Template-based pelvic model corridor positioning method and apparatus

Abstract

The present invention relates to the field of three-dimensional model simulation. Particularly disclosed is a template-based pelvic model corridor positioning method, comprising: on the basis of a pelvic model library, establishing a pelvic average statistical shape model; on the basis of the pelvic average statistical shape model, acquiring a corridor template, the corridor template comprising position information of a corridor in the pelvic average statistical shape model and the pelvic average statistical shape model; preprocessing a CT image, and using the preprocessed CT image to generate a target model of a pelvis; and on the basis of the corridor template, positioning a corridor in the target model. The present invention can realize automatic establishment of corridors, thus saving time for doctors, and simplifying operations of doctors.

Description

A pelvic model channel positioning method and device based on template Technical Field

[0001] The present invention relates to the field of three-dimensional model simulation, and in particular to a template-based pelvic model channel positioning method and device. Background Art

[0002] Pelvic fractures are often caused by severe trauma, resulting in a disability rate as high as 50%-60%. Improper treatment often results in a high mortality rate. To ensure early recovery and prevent further damage to the fracture site, patients undergo implant surgery, using implants such as steel nails, screws, and Kirschner pins to stabilize the fracture site. The effectiveness of this surgery is highly dependent on the fixation of these implants, which in turn depends on the accuracy of the implant's placement. Therefore, the positioning of pelvic implants has become a hot topic of research.

[0003] The implant position is the minimally invasive channel. In the existing technology, the minimally invasive channel is usually determined by the doctor in the preoperative planning stage. Its accuracy affects the patient's recovery effect, so it plays a vital role in the entire surgical cycle. The traditional method of determining the minimally invasive channel is based on the imaging data of the patient's fracture site and the subjective experience of the doctor as an objective basis. First, the degree of fracture is evaluated, and then the approximate position of the minimally invasive channel is determined on the image of the fracture site. Finally, the surgical process plan is refined. The judgment process of the existing technology relies heavily on the human experience of the physician, and the establishment of the channel relies on the doctor's manual operation and clinical experience. It is time-consuming and labor-intensive, and cannot achieve accurate positioning of the channel. In some personalized scenarios, 3D-printed patient bone models are also used to conduct multiple manual experiments to determine the specific location of the channel. However, this method is costly and time-consuming, and cannot meet the needs of quickly determining the channel position.

[0004] Summary of the Invention

[0005] The purpose of the present invention is to provide a template-based pelvic model channel positioning method and device to solve the problems raised in the above background technology.

[0006] The present invention provides a template-based pelvic model channel positioning method, comprising the following steps:

[0007] Establish the average statistical morphological model of the pelvis based on the pelvic model library;

[0008] Acquire a channel template based on the average statistical morphology model of the pelvis, wherein the channel template includes position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis;

[0009] Preprocessing the CT image, and generating a target model of the pelvis using the preprocessed CT image;

[0010] Channels are located in the target model based on the channel template.

[0011] Preferably, the template-based pelvic model channel positioning method includes an offline link and an online link. In the offline link, the channel template is obtained, and in the online link, a target model of the pelvis is generated, and the channel is positioned in the target model.

[0012] Preferably, the establishing of the average statistical morphological model of the pelvis based on the pelvic model library specifically includes:

[0013] Construct a pelvic model library: collect pelvic models of different ages and genders to obtain a pelvic model library;

[0014] Generate an average statistical morphological model: obtain the point and surface information of all models in the pelvic model library, use the statistical morphological model filter to perform statistical morphological analysis on the point and surface information of all models to obtain the fitting morphological error; use all models in the pelvic model library to fit the statistical model, and establish an average statistical morphological model based on the fitting morphological error and the statistical model.

[0015] Preferably, the pelvic minimally invasive channel is positioned on the average statistical morphology model of the pelvis, and the position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis are exported to serve as a channel template.

[0016] Preferably, the preprocessing of the CT image and generating a target model of the pelvis using the preprocessed CT image specifically includes:

[0017] There are multiple CT images, and the preprocessing includes extracting sections of the pelvic area from multiple CT volume data containing the entire lower body to obtain multiple pelvic images containing only the fracture area, compressing each of the pelvic images to a preset size, and registering the compressed multiple pelvic images to be registered based on the registration network to obtain a spatial registration deformation field from the pelvic image to be registered to the target image.

[0018] Preferably, the target image in the registration process is a pelvic image corresponding to any randomly selected patient, and the images to be registered are pelvic images corresponding to the remaining patients.

[0019] Preferably, the registration network specifically includes:

[0020] An encoder-decoder with skip connections and a first preset number of consecutive layers; in the encoder, an encoding operation with an image pyramid structure is adopted; in the decoder, upsampling, convolution and connection skipping are alternately used.

[0021] Preferably, generating a target model of the pelvis using the preprocessed CT image specifically includes:

[0022] Extract the affine matrix T from the CT volume data and transform it into the spatial transformation from the image coordinate system to the anatomical coordinate system according to the spatial registration deformation is a spatial registration deformation field; voxel information of the pelvis image to be registered and the target pelvis image is converted into corresponding anatomical three-dimensional models of the pelvis to be registered and the target pelvis.

[0023] Preferably, locating the channel in the target model based on the channel template specifically includes:

[0024] The average statistical morphological model and the target pelvic anatomical three-dimensional model are aligned to obtain the first transformation matrix between the two; the channel of the target pelvic three-dimensional model is obtained based on the position information of the channel in the channel template in the pelvic average statistical morphological model and the spatial transformation matrix between the two; the aligned pelvic three-dimensional model is calculated based on the spatial transformation from the image coordinate system to the anatomical coordinate system and the pelvic three-dimensional model to be aligned, and the channel of the aligned pelvic model is calculated based on the correspondence between the aligned pelvic model and the target pelvic model and the channel of the target pelvic model to serve as the channel of the pelvic model to be aligned.

[0025] The present invention also provides a template-based pelvic model channel positioning device, comprising:

[0026] A modeling module, used for establishing an average statistical morphological model of the pelvis based on a pelvic model library;

[0027] a template establishing module, configured to obtain a channel template based on an average statistical morphology model of the pelvis, wherein the channel template includes position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis;

[0028] A pelvic modeling module is used to preprocess the CT image and generate a target model of the pelvis using the preprocessed CT image;

[0029] A channel positioning module is used to locate the channel in the target model based on the channel template.

[0030] The present invention provides a pelvic model channel positioning method and device based on a template, which can obtain a channel template based on a statistical model, and use the template to automatically locate the channel in a target model of the pelvis generated based on a CT image. On the one hand, during the channel positioning process, positioning can be automatically completed based on the image without human intervention, which reduces the degree of dependence of the channel positioning process on manual experience and improves the automation process of channel positioning; on the other hand, the target model is directly generated based on the CT image, and the target model has the same shape and parameters as the real pelvis. Channel positioning is performed in the target model, avoiding the doctor's direct channel positioning through CT image viewing, thereby improving the accuracy of channel positioning. In traditional methods, the establishment of the channel depends on the doctor's manual operation and clinical experience, which is time-consuming and labor-intensive. Compared with the traditional preoperative matching method, the method of three-dimensional reconstruction of medical CT images that takes into account internal bone information can obtain an accurate target model of the pelvis, and can realize automatic channel establishment, saving the doctor's time and simplifying the doctor's operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] FIG1 is a flow chart of a template-based pelvic model channel positioning method provided by an embodiment of the present invention;

[0032] FIG2 is a schematic diagram of a registration network structure provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0034] The components of the embodiments of the present invention generally described and shown in the drawings herein may be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the invention.

[0035] Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative work shall fall within the scope of protection of the present invention.

[0036] The technical solution of the present invention is further described below with reference to the accompanying drawings.

[0037] An embodiment of the present invention provides a template-based pelvic model channel positioning method. As shown in FIG1 , the template-based pelvic model channel positioning method includes:

[0038] Establish the average statistical morphological model of the pelvis based on the pelvic model library;

[0039] Acquire a channel template based on the average statistical morphology model of the pelvis, wherein the channel template includes position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis;

[0040] Preprocessing the CT image, and generating a target model of the pelvis using the preprocessed CT image;

[0041] Channels are located in the target model based on the channel template.

[0042] The template-based pelvic model channel positioning method provided by the present invention can obtain a channel template based on a statistical model, and use the template to automatically locate the channel in the target model of the pelvis generated based on the CT image. On the one hand, in the process of channel positioning, positioning can be automatically completed based on the image without human intervention, which reduces the degree of dependence of the channel positioning process on manual experience and improves the automation process of channel positioning; on the other hand, the target model is directly generated based on the CT image, and the target model is the same as the real pelvic morphology and parameters. Channel positioning is performed in the target model, avoiding the doctor's direct channel positioning through CT image viewing, thereby improving the accuracy of channel positioning. In traditional methods, the establishment of the channel depends on the doctor's manual operation and clinical experience, which is time-consuming and laborious. Compared with the traditional preoperative matching method, the method of three-dimensional reconstruction of medical CT images taking into account internal bone information can obtain an accurate target model of the pelvis, which can realize the automatic establishment of the channel, save the doctor's time, and simplify the doctor's operation.

[0043] Furthermore, the template-based pelvic model channel positioning method includes an offline link and an online link. In the offline link, the channel template is obtained, in the online link, a target model of the pelvis is generated, and the channel is positioned in the target model. The present invention completes channel positioning by combining offline and online methods. The offline processing stage is supported by the doctor's clinical experience, and the online processing stage can be automatically obtained without the doctor's manual operation. The acquisition of models and templates in a statistical sense requires the analysis of a large amount of data, and the object of analysis is the model that already exists in the pelvic model library. Therefore, the above-mentioned entire process can be calculated in the offline link, which reduces the amount of calculation and calculation time of the online link, thereby speeding up the efficiency of pelvic model channel positioning.

[0044] As an optional embodiment, establishing the average statistical morphological model of the pelvis based on the pelvic model library specifically includes:

[0045] Constructing a pelvic model library: collecting pelvic models of different ages and genders to obtain a pelvic model library. Preferably, the pelvic model is a pelvic model of a person from a preset country, and the preset country is specifically China.

[0046] Generate an average statistical morphological model: obtain the point and surface information of all models in the pelvic model library, use the statistical morphological model filter to perform statistical morphological analysis on the point and surface information of all models to obtain the fitting morphological error; use all models in the pelvic model library to fit the statistical model, and establish an average statistical morphological model based on the fitting morphological error and the statistical model.

[0047] The template-based pelvic model channel positioning method provided by the present invention takes the models in the model library as input, first reads the point and surface information of all models, then uses the statistical morphological model filter to perform statistical morphological analysis on the point and surface information of all models to obtain the fitting morphological error, uses all models to fit the predicted statistical model, and finally establishes an average statistical morphological model. For the subjects with different characteristics, the pelvic contour shape is basically the same, but the structural parameters are different. In order to establish a template in advance, the method provided by the present invention uses the point and surface information of all models in the model library for fitting, thereby obtaining an average statistical morphological model, that is, obtaining an average statistical morphological model by fitting a large number of models, so as to obtain a channel template based on the average statistical morphological model. The average statistical morphological model integrates the point and surface information of all models in the model library, avoids manual labeling of each model, and simplifies the workload of template generation.

[0048] Preferably, the channel template is obtained based on the average statistical morphological model of the pelvis. According to the clinical experience provided by the doctor, the pelvic minimally invasive channel can be positioned on the average statistical morphological model of the pelvis, and the position information of the channel in the average statistical morphological model of the pelvis and the average statistical morphological model of the pelvis are exported as the channel template. The channel template includes the position information of the channel in the average statistical morphological model of the pelvis and the average statistical morphological model of the pelvis.

[0049] As an optional embodiment, the CT image is preprocessed and a target model of the pelvis is generated using the preprocessed CT image, specifically including: there are multiple CT images, and the preprocessing includes extracting sections of the pelvic region from multiple CT volume data containing the entire lower body to obtain multiple pelvic images containing only fracture sites, compressing each of the pelvic images to a preset size, and registering the multiple compressed pelvic images based on a registration network to obtain a spatial registration deformation field from the image to be registered to the target image.

[0050] Preferably, each of the pelvic images can be compressed to 440x256x256. The multiple pelvic images containing the fracture site after registration and compression are used to realize the registration between non-homologous CT images. The target image in the registration process is the pelvic image corresponding to any randomly selected patient, and the image to be registered is the pelvic image corresponding to the remaining patients. The registration network is shown in Figure 2. The registration network specifically includes: a jump-connected encoder-decoder and a first preset number of continuous layers. In the encoder, an encoding operation of an image pyramid structure is adopted. Preferably, the image pyramid structure uses a three-dimensional convolution with a convolution kernel size of 3x3x3 and a step size of 2. The spatial dimension is reduced to 1 / 2 of the original dimension each time under the action of the convolution step size until it is reduced to 1 / 16 of the original dimension; in the decoder, upsampling, convolution and connection jumps are used alternately. The registration network provided by the present invention adopts an encoder-decoder and a continuous layer structure, which has a finer receptive field for the image, thereby achieving more accurate alignment of anatomical structures.

[0051] Since the spatial registration deformation field of the image to be registered to the target image is the change of each voxel in the three directions of the CT volume data in the image coordinate system, and the channel template is located inside the three-dimensional model in the anatomical coordinate system, as an optional embodiment, the target model of the pelvis is generated using the preprocessed CT image, specifically including: extracting the affine matrix T in the CT volume data, and obtaining the spatial transformation from the image coordinate system to the anatomical coordinate system according to the spatial registration deformation is a spatial registration deformation field; voxel information of the image to be registered and the target image is converted into the corresponding pelvic three-dimensional model to be registered and the target pelvic three-dimensional model.

[0052] The channel is positioned in the target model based on the channel template, specifically including: aligning the average statistical morphological model with the target pelvic three-dimensional model to obtain a spatial transformation matrix between the two; obtaining the channel of the target pelvic model according to the position information of the channel in the channel template in the average statistical morphological model of the pelvis and the spatial transformation matrix; calculating the aligned pelvic model according to the spatial transformation from the image coordinate system to the anatomical coordinate system and the pelvic model to be aligned, and calculating the channel of the aligned pelvis according to the correspondence between the aligned pelvic model and the target pelvic model and the channel of the target pelvic model to serve as the channel of the pelvic model to be aligned.

[0053] An embodiment of the present invention further provides a template-based pelvic model channel positioning device, comprising:

[0054] A modeling module, used for establishing an average statistical morphological model of the pelvis based on a pelvic model library;

[0055] a template establishing module, configured to obtain a channel template based on an average statistical morphology model of the pelvis, wherein the channel template includes position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis;

[0056] A pelvic modeling module is used to preprocess the CT image and generate a target model of the pelvis using the preprocessed CT image;

[0057] A channel positioning module is used to locate the channel in the target model based on the channel template.

[0058] The above embodiments are intended only to illustrate the technical solutions of the present invention and are not intended to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art will appreciate that modifications may be made to the technical solutions described in the above embodiments, or that some of the technical features may be replaced with equivalents. Such modifications or replacements do not deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

[0059] Finally, it should be noted that the present invention is not limited to the aforementioned optional embodiments. Anyone can derive various other product forms based on the teachings of this invention. The aforementioned specific embodiments should not be construed as limiting the scope of protection of the present invention. The scope of protection of the present invention shall be determined by the claims, and the specification may be used to interpret the claims.

Claims

1. A template-based pelvic model channel positioning method, characterized in that: The template-based pelvic model channel positioning method includes: Establish the average statistical morphological model of the pelvis according to the pelvic model library; Acquire a channel template based on an average statistical morphology model of the pelvis, wherein the channel template includes position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis; Preprocessing the CT image, and using the preprocessed CT image to generate a target model of the pelvis; The preprocessing of the CT image and generating a target model of the pelvis using the preprocessed CT image specifically includes: There are multiple CT images, and the preprocessing includes extracting sections of the pelvic region from multiple CT volume data containing the entire lower body to obtain multiple pelvic images containing only the fracture region, compressing each of the pelvic images to a preset size, and registering the compressed multiple pelvic images based on a registration network to obtain a spatial registration deformation field from the image to be registered to the target image; The target model of the pelvis is generated using the preprocessed CT images, including: Extract the affine matrix T from the CT volume data and transform it into the spatial transformation from the image coordinate system to the anatomical coordinate system according to the spatial registration deformation is a spatial registration deformation field; converting the voxel information of the image to be registered and the target image into the corresponding pelvic three-dimensional model to be registered and the target pelvic three-dimensional model; locating a channel in the target model based on the channel template; The locating the channel in the target model based on the channel template specifically includes: The average statistical morphological model and the target pelvic model are registered to obtain the spatial transformation matrix between the two; the channel of the target pelvic model is obtained according to the position information of the channel in the channel template in the average statistical morphological model of the pelvis and the spatial transformation matrix; the registered pelvic model is calculated according to the spatial transformation from the image coordinate system to the anatomical coordinate system and the pelvic model to be registered, and the channel of the pelvic model to be registered is calculated according to the correspondence between the registered pelvic model and the target pelvic model and the channel of the second three-dimensional model, so as to serve as the channel of the pelvic model to be registered.

2. The template-based pelvic model channel positioning method according to claim 1, characterized in that: The template-based pelvic model channel positioning method includes an offline link and an online link. In the offline link, a channel template is obtained, and in the online link, a target model of the pelvis is generated and the channel is positioned in the target model.

3. The template-based pelvic model channel positioning method according to claim 1, characterized in that: The method of establishing the average statistical morphology model of the pelvis according to the pelvic model library specifically includes: Construct a pelvic model library: collect pelvic models of different ages and genders to obtain a pelvic model library; Generate an average statistical morphological model: obtain the point and surface information of all models in the pelvic model library, use the statistical morphological model filter to perform statistical morphological analysis on the point and surface information of all models to obtain the fitting morphological error; use all models in the pelvic model library to fit the statistical model, and establish an average statistical morphological model based on the fitting morphological error and the statistical model.

4. The template-based pelvic model channel positioning method according to claim 1, characterized in that: The pelvic minimally invasive channel is positioned on the average statistical morphology model of the pelvis, and the position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis are exported as a channel template.

5. The template-based pelvic model channel positioning method according to claim 1, characterized in that: The target image in the registration process is the pelvic image corresponding to any randomly selected patient, and the image to be registered is the pelvic image corresponding to the remaining patients.

6. The template-based pelvic model channel positioning method according to claim 1, characterized in that: The registration network specifically includes: An encoder-decoder with skip connections and a first preset number of consecutive layers; in the encoder, an encoding operation of an image pyramid structure is adopted; in the decoder, upsampling, convolution and connection skipping are used alternately.

7. A positioning device using the template-based pelvic model channel positioning method as described in any one of claims 1 to 6, characterized in that: The positioning device comprises: A modeling module, used for establishing an average statistical morphology model of the pelvis according to a pelvic model library; A template establishment module, used for acquiring a channel template based on an average statistical morphology model of the pelvis, wherein the channel template includes position information of the channel in the average statistical morphology model of the pelvis and the average statistical morphology model of the pelvis; A pelvic modeling module is used to preprocess the CT image and generate a target model of the pelvis using the preprocessed CT image; A channel positioning module is used to locate the channel in the target model based on the channel template.

Images