Fracture region instance processing method, device and equipment of medical image

Abstract

The application provides a fracture region instance processing method, device and equipment of a medical image, and relates to the technical field of medical image processing. The method comprises the following steps: obtaining a semantic segmentation result of a fracture region in a medical image. The fracture region voxels in the semantic segmentation result are subjected to connected domain analysis processing, the fracture region voxels that are spatially connected are divided into the same connected subset, and a plurality of connected subsets are generated. Effective connected subsets are determined from the connected subsets. Instance labels corresponding to each effective connected subset are respectively generated; and an instance-level division result of the fracture region is generated according to the instance labels. The method is used to achieve the effect of effective conversion from semantic-level segmentation to instance-level fragmentation.

Description

Technical Field

[0001] This application relates to the field of medical image processing technology, and more specifically, to a method, apparatus, and device for processing fracture areas in medical images. Background Technology

[0002] Currently, fractures are a common type of trauma in clinical practice, and accurate identification and differentiation of fracture fragments are crucial for preoperative planning, intraoperative navigation, and postoperative assessment. With the development of deep learning technology, medical image segmentation methods based on convolutional neural networks can achieve automated semantic segmentation of fracture regions, that is, marking voxels belonging to the fracture region from images such as CT and MRI to form a fracture region mask. However, semantic segmentation results can only distinguish between "fracture" and "non-fracture," and cannot differentiate between multiple spatially disconnected fracture fragments, which limits its application in fragment-level analysis and reduction planning.

[0003] In current technologies, the differentiation of fracture fragments typically relies on manual annotation by physicians or semi-automatic interactive operations, which are inefficient and highly subjective. A few studies have attempted to introduce instance segmentation branches into segmentation networks, but these networks are complex to train, computationally expensive, and require high-quality training data (pixel-by-pixel instance annotation), making them difficult to widely apply in clinical practice. Furthermore, while general connected component labeling algorithms can label connected components in binary images, their direct application to fracture regions can result in numerous invalid connected components due to noise, artifacts, or minor breaks in the semantic segmentation results, affecting the accuracy of fragment counting and the reliability of subsequent analysis.

[0004] Therefore, how to accurately and automatically distinguish individual fracture fragments from the semantic segmentation results of the fracture area, while effectively suppressing noise interference, has become a pressing technical problem to be solved in the field of computer-aided fracture diagnosis. Summary of the Invention

[0005] The purpose of this application is to provide a method, apparatus, and device for processing fracture region instances in medical images, which solves the above-mentioned problems existing in the prior art. It can solve the problems that semantic segmentation results cannot distinguish different fracture fragments and that direct application of connected component labels is easily affected by noise, and realize the effective conversion from semantic level segmentation to instance level fragments.

[0006] Firstly, a method for processing fracture region instances in medical images is provided, which may include: Obtain semantic segmentation results for fracture regions in medical images; Connectivity analysis is performed on the voxels of the fracture region in the semantic segmentation result to divide the spatially connected voxels of the fracture region into the same connected subset, generating multiple connected subsets; Determine valid connected subsets from the connected subsets; Instance identifiers are generated for each valid connected subset; based on the instance identifiers, instance-level distinction results for the fracture region are generated.

[0007] Secondly, a device for processing fracture region instances in medical images is provided, the device comprising: The acquisition module is used to acquire semantic segmentation results of fracture regions in medical images; The analysis module is used to perform connected component analysis on the voxels of the fracture region in the semantic segmentation result, divide the spatially connected voxels of the fracture region into the same connected subset, and generate multiple connected subsets. The determination module is used to determine a valid connected subset from the connected subset; The generation module is used to generate instance identifiers corresponding to each valid connected subset; and to generate instance-level differentiation results for the fracture region based on the instance identifiers.

[0008] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0009] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0010] This application provides a method, apparatus, and device for processing fracture region instances in medical images, obtaining semantic segmentation results of fracture regions in medical images. Connectivity analysis is performed on the voxels of the fracture region in the semantic segmentation results, dividing spatially connected voxels into the same connected subset, generating multiple connected subsets. Valid connected subsets are determined from these subsets. Instance identifiers are generated for each valid connected subset; based on the instance identifiers, instance-level differentiation results for the fracture region are generated. This solution addresses the problem that simply outputting semantic-level fracture segmentation results is insufficient to directly support subsequent fragment-level analysis and operations. It introduces connectivity analysis post-processing based on the semantic segmentation results predicted by the network to achieve instance-level differentiation of fracture fragments. This instance-level differentiation is crucial for subsequent fracture reduction: in computer-aided fracture reduction or preoperative planning, different fracture fragments typically need to be modeled and manipulated as independent rigid body units. Therefore, assigning a unique and consistent label to each fracture fragment is a prerequisite for fragment-level geometric modeling and reduction calculation. By introducing a connectivity analysis strategy, without changing the network structure and training process, an effective transition from semantic-level fracture segmentation to instance-level fragment differentiation is achieved. This fragment-level differentiation effectively suppresses noise interference and provides structured and distinguishable segmentation results to support the subsequent automatic repositioning and geometric analysis of fracture fragments. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating a method for processing an example of a fracture region in a medical image, provided in an embodiment of this application; Figure 2 A flowchart illustrating a method for processing an example of a fracture region in a medical image, provided in an embodiment of this application; Figure 3 A schematic diagram of the structure of a medical image fracture region processing device provided in this application embodiment; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art. The words "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are only used to distinguish different components. The words "comprising" or "including," etc., mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but do not exclude other elements or objects. The words "connected," "coupled," or "connected," etc., are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up," "down," "left," "right," etc., are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0014] The method for processing fracture area instances in medical images provided in this application embodiment can be applied to electronic devices, terminal devices, medical image fracture area instance processing devices or equipment, or other devices or equipment capable of executing this embodiment, and there are no limitations on this. In this embodiment, the execution subject is described as an electronic device.

[0015] The terminal can be a user equipment (UE) such as a mobile phone, smartphone, laptop computer, digital broadcast receiver, personal digital assistant (PDA), or tablet computer (PAD), handheld device, in-vehicle device, wearable device, computing device, or other processing device connected to a wireless modem, mobile station (MS), or mobile terminal. This terminal has the ability to communicate with one or more core networks via a radio access network (RAN).

[0016] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0017] Figure 1 This is a flowchart illustrating a method for processing an example of a fracture region in a medical image, provided as an embodiment of this application. Figure 1 As shown, the method may include: Step S101: Obtain the semantic segmentation results of the fracture region in the medical image.

[0018] For example, this application is used for post-processing the semantic segmentation results of fracture CT images to distinguish individual fracture fragments. Medical images include 3D CT images, MRI images, cone-beam CT images, or X-ray images, etc. For example, the fracture CT image is a pelvic fracture CT image. The semantic segmentation results of the fracture region output by a trained fracture segmentation network are obtained. The semantic segmentation results can be binary masks or equivalent representations of the fracture region obtained in any way, such as the output of a deep learning network, the results of traditional segmentation algorithms, or manual annotation. This step provides the data foundation for subsequent processing, demonstrating the decoupling of this application from the front-end segmentation algorithm.

[0019] For example, the semantic segmentation result can be a three-dimensional binary mask (1 represents a fracture, 0 represents the background), a probability map (which needs to be thresholded), or a subset of fracture categories from a multi-class segmentation result. This embodiment uses a three-dimensional binary mask as an example.

[0020] Step S102: Perform connected component analysis on the voxels of the fracture region in the semantic segmentation result, divide the spatially connected voxels of the fracture region into the same connected subset, and generate multiple connected subsets.

[0021] For example, connected component analysis is performed on all fracture region voxels (voxels with a value of 1) in the 3D binary mask. Specifically, using the 26-neighborhood connectivity criterion, each fracture voxel is traversed, and it is checked whether there are other voxels of the same fracture type among the 26 voxels surrounding each fracture voxel. If so, they are grouped into the same connected subset. After traversal, multiple spatially disjoint connected subsets are obtained, and each connected subset corresponds to a candidate fracture fragment. Candidate fracture fragments indicate that these subsets still need further verification to determine whether they are real fragments.

[0022] Step S103: Determine the valid connected subsets from the connected subsets.

[0023] For example, since semantic segmentation results may contain noise, artifacts, or minor breaks, directly using all connected subsets can lead to inaccurate fragment counting. A screening step improves the reliability of the results by retaining only clinically meaningful valid connected subsets. Screening criteria can include voxel count, geometry, location, etc., and are not limited thereto.

[0024] In this step, the number of voxels contained in each connected subset is calculated. First, a voxel count threshold is preset (e.g., the number of voxels corresponding to the minimum meaningful fragment volume calculated based on the slice thickness and interslice spacing of the CT image). Connected subsets with a voxel count less than the voxel count threshold are identified as noise or artifacts and removed. Connected subsets with a voxel count greater than or equal to the voxel count threshold are retained as valid connected subsets, and each valid connected subset corresponds to a real fracture fragment.

[0025] Step S104: Generate instance identifiers corresponding to each valid connected subset; based on the instance identifiers, generate instance-level differentiation results for the fracture region.

[0026] For example, for the retained valid connected subsets, unique instance labels (e.g., 1, 2, 3, ...) are assigned sequentially according to the coordinate order of the geometric centers of the valid connected subsets (e.g., from front to back or from left to right), resulting in instance-level differentiation of fracture fragments within the fracture region. This result is a 3D label map, where the value of each voxel represents its corresponding fragment number (0 represents background), which can be directly used for visualization or subsequent analysis. The coordinate order here is merely an example and is not limited thereto.

[0027] Therefore, compared to the original semantic segmentation results where multiple fragments are stuck together or mixed with noise, this application, after connected component analysis, divides the spatially separated fragments into different subsets; after filtering, minor noise is removed, and finally four clearly distinguishable fracture fragments are obtained.

[0028] The method provided in this application provides semantic segmentation results of fracture regions in medical images. Connectivity analysis is performed on the voxels of the fracture regions in the semantic segmentation results, dividing spatially connected voxels into the same connected subset, generating multiple connected subsets. Valid connected subsets are determined from these subsets. Instance identifiers are generated for each valid connected subset; based on the instance identifiers, instance-level differentiation results for the fracture region are generated. In this scheme, to address the problem that simply outputting semantic-level fracture segmentation results is insufficient to directly support subsequent fragment-level analysis and operations, connectivity analysis post-processing is introduced based on the semantic segmentation results predicted by the network to achieve instance-level differentiation of fracture fragments. This instance-level differentiation is crucial for subsequent fracture reduction: in computer-aided fracture reduction or preoperative planning, different fracture fragments typically need to be modeled and manipulated as independent rigid body units. Therefore, assigning a unique and consistent label to each fracture fragment is a prerequisite for conducting fragment-level geometric modeling and reduction calculations. By introducing a connectivity analysis strategy, without changing the network structure and training process, an effective transition from semantic-level fracture segmentation to instance-level fragment differentiation is achieved. This fragment-level differentiation effectively suppresses noise interference and provides structured and distinguishable segmentation results to support the subsequent automatic repositioning and geometric analysis of fracture fragments.

[0029] Figure 2 A flowchart illustrating a method for processing fracture regions in medical images provided in this application is shown below. Figure 2 As shown, in this embodiment... Figure 1 Based on the embodiments, the method is described in detail below, and the method includes: Step S201: Obtain the semantic segmentation results of the fracture region in the medical image.

[0030] In one example, the semantic segmentation result is any of the following: a 3D binary mask, a binary image obtained by thresholding the probabilistic map, and a subset of fracture categories in the multi-class segmentation result.

[0031] For example, suppose the obtained three-dimensional semantic segmentation result is represented as:

[0032] in, Represents the voxel coordinate space of a three-dimensional image. A collection of tags; This represents a three-dimensional integer coordinate space, where indices i, j, and k are all integers. For a given semantic category of fracture... The corresponding set of voxels can be defined as:

[0033] in, Representing voxel space Any voxel position (i.e., voxel coordinates) in the data; Representing voxel coordinates The corresponding label category in the semantic segmentation results.

[0034] Step S202: Perform connected component analysis on the voxels of the fracture region in the semantic segmentation result, divide the spatially connected voxels of the fracture region into the same connected subset, and generate multiple connected subsets.

[0035] In one example, S202 includes: traversing each fracture region voxel in the semantic segmentation result; for the currently traversed fracture region voxel, checking the region type of each neighboring voxel within the preset neighborhood range of the fracture region voxel; if there is a region type that is the fracture region type, then the currently traversed fracture region voxel and the neighboring voxels corresponding to the fracture region type are assigned to the same connected subset, generating multiple connected subsets.

[0036] In one example, the preset neighborhood range is any of the following: 26 neighborhoods, 18 neighborhoods, and 6 neighborhoods in three-dimensional space.

[0037] For example, each fracture region voxel in the semantic segmentation result is traversed; for the currently traversed fracture region voxel, the region type of each neighboring voxel within a preset neighborhood range of the fracture region voxel is checked; if there is a region type that is the fracture region type, then the currently traversed fracture region voxel and the neighboring voxels corresponding to the fracture region type are assigned to the same connected subset, generating multiple connected subsets. The preset neighborhood range is any one of the following: 26 neighbors, 18 neighbors, and 6 neighbors in three-dimensional space.

[0038] Optionally, given a three-dimensional connectivity criterion (e.g., using 26-neighborhood), for the voxel set... By partitioning the system into connected components, we can obtain multiple spatially disjoint connected subsets:

[0039] Each of them It represents a connected subset, that is, an independent instance of a fracture fragment; Indicates the semantic category of fracture The corresponding i-th connected subset; Indicates the semantic category of fracture The corresponding j-th connected subset; Indicates the semantic category of fracture The total number of connected subsets, i.e. the number of independent fracture fragment instances identified under this category.

[0040] Optionally, "26 neighborhood" means including 26 adjacent voxels in three directions around the current voxel (top, bottom, left, right, front, back, and diagonal directions), suitable for most medical image scenarios. "18 neighborhood" means excluding diagonal directions (i.e., only considering face and edge adjacency), which may be more suitable for certain specific applications. "6 neighborhood" means considering only face adjacency (top, bottom, left, right, front, and back), suitable for images with significant anisotropy.

[0041] Step S203: Based on the characteristics of each connected subset, select connected subsets that meet the preset valid conditions as valid connected subsets; wherein, the valid connected subset represents the fracture fragments; the characteristics include any one or more of the following: number of voxels, geometric shape and spatial location.

[0042] In one example, the feature is the number of voxels; step S203 includes: determining the connected subsets whose number of voxels is greater than or equal to a preset threshold as valid connected subsets.

[0043] For example, to suppress noise prediction, each connected subset is computed. voxel number And introduce a preset voxel number threshold. Remove pseudo-connected components with a small number of voxels, and filter from the connected subset to select components with a number of voxels greater than or equal to a preset valid condition (i.e., a preset voxel count threshold). A connected subset of the fragments is considered a valid connected subset. A valid connected subset represents a fracture fragment; its features include any one or more of the following: voxel count, geometry, and spatial location. The voxel count threshold is an adjustable parameter that can be adaptively set based on at least one of the following factors: the physical size of the voxels in the medical image, the image resolution, the slice thickness, or clinical experience.

[0044] The features include: Voxel count (size of connected subsets); subsets that are too small are likely noise or artifacts and can be removed; geometric shape (e.g., aspect ratio, compactness); spatial location (e.g., whether located inside or outside anatomical structures); and subsets outside the expected fracture area. This step can use a single feature or a combination of features for filtering.

[0045] Step S204: Generate instance identifiers corresponding to each valid connected subset.

[0046] In one example, S204 includes: generating instance identifiers corresponding to each valid connected subset according to a preset sorting rule; the sorting rule includes any of the following: the number of voxels, spatial location, geometric center, and preset order of the valid connected subset.

[0047] For example, according to a preset sorting rule, the retained valid connected subsets are sorted by a fixed offset. The initial relabeling strategy assigns a unique instance identifier to each fracture fragment. The sorting rules include any of the following: the number of voxels in a valid connected subset, spatial location, geometric center, and preset order.

[0048] Specifically, the voxel count is represented by sorting fragments by size, for example, assigning the largest fragment label 1, the second largest label 2, and so on. This rule facilitates subsequent statistical analysis of the fragments. Spatial location is represented by sorting fragments by their coordinate positions, such as along the X, Y, or Z axes. This helps maintain spatial logic during visualization. Geometric center representation calculates the centroid coordinates of each fragment and sorts them by centroid position, a compromise that balances size and location. Preset order representation covers any other custom rules, such as following the order of doctor annotations or the algorithm processing order.

[0049] Step S205: Generate instance-level differentiation results for the fracture area based on the instance identifier.

[0050] For example, based on the instance identifier, the instance-level segmentation result of the fracture region is obtained. The instance-level differentiation results of fracture fragments can be used for at least one clinical application in fracture reduction planning, fragment geometry modeling, fracture type classification, assessment of fracture comminutedness, or preoperative planning.

[0051] As a post-processing step, this application is applied after the output of the deep learning segmentation network and is executed without changing the network structure and training process. It can be widely applied to various fracture-related medical image analysis tasks. As an independent post-processing module, it can be seamlessly integrated into existing fracture segmentation workflows, elevating semantic segmentation results to instance-level fragment differentiation results, providing crucial data support for clinical work such as fracture reduction planning, fragment quantification analysis, and preoperative simulation. This application has advantages such as high automation, strong robustness, ease of deployment, modular design, and strong versatility.

[0052] The method provided in this application provides semantic segmentation results of fracture regions in medical images. Connectivity analysis is performed on the voxels of the fracture regions in the semantic segmentation results, dividing spatially connected voxels into the same connected subset, generating multiple connected subsets. Based on the characteristics of each connected subset, connected subsets that meet preset valid conditions are selected as valid connected subsets; wherein, a valid connected subset represents a fracture fragment; the characteristics include any one or more of the following: voxel count, geometric shape, and spatial location. Instance identifiers are generated for each valid connected subset. Based on the instance identifiers, instance-level differentiation results of the fracture region are generated. In this scheme, to address the problem that simply outputting semantic-level fracture segmentation results is insufficient to directly support subsequent fragment-level analysis and operations, connectivity analysis post-processing is introduced based on the semantic segmentation results predicted by the network to achieve instance-level differentiation of fracture fragments. This instance-level differentiation is of great significance for subsequent fracture reduction: in computer-aided fracture reduction or preoperative planning, different fracture fragments usually need to be modeled and manipulated as independent rigid body units. Therefore, assigning a unique and consistent label to each fracture fragment is a prerequisite for carrying out fragment-level geometric modeling and reduction calculations. By introducing a connectivity analysis strategy, without changing the network structure and training process, an effective transition from semantic-level fracture segmentation to instance-level fragment differentiation is achieved. This fragment-level differentiation effectively suppresses noise interference and provides structured and distinguishable segmentation results to support the subsequent automatic repositioning and geometric analysis of fracture fragments.

[0053] Corresponding to the above method, this application also provides a device for processing fracture area instances in medical images, such as... Figure 3 As shown, the device includes: Module 41 is used to acquire semantic segmentation results of fracture regions in medical images; Analysis module 42 is used to perform connected component analysis on the voxels of the fracture region in the semantic segmentation result, divide the spatially connected voxels of the fracture region into the same connected subset, and generate multiple connected subsets. Determining module 43 is used to determine a valid connected subset from the connected subset; The generation module 44 is used to generate instance identifiers corresponding to each valid connected subset; and to generate instance-level differentiation results of the fracture region based on the instance identifiers.

[0054] The functions of each functional unit of the medical image fracture area instance processing device provided in the above embodiments of this application can be implemented through the above method steps. Therefore, the specific working process and beneficial effects of each unit in the medical image fracture area instance processing device provided in the embodiments of this application will not be repeated here.

[0055] This application also provides an electronic device, such as... Figure 4 As shown, it includes a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other through the communication bus 540.

[0056] Memory 530 is used to store computer programs; The processor 510 performs the above steps when executing the program stored in the memory 530.

[0057] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0058] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0059] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0060] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0061] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 1 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0062] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform the fracture region processing method of any of the above embodiments.

[0063] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to perform the fracture region instance processing method of any of the above embodiments.

[0064] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0065] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0066] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0067] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0068] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0069] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of the claims in this application and their equivalents, then this application also intends to include these modifications and variations.

Claims

1. A method for processing fracture region instances in medical images, characterized in that, The method includes: Obtain semantic segmentation results for fracture regions in medical images; Connectivity analysis is performed on the voxels of the fracture region in the semantic segmentation result to divide the spatially connected voxels of the fracture region into the same connected subset, generating multiple connected subsets. Determine valid connected subsets from the connected subsets; Instance identifiers are generated for each valid connected subset; based on the instance identifiers, instance-level distinction results for the fracture region are generated.

2. The method as described in claim 1, characterized in that, Connectivity analysis is performed on the voxels of the fracture region in the semantic segmentation result to divide spatially connected voxels of the fracture region into the same connected subset, generating multiple connected subsets, including: Traverse each fracture region voxel in the semantic segmentation result; For the currently traversed fracture region voxel, check the region type of each neighboring voxel within the preset neighborhood range of the fracture region voxel. If there exists a region type that is a fracture region type, then the currently traversed fracture region voxel and the neighboring voxels corresponding to the fracture region type will be assigned to the same connected subset, generating multiple connected subsets.

3. The method as described in claim 2, characterized in that, The preset neighborhood range is any one of the following: 26-neighborhood, 18-neighborhood, and 6-neighborhood in three-dimensional space.

4. The method as described in claim 1, characterized in that, Determining a valid connected subset from the connected subset includes: Based on the characteristics of each connected subset, connected subsets that meet preset valid conditions are selected from the connected subsets as valid connected subsets; wherein, the valid connected subsets represent fracture fragments; The features include any one or more of the following: number of voxels, geometry, and spatial location.

5. The method as described in claim 4, characterized in that, The feature is the number of voxels; the step of selecting connected subsets that meet preset valid conditions from the connected subsets based on the features of each connected subset includes: A connected subset whose number of voxels is greater than or equal to a preset threshold is determined to be a valid connected subset.

6. The method as described in claim 1, characterized in that, The step of generating instance identifiers corresponding to each valid connected subset includes: Based on a preset sorting rule, instance identifiers corresponding to each valid connected subset are generated; the sorting rule includes any one of the following: The number of voxels, spatial location, geometric center, and preset order of the effective connected subset.

7. The method according to any one of claims 1-6, characterized in that, The semantic segmentation result is any of the following: The binary image obtained after thresholding of the 3D binary mask and probabilistic map, and the fracture category subset in the multi-class segmentation results.

8. A device for processing fracture region examples in medical images, characterized in that, The device includes: The acquisition module is used to acquire semantic segmentation results of fracture regions in medical images; The analysis module is used to perform connected component analysis on the voxels of the fracture region in the semantic segmentation result, divide the spatially connected voxels of the fracture region into the same connected subset, and generate multiple connected subsets. The determination module is used to determine a valid connected subset from the connected subset; The generation module is used to generate instance identifiers corresponding to each valid connected subset; and to generate instance-level differentiation results for the fracture region based on the instance identifiers.

9. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-7.

Images