CN115689971A - Pedicle screw implantation channel planning method and device based on deep learning - Google Patents
Pedicle screw implantation channel planning method and device based on deep learning Download PDFInfo
- Publication number
- CN115689971A CN115689971A CN202110857970.3A CN202110857970A CN115689971A CN 115689971 A CN115689971 A CN 115689971A CN 202110857970 A CN202110857970 A CN 202110857970A CN 115689971 A CN115689971 A CN 115689971A
- Authority
- CN
- China
- Prior art keywords
- pedicle
- image
- pedicle screw
- channel
- vertebral
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a method and a device for planning a pedicle screw implantation channel based on deep learning, wherein the method comprises the following steps: an extraction process, namely extracting a vertebral pedicle CT image in a spine CT image to be extracted; a layering process, wherein the vertebral pedicle CT image is sliced and layered along the cross section of the vertebral pedicle, the border of the vertebral pedicle is confirmed, and a plurality of layers of two-dimensional images of the vertebral pedicle are sequentially obtained; and planning a flow, sequentially obtaining the central point of the two-dimensional image of the pedicle of vertebral arch, fitting the central point, and obtaining a pedicle screw implantation channel. According to the method and the device for planning the pedicle screw implantation channel based on the deep learning, provided by the embodiment of the invention, the vertebral pedicle two-dimensional image is obtained by sequentially extracting the spine CT image, slicing, layering and confirming the boundary, the central point of the two-dimensional image is sequentially obtained, and the obtained central point is fitted to obtain the final pedicle screw implantation channel, so that the pedicle screw implantation channel is rapidly obtained, and the operation efficiency is improved.
Description
Technical Field
The invention relates to the technical field of medical images, in particular to a method and a device for planning a pedicle screw implantation channel based on deep learning.
Background
Currently, in human spine surgery, the pedicle screw internal fixation technology has a very important position and is considered as the most stable spine fixation mode. This fixation technique requires that the pedicle screws must be moved inside the pedicles; once the pedicle of a vertebral arch is punctured, serious damage such as spinal cord and vertebral vessel damage can be caused, and the life safety of a patient is even endangered. Therefore, ensuring accurate pedicle screw placement is a goal of study by researchers.
In the traditional pedicle implantation operation, a doctor adjusts an operation path according to an X-ray image shot by a C-shaped arm of an operating room by virtue of professional skills of the doctor, so that the pedicle implantation operation is completed. Different levels of expertise of doctors often lead to different degrees of sequelae in many operations. In the pedicle screw implantation operation based on the computer-aided navigation technology, the relationship among an operation space, an image space and a robot space is established by using bony marks and a position tracking sensor in the image according to a preoperative medical image or intraoperative medical image (CT medical image and X-ray image), and then surgical instruments and a robot end device in the operation space are converted into the image space, so that virtual reality is realized to assist a doctor to complete the pedicle screw implantation operation. The method is complex to operate, errors in the process can be amplified, the position relation seen on the navigation image is a false image, the relation between the tool and the pedicle of vertebral arch cannot be accurately and timely reflected, and real-time detection cannot be realized.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a method and a device for planning a pedicle screw implantation channel based on deep learning.
The invention provides a pedicle screw implantation channel planning method based on deep learning, which comprises the following steps:
an extraction process, namely extracting a vertebral pedicle CT image in a spine CT image to be extracted;
a layering process, wherein the vertebral pedicle CT image is sliced and layered along the cross section of the vertebral pedicle, the border of the vertebral pedicle is confirmed, and a plurality of layers of two-dimensional images of the vertebral pedicle are sequentially obtained;
and planning a flow, sequentially obtaining the central point of the two-dimensional image of the pedicle of vertebral arch, fitting the central point, and obtaining a pedicle screw implantation channel.
According to the pedicle screw implantation channel planning method based on deep learning provided by the invention, the method further comprises the following steps: and extending the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the herringbone ridge as the starting point of the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the vertebral body as the termination point of the pedicle screw-implanting channel, and acquiring the length of the pedicle screw-implanting channel based on the starting point and the termination point.
According to the deep learning-based pedicle screw implantation channel planning method provided by the invention, before the extraction process, the method further comprises the following steps: and the preprocessing process is used for normalizing the three-dimensional interlayer spacing in the spine CT image to be processed and performing 0-value centralization on the normalized data to obtain the spine CT image to be extracted.
According to the method for planning the pedicle screw implantation channel based on deep learning, provided by the invention, the pedicle boundary is confirmed, and a plurality of layers of two-dimensional images of the pedicle are sequentially obtained, and the method specifically comprises the following steps: performing switching operation on the pedicle CT image subjected to slice layering processing, counting the number of connected domains of the pedicle CT image subjected to slice layering processing, judging whether the number of the connected domains of the pedicle CT image subjected to slice layering processing is greater than a preset threshold value, and if the condition is met, reserving the connected domains as target connected domains; and if the condition is not met, abandoning the connected domain, and sequentially obtaining a multilayer pedicle two-dimensional image according to the target connected domain.
According to the deep learning-based pedicle screw implantation channel planning method provided by the invention, the central point of the two-dimensional pedicle image is obtained by utilizing a gray scale gravity center method, and the formula of the gray scale gravity center method is as follows:
wherein (u, v) represents the coordinates of the pixel points, f (u, v) represents the gray value of the pixel points (u, v), omega represents the pixel point set of the two-dimensional image of the pedicle of vertebral arch,
the abscissa representing the center point is shown as,
the ordinate of the center point is indicated.
According to the pedicle screw implantation channel planning method based on deep learning, provided by the invention, the central point is fitted to obtain a pedicle screw implantation channel, and the method specifically comprises the following steps: acquiring a fitted linear equation based on the fitted linear function of the central point and the minimum error sum of squares function, and acquiring the pedicle screw implantation channel based on the fitted linear equation; wherein the fitted straight line function is represented as: z = ax + by + c, the minimum sum of squared errors function being expressed as:
a. b and c represent parameters to be solved of the fitted straight line function, N represents the sequence of the central points, and N represents the number of the central points.
According to the deep learning-based pedicle screw implantation channel planning method provided by the invention, spine CT image data to be processed is divided into a training data set, a verification data set and a test data set; wherein the training data set is used for training a neural network, the validation data set is used for adjusting hyper-parameters of the neural network, and the test data set is used for validating the accuracy of the neural network.
The invention also provides a pedicle screw implantation channel planning device based on deep learning, which comprises:
an extraction module to: extracting a vertebral pedicle CT image in a spine CT image to be extracted;
a layering module to: carrying out layered processing on the vertebral pedicle CT images along the cross section of the vertebral pedicle to sequentially obtain a plurality of layers of vertebral pedicle two-dimensional images;
a planning module to: and sequentially obtaining the central point of the two-dimensional image of the pedicle of vertebral arch, fitting the central point, and obtaining a pedicle of vertebral arch screw implantation channel.
The invention also provides electronic equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor executes the program to realize the steps of the pedicle screw channel planning method based on deep learning.
The present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method for deep learning based pedicle screw channel planning as described in any of the above.
According to the method and the device for planning the pedicle screw implantation channel based on the deep learning, the vertebral pedicle CT image is obtained by performing extraction operation on the spinal CT image, then the vertebral pedicle CT image is subjected to layering processing and boundary confirmation processing to obtain the two-dimensional vertebral pedicle image, the central points of the two-dimensional image are sequentially obtained, the central points are fitted to obtain the final pedicle screw implantation channel, the whole process of obtaining the pedicle screw implantation channel depends on automatic calculation of an image processing technology, manual assistance of a doctor is not needed, automatic planning of the screw implantation channel is realized, and flow standardization and rapidity of the screw implantation channel obtaining process are realized.
Drawings
In order to more clearly illustrate the technical solutions of the present invention or the prior art, the drawings needed for the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
FIG. 1 is one of the flow diagrams of the deep learning-based pedicle screw channel planning method provided by the invention;
FIG. 2 is a second schematic flow chart of the method for planning a pedicle screw implantation channel based on deep learning according to the present invention;
FIG. 3 is a schematic view of a pedicle CT image and a landmark provided by the present invention;
FIG. 4 is a two-dimensional image of a pedicle with a center point provided by the present invention;
FIG. 5 is a schematic view of a pedicle screw channel provided by the invention;
FIG. 6 is a schematic structural diagram of a deep learning-based pedicle screw implantation channel planning device provided by the invention;
fig. 7 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is obvious that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments of the present invention, are within the scope of protection of the present invention.
Fig. 1 is a schematic flow diagram of a deep learning-based pedicle screw channel planning method provided by the invention, and as shown in fig. 1, the method includes:
s110, extracting a flow, namely extracting a vertebral pedicle CT image in the spine CT image to be extracted;
s120, a layering process, namely carrying out slicing layering processing on the vertebral pedicle CT image along the transverse section of the vertebral pedicle, confirming the boundary of the vertebral pedicle, and sequentially obtaining a multi-layer vertebral pedicle two-dimensional image;
s130, planning a process, sequentially obtaining the central point of the two-dimensional image of the pedicle, fitting the central point, and obtaining a pedicle screw implantation channel.
It should be noted that, the original spine CT image is preprocessed to obtain the spine CT image to be extracted, and compared with the original spine CT image, the interlayer spacing between the spine CT image data to be extracted realizes the unification processing, which is beneficial to the subsequent training process of the network model by using the data and accelerates the training speed of the network model.
Inputting the spine CT image data to be extracted into the 3D segmentation network model, and obtaining a pedicle CT image through extraction processing of the 3D segmentation network model; then, slice layering processing is carried out on the vertebral pedicle CT image along the cross section of the vertebral pedicle through a 3D segmentation network model, and multilayer slice images are sequentially obtained; and confirming the pedicle boundaries on the multilayer slice images in sequence to obtain a multilayer pedicle two-dimensional image.
The method comprises the steps of successively completing extraction processing and layering processing of a spine CT image depending on a 3D segmentation network model, then obtaining a plurality of layers of pedicle two-dimensional images which are sequentially and orderly distributed by confirming pedicle boundary processing, respectively calculating central points of the pedicle two-dimensional images, fitting the obtained central points by using a fitting linear equation, and taking the obtained fitting linear as a pedicle screw implantation channel.
According to the method for planning the pedicle screw implantation channel based on deep learning, the CT image of the pedicle is obtained by extracting the CT image of the spine, then the CT image of the pedicle is subjected to layering processing and boundary confirmation processing to obtain the two-dimensional image of the pedicle, the central points of the two-dimensional image are sequentially obtained, the central points are fitted to obtain the final pedicle screw implantation channel, the whole process of obtaining the pedicle screw implantation channel depends on automatic calculation of an image processing technology, manual assistance of a doctor is not needed, automatic planning of the screw implantation channel is achieved, and flow standardization and rapidity of the screw implantation channel obtaining process are achieved.
According to the pedicle screw implantation channel planning method based on deep learning, provided by the invention, the method further comprises the following steps: and extending the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the herringbone ridge as the starting point of the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the vertebral body as the termination point of the pedicle screw-implanting channel, and acquiring the length of the pedicle screw-implanting channel based on the starting point and the termination point.
It should be noted that, in the process of obtaining the pedicle screw implantation channel by using the 3D segmentation network model, the pedicle screw implantation channel output by the pedicle screw implantation channel is only limited to the length inside the pedicle, and the output result cannot be directly applied to the actual operation process.
According to the deep learning-based pedicle screw-planting channel planning method, the intersection point of the extension line of the pedicle screw-planting channel and the herringbone ridge obtained by the 3D network model is used as the starting point of the pedicle screw-planting channel, and the intersection point of the extension line of the pedicle screw-planting channel and the vertebral body is used as the ending point, so that the specific starting point and the specific ending point of the pedicle screw-planting channel are determined based on the intersection point, the length of the pedicle screw-planting channel is obtained simultaneously, further accurate planning of the pedicle screw-planting channel is realized based on the intersection point, the application of the pedicle screw-planting channel in the actual operation process is facilitated, and the accuracy of the actual screw-planting process is ensured.
According to the pedicle screw implantation channel planning method based on deep learning, provided by the invention, before the extraction process, the method further comprises the following steps: and the preprocessing process is used for normalizing the three-dimensional interlayer spacing in the spine CT image to be processed and performing 0-value centralization on the normalized data to obtain the spine CT image to be extracted.
The method comprises the steps of adjusting the window width and window position of an original spine CT image, removing interference outside a bone structure, changing the interlayer spacing of three dimensions of the CT image data into 1mm by a resampling method for original spine CT image data, and sequentially carrying out normalization operation and 0 value centralization processing on the data for convenience of data processing and accelerated convergence during program operation to obtain the spine CT image to be extracted. Normalization means that the mean value of the variable is subtracted from the mean value of the variable, then the mean value is divided by the standard deviation, 0-value centralization means that the mean value of the variable is subtracted from the mean value of the variable, namely a translation process is actually carried out, the centers of all data after translation are (0, 0), and data which are subjected to standard normal distribution and have the mean value of 0 and the standard deviation of 1 are finally obtained through centralization and standardization processing, so that errors caused by different dimensions, self-variation or large numerical value difference can be eliminated.
According to the pedicle screw implantation channel planning method based on deep learning, normalization processing and 0-value centralization processing are sequentially carried out on a spine CT image to be processed, standardization of image data is achieved, model convergence speed can be increased when a network model is processed based on the standard image data, and meanwhile when the data are input into the network model, the data processing speed can be improved by the network model.
According to the pedicle screw implantation channel planning method based on deep learning, provided by the invention, the pedicle boundary is confirmed, and a plurality of layers of pedicle two-dimensional images are sequentially obtained, and the method specifically comprises the following steps: performing switching operation on the pedicle CT image subjected to slice layering processing, counting the number of connected domains of the pedicle CT image subjected to slice layering processing, judging whether the number of the connected domains of the pedicle CT image subjected to slice layering processing is greater than a preset threshold value, and if the condition is met, reserving the connected domains as target connected domains; and if the condition is not met, abandoning the connected domain, and sequentially obtaining a multilayer pedicle two-dimensional image according to the target connected domain.
In the image processing process, the opening and closing operation is a common means for image processing, and is specifically divided into an opening operation and a closing operation, wherein the opening operation refers to an erosion operation and then an expansion operation, and the closing operation refers to an expansion operation and then an erosion operation. The corrosion operation refers to a process of eliminating boundary points and enabling boundaries to shrink inwards, and can be used for eliminating small and meaningless objects; the dilation operation refers to a process of merging all background points in contact with an object into the object and expanding the boundary to the outside, and can be used for filling a hole in the object. The on operation can remove small-particle noise in the image and break adhesion between objects, and the off operation can connect adjacent objects in the image, smooth the boundary and simultaneously not obviously change the area.
The connected domain refers to an image region which is formed by foreground pixel points with the same pixel value and adjacent positions in the image, the image with the number of the connected domains smaller than a preset threshold value indicates that the image is an interference image, the image is abandoned, the connected domain meeting the preset threshold value is reserved as a target connected domain, and the optimized pedicle two-dimensional image is obtained based on the target connected domain.
It should be noted that the preset threshold may be preset according to actual needs, and in the present invention, the preset threshold is set to be 2.
According to the pedicle screw-implanting channel planning method based on deep learning, provided by the invention, the opening and closing operation is carried out on the pedicle CT images subjected to slice layering processing in sequence, the screening process of the images is realized based on the number of the connected domains, the removal of irrelevant images is realized based on the opening and closing operation, the optimization of the images is finally realized, the interference of the irrelevant images on the pedicle screw-implanting channel planning process is favorably avoided, and the accurate planning is realized.
According to the pedicle screw implantation channel planning method based on deep learning, the center point of the two-dimensional image of the pedicle is obtained by using a gray scale gravity center method, and the formula of the gray scale gravity center method is as follows:
wherein (u, v) represents the coordinates of the pixel points, f (u, v) represents the gray value of the pixel points (u, v), omega represents the pixel point set of the two-dimensional image of the pedicle of vertebral arch,
the abscissa representing the center point is shown as,
the ordinate of the center point is indicated.
For targets with uneven brightness (such as light spots and light stripes), the gray scale gravity center method can work out light intensity weight mass center coordinates as tracking points according to the light intensity distribution of the targets, and is also called density mass center algorithm. In the invention, the characteristic that the pedicle boundary and the internal brightness in the two-dimensional image of the pedicle are different is fully considered, and the central point of the two-dimensional image of the pedicle is accurately obtained based on the characteristic.
The pedicle screw implantation channel planning method based on deep learning provided by the invention realizes accurate acquisition of the central point of a two-dimensional image of the pedicle of vertebral arch through a gray scale gravity center method, and is favorable for accurate planning of the pedicle screw implantation channel based on the central point subsequently.
According to the pedicle screw implantation channel planning method based on deep learning, the pedicle screw implantation channel is obtained by fitting the central point, and the method specifically comprises the following steps: acquiring a fitted linear equation based on the fitted linear function and the minimum error sum of squares function of the central point, and acquiring the pedicle screw implantation channel based on the fitted linear equation; wherein the fitted straight line function is represented as: z = ax + by + c, and the minimum sum of squared errors function is expressed as:
a. b and c represent parameters to be solved of the fitted straight line function, N represents the sequence of the central points, and N represents the number of the central points.
The central point is obtained by fitting with a least square method, and the basic principle of the least square is to minimize variance, specifically:after the central points of the two-dimensional images of the vertebral pedicle are obtained in sequence, a fitting straight line function z = ax + by + c is set, and the central point is specifically expressed as (x) i ,y i ,z i ) Meanwhile, the minimum error sum of squares function is used as the basis for solving the parameters a, b and c in the fitted straight line function, and specifically, the obtained central points are sequentially substituted into the minimum error sum of squares function
After specific numerical values of parameters a, b and c are obtained, the parameters are substituted into a fitting linear function to obtain a fitting linear equation, the fitting linear equation represents a specific path of a pedicle screw implantation channel, wherein the parameters to be solved of the fitting linear function are represented by a, b and c, N represents the sequence of the central points, N represents the number of the central points, and the value of i is [1, N ]]。
According to the deep learning-based pedicle screw-planting channel planning method, the fitted linear function is set, and the parameters in the fitted linear function are obtained through the least error sum of squares function, so that the fitted linear equation is finally obtained, the mathematical expression of the pedicle screw-planting channel is realized, and the visual quantitative description of the pedicle screw-planting channel is ensured.
According to the pedicle screw implantation channel planning method based on deep learning, provided by the invention, the CT image data of the spine to be processed is divided into a training data set, a verification data set and a test data set; wherein the training data set is used for training a neural network, the validation data set is used for adjusting hyper-parameters of the neural network, and the test data set is used for validating the accuracy of the neural network.
It should be noted that, in the present invention, the spine CT image data to be processed is divided into a training data set, a verification data set and a test data set, and meanwhile, the ratio of the data amount between the training data set and the test data set is 7: 3, and the data amount of the verification data set is 20% of the training data set.
According to the pedicle screw implantation channel planning method based on deep learning, provided by the invention, the CT image data of the spine to be processed are respectively divided into the training data set, the verification data set and the test data set, and the proportional relation among the data sets is reasonably limited, so that the quality of a network model obtained through the training of the data sets can be improved, and the network model is ensured to have good prediction accuracy.
Fig. 2 is a second schematic flow chart of the deep learning-based pedicle screw channel planning method provided by the invention, as shown in fig. 2, the method includes:
step1, acquiring spinal CT data;
step2,3D pedicle segmentation, wherein the 3D segmentation network model is used for extracting the CT data of the spine to obtain the CT data of the pedicle;
step3, layering the Mask cross section, and carrying out slice layering operation on the vertebral pedicle CT data along the vertebral pedicle cross section by using a 3D segmentation network model to obtain image slices of the vertebral pedicle on each layer;
and Step4, calculating the boundary of the pedicle of vertebral arch of the tangent plane, obtaining a target connected domain by utilizing the opening and closing operation and the connected domain screening process, and determining the boundary of the two-dimensional image of the pedicle of vertebral arch.
And Step5, calculating the center point of the Mask boundary of the tangent plane, and calculating the center point of the Mask boundary of the tangent plane by adopting a gray scale gravity center method, namely the center point of the two-dimensional image of the pedicle of vertebral arch on each layer.
Step6, fitting the pedicle anatomy axis, and fitting by using a least square method to obtain a central point obtained in Step5 to obtain the pedicle anatomy axis;
step7, calculating a starting point, a stopping point and a length, extending the pedicle anatomical axis obtained in Step6, taking the intersection point of the pedicle anatomical axis and the herringbone ridge as a starting point, taking the intersection point of the pedicle anatomical axis and the herringbone ridge as a stopping point, and taking the distance between the starting point and the stopping point as a final length;
and Step8, ending.
According to the method for planning the pedicle screw implantation channel based on deep learning, the vertebral pedicle CT image is obtained by performing extraction operation on the spine CT image, then the vertebral pedicle CT image is subjected to layering processing and boundary confirmation processing to obtain the two-dimensional vertebral pedicle image, the central points of the two-dimensional image are sequentially obtained, the central points are fitted to obtain the final pedicle screw implantation channel, the whole process of obtaining the pedicle screw implantation channel depends on automatic calculation of an image processing technology, manual assistance of a doctor is not needed, automatic planning of the screw implantation channel is realized, and the flow standardization and the rapidity of the screw implantation channel obtaining process are realized; meanwhile, the length of the pedicle screw implantation channel is determined by calculating the starting point and the ending point, the planning precision of the operation path is further improved, and the smooth operation of the operation process is ensured.
Fig. 3 is a schematic view of a pedicle CT image and a labeling diagram provided by the present invention, as shown in fig. 3, fig. 3-1 is a spine CT image, and fig. 3-2 is a pedicle portion, and in practical applications, the pedicle portion image shown in fig. 3-2 is displayed in different colors to achieve a labeling effect.
Fig. 4 is a two-dimensional image of the pedicle with a central point, as shown in fig. 4, a white area is a two-dimensional image representing the pedicle, a black-white adjacent area is a boundary of the two-dimensional image representing the pedicle, and a black point is a central point of the two-dimensional image representing the pedicle.
Fig. 5 is a schematic view of the pedicle screw implantation channel provided by the present invention, as shown in fig. 5, the pedicle screw implantation channel is extended, the intersection point of the extension line of the pedicle screw implantation channel and the herringbone ridge is used as the starting point of the pedicle screw implantation channel, the intersection point of the herringbone ridge and the vertebral body is used as the ending point of the pedicle screw implantation channel, and the distance between the starting point and the ending point is used as the length of the final pedicle screw implantation channel.
According to the method for planning the pedicle screw-planting channel based on deep learning, provided by the invention, the intersection point of the extension line of the pedicle screw-planting channel and the herringbone ridge is used as the starting point of the pedicle screw-planting channel, and the intersection point of the extension line of the pedicle screw-planting channel and the herringbone ridge is used as the ending point, so that the specific starting point and the specific stopping point of the pedicle screw-planting channel are determined based on the intersection point, and the length of the pedicle screw-planting channel is obtained simultaneously, so that the pedicle screw-planting channel is further accurately planned based on the intersection point, the application of the pedicle screw-planting channel in the actual operation process is facilitated, and the accuracy of the actual screw-planting process is ensured.
Fig. 6 is a schematic structural diagram of a deep learning-based pedicle screw channel planning device provided by the invention, and as shown in fig. 6, the device comprises: an extraction module 610, a layering module 620, and a planning module 630, wherein,
an extraction module 610 to: extracting a vertebral pedicle CT image in a spine CT image to be extracted;
a layering module 620 for: carrying out slice layering processing on the vertebral pedicle CT image along the cross section of the vertebral pedicle, confirming the vertebral pedicle boundary, and sequentially obtaining a plurality of layers of vertebral pedicle two-dimensional images;
a planning module 630 configured to: and sequentially obtaining the central point of the two-dimensional image of the pedicle of vertebral arch, fitting the central point, and obtaining a pedicle of vertebral arch screw implantation channel.
According to the deep learning-based pedicle screw-planting channel planning device, the spine CT image is subjected to extraction operation to obtain the pedicle CT image, then the pedicle CT image is subjected to layering processing and boundary confirmation processing to obtain the two-dimensional pedicle image, the central points of the two-dimensional image are sequentially obtained, the central points are fitted to obtain the final pedicle screw-planting channel, the whole process of obtaining the pedicle screw-planting channel depends on automatic calculation of an image processing technology, manual assistance of a doctor is not needed, automatic planning of the screw-planting channel is achieved, and flow standardization and rapidness of the screw-planting channel obtaining process are achieved.
According to the invention, the deep learning-based pedicle screw implantation channel planning device further comprises an intersection module, wherein the intersection module is used for: and extending the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the herringbone ridge as the starting point of the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the vertebral body as the termination point of the pedicle screw-implanting channel, and acquiring the length of the pedicle screw-implanting channel based on the starting point and the termination point.
According to the deep learning-based pedicle screw-planting channel planning device, the intersection point of the extension line of the pedicle screw-planting channel and the herringbone ridge obtained by the 3D network model is used as the starting point of the pedicle screw-planting channel, and the intersection point of the extension line of the pedicle screw-planting channel and the vertebral body is used as the ending point, so that the specific starting point and the specific ending point of the pedicle screw-planting channel are determined based on the intersection point, the length of the pedicle screw-planting channel is obtained simultaneously, further accurate planning of the pedicle screw-planting channel is realized based on the intersection point, the application of the pedicle screw-planting channel in the actual operation process is facilitated, and the accuracy of the actual screw-planting process is ensured.
According to the invention, the device for planning the pedicle screw implantation channel based on deep learning further comprises a preprocessing module, wherein the preprocessing module is used for: and normalizing the three-dimensional interlayer spacing in the spine CT image to be processed, and performing 0 value centralization on the normalized data to obtain the spine CT image to be extracted.
According to the pedicle screw implantation channel planning device based on deep learning, normalization processing and 0-value centralization processing are sequentially carried out on a spine CT image to be processed, standardization of image data is achieved, model convergence speed can be increased when a network model is processed based on the standard image data, and meanwhile when the data are input into the network model, the data processing speed can be improved by the network model.
According to the deep learning-based pedicle screw implantation channel planning device provided by the invention, the layering module 620 is specifically used for confirming the pedicle boundary and sequentially obtaining the multilayer pedicle two-dimensional images: performing switching operation on the pedicle CT image subjected to slice layering processing, counting the number of connected domains of the pedicle CT image subjected to slice layering processing, judging whether the number of the connected domains of the pedicle CT image subjected to slice layering processing is greater than a preset threshold value, and if the condition is met, reserving the connected domains as target connected domains; and if the condition is not met, abandoning the connected domain, and sequentially obtaining a multilayer pedicle two-dimensional image according to the target connected domain.
According to the pedicle screw-implanting channel planning device based on deep learning, the opening and closing operation is performed on the pedicle CT images subjected to slice layering processing successively, the screening process of the images is realized based on the number of the connected domains, the irrelevant images are removed based on the opening and closing operation, the optimization of the images is finally realized, the interference of the irrelevant images on the pedicle screw-implanting channel planning process is avoided, and the accurate planning is realized.
According to the pedicle screw implantation channel planning device based on deep learning, the device acquires the central point of the two-dimensional image of the pedicle by using a gray scale gravity center method, and the formula of the gray scale gravity center method is as follows:
wherein (u, v) represents the coordinates of the pixel points, f (u, v) represents the gray value of the pixel points (u, v), omega represents the pixel point set of the two-dimensional image of the pedicle of vertebral arch,
the abscissa representing the center point is shown as,
the ordinate of the center point is indicated.
The pedicle screw-implanting channel planning device based on deep learning provided by the invention realizes accurate acquisition of the central point of a two-dimensional image of the pedicle of vertebral arch through a gray scale gravity center method, and is beneficial to accurate planning of the pedicle screw-implanting channel based on the central point subsequently.
According to the deep learning-based pedicle screw channel planning device provided by the invention, the planning module 630 is specifically used for obtaining a fitting linear equation based on a fitting linear function and a minimum error sum of squares function of the central point and obtaining the pedicle screw channel when the planning module is used for fitting the central point and obtaining the pedicle screw channel; wherein the fitted straight line function is represented as: z = ax + by + c, and the minimum sum of squared errors function is expressed as:
a. b, c representing said fitted straight-line functionAnd (3) obtaining parameters, wherein N represents the sequence of the central points, and N represents the number of the central points.
According to the deep learning-based pedicle screw-planting channel planning device, the fitted linear function is set, and the parameters in the fitted linear function are obtained through the least error sum of squares function, so that the fitted linear equation is finally obtained, the mathematical expression of the pedicle screw-planting channel is realized, and the visual quantitative description of the pedicle screw-planting channel is ensured.
According to the invention, the device for planning the pedicle screw implantation channel based on deep learning further comprises a splitting module, wherein the splitting module is used for: splitting spine CT image data to be processed into a training data set, a verification data set and a test data set; wherein the training data set is used for training a neural network, the validation data set is used for adjusting hyper-parameters of the neural network, and the test data set is used for validating the accuracy of the neural network.
According to the pedicle screw implantation channel planning device based on deep learning, provided by the invention, the CT image data of the spine to be processed are respectively divided into the training data set, the verification data set and the test data set, and the proportional relation among the data sets is reasonably limited, so that the quality of a network model obtained through the training of the data sets can be improved, and the network model is ensured to have good prediction accuracy.
Fig. 7 illustrates a physical structure diagram of an electronic device, and as shown in fig. 7, the electronic device may include: a processor (processor) 710, a communication Interface (Communications Interface) 720, a memory (memory) 730, and a communication bus 740, wherein the processor 710, the communication Interface 720, and the memory 730 communicate with each other via the communication bus 740. The processor 710 may invoke logic instructions in the memory 730 to perform a deep learning based pedicle screw channel planning method comprising: an extraction process, wherein a vertebral pedicle CT image in a spine CT image to be extracted is extracted; a layering process, wherein the vertebral pedicle CT image is sliced and layered along the cross section of the vertebral pedicle, the vertebral pedicle boundary is confirmed, and a plurality of layers of vertebral pedicle two-dimensional images are sequentially obtained; and planning a flow, sequentially obtaining the central point of the two-dimensional image of the pedicle, fitting the central point, and obtaining a pedicle screw implantation channel.
In addition, the logic instructions in the memory 730 can be implemented in the form of software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions, which when executed by a computer, enable the computer to perform the method for deep learning based pedicle screw channel planning provided by the above methods, the method comprising: an extraction process, namely extracting a vertebral pedicle CT image in a spine CT image to be extracted; a layering process, wherein the vertebral pedicle CT image is sliced and layered along the cross section of the vertebral pedicle, the vertebral pedicle boundary is confirmed, and a plurality of layers of vertebral pedicle two-dimensional images are sequentially obtained; and planning a flow, sequentially obtaining the central point of the two-dimensional image of the pedicle of vertebral arch, fitting the central point, and obtaining a pedicle screw implantation channel.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the methods for deep learning based pedicle screw channel planning provided above, the methods comprising: an extraction process, wherein a vertebral pedicle CT image in a spine CT image to be extracted is extracted; a layering process, wherein the vertebral pedicle CT image is sliced and layered along the cross section of the vertebral pedicle, the vertebral pedicle boundary is confirmed, and a plurality of layers of vertebral pedicle two-dimensional images are sequentially obtained; and planning a flow, sequentially obtaining the central point of the two-dimensional image of the pedicle of vertebral arch, fitting the central point, and obtaining a pedicle screw implantation channel.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one position, or may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, and not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A pedicle screw implantation channel planning method based on deep learning is characterized by comprising the following steps:
an extraction process, namely extracting a vertebral pedicle CT image in a spine CT image to be extracted;
a layering process, wherein the vertebral pedicle CT image is sliced and layered along the cross section of the vertebral pedicle, the vertebral pedicle boundary is confirmed, and a plurality of layers of vertebral pedicle two-dimensional images are sequentially obtained;
and planning a flow, sequentially obtaining the central point of the two-dimensional image of the pedicle, fitting the central point, and obtaining a pedicle screw implantation channel.
2. The deep learning based pedicle screw channel planning method according to claim 1, further comprising:
and extending the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the herringbone ridge as the starting point of the pedicle screw-implanting channel, taking the intersection point of the pedicle screw-implanting channel and the vertebral body as the ending point of the pedicle screw-implanting channel, and acquiring the length of the pedicle screw-implanting channel based on the starting point and the ending point.
3. The deep learning based pedicle screw channel planning method according to claim 1, wherein before the extracting process, the method further comprises:
and the preprocessing process is used for normalizing the three-dimensional interlayer spacing in the spine CT image to be processed and performing 0-value centralization on the normalized data to obtain the spine CT image to be extracted.
4. A pedicle screw implantation channel planning method based on deep learning according to claim 1, wherein the confirming of the pedicle boundaries sequentially obtains a multi-layer pedicle two-dimensional image, specifically comprising:
performing switching operation on the pedicle CT image subjected to slice layering processing, counting the number of connected domains of the pedicle CT image subjected to slice layering processing, judging whether the number of the connected domains of the pedicle CT image subjected to slice layering processing is greater than a preset threshold value, and if the condition is met, reserving the connected domains as target connected domains; and if the condition is not met, abandoning the connected domain, and sequentially obtaining a multilayer pedicle two-dimensional image according to the target connected domain.
5. The deep learning-based pedicle screw implantation channel planning method according to claim 1, wherein a gray scale gravity center method is used for obtaining the center point of the two-dimensional pedicle image, and the formula of the gray scale gravity center method is as follows:
wherein (u, v) represents the coordinates of the pixel points, f (u, v) represents the gray value of the pixel points (u, v), omega represents the pixel point set of the two-dimensional image of the pedicle of vertebral arch,
the abscissa representing the center point is shown as,
the ordinate of the center point is indicated.
6. The deep learning-based pedicle screw channel planning method according to claim 1, wherein the fitting of the central point to obtain a pedicle screw channel specifically comprises:
acquiring a fitted linear equation based on the fitted linear function of the central point and the minimum error sum of squares function, and acquiring the pedicle screw implantation channel based on the fitted linear equation; wherein the fitted straight-line function is represented as: z = ax + by + c, the least square errorThe sum function is expressed as:
a. b and c represent parameters to be solved of the fitted straight line function, N represents the sequence of the central points, and N represents the number of the central points.
7. The deep learning-based pedicle screw implantation channel planning method according to claim 3, wherein spine CT image data to be processed is split into a training data set, a verification data set and a test data set; wherein the training data set is used for training a neural network, the validation data set is used for adjusting hyper-parameters of the neural network, and the test data set is used for validating the accuracy of the neural network.
8. The utility model provides a pedicle of vertebral arch is planted and is followed closely passageway planning device based on degree of depth study which characterized in that includes:
an extraction module to: extracting a vertebral pedicle CT image in a spine CT image to be extracted;
a layering module to: carrying out slice layering processing on the vertebral pedicle CT image along the cross section of the vertebral pedicle, confirming the vertebral pedicle boundary, and sequentially obtaining a plurality of layers of vertebral pedicle two-dimensional images;
a planning module to: and sequentially obtaining the central point of the two-dimensional image of the pedicle, fitting the central point, and obtaining a pedicle screw implantation channel.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps of the method for deep learning based pedicle screw channel planning as claimed in any one of claims 1 to 7.
10. A non-transitory computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the deep learning based pedicle screw channel planning method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110857970.3A CN115689971A (en) | 2021-07-28 | 2021-07-28 | Pedicle screw implantation channel planning method and device based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110857970.3A CN115689971A (en) | 2021-07-28 | 2021-07-28 | Pedicle screw implantation channel planning method and device based on deep learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115689971A true CN115689971A (en) | 2023-02-03 |
Family
ID=85058778
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110857970.3A Pending CN115689971A (en) | 2021-07-28 | 2021-07-28 | Pedicle screw implantation channel planning method and device based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115689971A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116650112A (en) * | 2023-07-24 | 2023-08-29 | 杭州键嘉医疗科技股份有限公司 | Automatic planning method, device, equipment and storage medium for pedicle screw path |
| CN116889467A (en) * | 2023-06-21 | 2023-10-17 | 北京长木谷医疗科技股份有限公司 | Intelligent self-nailing method, device, equipment and medium for vertebral column |
-
2021
- 2021-07-28 CN CN202110857970.3A patent/CN115689971A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116889467A (en) * | 2023-06-21 | 2023-10-17 | 北京长木谷医疗科技股份有限公司 | Intelligent self-nailing method, device, equipment and medium for vertebral column |
| CN116889467B (en) * | 2023-06-21 | 2024-04-02 | 北京长木谷医疗科技股份有限公司 | Intelligent self-nailing method, device, equipment and medium for vertebral column |
| CN116650112A (en) * | 2023-07-24 | 2023-08-29 | 杭州键嘉医疗科技股份有限公司 | Automatic planning method, device, equipment and storage medium for pedicle screw path |
| CN116650112B (en) * | 2023-07-24 | 2023-11-14 | 杭州键嘉医疗科技股份有限公司 | Automatic planning method, device, equipment and storage medium for pedicle screw path |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11937888B2 (en) | Artificial intelligence intra-operative surgical guidance system | |
| US10991070B2 (en) | Method of providing surgical guidance | |
| US20240096508A1 (en) | Systems and methods for using generic anatomy models in surgical planning | |
| CN110738681A (en) | automatic pedicle screw operation path planning method based on deep learning network | |
| US20200320697A1 (en) | Method, system, and device for lung lobe segmentation, model training, model construction and segmentation | |
| JPWO2020056086A5 (en) | ||
| US20180161102A1 (en) | Method and system for estimating a deflated lung shape for video assisted thoracic surgery in augmented and mixed reality | |
| US20080312663A1 (en) | Computer-assisted joint analysis using surface projection | |
| CN115689971A (en) | Pedicle screw implantation channel planning method and device based on deep learning | |
| EP3107477B1 (en) | Prediction of shapes | |
| CN109447960A (en) | A kind of object identifying method and device | |
| CN112509022A (en) | Non-calibration object registration method for preoperative three-dimensional image and intraoperative perspective image | |
| US11980491B2 (en) | Automatic recognition method for measurement point in cephalo image | |
| JP2022527525A (en) | Identification of lesion boundaries in image data | |
| CN112884792A (en) | Lung image segmentation method and device, electronic equipment and storage medium | |
| CN110559075B (en) | Intraoperative augmented reality registration method and device | |
| CN118453115A (en) | Real-time image guidance system based on surgery | |
| US20220207844A1 (en) | System and method for automatic transfer function | |
| CN113222886B (en) | Jugular fossa and sigmoid sinus groove positioning method and intelligent temporal bone image processing system | |
| CN117934689B (en) | Multi-tissue segmentation and three-dimensional rendering method for fracture CT image | |
| CN113177948B (en) | Medical image automatic window adjusting method based on semantic segmentation and application | |
| CN115100189B (en) | Three-dimensional medical image cutting method, electronic device and storage medium | |
| US11861804B1 (en) | Systems and method for smooth segmentation | |
| Yamada et al. | Extraction of median plane from facial 3D point cloud based on symmetry analysis using ICP algorithm | |
| CN110090076B (en) | Methods and systems for estimating deflated lung shape for video assisted thoracic surgery with augmented and mixed reality |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |