Method and device for correcting after liver segmentation, electronic equipment and storage medium
Through the post-liver segmentation correction method, the anatomical structure characteristics of the liver are temporarily merged and step-by-step correction, and the problem of liver segmentation error in the prior art is solved, improving the accuracy and reliability of segmentation.
Patent Information
- Application Number
- CN202411849002.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-16
- Publication Date
- 2025-03-18
AI Technical Summary
The existing liver segmentation model is prone to errors when dealing with complex liver structures, resulting in inaccurate segmentation results and affecting medical diagnosis and surgical planning.
A post-liver segmentation correction method is proposed. By obtaining the liver area segmented based on the model, removing external noise, temporarily combining liver lobe types, interleaved and intralobar correction treatments, and gradually optimizing the segmentation results.
It improves the accuracy of liver area segmentation, provides more reliable medical analysis and diagnostic basis, and comprehensively deals with different types of segmentation errors.
Smart Images
Figure CN119648733A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of medical image processing, and in particular to a method, device, electronic device and storage medium for correcting liver segmentation. Background Art
[0002] In the field of medical image processing, CT (computed tomography) data is widely used to diagnose and analyze various diseases, among which accurate segmentation of the liver region is crucial for assessing liver health, detecting lesions, and planning surgical treatment. Traditional practices rely on various learning models, such as machine learning or deep learning algorithms, to achieve automatic segmentation of liver regions in CT images. These models analyze large amounts of CT image data and learn the characteristic differences between the liver and surrounding tissues, thereby achieving preliminary recognition of the liver region.
[0003] However, existing model segmentation technologies face many challenges when dealing with complex liver structures. As a complex organ with multiple segments, the shape, size, and internal structure of the liver vary significantly between individuals. This high degree of anatomical variability makes it easy for existing models to make errors during the segmentation process, such as small segmented areas that are partially intertwined or separated from the liver area. Incorrect segmented areas will not only affect doctors' accurate diagnosis of liver disease, but may also mislead subsequent surgical planning and treatment decisions.
[0004] There is currently no effective technical solution to the above problems. Summary of the invention
[0005] The purpose of this application is to provide a liver segmentation post-correction method, device, electronic device and storage medium to improve the accuracy of liver region segmentation and provide a more reliable basis for subsequent medical analysis and diagnosis.
[0006] In a first aspect, the present application provides a liver segmentation correction method for correcting a liver region obtained based on model segmentation, the liver segmentation correction method comprising the following steps: S1. Obtaining a liver segmentation result, wherein the liver segmentation result is eight liver regions segmented based on CT data based on a preset segmentation model; S2, remove external noise from liver segmentation results; S3, temporarily merging the eight liver regions into five liver regions according to the liver lobe type; S4, performing inter-lobe correction processing on the five liver regions according to the contact relationship of the connected domains corresponding to each type of liver segment; S5, restoring the five-segment liver region after interlobar correction processing to eight-segment liver region; S6. Extract three single-lobe and two-segment regions of the eight restored liver regions, and perform intra-lobe correction processing on each region.
[0007] The core of the liver segmentation and post-correction method of the present application is to utilize the anatomical structural characteristics of the liver, gradually optimize the segmentation results through temporary merging and step-by-step correction, and finally obtain a more accurate eight-segment liver area division. This multi-step, multi-level processing method can comprehensively handle different types of segmentation errors, thereby improving the accuracy of liver area segmentation and providing a more reliable basis for subsequent medical analysis and diagnosis.
[0008] The liver segmentation and correction method, wherein step S2 comprises: S21, merging the eight liver regions into one category, and retaining the largest connected domain; S22: Synchronize the background area corresponding to the maximum connected domain into the segmentation result to remove the external noise.
[0009] In this example, step S21 merges the eight liver regions into one category and retains the largest connected domain, which helps to identify the main liver region and exclude possible small isolated regions or noise. By merging and retaining the largest connected domain, the region that is most likely to represent the real liver structure can be effectively screened out; then, step S22 synchronizes the background region corresponding to the largest connected domain to the segmentation result, that is, through the reverse operation, the part outside the identified main liver region is defined as the background, thereby removing external noise in the original segmentation result.
[0010] The liver segmentation and correction method, wherein step S4 comprises: S41, extracting five first maximum connected domains, where the five first maximum connected domains are respectively the maximum connected domains of each type of liver segment in the five liver regions; S42, correcting the first erroneous region to a liver segment category corresponding to the first maximum connected domain in contact with the first erroneous region, wherein the first erroneous region is a connected domain outside the first maximum connected domain and in contact with only one first maximum connected domain; S43: Correct the second erroneous region to a liver segment category corresponding to the first largest connected domain with the largest contact area, wherein the second erroneous region is a connected domain outside the first largest connected domain and in contact with at least two first largest connected domains.
[0011] In this example, step S41 establishes a reliable benchmark by extracting the maximum connected domain of each type of liver segment. Then, steps S42 and S43 respectively adopt different correction strategies for different types of error regions (only in contact with one or more maximum connected domains), ensuring the accuracy and comprehensiveness of the correction. This processing method can not only handle simple isolated error regions, but also complex error regions in contact with multiple liver segments, thereby significantly improving the accuracy of liver segmentation.
[0012] The liver segmentation and correction method, wherein in step S43, the contact area of the second error region is calculated based on traversing and calculating the length of the separation line between the second error region and the remaining four types of the first maximum connected domains in all corresponding slice planes.
[0013] The liver correction method after segmentation, wherein the intra-lobe correction process comprises: S61, extracting two second largest connected domains, where the two second largest connected domains are respectively the largest connected domains of each type of liver segment in the two-segment region of the single lobe; S62, correcting the third erroneous region to a liver segment category corresponding to the second largest connected domain with the largest contact area, wherein the third erroneous region is a connected domain that contacts two of the second largest connected domains and belongs to a liver segment category other than the two second largest connected domains; S63: Integrate the fourth error region into the corresponding second maximum connected region based on the category attribute of the exchanged non-maximum connected region, wherein the fourth error region is a connected region outside the two second maximum connected regions and belonging to a type of liver segment category in the single lobe two-segment region.
[0014] The liver segmentation and correction method, wherein the liver segmentation and correction method further comprises the steps of: S7. Merging the two segments of the single lobe after the intra-lobe correction processing with the remaining two segments of the eight-segment liver region to obtain the corrected eight-segment liver region.
[0015] The liver segmentation and correction method, wherein the three single lobe and two-segment areas are: the upper area of the left outer sector and the lower area of the left outer sector, the upper area of the right front sector and the lower area of the right front sector, the upper area of the right rear sector and the lower area of the right rear sector.
[0016] In a second aspect, the present application further provides a liver segmentation correction device for correcting a liver region obtained based on model segmentation, the liver segmentation correction device comprising: An acquisition module, used for acquiring a liver segmentation result, wherein the liver segmentation result is eight liver regions segmented based on CT data based on a preset segmentation model; A denoising module, used to remove external noise from liver segmentation results; A temporary merging module, used for temporarily merging the eight liver regions into five liver regions according to the type of liver lobes; An inter-lobe correction module, used for performing inter-lobe correction processing on the five liver regions according to the contact relationship of the connected domains corresponding to each type of liver segment; An unmerging module is used to restore the five-segment liver region to the eight-segment liver region after interlobar correction; The intra-lobe correction module is used to extract three single-lobe and two-segment regions of the restored eight-segment liver region and perform intra-lobe correction processing on each of them.
[0017] The liver segmentation post-correction device of the present application gradually optimizes the segmentation results by temporary merging and step-by-step correction, and finally obtains a more accurate eight-segment liver area division. This multi-step, multi-level processing method can comprehensively handle different types of segmentation errors, thereby improving the accuracy of liver area segmentation and providing a more reliable basis for subsequent medical analysis and diagnosis.
[0018] In a third aspect, the present application further provides an electronic device, comprising a processor and a memory, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the steps in the method provided in the first aspect are executed.
[0019] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps in the method provided in the first aspect are executed.
[0020] From the above, it can be seen that the present application provides a method, device, electronic device and storage medium for post-segmentation correction of the liver. The core of the post-segmentation correction method of the liver is to utilize the anatomical structure characteristics of the liver, and gradually optimize the segmentation results through temporary merging and step-by-step correction, and finally obtain a more accurate eight-segment liver area division. This multi-step, multi-level processing method can comprehensively handle different types of segmentation errors, thereby improving the accuracy of liver area segmentation and providing a more reliable basis for subsequent medical analysis and diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 A flowchart of a liver correction method after segmentation provided in some embodiments of the present application.
[0022] Figure 2 A flowchart of a liver segmentation correction method provided in some other embodiments of the present application.
[0023] Figure 3 A schematic diagram of the structure of a liver correction device after segmentation provided in some embodiments of the present application.
[0024] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.
[0025] Figure numerals: 201, acquisition module; 202, denoising module; 203, temporary merging module; 204, inter-leaf correction module; 205, unmerging module; 206, intra-leaf correction module; 207, merging module; 301, processor; 302, memory; 303, communication bus. DETAILED DESCRIPTION
[0026] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application claimed for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present application.
[0027] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.
[0028] First, please refer to Figure 1 Some embodiments of the present application provide a liver segmentation correction method for correcting a liver region obtained based on model segmentation. The liver segmentation correction method comprises the following steps: S1. Obtain liver segmentation results, where the liver segmentation results are eight liver regions segmented based on CT data based on a preset segmentation model; S2, remove external noise from liver segmentation results; S3, temporarily merging the eight liver regions into five liver regions according to the liver lobe type; S4, performing inter-lobe correction processing on the five liver regions according to the contact relationship of the connected domains corresponding to each type of liver segment; S5, restoring the five-segment liver region after interlobar correction processing to eight-segment liver region; S6. Extract three single-lobe and two-segment regions of the eight restored liver regions, and perform intra-lobe correction processing on each region.
[0029] Specifically, the post-segmentation correction method of the liver in the embodiment of the present application proposes a multi-step correction method that combines global processing and local optimization. This method first processes the entire liver area and then gradually refines it into local areas at different levels, taking into account both global consistency and fine-tuning local features.
[0030] More specifically, the liver segmentation result refers to the division result of the eight-segment liver area obtained after processing the CT data based on a preset segmentation model, which can be implemented by deep learning or machine learning models. The eight-segment liver area refers to a plurality of areas segmented based on the "eight segments" in the five-lobed and eight-segment structure of the liver tissue as segmentation targets. The eight segments in the division result refer to the areas corresponding to the eight-segment liver areas being divided into eight attribute categories, rather than directly referring to eight areas. The post-segmentation correction method of the liver in an embodiment of the present application aims to correct the abnormal small segmented areas in the eight-segment liver areas to obtain eight areas corresponding to the eight-segment liver areas that are accurate and can be used for medical diagnosis.
[0031] More specifically, in the embodiment of the present application, the segmentation result is preferably obtained by segmenting using the nnUNet model.
[0032] More specifically, external noise refers to erroneous segmentation results (small segmented areas) that are generated during the liver segmentation process and do not belong to the liver area and are separated from the liver area. Specifically, general denoising methods can be used to remove them.
[0033] More specifically, inter-lobe correction processing refers to the process of correcting the segmentation errors between different liver lobes, which can be achieved by using an analysis method based on the contact relationship of connected domains.
[0034] More specifically, intra-lobe correction processing refers to the process of correcting the segmentation error inside a single liver lobe, which can be achieved by using a method based on connected domain analysis and category attribute exchange.
[0035] Here’s how this application works: First, the segmentation results of the eight liver regions based on the preset segmentation model are obtained. This step uses a deep learning or machine learning model to perform a preliminary segmentation of the CT data, obtain the initial division of the eight liver segments, and remove external noise.
[0036] Next, the eight liver segments are temporarily merged into five segments according to the liver lobe type. This step can be simplified by configuring the temporary category attribute to merge the three single lobe and two segment regions into three temporary liver segment regions.
[0037] Then, based on the five liver segments, inter-lobe correction is performed. This step corrects the erroneous regions based on the contact relationship between connected domains.
[0038] Afterwards, the corrected five-segment liver area was restored to an eight-segment structure. This step maintained the original segmented structure of the liver, facilitating subsequent medical analysis.
[0039] Finally, the intra-leaf correction process is performed. In this step, two maximum connected regions are extracted for each of the three single-leaf two-segment regions, and then the error regions are corrected to correct the error regions of each single-leaf two-segment region.
[0040] The core of the liver segmentation and post-correction method of the embodiment of the present application is to utilize the anatomical structural characteristics of the liver, gradually optimize the segmentation results through temporary merging and step-by-step correction, and finally obtain a more accurate eight-segment liver area division. This multi-step, multi-level processing method can comprehensively handle different types of segmentation errors, thereby improving the accuracy of liver area segmentation and providing a more reliable basis for subsequent medical analysis and diagnosis.
[0041] In some preferred embodiments, step S2 comprises: S21, merge the eight liver regions into one category and retain the largest connected domain; S22. Synchronize the background area corresponding to the largest connected domain into the segmentation result to remove external noise.
[0042] Specifically, step S21 merges the eight liver regions into one category and retains the largest connected domain, which helps to identify the main liver region and exclude possible small isolated regions or noise. By merging and retaining the largest connected domain, the region that is most likely to represent the real liver structure can be effectively screened out; then, step S22 synchronizes the background region corresponding to the largest connected domain to the segmentation result, that is, through the reverse operation, the part outside the identified main liver region is defined as the background, thereby removing external noise in the original segmentation result. This method can effectively clean up the segmentation results and improve the accuracy of subsequent processing.
[0043] More specifically, the above steps solve the problem of external noise in the liver segmentation results in a simple and effective way. It utilizes the characteristics of the liver as a whole organ, assuming that the largest continuous area is most likely the real liver area, while other smaller or scattered areas may be noise caused by segmentation errors; this processing method has obvious advantages over traditional noise removal methods such as morphological operations or threshold filtering. Traditional methods may excessively remove effective information or fail to completely remove complex noise structures, while this processing method, by utilizing the overall characteristics of the liver, can more accurately identify and retain the real liver area, while effectively removing various forms of external noise, including discrete small pieces of noise and areas that are connected to the main body but do not belong to the liver. The liver segmentation post-correction method of the embodiment of the present application not only improves the accuracy of the segmentation results, but also provides a more reliable basis for subsequent liver analysis (such as liver segment division, lesion detection, etc.).
[0044] In some preferred embodiments, step S4 comprises: S41, extracting five first largest connected domains, where the five first largest connected domains are respectively the largest connected domains of each type of liver segment in the five liver regions; S42, correcting the first erroneous region to a liver segment category corresponding to the first largest connected domain in contact with the first erroneous region, wherein the first erroneous region is a connected domain outside the first largest connected domain and in contact with only one first largest connected domain; S43, correcting the second erroneous region to a liver segment category corresponding to the first largest connected domain with the largest contact area, wherein the second erroneous region is a connected domain outside the first largest connected domain and in contact with at least two first largest connected domains.
[0045] Specifically, step S41 establishes a reliable benchmark by extracting the maximum connected domain of each type of liver segment. Then, steps S42 and S43 respectively adopt different correction strategies for different types of error regions (only in contact with one or more maximum connected domains) to ensure the accuracy and comprehensiveness of the correction. This processing method can not only handle simple isolated error regions, but also complex error regions in contact with multiple liver segments, thereby significantly improving the accuracy of liver segmentation.
[0046] More specifically, the correction of the first error region can be done by using a regional growing algorithm or by directly defining the attribute category. Specifically, the growing algorithm starts from the boundary of the first error region and expands in the direction of the first maximum connected domain in contact with it until the first error region is completely covered. This processing method can ensure that the first error region is correctly merged into the adjacent first maximum connected domain; directly defining the attribute category directly defines the attribute category of the first error region as the liver segment category corresponding to the first maximum connected domain in contact with it, so that the first error region is merged into the corresponding first maximum connected domain. This processing method has the advantages of high processing efficiency and low computing power requirements.
[0047] More specifically, the correction of the second error region involves the calculation of the contact area of multiple first maximum connected domains. This can be achieved by calculating the contact area between the second error region and each first maximum connected domain in three-dimensional space. One possible method is to use voxel counting technology to count the number of voxels adjacent to each first maximum connected domain on the boundary of the second error region, thereby obtaining the contact area.
[0048] More specifically, in the embodiment of the present application, the correction strategy for the first error area and the second error area is based on these maximum connected domains, and the correction method is determined by different judgment criteria (single contact or multiple contacts). For isolated error areas (first error areas), the method can quickly and accurately merge them into the correct liver segment. For complex error areas (second error areas) located at the junction of multiple liver segments, the method makes a more accurate judgment by calculating the contact area, thereby improving the accuracy of segmentation. This hierarchical processing method enables the liver segmentation post-correction method of the embodiment of the present application to effectively process error areas of different complexities. In addition, the liver segmentation post-correction method of the embodiment of the present application does not require additional training data or complex model adjustments, and can be used as a post-processing step of the existing segmentation method to improve the overall segmentation performance.
[0049] In some preferred implementations, in step S43, the contact area of the second error region is calculated based on traversing and calculating the length of the separation line between the second error region and the other four types of first maximum connected domains in all corresponding slice planes.
[0050] Specifically, the present application provides a specific method for calculating the contact area between the second error region and the first maximum connected domain by introducing the concept of a slicing plane. The specific process is: first, traverse all slicing planes corresponding to the second error region. On each slicing plane, calculate the length of the dividing line between the second error region and the remaining four types of first maximum connected domains. The accumulation of these dividing line lengths can be used as an approximate measure of the contact area. By comparing the sum of the lengths of the dividing lines with different first maximum connected domains, it can be determined which first maximum connected domain the second error region has the largest contact area with. It is also possible to convert the contact area based on the length of the dividing line combined with the spacing thickness corresponding to the slicing plane, and then determine which first maximum connected domain the second error region has the largest contact area with.
[0051] More specifically, the liver segmentation and correction method of the embodiment of the present application has the following advantages: First, the three-dimensional problem is simplified into a two-dimensional problem, reducing the computational complexity. Second, by traversing multiple slice planes, the contact relationship of the entire three-dimensional structure is considered, rather than being limited to a single plane. Finally, the length of the separation line is used as a measure of the contact area, avoiding the difficulty of directly calculating the contact area of an irregular three-dimensional surface.
[0052] More specifically, in the present application, the concept of slice plane can be implemented in a variety of ways. For example, each cross section of the CT scan can be used as a slice plane. Alternatively, a specific slice direction can be selected as a slice plane based on the anatomical structure of the liver. In addition, the most appropriate slice plane direction can be dynamically selected based on the shape and position of the second error area.
[0053] More specifically, the length of the separation line can also be calculated using different methods. The simplest way is to directly calculate the length of the boundary line between the second error region and the first maximum connected domain on the plane. More complex methods may include considering the curvature of the boundary line or using weighted separation line length to more accurately reflect the contact relationship in three-dimensional space.
[0054] In some preferred embodiments, step S3 comprises: S31. According to the three single-lobe and two-segment liver division rule in the liver lobe type, three temporary category attributes are configured to merge the three single-lobe and two-segment liver regions in the eight-segment liver regions into three temporary liver segment regions to obtain five-segment liver regions.
[0055] Specifically, step S31 adopts the configuration mode of temporary category attributes to ensure the accuracy and traceability of the merging process. This merging mode not only realizes the conversion from eight segments to five segments, but also maintains the location information of the original area, providing a basis for subsequent correction processing; the use of temporary category attributes makes the merging process reversible, which is convenient for restoring the original eight-segment structure when needed.
[0056] More specifically, when configuring the temporary category attributes, a marking or labeling method can be used. For example, a corresponding temporary category label can be assigned to each pixel or voxel. This marking method can be implemented by using specific colors or values in image processing software, so that different temporary liver segment areas are visually easy to distinguish.
[0057] More specifically, the merging process can be performed by traversing each pixel or voxel in the eight-segment liver region and reallocating it to the corresponding temporary category according to the liver segment category to which it originally belongs. This process can be implemented through conditional judgment statements, for example, if a pixel originally belongs to the upper or lower segment of the left outer sector, it is assigned to temporary category A.
[0058] The liver post-segmentation correction method of the embodiment of the present application provides a flexible and reversible merging method. By using temporary category attributes, it can be easily restored to the original eight-segment structure when needed, which is very useful in subsequent analysis and processing; this method takes into account the actual anatomical structure of the liver, so that the merged five-segment liver area is more in line with the liver lobe division method in clinical practice, which helps to improve the accuracy of subsequent processing; thirdly, by reducing the number of areas to be processed (from eight to five), this method simplifies the subsequent inter-lobe correction processing steps and can improve the overall processing efficiency; finally, this method retains the detailed information of the original segmentation, while providing an intermediate level representation, which lays the foundation for subsequent liver area correction processing.
[0059] In some preferred embodiments, the intra-lobar correction treatment comprises: S61, extracting two second largest connected domains, where the two second largest connected domains are respectively the largest connected domains of each type of liver segment in the two segments of a single lobe; S62, correcting the third erroneous region to a liver segment category corresponding to the second largest connected domain with the largest contact area, wherein the third erroneous region is a connected domain that contacts two second largest connected domains and belongs to a liver segment category other than the two second largest connected domains; S63: Integrate the fourth error region into the corresponding second maximum connected region based on the category attribute of the exchanged non-maximum connected region, wherein the fourth error region is a connected region outside the two second maximum connected regions and belonging to a type of liver segment category in the two-segment region of a single lobe.
[0060] Specifically, step S61 extracts the maximum connected domain corresponding to the two types of liver segments in each single lobe two-segment area as a benchmark, providing a reliable reference for subsequent correction; step S62 and step S63 adopt different correction strategies for different types of error areas: for cross-category errors (the third error area), the correct category is determined by calculating the contact area; for errors in the same category (the fourth error area), they are fused by attribute exchange; this multi-level and targeted correction method can effectively improve the segmentation accuracy of the single lobe two-segment area, thereby improving the overall liver segmentation result.
[0061] More specifically, the contact area in step S62 can be calculated using the aforementioned dividing line length based on the slice plane, that is, the contact area is calculated by traversing and calculating all slice planes corresponding to the third error area, and the length of the dividing line between it and the other two types of second maximum connected domains; therefore, the implementation principle of step S62 is similar to that of step S43, and will not be repeated here.
[0062] More specifically, the processing of the fourth error region involves the exchange of category attributes. This can be achieved by label propagation or region merging. Specifically, it can start from the second largest connected domain and gradually expand outward to include the areas of the same category encountered but not belonging to the second largest connected domain. This process can be performed in an iterative manner until all fourth error regions are correctly classified, or the attribute category of the fourth error region can be directly switched between the two types of liver segment categories in the two-segment region of a single lobe, so that the fourth error region is directly integrated into the corresponding second largest connected domain. This processing method has the advantages of high processing efficiency and low computing power requirements.
[0063] In some preferred embodiments, Figure 2 As shown, the liver segmentation and correction method also includes the steps of: S7. Merge the two-segment region of the single lobe after the intra-lobe correction processing with the remaining two-segment region of the eight-segment liver region to obtain the corrected eight-segment liver region.
[0064] Specifically, the present application can better maintain the anatomical features of each region and reduce segmentation errors by first finely processing the two regions of a single lobe and then merging them with other regions.
[0065] It should be noted that the region refers to the model segmentation region composed of these liver regions to form eight liver regions, rather than the region fused into one whole.
[0066] More specifically, the merging operation can be performed using a region labeling method, that is, a unique identifier is assigned to each region, and the continuity and smoothness of the boundaries of each region are maintained when merging regions. During the merging process, the spatial relationship between adjacent regions is calculated to ensure the anatomical correctness of the merged region.
[0067] In some preferred embodiments, the three single-leaf two-section areas are: the upper section area of the left outer sector and the lower section area of the left outer sector, the upper section area of the right front sector and the lower section area of the right front sector, and the upper section area of the right rear sector and the lower section area of the right rear sector.
[0068] Specifically, the present application clearly defines three single-lobed and two-segment regions of the liver, including the left outer sector, the right anterior sector, and the right posterior sector, each of which is further divided into an upper segment and a lower segment. This division follows the anatomical characteristics of the liver and helps to accurately identify and locate the positions of different liver segments.
[0069] More specifically, the liver segmentation and correction method of the embodiment of the present application divides the liver region into three single lobes and two sections with clear anatomical positions, providing an accurate regional positioning basis for subsequent intra-lobe correction processing. This division method can better reflect the actual anatomical structure of the liver, which is conducive to improving the accuracy and reliability of liver segmentation. The divided regions are convenient for local refinement processing, thereby improving the overall segmentation effect.
[0070] In some preferred embodiments, the remaining two segments of the eight-segment liver region include the left inner lobe sector region and the caudate lobe region.
[0071] Specifically, the liver segmentation correction method of the embodiment of the present application makes the segmentation of the eight-segment liver region more complete and accurate by clearly specifying the specific anatomical locations of the remaining two segments, namely the left inner lobe sector region and the caudate lobe region; the accurate identification and segmentation of these two regions are of great significance for subsequent medical diagnosis and surgical planning. By clearly defining these two specific regions, the problem of accurate identification of the remaining regions during liver segmentation is effectively solved.
[0072] More specifically, in combination with the foregoing content, it can be known that the eight-segment liver area includes: the upper area of the left outer sector, the lower area of the left outer sector, the upper area of the right front sector, the lower area of the right front sector, the upper area of the right posterior sector, the lower area of the right posterior sector, the left inner lobe sector area and the caudate lobe area, a total of eight attribute types of areas; correspondingly, the five-segment liver area includes: the left outer sector area, the right front sector area, the right posterior sector area, the left inner lobe sector area and the caudate lobe area, a total of five attribute types of areas.
[0073] Second, please refer to Figure 3 Some embodiments of the present application further provide a liver segmentation correction device for correcting a liver region obtained based on model segmentation, the liver segmentation correction device comprising: An acquisition module 201 is used to acquire a liver segmentation result, where the liver segmentation result is eight liver regions segmented based on CT data based on a preset segmentation model; A denoising module 202, used to remove external noise from the liver segmentation result; A temporary merging module 203 is used to temporarily merge the eight liver regions into five liver regions according to the liver lobe type; An inter-lobe correction module 204 is used to perform inter-lobe correction processing on the five liver regions according to the contact relationship of the connected domains corresponding to each type of liver segment; The unmerging module 205 is used to restore the five-segment liver region after the interlobar correction process to the eight-segment liver region; The intra-lobe correction module 206 is used to extract three single-lobe and two-segment regions of the restored eight-segment liver region and perform intra-lobe correction processing on each of them.
[0074] The liver segmentation correction device of the embodiment of the present application gradually optimizes the segmentation results by temporary merging and step-by-step correction, and finally obtains a more accurate eight-segment liver area division. This multi-step and multi-level processing method can comprehensively handle different types of segmentation errors, thereby improving the accuracy of liver area segmentation and providing a more reliable basis for subsequent medical analysis and diagnosis.
[0075] In some preferred embodiments, the liver segmentation correction device further comprises: The merging module 207 is used to merge the two segments of the single lobe after the intra-lobe correction processing with the remaining two segments of the eight-segment liver region to obtain the corrected eight-segment liver region.
[0076] In some preferred embodiments, the liver correction device after segmentation according to the embodiment of the present application is used to execute the liver correction method after segmentation provided in the first aspect above.
[0077] Third, please refer to Figure 4Some embodiments of the present application also provide a structural diagram of an electronic device. The present application provides an electronic device, including: a processor 301 and a memory 302. The processor 301 and the memory 302 are interconnected and communicate with each other through a communication bus 303 and / or other forms of connection mechanisms (not marked). The memory 302 stores computer-readable instructions executable by the processor 301. When the electronic device is running, the processor 301 executes the computer-readable instructions to execute the method in any optional implementation of the above embodiments.
[0078] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method in any optional implementation of the above embodiment is executed. The computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, disk or optical disk.
[0079] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.
[0080] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0081] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.
[0082] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.
[0083] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A liver segmentation correction method for correcting a liver region obtained by model segmentation, characterized in that: The liver segmentation and correction method comprises the following steps: S1. Obtaining a liver segmentation result, wherein the liver segmentation result is eight liver regions segmented based on CT data based on a preset segmentation model; S2, remove external noise from liver segmentation results; S3, temporarily merging the eight liver regions into five liver regions according to the liver lobe type; S4, performing inter-lobe correction processing on the five liver regions according to the contact relationship of the connected domains corresponding to each type of liver segment; S5, restoring the five-segment liver region after interlobar correction processing to eight-segment liver region; S6. Extract three single-lobe and two-segment regions of the eight restored liver regions, and perform intra-lobe correction processing on each region.
2. The method for correcting liver segmentation according to claim 1, characterized in that: Step S2 includes: S21, merging the eight liver regions into one category, and retaining the largest connected domain; S22. Synchronize the background area corresponding to the maximum connected domain into the segmentation result to remove the external noise.
3. The liver segmentation and correction method according to claim 1, characterized in that: Step S4 includes: S41, extracting five first maximum connected domains, where the five first maximum connected domains are respectively the maximum connected domains of each type of liver segment in the five liver regions; S42, correcting the first erroneous region to a liver segment category corresponding to the first maximum connected domain in contact with the first erroneous region, wherein the first erroneous region is a connected domain outside the first maximum connected domain and in contact with only one first maximum connected domain; S43: Correct the second erroneous region to a liver segment category corresponding to the first largest connected domain with the largest contact area, wherein the second erroneous region is a connected domain outside the first largest connected domain and in contact with at least two first largest connected domains.
4. The method for correcting liver segmentation according to claim 3, characterized in that: In step S43, the contact area of the second error region is calculated by traversing and calculating the length of the separation line between the second error region and the other four types of the first maximum connected domains in all corresponding slice planes.
5. The liver segmentation and correction method according to claim 1, characterized in that: The intra-lobar correction process includes: S61, extracting two second largest connected domains, where the two second largest connected domains are respectively the largest connected domains of each type of liver segment in the two-segment region of the single lobe; S62, correcting the third erroneous region to a liver segment category corresponding to the second largest connected domain with the largest contact area, wherein the third erroneous region is a connected domain that contacts two of the second largest connected domains and belongs to a liver segment category other than the two second largest connected domains; S63: Integrate the fourth error region into the corresponding second maximum connected region based on the category attribute of the exchanged non-maximum connected region, wherein the fourth error region is a connected region outside the two second maximum connected regions and belonging to a type of liver segment category in the single lobe two-segment region.
6. The liver segmentation and correction method according to claim 1, characterized in that: The liver segmentation and correction method further comprises the steps of: S7. Merging the two segments of the single lobe after the intra-lobe correction processing with the remaining two segments of the eight-segment liver region to obtain the corrected eight-segment liver region.
7. The liver segmentation and correction method according to claim 1, characterized in that: The three single-leaf two-section areas are: the upper section area of the left outer sector and the lower section area of the left outer sector, the upper section area of the right front sector and the lower section area of the right front sector, and the upper section area of the right rear sector and the lower section area of the right rear sector.
8. A liver segmentation correction device, used to correct the liver region obtained based on model segmentation, characterized in that: The liver segmentation and correction device comprises: An acquisition module, used for acquiring a liver segmentation result, wherein the liver segmentation result is eight liver regions segmented based on CT data based on a preset segmentation model; A denoising module, used to remove external noise from liver segmentation results; A temporary merging module, used for temporarily merging the eight liver regions into five liver regions according to the type of liver lobes; An inter-lobe correction module, used for performing inter-lobe correction processing on the five liver regions according to the contact relationship of the connected domains corresponding to each type of liver segment; An unmerging module is used to restore the five-segment liver region to the eight-segment liver region after interlobar correction; The intra-lobe correction module is used to extract three single-lobe and two-segment regions of the restored eight-segment liver region and perform intra-lobe correction processing on each of them.
9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the steps in the method according to any one of claims 1 to 7 are executed.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are executed.