Bone segmentation result correction method and device, equipment and medium

Abstract

A bone segmentation result correction method, apparatus and device, and a medium, comprising: performing direction correction on a bone region having an orientation attribute in a bone segmentation result to obtain a bone segmentation result after direction correction, and when the number of connected domains of a rib mask image does not meet a normal rib number range, determining that the rib mask image is in the normal rib number range; and repairing each rib mask in the rib region in the bone segmentation result after direction correction according to the connected domain analysis result of the image region corresponding to each rib mask. And when the number of the connected domains of the image region corresponding to a certain vertebra mask is greater than a first threshold value, repairing the vertebra mask according to a judgment result that an abnormal connected domain in the image region corresponding to the vertebra mask should be fused with which one of the adjacent vertebra regions of the vertebra mask. And when the number of the connected domains in the upper limb region after the pixel value updating is completed accords with a preset number, repairing the skeleton label of the abnormal skeleton mask in the upper limb region into an upper limb skeleton label. And the accuracy of correcting the bone segmentation result is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image segmentation, and particularly relates to a method, device, equipment and medium for correcting bone segmentation results. Background Art

[0002] Computed Tomography (CT) is a medical imaging technique that uses X-rays to generate images of the internal structures of the body. Performing bone segmentation in CT images is an important medical image processing task, which is of great significance in clinical diagnosis, surgical planning, fracture assessment, etc.

[0003] Modern medical image segmentation technologies have been developing rapidly. For example, in deep learning, classic segmentation networks such as Unet and Transformer use a large number of network parameters through convolutional networks and attention networks, and both exceed traditional medical image segmentation technologies in terms of segmentation accuracy and speed. However, deep learning models usually contain multiple hyperparameters, and each hyperparameter has multiple possible values, which leads to an exponential growth in the number of potential hyperparameter combinations. Each time the model is trained and evaluated, it requires a large amount of computing resources. Especially when dealing with large-scale medical image datasets, it is difficult to comprehensively explore the hyperparameter space. To reduce the difficulty of manual parameter tuning and improve the clinical usability of deep learning model segmentation, automated segmentation frameworks are also actively developing. For example, the nnUnet framework can select appropriate network architectures, data augmentation methods, training parameters, and other hyperparameters through highly automated configuration, and achieve excellent results in conventional medical image segmentation tasks. Based on nnUnet, the whole-body segmentation framework TotalSegmentor trains a segmentation model with a certain accuracy using a labeled dataset, realizing the automatic combined segmentation of the whole body organs and bones of the entire human body, and at the same time can be used out of the box, significantly reducing the threshold of CT image bone segmentation. However, since it is a general human segmentation model, the segmentation of bone types is not perfect, and the CT data of bones in different parts also have different requirements for hyperparameters, resulting in the TotalSegmentor framework being unable to be accurately applied to bone segmentation. In addition, the SegmentAnyBone framework focuses on bone segmentation training using a large medical bone dataset based on SAM, enabling it to segment the bones of 17 parts of the whole body with relatively high accuracy, but it is mainly applied to MRI images and has fewer segmentation categories.

[0004] In deep learning methods, due to the limitation of hardware video memory, large-sized medical images cannot be directly input into the device for training and inference. Instead, the images need to be processed in chunks sequentially. This method causes the network to lack the context information in the images. At the same time, when there are situations such as blurred bone edges, overlapping adjacent bones, and differences in image quality caused by different CT devices and scanning parameters, the final segmentation result will degenerate to varying degrees compared to the true mask image. Therefore, it is necessary to use appropriate post-processing methods to correct the results of the deep learning segmentation model and further improve the accuracy and reliability of bone segmentation. The post-processing method can automatically correct the possible errors of the segmentation algorithm at the software level when the video memory is insufficient, without the need to manually edit the segmentation file again. This method can significantly improve the speed of disease diagnosis and treatment planning. Its core is to conduct in-depth statistical analysis on the original segmentation result, identify and calibrate outliers and abnormal regions, and gradually optimize these abnormalities, ultimately making the corrected segmentation result more in line with the actual distribution law. Commonly used post-processing methods include morphological operations, connected component analysis, and feature engineering, etc. Traditional post-processing methods for bone segmentation usually use global thresholds to process images without considering the spatial relationship and context connection between pixels, which may lead to blurred boundaries or errors. Existing hardware cannot support the segmentation model to directly infer full-body high-resolution CT images. The images need to be cut into pieces according to certain rules and then processed in batches. This method causes a loss of some spatial correlation between the segmentation result blocks, manifested in the segmentation result as misaligned bone masks, indistinguishable left and right of bone masks, increased abnormal regions, etc. Simple image post-processing such as threshold processing cannot solve the above problems and requires manual correction of the mask again, which is time-consuming and laborious. To sum up, some existing post-processing schemes only process abnormal bone masks, without considering the factors of the human bone structure and other potentially related bone masks. At the same time, anatomical prior knowledge is not used to guide the design and optimization of the post-processing algorithm, and the shape range and numerical size during the correction process are not restricted by prior knowledge. Therefore, the accuracy of correcting the bone segmentation result through this post-processing method is not high. Summary of the Invention

[0005] The main technical problem to be solved by the present invention is that global thresholds are usually used to post-process the bone segmentation results obtained by using a segmentation model without considering the spatial relationship and context connection between pixels, resulting in relatively low accuracy of the bone segmentation results.

[0006] According to a first aspect, in one embodiment, a method for correcting a bone segmentation result is provided, including:

[0007] Obtain the bone segmentation result, wherein the bone segmentation result includes multiple different bone regions segmented from the human body scan image. Each of the bone regions contains a different bone mask, and the bone mask is the bone identified from the human body scan image. Each bone mask has a bone label, and the bone regions include a vertebral region, a rib region, and an upper limb region;

[0008] Calculate the centroid coordinate point of the vertebral region, and use the centroid coordinate point as a demarcation point to correct the direction of the bone regions with orientation attributes in the bone segmentation result, obtaining the direction-corrected bone segmentation result, wherein the bone regions with orientation attributes refer to the regions where the bones that distinguish left and right are located;

[0009] For the rib region in the direction-corrected bone segmentation result, unify the bone labels of each rib mask therein to obtain a rib mask image, perform connected component analysis on the rib mask image to obtain the number of connected components of the rib mask image. When the number of connected components does not meet the normal rib number range, perform connected component analysis on the image region corresponding to each rib mask respectively to obtain the connected component analysis result of the image region corresponding to each rib mask, and perform repair processing on each rib mask in the rib region of the direction-corrected bone segmentation result according to the connected component analysis result, obtaining the repaired rib region;

[0010] For the vertebral region in the direction-corrected bone segmentation result, perform connected component analysis on the image region corresponding to each vertebral mask to obtain the number of connected components of the image region corresponding to each vertebral mask. When the number of connected components of the image region corresponding to a certain vertebral mask is greater than a preset first threshold, regard the connected components other than the connected component with the largest area in the image region corresponding to this vertebral mask as abnormal connected components, judge which of the adjacent vertebral regions of this vertebral mask the abnormal connected components should be fused with, and perform repair processing on this vertebral mask according to the judgment result, obtaining the repaired vertebral region, wherein the vertebral region includes multiple different vertebral masks connected in a spatial order from top to bottom, and the adjacent vertebral regions include the image regions corresponding to the vertebral masks that have an adjacent relationship with the vertebral mask with the number of connected components greater than the preset first threshold in the upper and lower spaces;

[0011] For the upper limb region in the bone segmentation result after direction correction, use a randomly selected pixel point in the upper limb region as the starting point, and use the image region corresponding to the abnormal bone mask in the upper limb region as the end point. Traverse the pixel points in the upper limb region starting from the starting point and perform pixel value update processing on the pixel value of each pixel point until the number of connected components in the upper limb region after pixel value update processing meets the preset number. Repair the bone label of the abnormal bone mask to the upper limb bone label, where the bone label of the abnormal bone mask is different from the upper limb bone label.

[0012] In some embodiments, the connected component analysis result of the image region corresponding to each rib mask includes the connected component image of the image region corresponding to each rib mask and the number of connected components corresponding to the connected component image. The repairing process of each rib mask in the rib region of the bone segmentation result after direction correction according to the connected component analysis result to obtain the repaired rib region includes:

[0013] When the number of connected components corresponding to the connected component image is greater than a preset second threshold, retain the connected component with the largest area in the connected component image, and eliminate all connected components other than the connected component with the largest area to obtain the connected component image of the image region corresponding to each processed rib mask;

[0014] Calculate the reference coordinates of the connected component image of the image region corresponding to each processed rib mask, sort the connected component images of the image regions corresponding to each processed rib mask according to the size of the reference coordinates, and assign the predefined rib mask values to the connected component images of the image regions corresponding to each processed rib mask in the order of the sorting process to obtain the repaired rib region, where the reference coordinate is the coordinate of the centroid point of the connected component image of the image region corresponding to each processed rib mask on the Z axis.

[0015] In some embodiments, the direction correction of the bone region with orientation attributes in the bone segmentation result using the centroid coordinate point as the demarcation point to obtain the bone segmentation result after direction correction includes:

[0016] Assign the mask value of all bone masks in the bone region with orientation attributes to a preset first mask value, where the first mask value is the mask value of the left bone mask or the mask value of the right bone mask;

[0017] When the first mask value is the mask value of the left bone mask, assign the mask value of the bone mask located on the right side of the centroid coordinate point in the bone region with orientation attributes to the mask value of the right bone mask;

[0018] When the first mask value is the mask value of the right - hand bone mask, assign the mask value of the bone mask on the left side of the centroid coordinate point in the bone region with orientation attributes as the mask value of the left - hand bone mask; wherein, the mask value is used to represent the bone label of the bone mask.

[0019] In some embodiments, the adjacent vertebra regions include a first adjacent vertebra region and a second adjacent vertebra region. The first adjacent vertebra region is the image region corresponding to the vertebra mask that has an adjacent relationship with the vertebra mask whose number of connected components is greater than a preset first threshold in the upper - layer space. The second adjacent vertebra region is the image region corresponding to the vertebra mask that has an adjacent relationship with the vertebra mask whose number of connected components is greater than a preset first threshold in the lower - layer space. Determining which of the adjacent vertebra regions of the vertebra mask the abnormal connected component should be fused with includes:

[0020] Replace the mask value of the abnormal connected component with the mask value of the first adjacent vertebra region, and perform connected - component analysis on the first adjacent vertebra region to obtain the number of connected components of the first adjacent vertebra region;

[0021] Determine whether the number of connected components of the first adjacent vertebra region is consistent with the number of connected components of the first adjacent vertebra region obtained when performing connected - component analysis on the image region corresponding to each vertebra mask, and determine whether the number of connected components of the first adjacent vertebra region is less than a preset third threshold;

[0022] When the number of connected components of the first adjacent vertebra region is consistent with the number of connected components of the first adjacent vertebra region obtained when performing connected - component analysis on the image region corresponding to each vertebra mask and the number of connected components of the first adjacent vertebra region is less than the third threshold, it is determined that the abnormal connected component should be fused with the first adjacent vertebra region;

[0023] Otherwise, replace the mask value of the abnormal connected component with the mask value of the second adjacent vertebra region, and perform connected - component analysis on the second adjacent vertebra region to obtain the number of connected components of the second adjacent vertebra region;

[0024] Determine whether the number of connected components of the second adjacent vertebra region is consistent with the number of connected components of the second adjacent vertebra region obtained when performing connected - component analysis on the image region corresponding to each vertebra mask, and determine whether the number of connected components of the second adjacent vertebra region is less than a preset fourth threshold;

[0025] When the number of connected components in the second adjacent vertebral region is consistent with the number of connected components in the second adjacent vertebral region obtained by performing connected component analysis on the image region corresponding to each vertebral mask, and the number of connected components in the second adjacent vertebral region is less than the fourth threshold, it is determined that the abnormal connected component should be fused with the second adjacent vertebral region;

[0026] Otherwise, calculate the centroid coordinates of the abnormal connected component, the centroid coordinates of the first adjacent vertebral region, and the centroid coordinates of the second adjacent vertebral region respectively, calculate the first centroid distance between the centroid coordinates of the abnormal connected component and the centroid coordinates of the first adjacent vertebral region, calculate the second centroid distance between the centroid coordinates of the abnormal connected component and the centroid coordinates of the second adjacent vertebral region, and determine that the abnormal connected component should be fused with the adjacent vertebral region corresponding to the minimum centroid distance among the first centroid distance and the second centroid distance.

[0027] In some embodiments, the repairing the vertebral mask according to the judgment result to obtain a repaired vertebral region includes:

[0028] Modify the bone mask of the abnormal connected component to the bone mask of the adjacent vertebral region with which the abnormal connected component should be fused to obtain a repaired vertebral region.

[0029] In some embodiments, the pixel value updating process for each pixel point includes:

[0030] Judge the magnitude relationship between the pixel value of each pixel point and a preset upper limb threshold, where the upper limb threshold is the difference between the mask value of the upper limb region and the preset background mask value;

[0031] When the pixel value of the pixel point is less than or equal to the upper limb threshold, assign the pixel value of the pixel point to the mask value of the upper limb region;

[0032] When the pixel value of the pixel point is greater than the upper limb threshold, stop executing the pixel updating process.

[0033] In some embodiments, the background mask value is 0. For the upper limb region in the bone segmentation result after direction correction, before repairing the abnormal bone mask in the upper limb region, save the mask value of the scapula mask of the scapula region connected to the upper limb region, and set the mask value of the scapula mask to zero after saving the mask value of the scapula mask, so that when traversing the pixel points in the upper limb region from the starting point and performing pixel value updating process on each pixel point, the mask value of the scapula mask in the scapula region is equal to the background mask value.

[0034] According to a second aspect, in one embodiment, a bone segmentation result correction device is provided, including:

[0035] A result acquisition module, configured to acquire a bone segmentation result. Among them, in the bone segmentation result, multiple different bone regions segmented from a human body scan image are included. In each of the bone regions, different bone masks are included. The bone mask is the bone recognized from the human body scan image. Each bone mask has a bone label. The bone regions include a vertebral region, a rib region, and an upper limb region;

[0036] A direction correction module, configured to calculate the centroid coordinate point of the vertebral region, and use the centroid coordinate point as a demarcation point to correct the direction of the bone regions with orientation attributes in the bone segmentation result, so as to obtain a bone segmentation result with corrected direction. Among them, the bone regions with orientation attributes refer to the regions where the bones that distinguish left and right are located;

[0037] A rib correction module, for the rib region in the bone segmentation result with corrected direction, unify the bone labels of each rib mask therein to obtain a rib mask image, perform connected component analysis on the rib mask image to obtain the number of connected components of the rib mask image. When the number of connected components does not meet the normal rib number range, perform connected component analysis on the image region corresponding to each rib mask respectively to obtain the connected component analysis result of the image region corresponding to each rib mask, and perform repair processing on each rib mask in the rib region in the bone segmentation result with corrected direction according to the connected component analysis result to obtain a repaired rib region;

[0038] A vertebral correction module, for the vertebral region in the bone segmentation result with corrected direction, perform connected component analysis on the image region corresponding to each vertebral mask to obtain the number of connected components of the image region corresponding to each vertebral mask. When the number of connected components of the image region corresponding to a certain vertebral mask is greater than a preset first threshold, use the connected components other than the connected component with the largest area in the image region corresponding to the vertebral mask as abnormal connected components, and judge which of the adjacent vertebral regions of the vertebral mask the abnormal connected components should be fused with and perform repair processing on the vertebral mask according to the judgment result to obtain a repaired vertebral region. Among them, the vertebral region includes multiple different vertebral masks connected in a top-down spatial order. The adjacent vertebral regions include the image regions corresponding to the vertebral masks that have an adjacent relationship with the vertebral mask with the number of connected components greater than the preset first threshold in the upper and lower spaces;

[0039] An upper limb correction module, for the upper limb region in the bone segmentation result after direction correction, taking a randomly selected pixel point in the upper limb region as the starting point and the image region corresponding to the abnormal bone mask in the upper limb region as the end point, traversing the pixel points in the upper limb region starting from the starting point and performing pixel value update processing on the pixel value of each pixel point until the number of connected components in the upper limb region after pixel value update processing meets the preset number, and repairing the bone label of the abnormal bone mask to the upper limb bone label, where the bone label of the abnormal bone mask is different from the upper limb bone label.

[0040] According to a third aspect, in one embodiment, a bone segmentation result correction device is provided, including:

[0041] A memory, for storing programs;

[0042] A processor, for implementing the bone segmentation result correction method by executing the programs stored in the memory.

[0043] According to a fourth aspect, in one embodiment, a computer-readable storage medium is provided, on which a program is stored, and the program can be executed by a processor to implement the bone segmentation result correction method.

[0044] Based on the bone segmentation result correction method, device, device and computer-readable storage medium of the above embodiments, since the centroid coordinate point of the vertebral region is used as the demarcation point to correct the direction of the bone region with orientation attributes in the bone segmentation result, the direction of the bone region with orientation attributes in the bone segmentation result can be accurately distinguished. When repairing the rib region, when the number of connected components in the rib mask image does not meet the normal rib number range, the connected component analysis is performed on the image region corresponding to each rib mask respectively, and each rib mask is repaired according to the connected component analysis result, improving the accuracy of the repair. When repairing the vertebral region, the adjacent vertebral regions with an adjacent relationship in the spatial level are referred to, and the repair is performed in the order of connection of the vertebral masks, making the repair result more reasonable and accurate. When repairing the upper limb region, the pixel points in the upper limb region are traversed and the pixel value update processing is performed on the pixel value of the pixel points, so as to realize the comprehensive repair of the upper limb region. In summary, the technical solution of the present application improves the accuracy of bone segmentation result correction. Description of the Drawings

[0045] Figure 1 is a flowchart of the bone segmentation result correction method of the embodiment of the present application;

[0046] Figure 2 is a reference diagram of a bone segmentation result of an embodiment;

[0047] Figure 3 A flowchart for direction correction of a bone region with an orientation attribute in a bone segmentation result using the centroid coordinate point as a demarcation point to obtain a bone segmentation result with corrected direction for an embodiment;

[0048] Figure 4 A comparison diagram before and after orientation correction of the scapula for an embodiment;

[0049] Figure 5 A comparison diagram before and after orientation correction of the lower limb region for an embodiment;

[0050] Figure 6 A flowchart for repairing each rib mask in the rib region of the bone segmentation result with corrected direction according to the connected component analysis result to obtain a repaired rib region for an embodiment;

[0051] Figure 7 A comparison diagram of the rib region before and after repair for an embodiment;

[0052] Figure 8 A flowchart for determining which adjacent vertebral region of the vertebral mask an abnormal connected component should be fused with for an embodiment;

[0053] Figure 9 A comparison diagram of the vertebral region before and after repair for an embodiment;

[0054] Figure 10 A flowchart for pixel value update processing of the pixel value of each pixel point for an embodiment;

[0055] Figure 11 A comparison diagram of the upper limb region before and after repair for an embodiment;

[0056] Figure 12 A comparison diagram of the bone segmentation result before and after repair for an embodiment;

[0057] Figure 13 A structural schematic diagram of a bone segmentation result correction device for an embodiment;

[0058] Figure 14 A structural schematic diagram of a bone segmentation result correction device for an embodiment. Detailed implementation manners

[0059] The present invention will be further described in detail below with reference to specific embodiments in conjunction with the accompanying drawings. Similar elements in different embodiments are denoted by related similar element numbers. In the following embodiments, many detailed descriptions are provided to enable a better understanding of the present application. However, those skilled in the art can easily recognize that some of the features can be omitted in different situations, or can be replaced by other elements, materials, or methods. In some cases, some operations related to the present application are not shown or described in the specification to avoid obscuring the core part of the present application with excessive descriptions. For those skilled in the art, it is not necessary to describe these related operations in detail, and they can fully understand the related operations based on the descriptions in the specification and the general technical knowledge in the art.

[0060] In addition, the features, operations, or characteristics described in the specification can be combined in any suitable manner to form various embodiments. At the same time, the steps or actions in the method description can also be reordered or adjusted in a manner obvious to those skilled in the art. Therefore, the various sequences in the specification and the drawings are only for clearly describing a certain embodiment and do not mean that they are the necessary sequences, unless it is stated that a certain sequence must be followed.

[0061] The serial numbers assigned to the components herein, such as "first", "second", etc., are only used to distinguish the described objects and do not have any sequential or technical meaning. The terms "connected" and "coupled" used in the present application, unless otherwise specified, both include direct and indirect connections (couplings).

[0062] Please refer to Figure 1 , in an embodiment of the present invention, a method for correcting bone segmentation results is provided, including steps S10 to S50, which will be specifically described below.

[0063] Step S10: Obtain the bone segmentation result.

[0064] In some embodiments, a deep learning segmentation model is used to perform bone segmentation on a computed tomography (CT) image of a human body to obtain the bone segmentation result.

[0065] In some embodiments, the bone segmentation result includes multiple different bone regions segmented from the human body scan image. The bone regions include vertebral regions, rib regions, and upper limb regions. Each bone region contains different bone masks. The bone mask is the bone recognized from the human body scan image, and each bone mask has a bone label for uniquely identifying the bone mask. In some embodiments, to facilitate distinguishing the bone masks from the image, a mask value is assigned to each bone mask to represent the bone label of the bone mask, and each bone label corresponds to a mask value.

[0066] Please refer to Figure 2 , in some embodiments, the bone region in the human body scan image outlined by the red frame is the rib region. Different bone masks are included in the rib region, that is, multiple different ribs are identified in the rib region. Each rib has a bone label. In Figure 2 the rib region of

[0067] Step S20: Calculate the centroid coordinate point of the vertebral region, and use the centroid coordinate point as a demarcation point to correct the direction of the bone region with orientation attributes in the bone segmentation result, so as to obtain the bone segmentation result after direction correction.

[0068] In some embodiments, since the training data set used in the TotalSegmentor segmentation framework does not distinguish between the left and right orientations during the training of the model, the bone masks in the bone segmentation result inferred do not distinguish between orientations either. In subsequent medical diagnosis tasks, it is impossible to quickly locate only based on the bone masks. Therefore, it is necessary to correct the direction of the bone region with orientation attributes in the bone segmentation result.

[0069] In some embodiments, the expression of the centroid coordinate point of the vertebral region is as follows:

[0070]

[0071] where C(X,Y,Z) represents the centroid coordinate point of the vertebral region, R represents the sum of the masses of each point in the vertebral region, and ∑ (x,y,z)∈Rx represents the sum of the products obtained by multiplying the mass of each point in the vertebral region by the coordinate of the point on the x-axis, and the sum of the products of each point is obtained. ∑ (x,y,z)∈Ry represents the sum of the products obtained by multiplying the mass of each point in the vertebral region by the coordinate of the point on the y-axis, and the sum of the products of each point is obtained. ∑ (x,y,z)∈Rz represents the sum of the products obtained by multiplying the mass of each point in the vertebral region by the coordinate of the point on the z-axis, and the sum of the products of each point is obtained.

[0072] Please refer to Figure 3 , in some embodiments, step S20 uses the centroid coordinate point as a demarcation point to correct the direction of the bone region with orientation attributes in the bone segmentation result, so as to obtain the bone segmentation result after direction correction, including steps S21 to S23, which are specifically described below.

[0073] Step S21: Assign the mask value of all bone masks in the bone region with orientation attributes to a preset first mask value.

[0074] In some embodiments, the bone region with an orientation attribute refers to the region where the bones that distinguish left and right are located. The first mask value is the mask value of the left bone mask or the mask value of the right bone mask.

[0075] Step S22: When the first mask value is the mask value of the left bone mask, assign the mask value of the bone mask on the right side of the centroid coordinate point in the bone region with an orientation attribute to the mask value of the right bone mask.

[0076] Step S23: When the first mask value is the mask value of the right bone mask, assign the mask value of the bone mask on the left side of the centroid coordinate point in the bone region with an orientation attribute to the mask value of the left bone mask.

[0077] Please refer to Figure 4 , Figure 4 which contains two pairs of scapulas. Among them, Figure 4 the left pair of scapulas corresponds to the scapula region in the bone segmentation result. In this scapula, except for the regions outlined by the two red frames, the orientations of other regions are disordered. In Figure 4 the bone labels of the bone masks are identified by different colors. It can be seen from the figure that there is more than one bone mask in the left scapula of the left pair of scapulas, and the correct bone mask should be the mask identified by the lower left red frame in this scapula. Therefore, the orientation of this scapula needs to be corrected. Figure 4 The right pair of scapulas corresponds to the scapula region in the bone segmentation result after orientation correction. It can be known that the region with disordered orientation outlined by the red frame in the left scapula has been corrected.

[0078] Please refer to Figure 5 , Figure 5 which contains the lower limb regions in the bone segmentation results before and after direction correction. Among them, Figure 5 the left side is the lower limb region in the bone segmentation result, Figure 5 and the right side is the lower limb region after orientation correction.

[0079] Step S30: For the rib region in the bone segmentation result after direction correction, unify the bone labels of each rib mask therein to obtain a rib mask image. Perform connected component analysis on the rib mask image to obtain the number of connected components of the rib mask image. When the number of connected components does not meet the normal rib number range, perform connected component analysis on the image region corresponding to each rib mask respectively to obtain the connected component analysis results of the image region corresponding to each rib mask. Repair each rib mask in the rib region in the bone segmentation result after direction correction according to the connected component analysis results to obtain the repaired rib region.

[0080] In some embodiments, when performing segmentation using a depth segmentation model, the training images are cut into small pieces and then input into the depth segmentation model in batches for calculation. As a result, the depth segmentation model is unable to understand the spatial position relationship between the ribs to a certain extent, leading to large-scale misalignment of the rib masks and the appearance of segmentation noise. Therefore, it is necessary to repair the rib regions in the bone segmentation result after direction correction.

[0081] In some embodiments, the bone labels of each rib mask in the rib region are unified to obtain a rib mask image. Then, connected component analysis is performed on the rib mask image to obtain a connected component image of the rib mask image. According to the connected component image of the rib mask image, the number of connected components of the rib mask image can be obtained. When the number of connected components does not fall within the normal rib number range, it indicates that there is a rib segmentation error in the rib mask image. Therefore, it is necessary to perform connected component analysis on the image region corresponding to each rib mask separately to accurately locate the rib mask with the error for subsequent rib repair.

[0082] In some embodiments, the normal rib number range is [22, 24].

[0083] In some embodiments, connected component analysis is performed on the image region corresponding to each rib mask separately to obtain the connected component analysis result of the image region corresponding to each rib mask. Among them, the connected component analysis result of the image region corresponding to each rib mask includes the connected component image of the image region corresponding to each rib mask and the number of connected components corresponding to the connected component image.

[0084] Please refer to Figure 6 , in some embodiments, step S30 repairs each rib mask in the rib region of the bone segmentation result after direction correction according to the connected component analysis result to obtain a repaired rib region, including steps S31 to S32, which are specifically described below.

[0085] Step S31: When the number of connected components corresponding to the connected component image is greater than a preset second threshold, the connected component with the largest area in the connected component image is retained, and all connected components other than the connected component with the largest area are eliminated to obtain the connected component image of the image region corresponding to each rib mask after processing.

[0086] In some embodiments, the second threshold is 1. When the number of connected regions in the connected region image is greater than 1, it indicates that there is more than one connected region in the connected region image of the image area corresponding to the rib mask, while there is only one connected region in the connected region image of the image area corresponding to a normal rib mask, that is, the image area corresponding to the rib mask is taken as a complete connected region. At this time, for the rib mask with the number of connected regions greater than 1, the connected region with the largest area in its connected region image is retained, and all connected regions other than the connected region with the largest area are eliminated, that is, the abnormally segmented regions that appear are eliminated, and only the part with the largest connected region area is retained as the connected region image of the image area corresponding to the processed rib mask.

[0087] Step S32: Calculate the reference coordinates of the connected region image of the image area corresponding to each processed rib mask, sort the connected region images of the image area corresponding to each processed rib mask according to the magnitudes of the reference coordinates, and assign mask values to the connected region images in sequence.

[0088] Specifically, first calculate the reference coordinates of the connected region image of the image area corresponding to each processed rib mask, then perform a sorting process on the connected region images of the image area corresponding to each processed rib mask according to the magnitudes of the reference coordinates, and sequentially assign the pre-defined rib mask values to the connected region images of the image area corresponding to each processed rib mask in the order of the sorting process to obtain the repaired rib region.

[0089] In some embodiments, the reference coordinate is the Z-axis coordinate of the centroid point of the connected region image of the image area corresponding to each processed rib mask. Since the arrangement order of multiple rib masks is disrupted during the above-mentioned connected region analysis, it is necessary to sort the connected region images of the image area corresponding to each processed rib mask according to the magnitudes of the reference coordinates. The rib mask value is used to represent the bone label of the rib mask. By sequentially assigning the pre-defined rib mask values to the connected region images of the image area corresponding to each processed rib mask in the order of the sorting process, the obtained repaired rib region follows the normal arrangement order of the ribs in the anatomical prior knowledge.

[0090] Please refer to Figure 7 , Figure 7 On the left is the rib region in the bone segmentation result, where different bone labels are identified by different colors. Normally, each rib should correspond to one rib mask, with only one bone label, identified by one color, while Figure 7 there are errors in the bone segmentation result in [reference], and some ribs are segmented into multiple bone masks, with more than one bone label, that is, their masks are identified by more than one color. Figure 7On the right side is the repaired rib region, where each rib contains only one bone label, that is, the corresponding bone label is identified by a color in each rib.

[0091] Step S40: For the vertebral region in the bone segmentation result after direction correction, perform connected component analysis on the image region corresponding to each vertebral mask to obtain the number of connected components in the image region corresponding to each vertebral mask. When the number of connected components in the image region corresponding to a certain vertebral mask is greater than a preset first threshold, the connected components other than the largest connected component in the image region corresponding to the vertebral mask are regarded as abnormal connected components, and it is determined with which of the adjacent vertebral regions of the vertebral mask the abnormal connected components should be fused, and the vertebral mask is repaired according to the determination result to obtain the repaired vertebral region.

[0092] In some embodiments, in the vertebral region of the bone segmentation result after direction correction, each vertebral mask will also be misaligned and abnormally segmented to varying degrees. Different from the smooth and separated characteristics of each rib mask in the rib region, the segmentation result in the vertebral region has a complex structure, and each vertebral mask contacts each other. Therefore, it is impossible to use the conventional connected component analysis method for processing.

[0093] In some embodiments, the vertebral region includes a plurality of different vertebral masks connected in a top-down spatial order. Perform connected component analysis on the image region corresponding to each vertebral mask to obtain the connected component image of the image region corresponding to each vertebral mask, and the number of connected components can be known according to the connected component image. There is only one connected component in the connected component image of the image region corresponding to the vertebral mask after normal bone segmentation. Therefore, the first threshold is set to 1. When the number of connected components in the image region corresponding to a certain vertebral mask is greater than 1, the connected components other than the largest connected component in the image region corresponding to the vertebral mask are regarded as abnormal connected components.

[0094] In some embodiments, the adjacent vertebral region includes the image regions corresponding to the vertebral masks that have an adjacent relationship with the vertebral mask whose number of connected components is greater than the preset first threshold in the upper and lower spaces.

[0095] In some embodiments, the adjacent vertebral region includes a first adjacent vertebral region and a second adjacent vertebral region. The first adjacent vertebral region is the image region corresponding to the vertebral mask that has an adjacent relationship with the vertebral mask whose number of connected components is greater than the preset first threshold in the upper space, and the second adjacent vertebral region is the image region corresponding to the vertebral mask that has an adjacent relationship with the vertebral mask whose number of connected components is greater than the preset first threshold in the lower space.

[0096] Please refer to Figure 8, in some embodiments, step S40 of determining which of the adjacent vertebral regions of the vertebral mask the abnormal connected region should be fused with includes steps S41 to S47, which are specifically described below.

[0097] Step S41: Replace the mask value of the abnormal connected region with the mask value of the first adjacent vertebral region, and perform connected component analysis on the first adjacent vertebral region to obtain the number of connected components of the first adjacent vertebral region.

[0098] Step S42: Determine whether the number of connected components of the first adjacent vertebral region is consistent with the number of connected components of the first adjacent vertebral region obtained when performing connected component analysis on the image region corresponding to each vertebral mask, and determine whether the number of connected components of the first adjacent vertebral region is less than a preset third threshold.

[0099] Step S43: When the number of connected components of the first adjacent vertebral region is consistent with the number of connected components of the first adjacent vertebral region obtained when performing connected component analysis on the image region corresponding to each vertebral mask and the number of connected components of the first adjacent vertebral region is less than the third threshold, it is determined that the abnormal connected region should be fused with the first adjacent vertebral region.

[0100] Step S44: Otherwise, replace the mask value of the abnormal connected region with the mask value of the second adjacent vertebral region, and perform connected component analysis on the second adjacent vertebral region to obtain the number of connected components of the second adjacent vertebral region.

[0101] Step S45: Determine whether the number of connected components of the second adjacent vertebral region is consistent with the number of connected components of the second adjacent vertebral region obtained when performing connected component analysis on the image region corresponding to each vertebral mask, and determine whether the number of connected components of the second adjacent vertebral region is less than a preset fourth threshold.

[0102] Step S46: When the number of connected components of the second adjacent vertebral region is consistent with the number of connected components of the second adjacent vertebral region obtained when performing connected component analysis on the image region corresponding to each vertebral mask and the number of connected components of the second adjacent vertebral region is less than the fourth threshold, it is determined that the abnormal connected region should be fused with the second adjacent vertebral region.

[0103] Step S47: Otherwise, calculate the centroid coordinates of the abnormal connected region, the centroid coordinates of the first adjacent vertebral region, and the centroid coordinates of the second adjacent vertebral region respectively, calculate the first centroid distance between the centroid coordinates of the abnormal connected region and the centroid coordinates of the first adjacent vertebral region, calculate the second centroid distance between the centroid coordinates of the abnormal connected region and the centroid coordinates of the second adjacent vertebral region, and determine that the abnormal connected region should be fused with the adjacent vertebral region corresponding to the smallest centroid distance among the first centroid distance and the second centroid distance.

[0104] In some embodiments, the vertebral mask is repaired according to the judgment result to obtain the repaired vertebral region, including:

[0105] Modify the bone mask of the abnormal connected domain to the bone mask of the adjacent vertebral region where the abnormal connected domain should be fused to obtain the repaired vertebral region.

[0106] Please refer to Figure 9 , Figure 9 The vertebral region on the left is the vertebral region in the bone segmentation result, and the selected part is the abnormal connected domain. Figure 9 The right side is the repaired vertebral region.

[0107] Step S50: For the upper limb region in the bone segmentation result after direction correction, use the pixel point randomly selected in the upper limb region as the starting point and the image region corresponding to the abnormal bone mask in the upper limb region as the end point. Traverse the pixel points in the upper limb region starting from the starting point and perform pixel value update processing on the pixel value of each pixel point until the number of connected domains in the upper limb region after pixel value update processing meets the preset number, and repair the bone label of the abnormal bone mask to the upper limb bone label.

[0108] In some embodiments, since the bone structure in the upper limb region is small and similar to other bone tissue structures such as the lower limb or the end of the rib, the depth segmentation model cannot correctly distinguish the above bone tissues when reasoning in the form of image patches. Specifically, the bone mask at the end of the upper limb region in the bone segmentation result will be wrongly segmented into the mask of other similar bone tissues.

[0109] In some embodiments, for the upper limb region in the bone segmentation result after direction correction, before repairing the abnormal bone mask in the upper limb region, save the mask value of the scapula mask in the scapula region connected to the upper limb region, and set the mask value of the scapula mask to zero after saving the mask value of the scapula mask, so that when traversing the pixel points in the upper limb region starting from the starting point and performing pixel value update processing on the pixel value of each pixel point, the mask value of the scapula mask in the scapula region is equal to the background mask value. Therefore, the scapula region will not be wrongly segmented into the upper limb region. At the same time, after repairing the bone label of the abnormal bone mask to the upper limb bone label, the mask value of the scapula mask will be restored to the saved mask value of the scapula mask. Among them, the bone label of the abnormal bone mask is different from the upper limb bone label.

[0110] Please refer to Figure 10 , in some embodiments, step S50 performs pixel value update processing on the pixel value of each pixel point, including steps S51 to S53, which are specifically described below.

[0111] Step S51: Determine the magnitude relationship between the pixel value of each pixel and a preset upper limb threshold value.

[0112] In some embodiments, the upper limb threshold value is the difference between the mask value of the upper limb region and a preset background mask value.

[0113] In some embodiments, the background mask value is 0.

[0114] Step S52: When the pixel value of a pixel is less than or equal to the upper limb threshold value, assign the pixel value of this pixel to the mask value of the upper limb region. Among them, the mask value of the upper limb region is the mask value corresponding to the upper limb bone label.

[0115] Step S53: When the pixel value of a pixel is greater than the upper limb threshold value, stop performing pixel update processing.

[0116] In some embodiments, perform connected component analysis on the upper limb region after pixel value update processing to obtain the connected components of the upper limb region. If the number of connected components meets the preset number, it indicates that the abnormal bone mask in the upper limb region has been repaired to the upper limb bone label at this time, thereby completing the repair of the upper limb region.

[0117] In some embodiments, the preset number is 1.

[0118] Please refer to Figure 11 , in some embodiments, Figure 11 The left side is the upper limb region in the bone segmentation result. Figure 11 The right side is the repaired upper limb region. Among them, the region outlined by the red box is the image region corresponding to the abnormal bone mask.

[0119] Please refer to Figure 12 , in some embodiments, Figure 12 The left side is the bone segmentation result. Figure 12 The right side is the repaired bone segmentation result. For the bone segmentation result after azimuth correction, fuse the repaired rib region, the repaired vertebra region, and the repaired upper limb region to obtain the repaired bone segmentation result. Compared with the bone segmentation result output by the depth segmentation model, the repaired bone segmentation result is more accurate.

[0120] In some embodiments, there are two core technologies in the bone segmentation result correction method of the embodiments of the present application: 1) The processing process combines skeletal anatomical features: In each bone mask processing process, not only the skeletal structure corresponding to the abnormal bone mask is referred to, but also the skeletal structures associated with it are referred to, and the morphological sequential relationship between the bone masks is effectively combined, making the processing result more reasonable and accurate. 2) Using anatomical prior knowledge to constrain the processing process: In the correction process, the range of change of the image mask is constrained in terms of these two features of shape and position by using anatomical prior knowledge, thereby effectively limiting the error in the post-processing. Compared with the existing post-processing methods for whole-body bone segmentation, the bone segmentation result correction method of the embodiments of the present application sets a reasonable correction order, ensuring the coherence and efficiency of the entire correction process, and realizing the optimization process from rough to fine. The correction accuracy is higher, the edges of the correction result are smoother and more fluent, avoiding unnatural mutation phenomena, so that the final bone segmentation correction result is more in line with the real skeletal structure. At the same time, the processing process combines skeletal structure features, avoiding ineffective iterations, improving the processing speed and efficiency, and realizing the rapid processing of images.

[0121] Please refer to Figure 13 , in some embodiments, a bone segmentation result correction device includes a result acquisition module 10, an orientation correction module 20, a rib correction module 30, a vertebra correction module 40, and an upper limb correction module 50, which will be specifically described below.

[0122] The result acquisition module 10 is used to acquire the bone segmentation result.

[0123] In some embodiments, the bone segmentation result includes multiple different bone regions segmented from a human body scan image. Each bone region contains different bone masks. The bone mask is the bone identified from the human body scan image. Each bone mask has a bone label. The bone regions include vertebra regions, rib regions, and upper limb regions.

[0124] The orientation correction module 20 is used to calculate the centroid coordinate point of the vertebra region, and use the centroid coordinate point as a demarcation point to correct the orientation of the bone regions with orientation attributes in the bone segmentation result, so as to obtain the bone segmentation result after orientation correction.

[0125] In some embodiments, the bone region with orientation attributes refers to the region where the bones that distinguish left and right are located.

[0126] The rib correction module 30 is used to unify the bone labels of each rib mask in the rib region of the bone segmentation result after direction correction, obtain a rib mask image, perform connected component analysis on the rib mask image to obtain the number of connected components of the rib mask image. When the number of connected components does not meet the normal rib number range, perform connected component analysis on the image region corresponding to each rib mask respectively to obtain the connected component analysis results of the image region corresponding to each rib mask, and perform repair processing on each rib mask in the rib region of the bone segmentation result after direction correction according to the connected component analysis results to obtain the repaired rib region.

[0127] The vertebra correction module 40 is used to perform connected component analysis on the image region corresponding to each vertebra mask in the vertebra region of the bone segmentation result after direction correction to obtain the number of connected components of the image region corresponding to each vertebra mask. When the number of connected components of the image region corresponding to a certain vertebra mask is greater than a preset first threshold, the connected components other than the connected component with the largest area in the image region corresponding to the vertebra mask are used as abnormal connected components, and it is determined which of the adjacent vertebra regions of the vertebra mask the abnormal connected components should be fused with, and the vertebra mask is repaired according to the determination result to obtain the repaired vertebra region.

[0128] In some embodiments, the vertebra region includes a plurality of different vertebra masks connected in a top-down spatial order, and the adjacent vertebra regions include the image regions corresponding to the vertebra masks having an adjacent relationship with the vertebra mask whose number of connected components is greater than the preset first threshold in the upper and lower spaces.

[0129] The upper limb correction module 50 is used to, for the upper limb region in the bone segmentation result after direction correction, use a randomly selected pixel point in the upper limb region as the starting point and the image region corresponding to the abnormal bone mask in the upper limb region as the end point, traverse the pixel points in the upper limb region starting from the starting point and perform pixel value update processing on the pixel value of each pixel point until the number of connected components in the upper limb region after pixel value update processing meets the preset number, and repair the bone label of the abnormal bone mask to the upper limb bone label.

[0130] In some embodiments, the bone label of the abnormal bone mask is different from the upper limb bone label.

[0131] Please refer to Figure 14 , in some embodiments, a bone segmentation result correction device is provided, including:

[0132] A memory 1000 for storing programs;

[0133] A processor 1100 for implementing the bone segmentation result correction method according to any embodiment of the present application by executing the program stored in the memory.

[0134] Those skilled in the art can understand that all or part of the functions of the various methods in the above embodiments can be implemented in a hardware manner or in a computer program manner. When all or part of the functions in the above embodiments are implemented in a computer program manner, the program can be stored in a computer-readable storage medium, and the storage medium can include: read-only memory, random access memory, magnetic disk, optical disk, hard disk, etc. The above functions can be realized by a computer executing the program. For example, the program is stored in the memory of the device, and when the processor executes the program in the memory, the above all or part of the functions can be realized. In addition, when all or part of the functions in the above embodiments are implemented in a computer program manner, the program can also be stored in a storage medium such as a server, another computer, magnetic disk, optical disk, flash drive or mobile hard disk, and saved to the memory of the local device by downloading or copying, or the system of the local device is updated in version. When the processor executes the program in the memory, all or part of the functions in the above embodiments can be realized.

[0135] The above uses specific examples to elaborate on the present invention, which is only used to help understand the present invention and is not intended to limit the present invention. For those skilled in the art of the present invention, according to the idea of the present invention, several simple deductions, deformations or substitutions can also be made.

Claims

1. A method for correcting bone segmentation results, characterized in that: include: Obtaining a bone segmentation result, wherein the bone segmentation result includes a plurality of different bone regions segmented from a human body scan image, each of the bone regions includes a different bone mask, the bone mask is a bone identified from the human body scan image, each of the bone masks has a bone label, and the bone regions include a vertebra region, a rib region, and an upper limb region; Calculating the centroid coordinate point of the vertebral region, and using the centroid coordinate point as a demarcation point to correct the direction of the bone region with orientation attributes in the bone segmentation result, to obtain a bone segmentation result after direction correction, wherein the bone region with orientation attributes refers to the region where the left and right bones are located; For the rib region in the direction-corrected bone segmentation result, unify the bone label of each rib mask therein to obtain a rib mask image, perform connected domain analysis on the rib mask image to obtain the number of connected domains of the rib mask image, when the number of connected domains does not meet the normal rib number range, perform connected domain analysis on the image region corresponding to each rib mask respectively to obtain the connected domain analysis result of the image region corresponding to each rib mask, and repair each rib mask in the rib region in the direction-corrected bone segmentation result according to the connected domain analysis result to obtain a repaired rib region; For the vertebral region in the direction-corrected bone segmentation result, a connected domain analysis is performed on the image region corresponding to each vertebral mask to obtain the number of connected domains in the image region corresponding to each vertebral mask, when the number of connected domains in the image region corresponding to a certain vertebral mask is greater than a preset first threshold, the connected domains in the image region corresponding to the vertebral mask except for the connected domain with the largest area are taken as abnormal connected domains, and it is determined which of the adjacent vertebral regions of the vertebral mask the abnormal connected domain should be fused with, and the vertebral mask is repaired according to the determination result to obtain a repaired vertebral region, wherein the vertebral region includes a plurality of different vertebral masks connected in a spatial order from top to bottom, and the adjacent vertebral region includes the image region corresponding to the vertebral mask having an adjacent relationship in the upper space and the lower space with the vertebral mask having the number of connected domains greater than the preset first threshold; For the upper limb area in the bone segmentation result after the direction correction, a pixel point randomly selected in the upper limb area is taken as the starting point, and the image area corresponding to the abnormal bone mask in the upper limb area is taken as the end point. The pixel points in the upper limb area are traversed from the starting point and the pixel value of each pixel point is updated until the number of connected domains in the upper limb area after the pixel value update processing meets the preset number, and the bone label of the abnormal bone mask is repaired to the upper limb bone label, wherein the bone label of the abnormal bone mask is different from the upper limb bone label.

2. The bone segmentation result correction method according to claim 1, characterized in that: The connected domain analysis result of the image area corresponding to each rib mask includes a connected domain image of the image area corresponding to each rib mask and the number of connected domains corresponding to the connected domain image, and the repairing process is performed on each rib mask in the rib area in the direction-corrected bone segmentation result according to the connected domain analysis result to obtain a repaired rib area, including: When the number of connected domains corresponding to the connected domain image is greater than a preset second threshold, the connected domain with the largest area in the connected domain image is retained, and all connected domains except the connected domain with the largest area are eliminated to obtain a connected domain image of the image area corresponding to each rib mask after processing; Calculate the reference coordinates of the connected domain image of the image area corresponding to each rib mask after processing, sort the connected domain images of the image area corresponding to each rib mask after processing according to the size of the reference coordinates, and assign predefined rib mask values ​​to the connected domain images of the image area corresponding to each rib mask after processing in the order of the sorting process to obtain the repaired rib area, wherein the reference coordinates are the coordinates of the centroid of the connected domain image of the image area corresponding to each rib mask after processing on the Z axis.

3. The bone segmentation result correction method according to claim 1, characterized in that: The method of performing direction correction on the bone region with orientation attributes in the bone segmentation result by using the centroid coordinate point as a dividing point to obtain the bone segmentation result after direction correction includes: Assigning the mask values ​​of all the bone masks in the bone region with the orientation attribute to a preset first mask value, wherein the first mask value is the mask value of the left bone mask or the mask value of the right bone mask; When the first mask value is the mask value of the left skeleton mask, assigning the mask value of the skeleton mask located on the right side of the centroid coordinate point in the skeleton region with the orientation attribute as the mask value of the right skeleton mask; When the first mask value is the mask value of the right bone mask, the mask value of the bone mask located on the left side of the centroid coordinate point in the bone area with orientation attributes is assigned to the mask value of the left bone mask; wherein the mask value is used to represent the bone label of the bone mask.

4. The bone segmentation result correction method according to claim 1, characterized in that: The adjacent vertebral regions include a first adjacent vertebral region and a second adjacent vertebral region, the first adjacent vertebral region is an image region corresponding to a vertebral mask having an adjacent relationship with the vertebral mask having a number of connected domains greater than a preset first threshold in an upper space, and the second adjacent vertebral region is an image region corresponding to a vertebral mask having an adjacent relationship with the vertebral mask having a number of connected domains greater than a preset first threshold in a lower space, and the determining which of the adjacent vertebral regions of the vertebral mask the abnormal connected domain should be fused with includes: Replacing the mask value of the abnormal connected domain with the mask value of the first adjacent vertebrae region, and performing connected domain analysis on the first adjacent vertebrae region to obtain the number of connected domains in the first adjacent vertebrae region; Determining whether the number of connected domains of the first adjacent vertebrae region is consistent with the number of connected domains of the first adjacent vertebrae region obtained by performing connected domain analysis on the image region corresponding to each vertebrae mask, and determining whether the number of connected domains of the first adjacent vertebrae region is less than a preset third threshold; When the number of connected domains of the first adjacent vertebrae region is consistent with the number of connected domains of the first adjacent vertebrae region obtained by performing connected domain analysis on the image region corresponding to each vertebrae mask and the number of connected domains of the first adjacent vertebrae region is less than the third threshold, it is determined that the abnormal connected domain should be fused with the first adjacent vertebrae region; Otherwise, the mask value of the abnormal connected domain is replaced with the mask value of the second adjacent vertebrae region, and a connected domain analysis is performed on the second adjacent vertebrae region to obtain the number of connected domains of the second adjacent vertebrae region; Determining whether the number of connected domains of the second adjacent vertebrae region is consistent with the number of connected domains of the second adjacent vertebrae region obtained by performing connected domain analysis on the image region corresponding to each vertebrae mask, and determining whether the number of connected domains of the second adjacent vertebrae region is less than a preset fourth threshold; When the number of connected domains of the second adjacent vertebrae region is consistent with the number of connected domains of the second adjacent vertebrae region obtained by performing connected domain analysis on the image region corresponding to each vertebrae mask and the number of connected domains of the second adjacent vertebrae region is less than the fourth threshold, it is determined that the abnormal connected domain should be fused with the second adjacent vertebrae region; Otherwise, the centroid coordinates of the abnormal connected domain, the centroid coordinates of the first adjacent vertebral region, and the centroid coordinates of the second adjacent vertebral region are calculated respectively, and a first centroid distance between the centroid coordinates of the abnormal connected domain and the centroid coordinates of the first adjacent vertebral region is calculated, and a second centroid distance between the centroid coordinates of the abnormal connected domain and the centroid coordinates of the second adjacent vertebral region is calculated, and it is determined that the abnormal connected domain should be fused with the adjacent vertebral region corresponding to the smallest centroid distance between the first centroid distance and the second distance.

5. The bone segmentation result correction method according to claim 1, characterized in that: The step of repairing the vertebral mask according to the judgment result to obtain a repaired vertebral region includes: The bone mask of the abnormal connected domain is modified to be the bone mask of the adjacent vertebral region to which the abnormal connected domain should be fused, so as to obtain a repaired vertebral region.

6. The bone segmentation result correction method according to claim 1, characterized in that: The pixel value updating process for each pixel point includes: Determine the difference between the pixel value of each pixel and a preset upper limb threshold, wherein the upper limb threshold is the difference between the mask value of the upper limb area and a preset background mask value; When the pixel value of the pixel point is less than or equal to the upper limb threshold, assigning the pixel value of the pixel point as the mask value of the upper limb area; When the pixel value of the pixel point is greater than the upper limb threshold, the pixel update process is stopped.

7. The bone segmentation result correction method according to claim 6, characterized in that: The background mask value is 0. For the upper limb area in the bone segmentation result after the direction correction, before the abnormal bone mask in the upper limb area is repaired, the mask value of the scapula mask of the scapula area connected to the upper limb area is saved, and after saving the mask value of the scapula mask, the mask value of the scapula mask is set to zero, so that when traversing the pixel points in the upper limb area from the starting point and updating the pixel value of each pixel point, the mask value of the scapula mask of the scapula area is equal to the background mask value.

8. A bone segmentation result correction device, characterized in that: include: A result acquisition module is used to acquire a bone segmentation result, wherein the bone segmentation result includes a plurality of different bone regions segmented from a human body scan image, each of the bone regions includes a different bone mask, the bone mask is a bone identified from the human body scan image, each of the bone masks has a bone label, and the bone regions include a vertebra region, a rib region, and an upper limb region; A direction correction module, used for calculating the centroid coordinate point of the vertebral region, and using the centroid coordinate point as a demarcation point to perform direction correction on the bone region with orientation attributes in the bone segmentation result, to obtain a direction-corrected bone segmentation result, wherein the bone region with orientation attributes refers to the region where the left and right bones are located; A rib correction module is used to unify the bone labels of each rib mask in the rib area in the direction-corrected bone segmentation result to obtain a rib mask image, perform a connected domain analysis on the rib mask image to obtain the number of connected domains of the rib mask image, and when the number of connected domains does not meet the normal range of the number of ribs, perform a connected domain analysis on the image area corresponding to each rib mask to obtain the connected domain analysis result of the image area corresponding to each rib mask, and repair each rib mask in the rib area in the direction-corrected bone segmentation result according to the connected domain analysis result to obtain a repaired rib area; A vertebra correction module is used to perform a connected domain analysis on the image area corresponding to each vertebra mask for the vertebra area in the bone segmentation result after the direction correction, to obtain the number of connected domains of the image area corresponding to each vertebra mask, and when the number of connected domains of the image area corresponding to a certain vertebra mask is greater than a preset first threshold, to use the connected domains other than the connected domain with the largest area in the image area corresponding to the vertebra mask as abnormal connected domains, to determine which of the adjacent vertebra areas of the vertebra mask the abnormal connected domain should be fused with, and to perform a repair process on the vertebra mask according to the determination result, to obtain a repaired vertebra area, wherein the vertebra area includes a plurality of different vertebra masks connected in a spatial order from top to bottom, and the adjacent vertebra area includes the image area corresponding to the vertebra mask having an adjacent relationship in the upper space and the lower space with the vertebra mask having the number of connected domains greater than the preset first threshold; An upper limb correction module is used to traverse the pixel points in the upper limb area in the bone segmentation result after the direction correction, with a pixel point randomly selected in the upper limb area as the starting point, and an image area corresponding to the abnormal bone mask in the upper limb area as the end point, starting from the starting point to traverse the pixel points in the upper limb area and perform pixel value update processing on the pixel value of each pixel point until the number of connected domains in the upper limb area after the pixel value update processing meets the preset number, and repair the bone label of the abnormal bone mask to the upper limb bone label, wherein the bone label of the abnormal bone mask is different from the upper limb bone label.

9. A bone segmentation result correction device, characterized in that: include: Memory, used to store programs; A processor, configured to implement the bone segmentation result correction method according to any one of claims 1 to 7 by executing the program stored in the memory.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a program, and the program can be executed by a processor to implement the bone segmentation result correction method according to any one of claims 1 to 7.

Images