provide a visualization of the anatomical structure based on the 3D model and the detected single-layer anatomical features in the anatomical structure.

[0013] Another exemplary non-transitory computer-readable medium stores instructions that, when executed, direct at least one processor of a computing device to apply a first detection process to a set of images to generate a three-dimensional (3D) model of an anatomical structure of a patient, apply a second detection process different from the first detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, and provide a visualization of the anatomical structure based on the 3D model and the detected single-layer anatomical features in the anatomical structure.

[0014] Another exemplary system includes a processor configured to access a three-dimensional (3D) model of an anatomical structure, apply a first detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, apply a second detection process different from the first detection process to the 3D model to detect non-single-layer anatomical features in the anatomical structure, and provide a representation of the anatomical structure based on the 3D model, the detected single-layer anatomical features, and the detected non-single-layer anatomical features. The system further includes a display communicatively coupled to the processor and configured to display the representation of the anatomical structure and a representation of a potential path for a surgical instrument to traverse in the anatomical structure. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The accompanying drawings illustrate various embodiments and are part of the specification. The illustrated embodiments are examples only and do not limit the scope of the present disclosure. Throughout the drawings, the same or similar reference numerals represent the same or similar elements.

[0016] Figure 1 An exemplary processing system according to the principles described herein is illustrated.

[0017] Figure 2 An exemplary method for generating a single-layer mesh according to the principles described herein is illustrated.

[0018] Figure 3 Illustrated is an exemplary configuration for detecting single-layer anatomical features according to the principles described herein.

[0019] Figure 4 Illustrated is an exemplary configuration for generating an initial single-layer mesh representing an anatomical feature according to the principles described herein.

[0020] Figure 5 Illustrated are exemplary configurations for refining an initial single-layer mesh representing anatomical features according to the principles described herein.

[0021] Figure 6 An exemplary probability map for generating a single-layer grid according to the principles described herein is illustrated.

[0022] Figure 7 Illustrated is an exemplary signed distance map for generating a single-layer grid according to the principles described herein.

[0023] Figure 8 An exemplary zero-crossing diagram for generating a single-layer grid according to the principles described herein is illustrated.

[0024] Figure 9 An exemplary initial single-layer mesh according to the principles described herein is illustrated.

[0025] Figure 10 An exemplary refinement process for generating a single-layer mesh according to the principles described herein is illustrated.

[0026] Figures 11 to 13 An exemplary single-layer mesh according to the principles described herein is illustrated.

[0027] Figure 14 and Figure 15 Illustrated is an exemplary configuration for generating visualizations of anatomical structures using a single layer mesh according to the principles described herein.

[0028] Figure 16 An exemplary computing system according to the principles described herein is illustrated. DETAILED DESCRIPTION

[0029] Systems and methods for detecting and representing anatomical features of an anatomical structure are described herein. In certain embodiments, for example, a processing system accesses a three-dimensional (3D) model of a patient's anatomical structure. The processing system applies a detection process to the 3D model to detect single-layer anatomical features in the anatomical structure. The detection process includes generating a probability map of candidate points for the single-layer anatomical features from the 3D model and generating a single-layer mesh representing the single-layer anatomical features based on the probability map of the candidate points.

[0030] In some examples, generating a single-layer mesh may include generating an initial single-layer mesh based on the probability map and generating a refined single-layer mesh based on the initial single-layer mesh. Generating the initial single-layer mesh may include processes such as reducing candidate points to a subset of candidate points, applying a signed distance transform to the subset of candidate points to generate a signed distance map, and connecting zero-crossing points in the signed distance map to form the initial single-layer mesh. Generating the refined single-layer mesh may include processes such as refining the initial single-layer mesh based on attributes of anatomical features. For example, refinement may include smoothing the initial single-layer mesh, filling holes in the initial single-layer mesh, and / or tailoring the initial single-layer mesh based on anatomical structure.

[0031] The processing system can also provide visualization of the anatomical structure, which can include single-layer anatomical features and one or more non-single-layer anatomical features. The processing system can use the detected anatomical features to determine or help the surgeon determine various aspects of the surgical procedure.

[0032] As an example, a processing system can be used to plan an instrument path to perform a biopsy in a patient's anatomical structure (e.g., a lung). The processing system can access a 3D model of the lung (e.g., generating a 3D model based on preoperative images, receiving a 3D model, etc.). A surgeon may want to determine an instrument path that will avoid anatomical features that may be damaged by an instrument. Such anatomical features may include single-layer anatomical features (e.g., lung fissures) and non-single-layer anatomical features (e.g., blood vessels, nerves, certain airways). As used herein, a single-layer anatomical feature may include any thin film having the same or substantially similar media on both sides (e.g., the ligaments that divide the liver into lobes, the membrane between muscle fibers, etc.). Non-single-layer anatomical features may include any other anatomical features. The processing system can apply a detection process to the 3D model to detect single-layer anatomical features and non-single-layer anatomical features, which will be further described herein. The processing system can provide visualization of anatomical structures, including anatomical features detected by the detection process for use by surgeons when planning or performing surgical procedures. Additionally or alternatively, the processing system may use the representation of the anatomy (including anatomical features detected by the detection process) to determine various aspects of the surgical procedure (eg, potential instrument paths).

[0033] The methods and systems described herein for detecting and representing anatomical features of anatomical structures may provide various advantages and benefits. For example, a single-layer mesh may provide an accurate representation of single-layer anatomical features, which may assist surgeons in planning and / or performing surgical procedures. Furthermore, using an automated process to detect single-layer features may save time and resources that could be allocated elsewhere to allow for more efficient and effective surgical procedures.

[0034] Various embodiments will now be described in greater detail with reference to the accompanying drawings.The disclosed methods and systems can provide one or more of the benefits described above and / or various additional and / or alternative benefits that will become apparent herein.

[0035] Figure 1 An exemplary processing system 100 (also referred to herein as "system 100") is illustrated that is configured to detect and represent anatomical features of an anatomical structure. The processing system 100 may be implemented within one or more components of a computing system and / or a computer-assisted surgical system. Figure 1 As shown, processing system 100 includes a storage facility 102 and a processing facility 104 selectively and communicatively coupled to each other. Each of facilities 102 and 104 may include or be implemented by one or more physical computing devices including hardware and / or software components, such as processors, memories, storage drives, communication interfaces, instructions stored in memories for execution by processors, etc. Although facilities 102 and 104 are Figure 1 102 and 104 may be combined into fewer facilities, such as a single facility, or divided into more facilities to serve a particular implementation. In some examples, each of facilities 102 and 104 may be distributed across multiple devices and / or multiple locations to serve a particular implementation.

[0036] The storage facility 102 may maintain (e.g., store) executable data used by the processing facility 104 to perform any of the operations described herein. For example, the storage facility 102 may store instructions 106 that are executable by the processing facility 104 to perform any of the operations described herein. The instructions 106 may be implemented by any suitable application, software, code, and / or other executable data instance.

[0037] The storage facility 102 may also maintain any data received, generated, managed, used, and / or transmitted by the processing facility 104. For example, as will be described in greater detail below, the storage facility 102 may maintain model data, image data, mesh data, surgical planning data, anatomical structure data, anatomical feature data, and the like.

[0038] The processing facility 104 may be configured to perform (eg, execute instructions 106 stored in the storage facility 102 to perform) various processing operations associated with detecting and representing anatomical features of anatomical structures in any manner described herein.

[0039] Figure 2 An exemplary method 200 for detecting and representing anatomical features of an anatomical structure is illustrated. Figure 2Exemplary operations according to one embodiment are illustrated, but other embodiments may omit, add to, reorder, and / or modify Figure 2 Any of the actions shown in . Figure 2 One or more of the operations shown in may be performed by processing system 100, any component included in the processing system, and / or any implementation thereof.

[0040] In operation 202, the processing system accesses a 3D model of the anatomical structure. In some examples, the 3D model can be generated based on preoperative images, such as CT scans, MRI scans, X-ray images, etc. The 3D model can be generated by the processing system 100 or generated by an external system and provided to the processing system 100. Example ways in which operation 202 can be performed are described herein.

[0041] In operation 204, the processing system generates a probability map of candidate points for the single-layer anatomical feature from the 3D model. A candidate point may be a point whose probability is greater than a threshold probability that the point is part of the single-layer anatomical feature. For example, the processing system 100 may analyze the 3D model and generate a set of candidate points whose probability is greater than a threshold probability (e.g., 20%, 50%, etc.). The set of candidate points may constitute a probability map of the presence of the single-layer anatomical feature in the 3D model. Example manners in which operation 204 may be performed are described herein.

[0042] In operation 206, the processing system generates a single-layer mesh representing the single-layer anatomical feature based on the probability map of the candidate points. For example, generating the single-layer mesh may include generating an initial single-layer mesh based on the probability map and generating a refined single-layer mesh based on the initial single-layer mesh. Additionally or alternatively, generating the single-layer mesh may include reducing the candidate points to a subset of candidate points, applying a signed distance transform to the subset of candidate points to generate a signed distance map, and connecting zero-crossing points in the signed distance map to form the single-layer mesh. Example manners in which operation 206 may be performed are described herein.

[0043] Figure 3 An exemplary configuration 300 for detecting single-layer anatomical features is illustrated, for example, for performing various processes or operations within a method such as method 200. Configuration 300 includes modules for accessing a 3D model 302 of an anatomical structure and generating a final single-layer mesh 320 representing anatomical features of the anatomical structure. Configuration 300 can be implemented by any suitable component of processing system 100 and / or a computer-assisted surgery system, surgical planning system, or other computer system.

[0044] As shown, configuration 300 includes a probability map module 304 that accesses a 3D model 302 of an anatomical structure. As described above, 3D model 302 may be generated by processing system 100 or generated and provided to probability map module 304 by an external system.

[0045] The probability map module 304 accesses the 3D model 302 of the anatomical structure and generates a set of candidate points 306. The probability map module 304 can use image processing techniques to detect elements in the 3D model 302 that indicate a single-layer anatomical feature. For each point, the probability map module 304 can generate a probability (e.g., between 0 and 1) that the point is part of an element that indicates a single-layer anatomical feature. The set of candidate points 306 can be an aggregation (e.g., a point cloud) of all points whose probability is greater than a threshold.

[0046] In some examples, the probability map module 304 can use machine learning techniques combined with image processing techniques to generate a set of candidate points 306. For example, the probability map module 304 can use a convolutional neural network (CNN) approach, such as U-Net, V-Net, Residual U-Net, etc., or any other suitable deep learning or machine learning technique.

[0047] The probability map module 304 provides the candidate points 306 to the single-layer mesh generation module 308, which generates a final single-layer mesh 320 based on the candidate points 306. As used herein, a mesh can refer to any representation (e.g., a representation of an anatomical feature) of a surface and / or elements using smaller discrete shapes (e.g., polygons), which can be defined by interconnections between a set of points, 3D coordinate positions, voxels, and / or equations. The mesh can be a single-layer mesh representing a single-layer anatomical feature, a non-single-layer mesh representing a non-single-layer anatomical feature, or a probability map, point cloud, or other intermediate element used to generate a single-layer mesh (e.g., the final single-layer mesh 320).

[0048] like Figure 3As shown, the single-layer grid generation module 308 includes a reduced candidate point module 310, an initial grid module 314, and a grid refinement module 318. The reduced candidate point module 310 receives the candidate points 306 from the probability map module 304 and generates a reduced set of candidate points (reduced candidate points 312) based on the candidate points 306. The reduced candidate point module 310 can reduce the set of candidate points 306 in any suitable manner. For example, the reduced candidate point module 310 can use a threshold probability value (e.g., 0.5 or 0.7 or any other suitable threshold) that is higher than the threshold probability for the candidate points 306 to generate the reduced candidate points 312. The reduced candidate point module 310 can also use a threshold number of points (e.g., 100 or 1000 or any other suitable number of points) or a percentage of points (e.g., 50% or 25% or any other suitable percentage) to include or exclude the candidate points 306 to reduce the candidate points 306 to the reduced candidate points 312. The reduced candidate point module 310 may also determine an optimal threshold based on a statistical analysis (e.g., Dice coefficient, Jaccard similarity coefficient, or any other suitable similarity analysis). The reduced candidate point module 310 may also use feature detection algorithms and techniques (e.g., Hessian matrix) to estimate initial ridge (and / or other feature) points. The reduced candidate point module 310 may also generate the reduced candidate points 312 based on any suitable combination of such thresholds.

[0049] The reduced candidate point module 310 provides the reduced candidate points 312 to the initial mesh module 314. The initial mesh module 314 generates an initial single-layer mesh 316 based on the reduced candidate points 312. The initial single-layer mesh 316 may include a mesh having a single layer thickness, but may also include various inaccuracies and / or uncertainties. Because the generation of the initial single-layer mesh 316 may be based on the probabilistic reduced candidate points 312 (or candidate points 306), inaccuracies and / or uncertainties may be introduced. Inaccuracies and / or uncertainties may also be introduced due to artifacts and / or missing information in the 3D model and / or preoperative images. Inaccuracies and / or uncertainties may also be introduced based on some mathematical algorithm used in generating the initial single-layer mesh 316. The mesh refinement module 318 may refine the initial single-layer mesh 316 to correct for such inaccuracies and / or uncertainties (e.g., based on known properties of the single-layer anatomical features, the 3D model 302, etc.) and generate a final single-layer mesh 320. The initial mesh module 314 and the mesh refinement module 318 can use any suitable techniques and algorithms to generate the initial single-layer mesh 316 and the final single-layer mesh 320, respectively. Examples of such techniques and algorithms are further described herein.

[0050] While configuration 300 illustrates a single-layer mesh generation module 308 including a reduced candidate point module 310 , an initial mesh module 314 , and a mesh refinement module 318 , in some examples, the single-layer mesh generation module 308 may include more modules, fewer modules, a combination of modules, etc. For example, the reduced candidate point module 310 may be optional, and the initial mesh module 314 may receive the candidate points 306 and generate the initial single-layer mesh 316 based on the complete set of candidate points 306 .

[0051] Figure 4 An exemplary configuration 400 for generating an initial single-layer mesh representing an anatomical feature, such as for performing various processes or operations within a method such as method 200 and / or method 300, is illustrated. Figure 3 Configuration 400 may be implemented by any suitable component of processing system 100 and / or a computer-assisted surgery system, surgical planning system, or other computer system.

[0052] Similar to Figure 3 , configuration 400 illustrates a reduced candidate point module 310, reduced candidate points 312, an initial grid module 314, and an initial single-layer grid 316. In some embodiments, the reduced candidate point module 310 can receive candidate points generated from the probability map module 304 by accessing the 3D model 302. Configuration 400 further illustrates the initial grid module 314 including a signed distance map module 402 and a zero-crossing point module 404.

[0053] The signed distance map module 402 can receive the reduced candidate points 312 from the reduced candidate point module 310. The signed distance map module 402 can generate a signed distance map 406 based on the reduced candidate points 312. The signed distance map module 402 can generate the signed distance map 406 using any suitable technique and algorithm. For example, the signed distance map module 402 can use principal component analysis (PCA), a signed distance function, etc. to generate the signed distance map 406 based on the reduced candidate points 312. An example signed distance map is described further below.

[0054] The signed distance map module 402 can provide the signed distance map 406 to the zero-crossing module 404. The zero-crossing module 404 can generate the initial single-layer grid 316 based on the signed distance map 406. The zero-crossing module 404 can generate the initial single-layer grid 316 by analyzing the signed distance map 406 to determine zero-crossings and connecting the zero-crossings to generate the initial single-layer grid 316. An example connection of the zero-crossings used to generate the initial single-layer grid is described further below.

[0055] Figure 5An exemplary configuration 500 for refining an initial single-layer mesh representing anatomical features, for example, for performing various processes or operations within a method such as method 200, method 300, and / or method 400, is illustrated. Figure 3 Configuration 500 may be implemented by any suitable component of processing system 100 and / or a computer-assisted surgery system, surgical planning system, or other computer system.

[0056] Similar to Figure 3 , configuration 500 illustrates a mesh refinement module 318 that accesses the initial single-layer mesh 316 and the 3D model 302 and generates a final single-layer mesh 320. In some embodiments, the initial single-layer mesh 316 can be generated using a configuration (e.g., configuration 400). Configuration 500 further illustrates a mesh refinement module 318 that includes a clipping module 502, a connected component module 504, a hole filling module 506, a small region module 508, a smoothing module 510, and a curled edge module 512. Such modules can be exemplary modules for refining the initial single-layer mesh based on one or more properties of the anatomical feature and / or anatomical structure. As an example, it can be known that the anatomical feature is completely contained within the anatomical structure. Therefore, the boundary of the anatomical feature can be constrained by the shape of the anatomical feature, and the initial single-layer mesh can be refined to conform to such constraints. Other such suitable properties of the anatomical feature and / or anatomical structure can be used to refine the initial single-layer mesh.

[0057] For example, the clipping module 502 can receive the initial single-layer mesh 316 and the 3D model 302. The clipping module 502 can refine the initial single-layer mesh 316 based on properties of anatomical features (anatomical features contained within the anatomical structure). Thus, the clipping module 502 can use the 3D model 302 to clip portions of the initial single-layer mesh 316 that extend beyond the boundaries of the anatomical structure, as provided by the 3D model 302. Additionally or alternatively, the clipping module 502 can extend portions of the initial single-layer mesh 316 to more closely align with the boundaries of the anatomical structure.

[0058] The connected components module 504 can refine the initial single-layer mesh 316 based on the contiguousness of the anatomical features. Based on this property of the anatomical features, the connected components module 504 can combine portions of the initial single-layer mesh 316 that were initially unconnected to each other. Additionally or alternatively, the connected components module 504 can separate portions of the initial single-layer mesh that were initially sparsely connected to each other.

[0059] The hole filling module 506 can refine the initial single-layer mesh 316 based on the continuity of the anatomical features. Based on this property of the anatomical features, the hole filling module 506 can search for holes in the initial single-layer mesh 316 and interpolate the initial single-layer mesh 318 to fill these holes. Additionally or alternatively, the hole filling module 506 may not fill large holes if such holes are expected in the anatomical features.

[0060] The small region module 508 can refine the initial single-layer mesh 316 based on the typical size of the anatomical feature. Based on this property of the anatomical feature, the small region module 508 can detect portions of the initial single-layer mesh 316 that appear too small and / or too far away from the main portion of the initial single-layer mesh 316 to be part of the anatomical feature. Based on this property of the anatomical feature, the small region module 508 can determine that such small, distant portions are false positives and remove such portions from the initial single-layer mesh 318.

[0061] The smoothing module 510 can refine the initial single-layer mesh 316 based on the typical uniformity of the surface and / or edges of the anatomical feature. Based on this property of the anatomical feature, the smoothing module 510 can modify the surface and / or edges, for example, by smoothing, roughening, patterning, or otherwise changing the surface and / or edges of the initial single-layer mesh 316.

[0062] The curled edge module 512 can refine the initial single layer mesh 316 based on the typical structure of the anatomical feature. Based on such properties of the anatomical feature, the curled edge module 512 can remove or interpolate portions of the initial single layer mesh 316. Examples of such a process are further described herein.

[0063] Although Figure 5 The mesh refinement module 318 is shown as including each of the modules 502-512, but other examples or implementations of the mesh refinement module 318 may include additional modules (e.g., based on other properties of the anatomical features), fewer modules, combinations of modules, etc. Figure 5 The flow from one module to the next is shown, but in some examples, a different order of modules may be implemented. Based on some or all of these refinements, mesh refinement module 318 may generate a final single-layer mesh 320 from initial single-layer mesh 318 .

[0064] Figures 6 to 12 An example application of the methods and systems described herein is illustrated. This example illustrates detecting and representing a pulmonary fissure in a patient's lung. Figure 6An image slice 600 with an exemplary probability map for generating a single-layer mesh is shown. The image slice 600 shows a two-dimensional (2D) slice of a 3D model of a patient's pair of lungs 602 (shown as a right lung 602-R and a left lung 602-L). The right lung 602-R has two fissures that divide the right lung 602-R into three lobes. The left lung 602-L has one fissure that divides the left lung 602-L into two lobes.

[0065] Image slice 600 shows a right lung 602-R having a first probability map 604 (shown as three separate segments 604-1, 604-2, and 604-3) and a second probability map 606. The left lung 602-L is shown having a third probability map 608. Each of the probability maps 604, 606, 608 can be a representation of the candidate points 306 or an implementation of a reduced representation of the candidate points 312. The image slice 600 shows each of the probability maps 604, 606, 608 as a band for each fissure. Although the image slice 600 shows bands of the probability maps 604, 606, 608, the bands can represent a simplified visualization of a mesh structure (e.g., a non-single-layer mesh that can be referred to as a "tongue mesh"). In other examples, any other suitable representation can be used for the probability maps 604, 606, 608, such as a point cloud of candidate points, a connected mesh of candidate points, or any other mesh-based representation of candidate points, a voxel-based representation of candidate points, etc.

[0066] As described above, each probability map 604, 606, 608 can represent a set of candidate points 306 or a reduced set of candidate points 312 for each fissure. For example, the third probability map 608 can represent an aggregation of points where the probability of each point being part of a left lung fissure is greater than a threshold. The probability of each point can be determined using image processing and machine learning techniques as described herein. Alternatively, the third probability map 608 can represent a reduced set of candidate points 312 for the presence of a lung fissure. The candidate points 306 can be reduced in any manner described herein to generate the reduced candidate points 312.

[0067] As shown, the first probability map 604 includes three separate segments 604-1, 604-2, and 604-3. The segmentation of the first probability map 604 can illustrate an example of how the initial single-layer mesh generated based on the first probability map 604 may be inaccurate. Because lung fissures are typically continuous and contiguous, the segmentation may be inaccurate. This inaccuracy can be corrected by the mesh refinement module as described herein.

[0068] The system 100 can generate a single-layer mesh representing a lung fissure based on the probability maps 604, 606, and 608 in any manner described herein. For example, based on the probability maps 604, 606, or 608, the system 100 can apply a signed distance transform to generate a signed distance map, and generate an initial single-layer mesh representing the fissure from the signed distance map. The system 100 can refine the initial single-layer mesh to generate a final single-layer mesh representing the fissure. Examples of such operations will now be described in greater detail.

[0069] Figure 7 An exemplary signed distance map 700 that can be generated by the system 100 and used by the system 100 to generate a single layer mesh is illustrated. The signed distance map 700 can be an embodiment of the signed distance map 406. The signed distance map 700 shows a tongue mesh 702 (e.g., the third probability map 608) of a set of candidate points (or a reduced set of candidate points) representing a lung fissure. The signed distance map 700 also includes vectors 704 (shown as vectors 704-1, 704-2...704-N). Using vector 704-1 as an example, each vector 704 (e.g., vector 704-1) shows a point 706-1 represented by X, through which the vector 704-1 passes orthogonally through the tongue mesh 702. The orthogonality of the vector 704-1 through the tongue mesh 702 can be determined in any suitable manner, such as using a Hessian matrix to estimate and / or calculate normal information for a point on the tongue mesh 702.

[0070] Vector 704-1 further illustrates distance information 708-1 for points on vector 704-1. Distance information 708-1 is represented as a signed number (e.g., +3, +2, +1, -1, -2, -3, etc.). The signed number can indicate the magnitude of the distance from point 706-1 and the direction relative to tongue grid 702, where positive values ​​are on one side of tongue grid 702 and negative values ​​are on the other side of tongue grid 702. As described herein, system 100 can analyze signed distance map 700 to determine zero crossings to generate an initial single-layer grid.

[0071] Figure 8 An exemplary zero-crossing diagram 800 that can be generated by the system 100 and used by the system 100 to generate an initial single-layer grid 802 is illustrated. The zero-crossing diagram 800 shows number columns 804 (shown as number columns 804-1, 804-2, ..., 804-N) generated from the signed distance diagram 700. Using number column 804-1 as an example, each number column 804 (e.g., number column 804-1) includes zeros 806-1, where the numbers in number column 804-1 cross from positive to negative (or vice versa). Connecting the zero-crossing points (e.g., 806-1 and other zero-crossing points) can generate a surface grid that can be provided as the initial single-layer grid 802.

[0072] Figure 9An exemplary initial single-layer mesh that may be generated by system 100 and used by system 100 to generate a final single-layer mesh is illustrated. Figure 9 A first initial single-layer mesh 902 and a second initial single-layer mesh 904 representing two fissures of the right lung are shown, as well as a third initial single-layer mesh 906 representing a fissure of the left lung. The representations of the initial single-layer meshes 902, 904, 906 can be generated by the system 100 in any suitable manner, such as by using a marching cubes algorithm or any other suitable computer graphics algorithm or technique to generate meshes from information such as the zero-crossing map 800.

[0073] In some examples, the initial single-layer meshes 902, 904, 906 may include various inaccuracies that can be refined by the mesh refinement module. For example, the first initial single-layer mesh 902 displays a jagged edge 908, which may be an uncertain representation of a typical edge of a lung fissure. To refine the jagged edge 908, the smoothing module may use any suitable algorithm or technique to smooth the jagged edge 908. For example, the smoothing module may average, interpolate, etc., points on the jagged edge 908 to generate a smoother edge, which may be a more accurate representation of a lung fissure. Additionally or alternatively, the clipping module may clip or align the jagged edge 908 using the boundary of the right lung, which may also be a more accurate representation of a lung fissure (because portions of a lung fissure may typically be attached to the inner surface of the lung). As described herein, other such inaccuracies may be refined by the mesh refinement module.

[0074] Figure 10 An exemplary refinement process for generating a single-layer mesh is illustrated. Figure 10 Two exemplary initial single-layer meshes are shown, a first initial single-layer mesh 1002 representing two fissures of the right lung and a second initial single-layer mesh 1004. Similar to initial single-layer meshes 902 and 904, initial single-layer meshes 1002 and 1004 can be generated by system 100 in any suitable manner, such as by using a marching cubes algorithm or any other suitable computer graphics algorithm or technique to generate meshes from information such as zero-crossing map 800.

[0075] Second initial single-layer mesh 1004 shows an exemplary inaccuracy of curled edges 1006. Curled edges 1006 may be an inaccurate representation of a fissure, as fissures typically may not have such edges. Inaccuracies such as curled edges 1006 may be generated as a byproduct of the algorithm used to generate the initial single-layer mesh. For example, curled edges 1006 may be a byproduct of the signed distance function applied to the tongue mesh, representing portions of the fissure that are close together.

[0076] A mesh refinement module (e.g., curled edge module 512) can refine the second initial single-layer mesh 1004 to remove curled edges 1006. The mesh refinement module can use the typical structure of fissures (e.g., not having curled edges) and an understanding of the byproducts of the algorithm used to generate the initial single-layer meshes 1002, 1004 to identify and remove inaccuracies (e.g., curled edges 1006). The mesh refinement module can also use a probability map (e.g., a tongue mesh) and / or the shape and / or structure of an anatomical structure (e.g., a lung) from which the initial single-layer meshes 1002, 1004 were generated to remove curled edges 1006. For example, the mesh refinement module can compare the second initial single-layer mesh 1002 to the probability map to detect portions that are inaccurately aligned with the probability map (e.g., portions that extend beyond the probability map, etc.) and remove these portions.

[0077] Figure 11 and Figure 12 An exemplary single-layer grid that may be generated by system 100 is illustrated. Figure 11 An image slice 1100 is shown having an exemplary single-layer mesh representing fissures in a patient's lungs. The image slice 1100 shows a two-dimensional (2D) slice of a 3D model of a pair of lungs 1102 (shown as a right lung 1102-R and a left lung 1102-L) of the patient. The image slice 1100 also shows a first single-layer mesh 1104 and a second single-layer mesh 1106 representing two fissures in the right lung 1102-R. The image slice 1100 also shows a third single-layer mesh 1108 representing a fissure in the left lung 1102-L. The single-layer meshes 1104, 1106, 1108 can be generated by the system 100 in any suitable manner, such as by generating an initial single-layer mesh based on a probability map of candidate points and refining the initial single-layer mesh (as described herein).

[0078] Compare image slice 1100 with Figure 6 For the image slice 600 of FIG. 600 , the single-layer grids 1104, 1106, 1108 may provide a more accurate representation of the lung fissures than the tongue grid provided by the probability maps 604, 606, 608. As surgeons and surgical team members may desire a single-layer representation of a single anatomical feature (e.g., a fissure), the single-layer grids 1104, 1106, 1108 may more accurately display the actual location of the fissures and provide a more intuitive representation.

[0079] Figure 12 A perspective view of a 3D visualization 1200 is shown that includes single-layer meshes, such as a first single-layer mesh 1104, a second single-layer mesh 1106, and a third single-layer mesh 1108. The 3D visualization 1200 also shows an airway structure 1202 of a patient.

[0080] In some examples, 3D visualization 1200 can be an interactive visualization that allows a user to view different aspects of a patient's lungs and fissures. For example, a user can interact with 3D visualization 1200 to move, rotate, zoom, or otherwise manipulate 3D visualization 1200 to view the lungs, fissures, airways, and other structures and features from different angles and viewpoints. Additionally or alternatively, a user can interact with 3D visualization 1200 to hide, show, change the opacity of different structures and / or features, such as fissures, airways, lung structures (e.g., pleura), blood vessels, and the like. For example, while Figure 12 The single layer meshes 1104, 1106, and 1108 are displayed opaquely so that the airway structure 1202 is not visible underneath, but the 3D visualization 1200 can be manipulated to display the single layer meshes 1104, 1106, 1108 partially or completely translucently, outlined, or in any other suitable manner so that the airway structure 1202 may also be visible or only visible. For example, Figure 13 A 3D visualization 1300 is shown having the same elements but a more semi-transparent representation of the single layer meshes 1104 , 1106 , and 1108 .

[0081] Although certain examples described herein relate to detecting and representing lung fissures, processing system 100 may be configured to detect and represent any suitable single-layer anatomical feature included in a patient's anatomy (eg, a single membrane, ligament, etc.).

[0082] In some examples, the system 100 can also apply a detection process to generate a representation of an anatomical feature such as an airway structure 1202. For example, the system 100 can apply a first detection process to detect non-monolayer anatomical features (e.g., airways, blood vessels, etc.) and a second detection process (e.g., different from the first detection process) to detect monolayer anatomical features (e.g., lung fissures). The example second detection process can include the above-described processes, techniques, and algorithms. An example first detection process is described herein.

[0083] Figure 14 Illustrated is an exemplary configuration 1400 for generating a visualization of an anatomical structure using a single-layer mesh that can be generated by system 100. Configuration 1400 includes modules that access the 3D model 302 of the anatomical structure and generate a 3D visualization 1410 representing the anatomical structure. Configuration 1400 can be implemented by any suitable component of processing system 100 and / or a computer-assisted surgery system, surgical planning system, or other computer system.

[0084] As shown, configuration 1400 includes a first detection process module 1402 and a second detection process module 1404 that access a 3D model 302 of an anatomical structure. As described above, the 3D model 302 can be generated based on preoperative images, such as CT scans, MRI scans, X-ray images, etc. The first detection process module 1402 can analyze the 3D model 302 to detect a first type of anatomical feature (e.g., a non-monolayer anatomical feature 1406). The first detection process module 1402 can detect the non-monolayer anatomical feature 1406 in any suitable manner, such as by utilizing a segmentation process applied to the 3D model 302 or any other such suitable process, technique, and / or algorithm.

[0085] The second detection process module 1404 can analyze the 3D model 302 to detect a second type of anatomical feature (eg, a single layer anatomical feature 1408). The second detection process module 1404 can detect the single layer anatomical feature 1408 using any of the processes, techniques, and / or algorithms described herein.

[0086] The system 100 can access and use data representing non-monolayer anatomical features 1406 and monolayer anatomical features 1408 to generate a 3D visualization 1410 of the anatomical structure, wherein the 3D visualization 1410 can include representations of the non-monolayer anatomical features 1406 and representations of the monolayer anatomical features 1408. For example, the 3D visualization 1410 can include Figure 12 1200 and illustrates an exemplary 3D visualization that includes both representations of non-single-layer anatomical features of an anatomical structure (e.g., representations of airway structures 1202) and representations of single-layer anatomical features (e.g., single-layer meshes 1104, 1106, and 1108 representing lung fissures). System 100 can generate 3D visualization 1410 using the detected features in any suitable manner, such as by using marching cubes or any other suitable volume rendering process, technique, and / or algorithm.

[0087] Figure 15 Illustrated is an exemplary configuration 1500 for generating a visualization of an anatomical structure using a single-layer mesh that can be generated by system 100. Configuration 1500 includes modules for generating a 3D visualization 1504 representing the anatomical structure. Configuration 1500 can be implemented by any suitable component of processing system 100 and / or a computer-assisted surgery system, surgical planning system, or other computer system.

[0088] As shown, the configuration 1500 includes a first detection process module 1502 that generates a 3D model 302 of the anatomical structure. As described above, the 3D model 302 can be generated based on pre-operative images, such as CT scans, MRI scans, X-ray images, etc. The first detection process module 1502 can generate the 3D model 302 based on the pre-operative images using any suitable techniques and / or algorithms, such as the segmentation and volume rendering techniques described above or any other suitable techniques and / or algorithms.

[0089] Configuration 1500 also includes a second detection process module 1404. As in configuration 1400, second detection process module 1404 can analyze 3D model 302 to detect single layer anatomical features 1408. Second detection process module 1404 can detect single layer anatomical features 1408 using any of the processes, techniques, and / or algorithms described herein.

[0090] System 100 can access and use data representing 3D model 302 and single layer of anatomical features 1408 to generate 3D visualization 1504 of the anatomical structure, where 3D visualization 1504 can include a representation of the anatomical structure and a representation of single layer of anatomical features 1408. System 100 can generate 3D visualization 1504 using the detected features in any suitable manner, such as by using marching cubes or any other suitable volume rendering process, technique, and / or algorithm.

[0091] In some examples, in conjunction with aspects of configurations 1400 and 1500, system 100 may also apply another first detection process to 3D model 302 to additionally detect non-monolayer anatomical features. In other cases, system 100 may apply a first detection process to generate a 3D model, a second detection process to detect monolayer anatomical features, and a third detection process (e.g., different from the first and second detection processes) to detect non-monolayer anatomical features.

[0092] In some examples, the processing system 100 can be implemented by or communicatively connected to one or more components of a surgical system that performs or is used to perform a surgical procedure. A surgical procedure can include any procedure that uses manual and / or instrumental techniques on a patient to investigate or treat a physical condition of the patient. The surgical system can use information generated by the processing system 100, such as detection and / or representation of anatomical features associated with the surgical procedure (e.g., for planning, execution, analysis, and / or other aspects of the surgical procedure). For example, the surgical system can be configured to detect anatomical features of an anatomical structure and represent them to one or more users of the surgical system. The users of the surgical system can use the representation of the detected anatomical features for various aspects of the surgical procedure. For example, using the lung fissure example described above, the user can use the representation of the detected anatomical features to plan potential paths for a surgical instrument to traverse to obtain a biopsy of a lung nodule, wherein the potential path avoids the detected fissure. Additionally or alternatively, the surgical system can be configured to detect anatomical features and use the detection of the anatomical features to determine aspects of the surgical procedure. For example, using the lung fissure example described above, a surgical system may use detection of the fissure to determine potential paths for a surgical instrument to traverse to obtain a biopsy of a lung nodule, where the potential paths avoid the detected fissure.

[0093] The data generated and output by the processing system 100 (e.g., data representing a single layer of mesh representing a single layer of anatomical features) can be accessed and / or used in any suitable manner by any appropriately configured computer-assisted surgical system (e.g., a teleoperated surgical system, a sensor-guided surgical system (e.g., a vision-guided surgical system), etc.). For example, the data generated by the processing system 100 can be used by a computer-assisted surgical system configured for minimally invasive surgical procedures. For example, the data generated by the processing system 100 can be used by a computer-assisted surgical system in the performance of a bronchoscopic procedure (e.g., a biopsy of a lung nodule), for example, to plan or facilitate the planning of potential paths traversed by a surgical instrument (e.g., a teleoperated catheter) to reach a particular location with the lung. Examples of computer-assisted surgical systems that can use the data generated by the processing system 100 include, but are not limited to, those provided by Intuitive Surgical Operations, Inc. System and DA system.

[0094] In some examples, a non-transitory computer-readable medium storing computer-readable instructions can be provided according to the principles described herein. When executed by a processor of a computing device, these instructions can direct the processor and / or computing device to perform one or more operations, including one or more operations described herein. Any of a variety of known computer-readable media can be used to store and / or transmit such instructions. Such a non-transitory computer-readable medium storing computer-readable instructions can be implemented by one or more components of a surgical system.

[0095] As used herein, non-transitory computer-readable media may include any non-transitory storage medium that participates in providing data (e.g., instructions) that can be read and / or executed by a computing device (e.g., by a processor of a computing device). For example, non-transitory computer-readable media may include, but are not limited to, any combination of non-volatile storage media and / or volatile storage media. Exemplary non-volatile storage media include, but are not limited to, read-only memory, flash memory, solid-state drives, magnetic storage devices (e.g., hard disks, floppy disks, magnetic tape, etc.), ferroelectric random access memory ("RAM"), and optical disks (e.g., compact disks, digital video disks, Blu-ray disks, etc.). Exemplary volatile storage media include, but are not limited to, RAM (e.g., dynamic RAM).

[0096] In certain embodiments, one or more of the systems, components, and / or processes described herein may be implemented and / or performed by one or more appropriately configured computing devices. To this end, one or more of the above-described systems and / or components may include or be implemented by any computer hardware and / or computer-implemented instructions (e.g., software) embodied on at least one non-transitory computer-readable medium configured to perform one or more processes described herein. In particular, system components may be implemented on one physical computing device or may be implemented on more than one physical computing device. Thus, system components may include any number of computing devices and may employ any of a number of computer operating systems.

[0097] In some embodiments, one or more processes described herein may be implemented at least in part as instructions embodied in a non-transitory computer-readable medium and executable by one or more computing devices. Generally speaking, a processor (e.g., a microprocessor) receives instructions from a non-transitory computer-readable medium (e.g., a memory, etc.) and executes those instructions to perform one or more processes (including one or more processes described herein). Any of a variety of known computer-readable media may be used to store and / or transmit such instructions.

[0098] Computer-readable media (also known as processor-readable media) include any non-transitory media that participates in providing data (e.g., instructions) that can be read by a computer (e.g., by a processor of a computer). Such media can take a variety of forms, including but not limited to non-volatile media and / or volatile media. Non-volatile media can include, for example, optical or magnetic disks and other persistent memory. Volatile media can include, for example, dynamic random access memory ("DRAM"), which typically constitutes main memory. Common forms of computer-readable media include, for example, magnetic disks, hard disks, tapes, any other magnetic media, compact disk read-only memory ("CD-ROM"), digital video disks ("DVD"), any other optical media, random access memory ("RAM"), programmable read-only memory ("PROM"), electrically erasable programmable read-only memory ("EPROM"), FLASH-EEPROM, any other memory chip or cartridge, or any other tangible medium from which a computer can read.

[0099] Figure 16 An exemplary computing device 1600 is illustrated that may be specifically configured to perform one or more of the processes described herein. Figure 16 As shown, computing device 1600 may include a communication interface 1602, a processor 1604, a storage device 1606, and an input / output ("I / O") module 1608 communicatively connected via a communication infrastructure 1610. Figure 16 shown in, but Figure 16 The components shown in FIG are not intended to be limiting. Additional or alternative components may be used in other embodiments. Figure 16 Components of computing device 1600 are shown.

[0100] Communication interface 1602 can be configured to communicate with one or more computing devices. Examples of communication interface 1602 include, but are not limited to, a wired network interface (e.g., a network interface card), a wireless network interface (e.g., a wireless network interface card), a modem, an audio / video connection, and any other suitable interface.

[0101] The processor 1604 generally represents any type or form of processing unit capable of processing data or interpreting, executing, and / or directing the execution of one or more of the instructions, processes, and / or operations described herein. The processor 1604 may direct the execution of operations according to computer-executable instructions 1612 (e.g., one or more application programs), such as may be stored in the storage device 1606 or another computer-readable medium.

[0102] Storage device 1606 may include one or more data storage media, devices, or configurations and may take any type, form, and combination of data storage media and / or devices. For example, storage device 1606 may include, but is not limited to, a hard drive, a network drive, a flash drive, a magnetic disk, an optical disk, RAM, dynamic RAM, other non-volatile and / or volatile data storage units, or combinations or subcombinations thereof. Electronic data (including the data described herein) may be temporarily and / or permanently stored in storage device 1606. For example, data representing executable instructions 1612 configured to direct processor 1604 to perform any of the operations described herein may be stored within storage device 1606. In some examples, the data may be arranged in one or more databases residing within storage device 1606. In some embodiments, instructions 1612 may include instructions 106 of processing system 100, processor 1604 may include or implement processing facility 104, and storage device 1606 may include or implement storage facility 102.

[0103] I / O modules 1608 may include one or more I / O modules configured to receive user input and provide user output. One or more I / O modules may be used to receive input for a single virtual reality experience. I / O modules 1608 may include any hardware, firmware, software, or combination thereof that supports input and output capabilities. For example, I / O modules 1608 may include hardware and / or software for capturing user input, including but not limited to a keyboard or keypad, a touch screen component (e.g., a touch screen display), a receiver (e.g., an RF or infrared receiver), a motion sensor, and / or one or more input buttons.

[0104] The I / O module 1608 may include one or more devices for presenting output to a user, including but not limited to a graphics engine, a display (e.g., a display screen), one or more output drivers (e.g., a display driver), one or more audio speakers, and one or more audio drivers. In some embodiments, the I / O module 1608 is configured to provide graphical data to the display for presentation to the user. The graphical data may represent one or more graphical user interfaces and / or any other graphical content to serve a particular embodiment.

[0105] The display can be configured to display a representation of the anatomical structure in any suitable manner. The displayed representation of the anatomical structure can represent one or more features of the anatomical structure, such as a single-layer anatomical structure and / or a non-single-layer anatomical structure. The display can be further configured to display, along with the representation of the anatomical structure, a representation of a potential path for a surgical instrument to traverse in the anatomical structure. The representation of the potential path can depict that the potential path avoids one or more anatomical features of the anatomical structure, such as a single-layer anatomical feature (e.g., by not contacting, crossing, or entering within a proximity threshold of a single-layer anatomical feature). The representation of the potential path can additionally or alternatively depict that at least a portion of the potential path is within an anatomical feature of the anatomical structure, such as a non-single-layer anatomical feature.

[0106] In some examples, a display can present a graphical user interface that includes one or more of the visual representations described herein. For example, the graphical user interface can present a representation of an anatomical structure, a single-layer anatomical feature of the anatomical structure, and / or a representation of a non-single-layer anatomical feature of the anatomical structure. The representation of the single-layer anatomical feature can be presented as a feature that is avoided by a potential path to be traversed by a surgical instrument in the anatomical structure. The representation of the non-single-layer anatomical feature can be presented as a feature that is avoided by a potential path to be traversed by a surgical instrument in the anatomical structure, or as a potential pathway for at least a portion of a potential path to be traversed by a surgical instrument in the anatomical structure.

[0107] In some examples, processor 1604 can be configured to receive user input via a graphical user interface. The user input can define potential paths through the anatomical structure for a surgical instrument to traverse. For example, the user input can create and / or modify potential paths through the anatomical structure for a surgical instrument to traverse. Representations of the defined potential paths can be displayed in the graphical user interface and can assist in planning a medical procedure to be performed on a patient.

[0108] Computing device 1600 may be implemented by or communicatively coupled to a computer-assisted surgical system. As an example, computing device 1600 may include or be implemented by a computing system, such as a surgical procedure planning computer communicatively coupled to a computer-assisted surgical system. As another example, computing device 1600 may be integrated within a computer-assisted surgical system.

[0109] In the foregoing description, various exemplary embodiments have been described with reference to the accompanying drawings. However, it will be apparent that various modifications and variations may be made thereto, and that additional embodiments may be implemented, without departing from the scope of the present invention as set forth in the appended claims. For example, certain features of one embodiment described herein may be combined with or substituted for features of another embodiment described herein. Accordingly, the description and drawings are to be regarded as illustrative rather than restrictive.

Claims

1. A system comprising: a memory that stores instructions; and a processor communicatively coupled to the memory and configured to execute the instructions to: Access three-dimensional models of anatomical structures, i.e. 3D models; and Applying a detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, the detection process comprising: determining a set of candidate points of the single-layer anatomical feature from the 3D model; applying a signed distance transform to the set of candidate points to generate a signed distance map, wherein the signed distance map comprises vectors passing through the set of candidate points, the vectors being orthogonal to the single-layer anatomical feature and representing distances of the candidate points from the single-layer anatomical feature; and A single-layer grid representing the single-layer anatomical features is generated by connecting the zero-crossing points in the signed distance map.

2. The system of claim 1 , wherein generating the single-layer mesh representing the single-layer anatomical features comprises: Generate an initial single-layer grid; and A refined single-layer mesh is generated based on the initial single-layer mesh. 3 . The system of claim 2 , wherein generating the refined single-layer mesh comprises refining the initial single-layer mesh based on properties of the single-layer anatomical features.

4. The system of claim 2, wherein generating the refined single-layer mesh comprises at least one of the following: smoothing the initial single-layer mesh; filling the holes in the initial single layer mesh; or The initial single-layer mesh is trimmed based on the 3D model of the anatomical structure.

5. The system of any one of claims 1-4, wherein the single-layer anatomical features comprise clefts in the anatomical structure.

6. The system of any one of claims 1-4, wherein the processor is further configured to execute the instructions to apply an additional detection process different from the detection process to the 3D model to detect non-single-layer anatomical features in the anatomical structure.

7. The system of claim 6, wherein the additional detection process comprises a segmentation process to detect the non-single-layer anatomical features.

8. The system of claim 6, wherein the non-monolayer anatomical feature comprises one or more blood vessels.

9. The system of claim 6, wherein the non-monolayer anatomical feature comprises one or more airways.

10. The system of any one of claims 1-4, wherein the processor is further configured to execute the instructions to provide a visual representation of the anatomical structure based on the 3D model and single-layer anatomical features detected in the anatomical structure.

11. A system comprising: A processor configured to: Access three-dimensional models of anatomical structures, i.e. 3D models; Applying a first detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, the first detection process comprising: determining a set of candidate points of the single-layer anatomical feature from the 3D model; applying a signed distance transform to the set of candidate points to generate a signed distance map, wherein the signed distance map comprises vectors passing through the set of candidate points, the vectors being orthogonal to the single-layer anatomical feature and representing distances of the candidate points from the single-layer anatomical feature; and generating a single-layer grid representing the single-layer anatomical features by connecting the zero-crossing points in the signed distance map; applying a second detection process different from the first detection process to the 3D model to detect non-monolayer anatomical features in the anatomical structure; and providing a representation of the anatomical structure based on the 3D model, the detected single-layer anatomical features, and the detected non-single-layer anatomical features; and A display is communicatively coupled to the processor and configured to display the representation of the anatomical structure and a representation of a potential path for a medical instrument to traverse in the anatomical structure.

12. The system of claim 11, wherein: said representation of said anatomical structure represents a single layer of said detected anatomical features; and The potential paths avoid the detected single-layer anatomical features.

13. The system according to claim 11 or 12, wherein: said representation of said anatomical structure representing said detected non-monolayer anatomical features; and At least a portion of the potential path is within the detected non-monolayer anatomical feature.

14. The system according to claim 11 or 12, wherein: The single-layer anatomical features include pulmonary fissures; and The non-monolayer anatomical features include airways.

15. The system of claim 11 or 12, wherein the representation of the potential path to be traversed by the medical instrument in the anatomical structure represents a path to be traversed by the medical instrument to obtain a biopsy of a lung nodule.

16. The system of claim 11 , wherein the processor is configured to generate the representation of the potential path for the medical device to traverse in the anatomical structure based on the 3D model, the detected single-layer anatomical features, and the detected non-single-layer anatomical features, the potential path avoiding the detected single-layer anatomical features.

17. The system of claim 11, wherein: the display being configured to display the representation of the anatomical structure in a graphical user interface; and The processor is configured to receive user input, via the graphical user interface, defining the potential path in the anatomical structure for the medical instrument to traverse.

18. The system of claim 17, wherein the graphical user interface presents one or more of the following: a representation of the single layer of anatomical features in the anatomical structure that are to be avoided by the potential path to be traversed by the medical device; or A representation of the non-single-layer anatomical feature as a potential pathway for at least a portion of the potential path to be traversed by the medical device in the anatomical structure.

19. A method comprising: A three-dimensional model of the anatomical structure, i.e., a 3D model, is accessed by a processor; and Applying, by the processor, a detection process to the 3D model to detect single-layer anatomical features in the anatomical structure, the detection process comprising: determining a set of candidate points of the single-layer anatomical feature from the 3D model; applying a signed distance transform to the set of candidate points to generate a signed distance map, wherein the signed distance map comprises vectors passing through the set of candidate points, the vectors being orthogonal to the single-layer anatomical feature and representing distances of the candidate points from the single-layer anatomical feature; and A single-layer grid representing the single-layer anatomical features is generated by connecting the zero-crossing points in the signed distance map.

Images