A hip joint image segmentation analysis system and method based on SEGVOL

Abstract

The present invention discloses a hip joint image segmentation and analysis system and method based on SEGVOL, which specifically relates to the technical field of medical image processing, including a hip joint image segmentation data acquisition module, a hip joint image segmentation data analysis module, a hip joint image segmentation evaluation module, a hip joint image segmentation comprehensive analysis module and a hip joint image segmentation human-computer interaction module; the hip joint image segmentation data acquisition module is used to collect hip joint image segmentation data, obtain various parameters for analyzing hip joint image segmentation performance indicators, and transmit the various parameters to the hip joint image segmentation data analysis module; the present invention can analyze the ability of the SEGVOL model to retain details while accurately segmenting the hip joint structure by analyzing the image segmentation structure retention parameters, thereby helping to more comprehensively analyze the performance of the model in hip joint image segmentation and formulate corresponding optimization strategies.

Description

Technical Field

[0001] The present invention relates to the technical field of medical image processing, and in particular to a hip joint image segmentation and analysis system and method based on SEGVOL. Background Art

[0002] The hip joint is composed of the femoral head and acetabulum. It is one of the important joints of the human body. Its health directly affects people's daily quality of life. The SEGVOL model is a general interactive medical image segmentation model based on voxels. By integrating spatial and semantic cues, it has powerful segmentation capabilities and flexibility, achieves high-precision medical image segmentation, and provides an efficient and reliable solution for the accurate segmentation of hip joint images.

[0003] Hip joint disease is one of the common orthopedic diseases. In medical image analysis, the performance of hip joint image segmentation methods is of great significance for disease diagnosis, surgical planning and treatment effect evaluation. Traditional hip joint image segmentation methods mainly rely on manual operations, which is not only time-consuming and laborious, but also easily affected by subjective factors. In addition, due to the complexity of the hip joint structure and the diversity of images, traditional image segmentation methods are often difficult to accurately segment and obtain reliable results. Therefore, the performance analysis of hip joint image segmentation methods is very important.

[0004] Therefore, the present invention proposes a hip joint image segmentation and analysis system and method based on SEGVOL to solve the above problems. By introducing the SEGVOL model, the key features of the hip joint can be automatically identified and extracted to achieve accurate segmentation. The system and method use computer technology to analyze the performance of the hip joint image segmentation method. It has good versatility and scalability, can adapt to hip joint images of different types and qualities, and thus help to improve medical efficiency. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a hip joint image segmentation and analysis system and method based on SEGVOL to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions: a hip joint image segmentation and analysis system based on SEGVOL, comprising a hip joint image segmentation data acquisition module, a hip joint image segmentation data analysis module, a hip joint image segmentation evaluation module, a hip joint image segmentation comprehensive analysis module and a hip joint image segmentation human-computer interaction module;

[0007] The hip joint image segmentation data acquisition module is used to collect hip joint image segmentation data, obtain various parameters for analyzing hip joint image segmentation performance indicators, and transmit the various parameters to the hip joint image segmentation data analysis module;

[0008] The hip joint image segmentation data analysis module includes an image segmentation structure retention index analysis unit, an image segmentation structure space consistency index analysis unit, an interactive prompt image segmentation effect index analysis unit and an image segmentation efficiency index analysis unit, which are respectively used to calculate various parameters and transmit the corresponding calculation results to the hip joint image segmentation evaluation module;

[0009] The hip joint image segmentation evaluation module is used to compare the various indicators of hip joint image segmentation performance obtained by the data analysis module with the set values, and transmit the comparison results to the hip joint image segmentation comprehensive analysis module;

[0010] The hip joint image segmentation comprehensive analysis module is used to comprehensively analyze the effective factors of each indicator calculated by the evaluation module, and transmit the comprehensive analysis results to the hip joint image segmentation human-computer interaction module;

[0011] The hip joint image segmentation human-computer interaction module is used to transmit the comprehensive analysis results input by the comprehensive analysis module to the system administrator information terminal.

[0012] Preferably, the parameters in the hip joint image segmentation data acquisition module include system administrator authentication information, image segmentation structure retention parameters, image segmentation structure spatial position parameters, interactive prompt image segmentation parameters and image segmentation efficiency parameters, wherein the image segmentation structure retention parameters specifically include the area of ​​fine structures in the segmentation result, the annotated fine structure area, the total number of matched gradient vector pairs, the gradient vectors on the segmented edge and the annotated edge, the hip joint volume obtained by segmentation, the true volume of the hip joint, the high-frequency components of the image before segmentation and the high-frequency components of the image after segmentation; the image segmentation structure spatial position parameters specifically include the number of segmentation positioning points, the segmentation three-dimensional space, the image segmentation structure spatial position parameters ... dimensional coordinates and real 3D coordinates, the number of symmetrical feature points, the left features and right features of the symmetrical points, the hip joint segmentation angle and the standard segmentation angle, the number of sampling points in the adjacent structure area, and the hip joint segmentation and the real adjacent structure area; the image segmentation parameters under interactive prompts are specifically the image segmentation overlap index with text prompts and without text prompts, the segmentation space position deviation index with point prompts and without point prompts, and the average hip joint segmentation time with frame prompts and without frame prompts; the image segmentation efficiency parameters are specifically the actual segmentation speed, the theoretical segmentation speed, the maximum memory occupied by the SEGVOL model, the total system memory and the computing resource utilization of the GPU.

[0013] Preferably, the model for obtaining the image segmentation structure preservation performance index in the image segmentation structure preservation index analysis unit is: , where Pri represents the image segmentation structure retention performance index, Dsc represents the image segmentation overlap index, , X represents the area of ​​fine structure in the segmentation result, Y represents the marked area of ​​fine structure, |X∩Y| represents the size of the intersection area between X and Y, |X| and |Y| represent the size of X and Y area respectively, Egc represents the edge gradient consistency index, , n represents the total number of matching gradient vector pairs, ▽S i represents the gradient vector of the i-th point on the segmentation edge, ▽G i represents the gradient vector of the i-th point on the labeled edge, |▽S i | and |▽G i |represents ▽S i and ▽G i The modulus length, Vpr represents the image segmentation volume retention rate index, , V seg represents the volume of the hip joint obtained by segmentation, V gt represents the true volume of the hip joint, Dlr represents the detail loss rate index, , H(Iorig) represents the high-frequency component of the image before segmentation, H(Iseg) represents the high-frequency component of the image after segmentation, a1, a2, a3 and a4 represent the weights of image segmentation overlap, edge gradient consistency, image segmentation volume retention rate and detail loss rate respectively.

[0014] Preferably, the model for obtaining the image segmentation structure spatial consistency performance index in the image segmentation structure spatial consistency index analysis unit is: , where Ssci represents the image segmentation structure spatial consistency performance index, Spd represents the segmentation space position deviation index, , N represents the number of segmentation points, (x I seg ,y I seg ,z I seg ) and (x I gt ,y I gt ,z I gt ) represent the segmentation 3D coordinates and the real 3D coordinates of the I-th positioning point, respectively, Ssi represents the structural segmentation symmetry index, , m represents the number of symmetrical feature points on both sides of the hip joint segmentation result, F j left and F j right represents the left and right features of the jth symmetric point, Dc represents the segmentation angle consistency index, , l represents the number of relevant angles, θ k seg and θk atl They represent the kth segmentation angle and standard segmentation angle of the hip joint, Asr represents the continuity index of the adjacent structure segmentation, ; M represents the number of sampling points in the adjacent structure area, s J and g J Represent the hip joint segmentation and the true adjacent structure area respectively, b1, b2, b3 and b4 represent the weights of the segmentation space position deviation index, the structure segmentation symmetry index, the segmentation angle consistency index and the adjacent structure segmentation continuity index respectively.

[0015] Preferably, the model for obtaining the interactive prompt image segmentation effect performance index in the interactive prompt image segmentation effect index analysis unit is: , where Seci represents the performance index of interactive prompt image segmentation effect, Dsc tp and Dsc Ntp Represents the image segmentation overlap index with and without text prompts, Spd Nh and Spd h They represent the spatial position deviation index of segmentation with and without point cues, t Nbp and t bp denote the average segmentation time of hip joints with and without frame prompts, respectively. c1, c2, and c3 denote the weights of text prompts, point prompts, and frame prompts, respectively.

[0016] Preferably, the model for obtaining the image segmentation efficiency performance index in the image segmentation efficiency index analysis unit is: , where Sci represents the image segmentation efficiency performance index, V represents the actual segmentation speed, and V 0 represents the theoretical segmentation speed, Mc represents the maximum memory occupied by the SEGVOL model, T_mc represents the total system memory, and Ur represents the computing resource utilization of the GPU.

[0017] Preferably, the evaluation module in the hip joint image segmentation evaluation module comprises the following steps:

[0018] Step 1: Substitute the image segmentation structure retention performance index Pri obtained in the image segmentation structure retention index analysis unit into the formula: , where η(Pri) represents the effective factor of the image segmentation structure retention performance index, Pri 0 Indicates the set value;

[0019] Step 2: Substitute the image segmentation structure spatial consistency performance index Ssci obtained from the image segmentation structure spatial consistency index analysis unit into the formula: , where η(Ssci) represents the effective factor of the image segmentation structure spatial consistency performance index, Ssci0 Indicates the set value;

[0020] Step 3: Substitute the interactive prompt image segmentation effect performance index Seci obtained from the interactive prompt image segmentation effect index analysis unit into the formula: , where η(Seci) represents the effective factor of the performance index of the interactive prompt image segmentation effect, Seci 0 Indicates the set value;

[0021] Step 4: Substitute the image segmentation efficiency performance index Sci obtained from the image segmentation efficiency index analysis unit into the formula: , where η(Sci) represents the effective factor of the image segmentation efficiency performance index, Sci 0 Indicates the set value.

[0022] Preferably, the comprehensive analysis model in the hip joint image segmentation comprehensive analysis module is: , where Isci represents the comprehensive index of hip joint image segmentation performance.

[0023] A hip joint image segmentation and analysis method based on SEGVOL comprises the following steps:

[0024] Step 1: Collect hip joint image segmentation data to obtain various parameters for analyzing hip joint image segmentation performance indicators;

[0025] Step 2: According to the parameters in step 1, the image segmentation structure retention parameters, image segmentation structure spatial position parameters, interactive prompt image segmentation parameters and image segmentation efficiency parameters are calculated respectively to obtain various indicators for analyzing the hip joint image segmentation performance;

[0026] Step 3: Compare the calculation results of step 2 with the set values ​​to obtain the effective factors of each index of hip joint image segmentation performance;

[0027] Step 4: Perform comprehensive analysis based on the comparison results of step 3 to obtain comprehensive indicators of hip joint image segmentation performance;

[0028] Step 5: Perform human-computer interaction based on the comprehensive analysis results of step 4 and the comprehensive performance index of hip joint image segmentation.

[0029] Technical effects and advantages of the present invention:

[0030] 1. By analyzing the image segmentation structure retention parameters, the present invention can analyze the ability of the SEGVOL model to accurately segment the hip joint structure while retaining details, thereby helping to more comprehensively analyze the performance of the model in hip joint image segmentation and formulate corresponding optimization strategies, providing a reference for medical imaging diagnosis;

[0031] 2. The present invention can analyze the consistency of the segmentation results in spatial position and the symmetry of the hip joint structure by analyzing the spatial position parameters of the image segmentation structure, and analyze the accuracy and reliability of the segmentation results from multiple aspects such as spatial position, structural symmetry, segmentation angle consistency and adjacent structure relationship, thereby helping to formulate more accurate diagnosis and treatment plans;

[0032] 3. By analyzing the image segmentation parameters under interactive prompts, the present invention can analyze the segmentation effect of the SEGVOL model after receiving different interactive prompts such as text, points, and boxes. It can identify the performance differences of the SEGVOL model under different prompting methods, and then guide the further optimization of the model, reduce the image segmentation processing time, and improve the efficiency of medical services. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a schematic diagram of the overall process of the present invention. DETAILED DESCRIPTION

[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0035] Example 1

[0036] See also Figure 1 As shown, the present invention provides a hip joint image segmentation and analysis system based on SEGVOL, including a hip joint image segmentation data acquisition module, a hip joint image segmentation data analysis module, a hip joint image segmentation evaluation module, a hip joint image segmentation comprehensive analysis module and a hip joint image segmentation human-computer interaction module;

[0037] The hip joint image segmentation data acquisition module is connected to the hip joint image segmentation data analysis module, the hip joint image segmentation evaluation module is respectively connected to the hip joint image segmentation data analysis module, the hip joint image segmentation comprehensive analysis module and the hip joint image segmentation human-computer interaction module, and the human-computer interaction module is respectively connected to the hip joint image segmentation comprehensive analysis module and the hip joint image segmentation data analysis module.

[0038] The hip joint image segmentation data acquisition module is used to collect hip joint image segmentation data, obtain various parameters for analyzing hip joint image segmentation performance indicators, and transmit the various parameters to the hip joint image segmentation data analysis module;

[0039] The hip joint image segmentation data analysis module includes an image segmentation structure retention index analysis unit, an image segmentation structure space consistency index analysis unit, an interactive prompt image segmentation effect index analysis unit and an image segmentation efficiency index analysis unit, which are respectively used to calculate various parameters and transmit the corresponding calculation results to the hip joint image segmentation evaluation module;

[0040] The hip joint image segmentation evaluation module is used to compare the various indicators of hip joint image segmentation performance obtained by the data analysis module with the set values, and transmit the comparison results to the hip joint image segmentation comprehensive analysis module;

[0041] The hip joint image segmentation comprehensive analysis module is used to comprehensively analyze the effective factors of each indicator calculated by the evaluation module, and transmit the comprehensive analysis results to the hip joint image segmentation human-computer interaction module;

[0042] The hip joint image segmentation human-computer interaction module is used to transmit the comprehensive analysis results passed in by the comprehensive analysis module to the system administrator information terminal. If the comprehensive index of the hip joint image segmentation performance belongs to the set allowable range, it means that the hip joint image segmentation is good. Otherwise, the system administrator is prompted to manage in time, adjust the model structure, and increase the hip joint image segmentation training data, so as to improve the accuracy and reliability of hip joint image segmentation and provide a reference basis for medical imaging diagnosis and treatment planning.

[0043] In this embodiment, it should be specifically explained that the various parameters in the hip joint image segmentation data acquisition module include system administrator identity authentication information, image segmentation structure retention parameters, image segmentation structure spatial position parameters, interactive prompt image segmentation parameters and image segmentation efficiency parameters, wherein the image segmentation structure retention parameters are specifically the area of ​​fine structures in the segmentation results, the annotated fine structure areas, the total number of matched gradient vector pairs, the gradient vectors on the segmented edges and the annotated edges, the hip joint volume obtained by segmentation, the true volume of the hip joint, the high-frequency components of the image before segmentation and the high-frequency components of the image after segmentation; the image segmentation structure spatial position parameters are specifically the number of segmentation positioning points. The image segmentation parameters include the image segmentation overlap index with and without text prompts, the segmentation space position deviation index with and without point prompts, and the average segmentation time of hip joints with and without frame prompts; the image segmentation efficiency parameters include the actual segmentation speed, the theoretical segmentation speed, the maximum memory occupied by the SEGVOL model, the total system memory, and the GPU computing resource utilization.

[0044] In this embodiment, it should be specifically explained that the image segmentation structure retention index analysis unit in the hip joint image segmentation data analysis module is used to calculate the image segmentation structure retention parameters, the image segmentation structure space consistency index analysis unit is used to calculate the image segmentation structure space position parameters, the interactive prompt image segmentation effect index analysis unit is used to calculate the image segmentation parameters under the interactive prompt, and the image segmentation efficiency index analysis unit is used to calculate the image segmentation efficiency parameters, and its data analysis includes the following steps:

[0045] Step 1: By analyzing the image segmentation structure retention parameters, we can analyze the ability of the SEGVOL model to accurately segment the hip joint structure while retaining details. Taking into account the fine structures in the hip joint image, the image segmentation edges, the image details lost due to excessive smoothing during the segmentation process, and the hip joint volume obtained by segmentation, the model for obtaining the image segmentation structure retention performance index in the image segmentation structure retention index analysis unit is: , where Pri represents the image segmentation structure retention performance index, Dsc represents the image segmentation overlap index, , X represents the area of ​​fine structure in the segmentation result, such as the joint surface, vascular channel area, etc., Y represents the annotated fine structure area, |X∩Y| represents the size of the intersection area between X and Y, that is, the part that both automatic segmentation and annotation consider to be the fine structure, |X| and |Y| represent the size of the X and Y areas, that is, the total number of pixels in each area, Egc represents the edge gradient consistency index, , n represents the total number of matching gradient vector pairs. When comparing the segmented edge (generated by the segmentation algorithm) and the annotated edge, each pair of corresponding edge points is recorded. i represents the gradient vector of the i-th point on the segmentation edge, ▽G i represents the gradient vector of the i-th point on the labeled edge, |▽S i | and |▽G i |represents ▽S i and ▽G i The modulus of the gradient vector is Vpr, which represents the image segmentation volume retention rate index. , V seg represents the volume of the hip joint obtained by segmentation, V gt represents the true volume of the hip joint, Dlr represents the detail loss rate index, that is, the proportion of details lost due to excessive smoothing during the segmentation process, , H(Iorig) represents the high-frequency component of the image before segmentation, H(Iseg) represents the high-frequency component of the image after segmentation, the high-frequency component extraction function is an existing technology, such as Fourier transform technology to extract the high-frequency component of the image, a1, a2, a3 and a4 represent the weights of image segmentation overlap, edge gradient consistency, image segmentation volume retention rate and detail loss rate, for example, a1=0.3, a2=0.3, a3=0.25, a4=0.15;

[0046] Step 2: By analyzing the spatial position parameters of the image segmentation structure, the consistency of the segmentation results in spatial position and the symmetry of the hip joint structure can be analyzed. The segmentation spatial position deviation, the symmetry of the structure segmentation results, the segmentation angle consistency and the continuity of the adjacent structure segmentation are comprehensively considered. Therefore, the model for obtaining the image segmentation structure spatial consistency performance index in the image segmentation structure spatial consistency index analysis unit is: , where Ssci represents the image segmentation structure spatial consistency performance index, Spd represents the segmentation space position deviation index, , N represents the number of segmentation points, such as the femoral head center, acetabulum center, etc., (x I seg ,y I seg ,z I seg ) and (x I gt ,y I gt ,z I gt ) represent the segmentation 3D coordinates and the real 3D coordinates of the I-th positioning point, respectively, Ssi represents the structural segmentation symmetry index, , m represents the number of symmetrical feature points on both sides of the hip joint segmentation result, F j left and F j right represents the left and right features of the jth symmetrical point, such as the symmetry of the left and right femoral heads. Dc represents the segmentation angle consistency index. , l represents the number of related angles, such as acetabular angle, neck-shaft angle, femoral neck anteversion angle, etc., θ k seg and θ k atl They represent the kth segmentation angle and standard segmentation angle of the hip joint, Asr represents the continuity index of the adjacent structure segmentation, ; M represents the number of sampling points in the adjacent structure area, s J and g JRespectively represent the hip joint segmentation and the true adjacent structure area, that is, analyze the adjacency relationship between the segmentation result and the surrounding structure (such as pelvis, femur) to ensure the continuity and accuracy of the segmentation. The IoU function is a metric to measure the overlap between the predicted bounding box and the true bounding box. It evaluates the accuracy of the prediction by calculating the ratio of the intersection area of ​​the predicted box and the true box to the union area. The value range of IoU is between 0 and 1, where 1 indicates perfect overlap and 0 indicates no overlap at all. b1, b2, b3 and b4 represent the weights of the segmentation space position deviation index, the structure segmentation symmetry index, the segmentation angle consistency index and the adjacent structure segmentation continuity index, respectively. For example, b1=0.2, b2=0.3, b3=0.25, b4=0.25;

[0047] Step 3: By analyzing the image segmentation parameters under interactive prompts, we can analyze the segmentation effect of the SEGVOL model after receiving different interactive prompts (such as text, point, box), and then evaluate the effectiveness of each prompt method in improving the segmentation results. Therefore, the model for obtaining the interactive prompt image segmentation effect performance index in the interactive prompt image segmentation effect index analysis unit is: , where Seci represents the performance index of interactive prompt image segmentation effect, Dsc tp and Dsc Ntp Represents the image segmentation overlap index with and without text prompts, Spd Nh and Spd h They represent the spatial position deviation index of segmentation with and without point cues, t Nbp and t bp denote the average segmentation time of hip joints with and without frame prompts, respectively. c1, c2, and c3 denote the weights of text prompts, point prompts, and frame prompts, respectively, for example, c1=0.3, c2=0.35, and c3=0.35;

[0048] Step 4: By analyzing the image segmentation efficiency parameters, we can analyze the computational efficiency and resource consumption of the SEGVOL model when processing hip joint image segmentation tasks. Taking into account the number of segmentation tasks completed per unit time, the maximum amount of memory occupied by the model during the segmentation process, and the percentage of GPU computing resources actually used during the segmentation process, the model for obtaining the image segmentation efficiency performance index in the image segmentation efficiency index analysis unit is: , where Sci represents the image segmentation efficiency performance index, V represents the actual segmentation speed, and V 0Indicates the theoretical segmentation speed, Mc indicates the maximum amount of memory occupied by the SEGVOL model, T_mc indicates the total system memory, and Ur indicates the GPU computing resource utilization. The system monitoring tool will track the GPU activity, including when it is in computing state and when it is in idle state, and then calculate the utilization based on this information, which is usually a percentage, such as 90%;

[0049] In this embodiment, it should be specifically explained that the evaluation module in the hip joint image segmentation evaluation module includes the following steps:

[0050] Step 1: Substitute the image segmentation structure retention performance index Pri obtained in the image segmentation structure retention index analysis unit into the formula: , where η(Pri) represents the effective factor of the image segmentation structure retention performance index, Pri 0 Indicates the set value;

[0051] What needs to be specifically explained in this embodiment is that if η(Pri)≥1, it means that the image segmentation structure retention performance is good, otherwise it means that it is abnormal. Feedback control is performed, the model structure is adjusted, hip joint image segmentation training data is increased, more complex feature extraction methods are used, etc., and the abnormal signal is output to the hip joint image segmentation human-computer interaction module.

[0052] Step 2: Substitute the image segmentation structure spatial consistency performance index Ssci obtained from the image segmentation structure spatial consistency index analysis unit into the formula: , where η(Ssci) represents the effective factor of the image segmentation structure spatial consistency performance index, Ssci 0 Indicates the set value;

[0053] What needs to be specifically explained in this embodiment is that if η(Ssci)≥1, it means that the spatial consistency performance of the image segmentation structure is good, otherwise it means that it is abnormal. Feedback control is performed, spatial constraints are added, the loss function is adjusted, symmetry is introduced as part of the model training, and the abnormal signal is output to the hip joint image segmentation human-computer interaction module.

[0054] Step 3: Substitute the interactive prompt image segmentation effect performance index Seci obtained from the interactive prompt image segmentation effect index analysis unit into the formula: , where η(Seci) represents the effective factor of the performance index of the interactive prompt image segmentation effect, Seci 0 Indicates the set value;

[0055] What needs to be specifically explained in this embodiment is that if η(Seci)≥1, it means that the interactive prompt image segmentation effect has good performance, otherwise it means it is abnormal. Feedback control is performed to determine the prompt method that can best improve the model performance, and then adjust the prompt method so that the model can use these prompt information more effectively and output abnormal signals to the hip joint image segmentation human-computer interaction module.

[0056] Step 4: Substitute the image segmentation efficiency performance index Sci obtained from the image segmentation efficiency index analysis unit into the formula: , where η(Sci) represents the effective factor of the image segmentation efficiency performance index, Sci 0 Indicates the set value;

[0057] What needs to be specifically explained in this embodiment is that if η(Sci)≥1, it means that the image segmentation efficiency performance is good, otherwise it means that it is abnormal. Feedback control is performed. When the segmentation efficiency is low, the model structure is optimized and the amount of calculation is reduced. When the calculation cost is high, model pruning, quantization or the use of a lighter network structure is considered, and the abnormal signal is output to the hip joint image segmentation human-computer interaction module.

[0058] In this embodiment, it should be specifically explained that the hip joint image segmentation comprehensive analysis module comprehensively analyzes the effective factors of each indicator calculated by the evaluation module, and its comprehensive analysis model is: , where Isci represents the comprehensive index of hip joint image segmentation performance.

[0059] Example 2

[0060] For the parts not described in detail in this embodiment, please refer to the description of Embodiment 1. A hip joint image segmentation and analysis method based on SEGVOL is provided, which includes the following steps:

[0061] Step 1: Collect hip joint image segmentation data to obtain various parameters for analyzing hip joint image segmentation performance indicators;

[0062] Step 2: According to the parameters in step 1, the image segmentation structure retention parameters, image segmentation structure spatial position parameters, interactive prompt image segmentation parameters and image segmentation efficiency parameters are calculated respectively to obtain various indicators for analyzing the hip joint image segmentation performance;

[0063] Step 3: Compare the calculation results of step 2 with the set values ​​to obtain the effective factors of each index of hip joint image segmentation performance;

[0064] Step 4: Perform comprehensive analysis based on the comparison results of step 3 to obtain comprehensive indicators of hip joint image segmentation performance;

[0065] Step 5: Perform human-computer interaction based on the comprehensive analysis results of step 4 and the comprehensive performance index of hip joint image segmentation.

[0066] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other;

[0067] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A hip joint image segmentation and analysis system based on SEGVOL, characterized by: It includes a hip joint image segmentation data acquisition module, a hip joint image segmentation data analysis module, a hip joint image segmentation evaluation module, a hip joint image segmentation comprehensive analysis module and a hip joint image segmentation human-computer interaction module; The hip joint image segmentation data acquisition module is used to collect hip joint image segmentation data, obtain various parameters for analyzing hip joint image segmentation performance indicators, and transmit the various parameters to the hip joint image segmentation data analysis module; The parameters in the hip joint image segmentation data acquisition module include system administrator authentication information, image segmentation structure retention parameters, image segmentation structure spatial position parameters, interactive prompt image segmentation parameters and image segmentation efficiency parameters, wherein the image segmentation structure retention parameters are specifically the area of ​​fine structures in the segmentation results, the annotated fine structure areas, the total number of matched gradient vector pairs, the gradient vectors on the segmented edges and the annotated edges, the hip joint volume obtained by segmentation, the real volume of the hip joint, the high-frequency components of the image before segmentation and the high-frequency components of the image after segmentation; the image segmentation structure spatial position parameters are specifically the number of segmentation positioning points, the segmentation three-dimensional coordinates, the number of segmentation positioning points, the number of segmentation positioning points, the number of segmentation three-dimensional coordinates ... The standard and real three-dimensional coordinates, the number of symmetrical feature points, the left and right features of the symmetrical points, the hip joint segmentation angle and the standard segmentation angle, the number of sampling points in the adjacent structure area, and the hip joint segmentation and the real adjacent structure area; the image segmentation parameters under interactive prompts are specifically the image segmentation overlap index with and without text prompts, the segmentation space position deviation index with and without point prompts, and the average hip joint segmentation time with and without frame prompts; the image segmentation efficiency parameters are specifically the actual segmentation speed, the theoretical segmentation speed, the maximum memory occupied by the SEGVOL model, the total system memory, and the GPU computing resource utilization; The hip joint image segmentation data analysis module includes an image segmentation structure retention index analysis unit, an image segmentation structure space consistency index analysis unit, an interactive prompt image segmentation effect index analysis unit and an image segmentation efficiency index analysis unit, which are respectively used to calculate various parameters and transmit the corresponding calculation results to the hip joint image segmentation evaluation module; The model for obtaining the image segmentation structure retention performance index in the image segmentation structure retention index analysis unit is: , where Pri represents the image segmentation structure retention performance index, Dsc represents the image segmentation overlap index, , X represents the area of ​​fine structure in the segmentation result, Y represents the marked area of ​​fine structure, |X∩Y| represents the size of the intersection area between X and Y, |X| and |Y| represent the size of X and Y area respectively, Egc represents the edge gradient consistency index, , n represents the total number of matching gradient vector pairs, ▽S i represents the gradient vector of the i-th point on the segmentation edge, ▽G i represents the gradient vector of the i-th point on the labeled edge, |▽S i | and |▽G i |represents ▽S i and ▽G i The modulus of Vpr represents the image segmentation volume retention index. , V seg represents the volume of the hip joint obtained by segmentation, V gt represents the true volume of the hip joint, Dlr represents the detail loss rate index, , H(Iorig) represents the high-frequency component of the image before segmentation, H(Iseg) represents the high-frequency component of the image after segmentation, a1, a2, a3 and a4 represent the weights of image segmentation overlap, edge gradient consistency, image segmentation volume retention rate and detail loss rate respectively; The model for obtaining the image segmentation structure spatial consistency performance index in the image segmentation structure spatial consistency index analysis unit is: , where Ssci represents the image segmentation structure spatial consistency performance index, Spd represents the segmentation space position deviation index, , N represents the number of segmentation points, (x I seg ,y I seg ,z I seg ) and (x I gt ,y I gt ,z I gt ) represent the segmentation 3D coordinates and the real 3D coordinates of the I-th positioning point, respectively, Ssi represents the structural segmentation symmetry index, , m represents the number of symmetrical feature points on both sides of the hip joint segmentation result, F j left and F j right represents the left and right features of the jth symmetric point, Dc represents the segmentation angle consistency index, , l represents the number of relevant angles, θ k seg and θ k atl They represent the kth segmentation angle and standard segmentation angle of the hip joint, Asr represents the continuity index of the adjacent structure segmentation, ; M represents the number of sampling points in the adjacent structure area, s J and g J denote the hip joint segmentation and the true adjacent structure area, respectively; b1, b2, b3, and b4 denote the weights of the segmentation space position deviation index, the structure segmentation symmetry index, the segmentation angle consistency index, and the adjacent structure segmentation continuity index, respectively; The model for obtaining the interactive prompt image segmentation effect performance index in the interactive prompt image segmentation effect index analysis unit is: , where Seci represents the performance index of interactive prompt image segmentation effect, Dsc tp and Dsc Ntp Represents the image segmentation overlap index with and without text prompts, Spd Nh and Spd h They represent the spatial position deviation index of segmentation with and without point cues, t Nbp and t bp denote the average segmentation time of hip joints with and without frame prompts, respectively; c1, c2, and c3 denote the weights of text prompts, point prompts, and frame prompts, respectively; The model for obtaining the image segmentation efficiency performance index in the image segmentation efficiency index analysis unit is: , where Sci represents the image segmentation efficiency performance index, V represents the actual segmentation speed, V0 represents the theoretical segmentation speed, Mc represents the maximum memory occupied by the SEGVOL model, T_mc represents the total system memory, and Ur represents the computing resource utilization of the GPU; The hip joint image segmentation evaluation module is used to compare the various indicators of hip joint image segmentation performance obtained by the data analysis module with the set values, and transmit the comparison results to the hip joint image segmentation comprehensive analysis module; The hip joint image segmentation evaluation module includes the following steps: Step 1: Substitute the image segmentation structure retention performance index Pri obtained in the image segmentation structure retention index analysis unit into the formula: , where η(Pri) represents the effective factor of the image segmentation structure retention performance index, and Pri0 represents the set value; Step 2: Substitute the image segmentation structure spatial consistency performance index Ssci obtained from the image segmentation structure spatial consistency index analysis unit into the formula: , where η(Ssci) represents the effective factor of image segmentation structure space consistency, and Ssci0 represents the set value; Step 3: Substitute the interactive prompt image segmentation effect performance index Seci obtained from the interactive prompt image segmentation effect index analysis unit into the formula: , where η(Seci) represents the effective factor of the performance index of the interactive prompt image segmentation effect, and Seci0 represents the set value; Step 4: Substitute the image segmentation efficiency performance index Sci obtained from the image segmentation efficiency index analysis unit into the formula: , where η(Sci) represents the effective factor of the image segmentation efficiency performance index, and Sci0 represents the set value; The hip joint image segmentation comprehensive analysis module is used to comprehensively analyze the effective factors of each indicator calculated by the evaluation module, and transmit the comprehensive analysis results to the hip joint image segmentation human-computer interaction module; The comprehensive analysis model in the hip joint image segmentation comprehensive analysis module is: , where Isci represents the comprehensive index of hip joint image segmentation performance; The hip joint image segmentation human-computer interaction module is used to transmit the comprehensive analysis results input by the comprehensive analysis module to the system administrator information terminal.

2. A hip joint image segmentation and analysis method based on SEGVOL, used for using a hip joint image segmentation and analysis system based on SEGVOL as described in any one of the above claims 1, characterized in that: The following steps are involved: Step 1: Collect hip joint image segmentation data to obtain various parameters for analyzing hip joint image segmentation performance indicators; Step 2: According to the parameters in step 1, the image segmentation structure retention parameters, image segmentation structure spatial position parameters, interactive prompt image segmentation parameters and image segmentation efficiency parameters are calculated respectively to obtain various indicators for analyzing the hip joint image segmentation performance; Step 3: Compare the calculation results of step 2 with the set values ​​to obtain the effective factors of each index of hip joint image segmentation performance; Step 4: Perform comprehensive analysis based on the comparison results of step 3 to obtain comprehensive indicators of hip joint image segmentation performance; Step 5: Perform human-computer interaction based on the comprehensive analysis results of step 4 and the comprehensive performance index of hip joint image segmentation.

Images