KR20250042108A - One-dimensional image-based spinal surgery planning method - Google Patents

Abstract

The present invention provides a one-dimensional image-based spinal surgery planning method that can minimize radiation exposure and calculate the shortest path for a target area and the insertion length and angle of a needle by using only a one-dimensional image without an adjustment process using a calibrator during spinal surgery planning.

Description

{One-dimensional image-based spinal surgery planning method}

The present invention relates to a method for planning spinal surgery, and more particularly, to a one-dimensional image-based method for planning spinal surgery.

The spine is the center of our body and acts as a pillar, supporting the head from above and connected to the pelvis from below. It is an organ that forms the center of the body and acts as a pillar, and our body is composed of 33 vertebrae. The muscles around the spine support and protect the spine and enable movement. There are 7 types of ligaments in the spine, which are made of elastic fibers and protect and balance the spine, and blood vessels are distributed and spinal nerves are distributed, transmitting sensory information to the brain and transmitting commands received from the brain to the tissues. The spine is not a straight line, but has a gentle S-shaped curve and is connected by intervertebral discs, which stretch and absorb shock depending on the direction of movement when the body moves. If the height between the vertebrae decreases or is pushed out due to various causes, each joint of the spine that is connected to each other becomes weak, causing various spinal diseases such as disc herniation, spinal stenosis, and spondylolisthesis. In order to correct the unstable spinal joints in this way, spinal fusion surgery is performed to connect and fix the joints into a single spine. To do this, it is necessary to determine the appropriate length and diameter of the pedicle screw depending on the patient and the surgical site, and to make a plan for the insertion location and path. In this regard, Korean Patent No. 2394901 discloses a device and method for planning spinal surgery based on two-dimensional medical images.

However, the above prior art may be inefficient because it requires a procedure for alignment using a calibrator for spinal surgery planning, which may expose the patient to excessive radiation.

The present invention is intended to solve various problems including the above-mentioned problems, and aims to provide a one-dimensional image-based spinal surgery planning method capable of minimizing radiation exposure and calculating the shortest path of a target area and the insertion length and angle of a needle by using only a one-dimensional image without an adjustment procedure using a corrector during spinal surgery planning. However, these tasks are exemplary and the scope of the present invention is not limited thereby.

According to one aspect of the present invention, there is provided a step of obtaining a one-dimensional spinal image of a patient requiring spinal surgery;

A feature point designation step of identifying the location of the vertebral body in the spinal image and designating feature points of the vertebral body using a pre-learned U-net structure;

Step of setting a virtual coordinate space for the target area based on the above-mentioned specified feature points:

A step of deriving a triangulation point by applying the coordinate values of the above virtual coordinate space to a trigonometric function;

A step of calculating the insertion length and angle of the needle from the entry point to the target area from the above triangulation point; and

A method for planning a spinal surgery is provided, which includes a planning completion step of determining the length, diameter, etc. of a pedicle screw and determining a plan for the insertion location and path.

As described above, the one-dimensional image-based spinal surgery planning method of the present invention finds anatomical landmarks on anterior-posterior (AP) X-ray without a corrector, thereby minimizing additional radiation exposure, enabling more accurate surgery planning based on knowledge of pre-learned structures, and enabling smooth performance of the targeted surgery according to the surgery planning, thereby enabling rapid treatment of spinal diseases. Of course, the scope of the present invention is not limited by these effects.

Figure 1 is a schematic diagram (left) and a photograph (right) showing a pedicle screw inserted into a spinal body in spinal fusion.
Figure 2 is a photograph showing an AP X-ray view of a patient requiring spinal fusion. The red dot indicates the pedicle, and the yellow dot indicates the spinous process.
Figure 3 is a schematic diagram showing the schematic structure of the vertebral body plane. α1 and α3 shown in the drawing represent pedicles, and α2 represents spinous processes. In the spinal cord surgery planning method of the present invention, a red dot (α3) represents a target site.
Figure 4 is a schematic diagram showing the structure of the posterior part of the vertebral body. In the drawing, the red dot (α3) indicates the target area.
Figure 5 is a schematic diagram schematically showing a configuration for calculating the insertion length and angle of a surgical needle using trigonometry based on a planar image of the vertebral body.
Figure 6 is a schematic diagram showing the shortest path to the target site when inserting a needle in the spinal surgery planning method of the present invention.
Figure 7 is a flow chart showing the procedure of the spinal surgery planning method of the present invention step by step.
Figure 8 is an AP image and an LL image showing the fixation of a pedicle screw in spinal fusion.
Figure 9 is a schematic diagram showing the screw thread angle for estimating the angle between the screw and the patient's vertebral plane based on the change in thread distance.
Figure 10 is a schematic diagram showing the screw angle for estimating the angle between the screw and the patient's vertebral plane based on the change in thread distance.

Glossary of terms:

The term "spinal fusion" used in this document refers to a method of fundamentally treating by removing all diseased discs between the vertebrae and the bones, inserting a disc-shaped support called a cage between them, and fixing it with screws to stabilize the unstable spinal joint. It is performed by surgically immobilizing two to several adjacent vertebrae, and is performed following the removal of lesions of spinal caries or spinal tumors, or to resolve spinal instability such as spinal dislocation fracture, spondylolysis, and spondylolisthesis. The latter is also performed in cases of herniated discs and cervical spondylosis. The methods include anterior fixation, which invades the vertebral body from the front, and posterior fixation, which invades the laminae, intervertebral joints, and spinous processes from the back, and recently, the posterolateral fixation method, which fixes the transverse process interval, is used as the latter method.

The term "U-net" used in this document refers to the most basic deep learning architecture in the field of image semantic segmentation. In the present invention, it is an algorithm optimized for specific object recognition as a variation of artificial intelligence CNN. In the situation of AP imaging of a C-arm, there may be various noises, and the algorithm informs the location of the spinous process and pedicle of interest. This model takes a 2D image as input and predicts a class (e.g., spine, intervertebral disc, etc.) for each pixel.

The term "intervertebral foramen" used in this document refers to the space between the vertebrae through which the spinal cord emerges. The spinal cord passes through the hole that connects the spinal canal with the corpora on both sides. In addition, in the sacrum, the upper and lower parts are fused together and cannot be seen from the outside, but the same-named hole leads from the sacral canal to both sides, and this connects to the anterior and posterior sacral foramina, through which the spinal cord passes through in the same way.

Detailed description of the invention:

According to one aspect of the present invention, there is provided a step of obtaining a one-dimensional spinal image of a patient requiring spinal surgery;

A feature point designation step of identifying the location of the vertebral body in the spinal image and designating feature points of the vertebral body using a pre-learned U-net structure;

Step of setting a virtual coordinate space for the target area based on the above-mentioned specified feature points:

A step of deriving a triangulation point by applying the coordinate values of the above virtual coordinate space to a trigonometric function;

A step of calculating the insertion length and angle of the needle from the entry point to the target area from the above triangulation point; and

A method for planning a spinal surgery is provided, which includes a planning completion step of determining the length, diameter, etc. of a pedicle screw and determining a plan for the insertion location and path.

In the above planning method, the feature point designation step

A step of training the model using spinal image data with specified feature points marked after preparing the UNET model;

A step of extracting feature point location information from the model by the above training;

A step of representing the above feature point location information as a specific pixel location in a spine image; and

It may include a step of identifying and analyzing the structure of the vertebral body by interpreting the location information of the above-mentioned feature points.

In the above planning method, the structure of the vertebral body may be a spinous process, a transverse process or a pedicle, and the feature point may be a position value of the spinous process, the first pedicle or the second pedicle of the vertebral body, and the planning completion step may include a shortest path from the entry point to the target site.

In the above planning method, the triangulation point refers to a point for calculating a path to a target point such as a pedicle or foramen based on the coordinates of the cutting edge of the spinous process and transverse process of the vertebral body, which are known in advance. For example, the path can be calculated by applying triangulation, and the triangulation method means "a method of finding the coordinates and distance to the point by measuring the angles formed by the base and the other two sides of a triangle formed by the point and two reference points, measuring the lengths of the sides, and then performing a series of calculations using the law of sines, etc., when a point and two reference points are given."

In the above planning method, the one-dimensional image may be an X-ray (C-arm) or a spine CT, and the spinal surgery may be a spinal fusion, an endoscopic spinal surgery, or a minimally invasive spinal surgery, and the spinal fusion may be selected from the group consisting of transforaminal lumbar interbody fusion (TLIF), posterior lumbar interbody fusion (PLIF), anterior lumbar interbody fusion (ALIF), and direct lumbar interbody fusion (DLIF).

Hereinafter, the present invention will be described in more detail through examples. However, the present invention is not limited to the examples disclosed below, but can be implemented in various different forms. The following examples are provided to ensure that the disclosure of the present invention is complete, and to fully inform a person having ordinary skill in the art of the scope of the invention.

In the case of patients with severe symptoms that are difficult to treat with non-surgical methods, such as spinal stenosis, herniated disc, scoliosis, or degenerative kyphosis, spinal fusion is generally performed by approaching from the back to restore and fix the vertebral body due to the characteristics of the thoracic spine being protected by the sternum. This is a surgery to prevent abnormal movement between the vertebrae and eliminate the pain caused by inserting pedicle screws into both ends of the vertebrae with the above disease and fixing them together. The thoracic vertebrae, unlike the cervical or lumbar vertebrae, are protected by the sternum, and the height of the intervertebral disc (disk) is low and the space for the neural passage is small, so it is generally approached from the back (back) rather than the front. Figure 1 is a drawing (left) and a photograph (right) showing a pedicle screw inserted into the vertebral body in spinal fusion. For these spinal diseases, it is desirable to avoid surgery and treat them if possible, but spinal fusion should be performed only in limited cases such as stenosis and spondylolysis with spinal instability. Spinal fusion is a widely used surgical method that uses screws to fix vertebrae that have deviated from their normal position to their normal position. It requires less skin incision, significantly less muscle damage related to the surgery compared to conventional surgeries, and the possibility of nerve damage is also low due to the use of a transforaminal approach. For this spinal fusion, it is necessary to determine the appropriate length and diameter of the pedicle screw according to the patient and the surgical site, and to plan the insertion location and path. When performing surgical planning for pedicle screw fixation surgery, the axial view of the spine is the most appropriate, so in the past, axial images were provided to the surgeon through preoperative computed tomography (CT), and surgical planning was performed based on the axial images. The conventional planning method based on the above CT images has the advantage of being able to perform planning on axial images, but has the problem of being harmful to the human body because it requires long-term exposure to radiation during CT scanning.

Specifically, in order to perform a root block (nerve root block), which is one of the pain blocks that injects painkillers, etc. into the relevant nerve segment to relieve pain when planning a spinal surgery to treat spinal diseases such as conventional spinal fusion, the conventional technique is to roughly identify the location based on the doctor's experience by observing the cross-section of the structure with ultrasound under the AP image of an X-ray (C-arm) or a sono guide, and target the anatomical landmark. After identifying the location of the affected area, a needle can be inserted through the back to avoid the lateral processes and inject the drug into the intervertebral foramen, which is the space between the vertebrae through which the spinal nerves emerge.

In general, a needle is inserted while viewing the ultrasound and X-ray (C-arm) in real time. X-rays are continuously exposed to radiation during the insertion process while checking the front and sides, and ultrasound must be performed with one hand, so it requires the doctor's skilled experience. In the case of conventional spinal surgery planning, a process of adjusting the image of the target area by taking a picture of the calibrator together, that is, a process of matching the 2D image with the surgical space is necessary. This is because the alignment is not accurate when the images are virtually superimposed after taking the pictures. However, there is a possibility of being exposed to excessive radiation during the alignment process using the calibrator, and in the case of sono-guided (ultrasound-guided) spinal nerve root block, when multiple segments need to be blocked, each segment must be targeted and a probe must be used to explore, so pain injections (pain blockade) must be placed in multiple places, which is a problem of time and resource consumption. However, since the one-dimensional image-based spinal surgery planning method of the present invention replaces the two-dimensional image of the above-mentioned registration process with only one-dimensional X-ray (C-arm) or a previously taken spine CT, the image for surgery planning needs to be taken only once, and since the CT image taken for preoperative diagnosis can be used, the length and angle of the insertion needle can be calculated while minimizing additional radiation exposure for surgery planning, and the spinal nerve root of the vertebral foramen can be targeted through the shortest path.

Fig. 2 is a photograph showing an X-ray AP view of a patient requiring spinal fusion. The red dot represents the pedicle, and the yellow dot represents the spinous process. The surgical planning method of the present invention uses U-net to find feature points, sets a virtual coordinate space, and calculates the insertion length and angle of a needle to effectively insert a pain block into the transverse process tip and intervertebral foramen or insert a percutaneous pedicle screw. Both the transverse process and the spinous process can be used for accurate alignment, and in the case of the cutting edge used as the feature point, the tip points of the transverse process and the spinous process, which are features of the spinal structure, are not limited to the above processes, but as an embodiment of the present invention, they can be used as calibration markers.

As described above, the surgical planning method of the present invention can target the intervertebral foramen using only AP, unlike the conventional technology that utilizes AP (anterior-posterior) and lateral images based on process feature points determined without a corrector, and can be used to calculate the position and distance of the insertion device during spinal fusion using a pedicle screw as well as endoscopic spinal surgery. The endoscopic spinal surgery removes the cause of pain by inserting an endoscope and a treatment device through two small incisions of about 5 mm, and then checking the lesion from various angles with the endoscope. Although the incisions are small, the lesion and its surrounding tissues can be magnified and checked using an endoscope, enabling much more precise surgery.

The present invention can set a virtual coordinate space of relative values (ratios) based on feature points designated in advance through a spine CT image for spinal surgery planning, and if two points at different locations are known on one plane (x, y-axis information), it is also possible to calculate the z-axis from where the other point is located to a target point. In addition, the CT or AP image of the patient in question can be reconstructed as a spine CT during spinal fusion surgery, and even if there is no CT image, the insertion length and angle of a needle capable of targeting the spinal nerve root through the shortest path can be estimated without a correction procedure using only the AP lateral image. At this time, the spinal fusion may be selected from the group consisting of, but is not limited to, transforaminal lumbar interbody fusion (TLIF), posterior lumbar interbody fusion (PLIF), anterior lumbar interbody fusion (ALIF), and direct lumbar interbody fusion (DLIF).

Hereinafter, a method for calculating the needle insertion length and angle of a one-dimensional image-based spinal surgery planning method of the present invention will be described through the following examples.

Example 1: Calculating the insertion length and angle of the needle

As an example of the one-dimensional image-based spinal surgery planning method of the present invention, the insertion length and angle of the needle can be calculated using trigonometric functions as follows. First, Fig. 3 is a plan view showing the structure of the vertebral body in a three-dimensional plane, in which α1 represents the right pedicle, α2 represents the spinous process, α3 represents the left pedicle, and the circular structure represents the vertebral body. In addition, the direction of the spinous process points toward the dorsal side, and the vertebral body points toward the ventral side. In addition, Fig. 4 is a front view looking from the dorsal side showing the structure of the vertebral body, in which the positions of α1 to α3 are the same as those in Fig. 3. If the y-axis information of α1 to α3 of Fig. 3 is known in advance, the y-axis change value according to the x-axis movement value in Fig. 4 can be mathematically predicted, and this can be calculated with reference to the trigonometric function configuration of Fig. 5. For example, if the above α3 (indicated by a red dot) is a target, the orthogonal plane constant (α2-α3) of Fig. 4 can be calculated by applying the following equation 1 according to the needle insertion length (Da) and the external incident angle (a).

(Formula 1)

Da · cos a = Xa - α3 = O (observed value)

Da · sin a = α2 _y - α3 _y = K (known value)

Since the shape of the vertebral body is slightly different for each surgical patient, if the triangulation point is derived according to the above formula, the needle entrance length and angle corresponding to the shortest distance from the back to the target area can be estimated. At this time, the vertebral body is divided into left and right based on the back, but it can be divided into the first and second vertebral bodies without distinction between left and right at points symmetrical to the left and right based on the center of the intervertebral foramen.

In addition, by the same method, the optimized root block skin incidence angle and length can be calculated by applying the principle of rotation transformation using the transformation and inverse matrix of the three-dimensional matrix based on the coordinates of the feature point at the y-axis end (top) of the spinous process and the center of the target intervertebral foramen in the three-dimensional structure. The content of applying the principle of rotation transformation using the transformation and inverse matrix of the above three-dimensional matrix will be explained. First, the coordinates of the feature point at the y-axis end (top) of the spinous process and the center of the intervertebral foramen are obtained. The coordinates of the feature point at the y-axis end (top) of the spinous process and the center of the intervertebral foramen are measured or calculated. Thereafter, the above coordinates are used to determine the incidence angle and length of the root block skin based on the rotation transformation.

If we explain the composition of the rotation transformation matrix, the rotation transformation is a transformation that rotates an object in three-dimensional space by a desired angle, and this can be expressed as a matrix. The rotation matrix is composed according to a given angle. For example, the transformation matrix in the case of rotation around the x-axis is as follows in Equation 2: Here, θ is the rotation angle.

(Formula 2)

R_x(θ) = | 1 0 0 |

| 0 cos(θ) -sin(θ) |

| 0 sin(θ) cos(θ) |

The transformation calculation uses the acquired rotation matrix to rotate the coordinates of the spinous process feature points based on the coordinates of the center of the intervertebral foramen. This allows us to obtain transformed coordinates. In addition, the application of the inverse matrix must perform a reverse rotation to return the calculated transformed coordinates back to the original coordinate space. Since the inverse matrix of the rotation matrix is equal to its transpose, the inverse rotation matrix is as follows in Equation 3:

(Formula 3)

R_x(-θ) = | 1 0 0 |

| 0 cos(θ) sin(θ) |

| 0 -sin(θ) cos(θ) |

The coordinates obtained through the above reverse rotation are the positions of the spinous process feature points in the original space. The incidence angle and length of the spinous process can be estimated through the above coordinates.

FIG. 6 is a schematic diagram showing the shortest path from the entry point (back) to the target site during needle insertion for spinal surgery planning of the present invention. As shown, the blue arrow represents the spinous process, the red arrow represents the pedicle, the pink line represents the foramen, and the horizontal blue line represents the spinal nerve root. In the case of a root block by the one-dimensional image-based spinal surgery planning method of the present invention, the nerve root extends like a wire, and A is the back, and from A to P can be the shortest path for inserting the needle while avoiding bone structures.

Figure 7 is a flow chart showing the procedure of the spinal surgery planning method of the present invention step by step. The procedure is explained step by step. A one-dimensional spinal image of a patient requiring spinal surgery is acquired (S1). At this time, the one-dimensional spinal image may be an X-ray image or a spinal CT image. Then, using a pre-learned U-net structure, feature points for the spinous process, pedicle, and intervertebral foramen of the vertebral body can be designated (S2) in the one-dimensional spinal image. The U-net consists of a contracting path and an expansive path, and has a bottleneck structure in the middle. The contracting path image is downsampled to extract abstract features, and in the expansive path, the extracted features are upsampled to restore them to the original image size. After preparing a U-net model for model training, the model is trained using preprocessing data on which predetermined feature points are marked. A spinal image is input, and a mask indicating the location of the feature points is output. During model training, a pixel-based loss function is mainly used. After that, the inference and postprocessing process is performed, and the trained model is used to perform inference on a new spinal image. The feature point locations are extracted from the output of the model. Since the extracted feature point location information indicates a specific pixel location in the image, a task of converting it into actual coordinates may be required. The feature point location information obtained as the inference result is interpreted to identify and analyze structures such as the spinous process and the transverse process, and through this, information on the spinal structure is extracted, so that stereotactic functional spinal surgery and operation can be performed on a one-dimensional plane during the surgery or procedure. Therefore, the height change according to the distance change of the incident angle at a predetermined angle can be calculated based on the distance between the position of the end of the spinous process and the center of the pedicle. In the step (S3) of setting a virtual coordinate space based on the specified feature points, the coordinate values of two points whose positions are different in advance on one plane can be applied to a trigonometric function to derive a triangulation point (S4). After calculating the insertion length and angle of the needle from the entry point to the target site through the above virtual coordinate space (S5), the planned insertion path of the spinal surgical instrument (pedicle screw) is determined, and the spinal surgical planning for the patient is completed (S6).

In addition, the spinal surgery planning method of the present invention enables more accurate surgery planning by providing the surgeon with three-dimensional information such as an axial image that can be obtained from a CT (Computed Tomography) from a two-dimensional spinal image of the patient. In order to implement this, it is specifically necessary to secure a two-dimensional spinal image of the surgical patient and align the two-dimensional spinal image and the spinal surgery site location using a surgery planning program. The alignment can be performed by applying a surgery planning program including an algorithm for alignment to derive spatial coordinates for the surgery site location. At this time, the alignment method can generate spatial coordinates of the two-dimensional spinal image and perform alignment of the surgery site location corresponding to the spatial coordinates using the program. In addition, alignment can be performed based on an image generated by backprojecting the two-dimensional spinal image onto another plane on the X-ray projection path. At this time, the two-dimensional spinal image can use an AP (Anterior-Posterior) or LL (Lateral-Lateral) image, but is not limited thereto.

Thereafter, based on the alignment information as described above, a surgical planning program is used to apply and superimpose a virtual stereoscopic figure to the surgical site of the two-dimensional spinal image so as to correspond to an anatomical landmark corresponding to the target during surgery, and then a step of adjusting the shape, size, and position of the stereoscopic figure to fit the actual surgical site is performed. The surgical site position can be automatically detected and adjusted by directly inputting the corresponding position information or performing deep learning using an algorithm that can recognize images or videos. Thereafter, a three-dimensional image is finally generated based on the alignment information and the adjusted stereoscopic figure information and can be confirmed through a display.

The above three-dimensional image is an axial image obtained from a two-dimensional spine image, and can provide the surgeon with an axial image of the spine most suitable for planning a pedicle screw fixation surgery, thereby increasing the accuracy and efficiency of the surgery. At this time, the algorithm can utilize a convolutional neural network (CNN), a convolutional deep belief network (CDBN), a deep belief network (DBN), or a residual network (ResNet), which are effective in extracting and classifying features from two-dimensional data of images or videos. In addition, the anatomical landmark may be a vertebra body, and the step of adjusting the shape, size, and position of the three-dimensional figure may be adjusted so that the boundary of the three-dimensional figure corresponds to the boundary (left, right, top, and bottom) of the vertebra body, and the virtual three-dimensional figure may be a figure having a columnar shape, and may be, for example, a cylinder or a polygonal column, but is not limited thereto.

Furthermore, the spinal surgery planning method of the present invention can implement a three-dimensional image from a two-dimensional spinal image using software, and perform accurate and efficient surgery planning based on the three-dimensional image. The spinal surgery planning method of the present invention can obtain various information necessary for spinal surgery by visualizing the structure and size of the bone at the surgical site by generating a three-dimensional image from a two-dimensional spinal image obtained from a patient using a deep learning model.

Specifically, in 2D-3D reconstruction that can extract anatomical landmarks from 2D images and implement 3D shapes, it is important to train structured data using a deep learning model. A representative example is a deep learning model optimized for structured data training, called convolutional neural networks (CNNs). The CNNs are a deep learning model mainly used for image processing, and are used to extract image features through convolution operations and generate 3D volumes by receiving 2D slice images as input. However, since it is realistically difficult to prepare a large amount of training data, there is a method of utilizing digitally reconstructed radiography (DRR) images. The DRR is efficient in extracting additional information, such as bone structures, by digitally reconstructing 2D radiographic images.

In order to learn the above deep learning model, a large amount of DRR images are generated using a generative adversarial network (GAN), which is an artificial intelligence model for data generation, and a conditional adversarial network (PNN) model for image-to-image conversion, pix2pix, is used to learn actual X-ray images and DRR images, and DRR-like images converted from actual X-ray images are generated. After that, a 3D image is reconstructed from the 2D DRR-like images based on the deep learning model CNNs. After that, axial information, etc. are confirmed through the 3D image, and the shortest path and the insertion length and angle of the needle of the surgical target site are calculated through an algorithm, and the length and diameter of the pedicle screw are determined, thereby obtaining basic information on the insertion location and optimal path.

Fig. 8 is an AP image (left) and an LL image (right) showing the shape of fixing the vertebral body using a pedicle screw. According to the one-dimensional image-based spinal surgery planning method of the present invention, the optimal path insertion of the pedicle screw can be guided according to the height change of the spinous process and the pedicle based on a predetermined entry point. In addition, the angle between the screw and the patient's vertebral plane can be estimated based on the thread of the screw and the change in the distance of the thread, and the error in the entry angle and direction between the guided image and the actual surgery can be notified, and the appropriate direction can be guided. Through this, the result of accurate screw fixation can be obtained even in the lateral view after surgery using only the AP image. The method of estimating the angle between the screw and the patient's vertebral plane based on the change in the distance of the thread first uses C-arm fluoroscopy to photograph an object with a screw, and acquires an image of the process of the thread length changing when rotating 360°. The images obtained in this way interpret the change in the mutual position of the screw or needle and the virtual fixed patient plane according to the change in the thread. By combining the estimated screw angle and the plane angle, it is possible to interpret how the screw rotates on the one-dimensional plane, and understand the spatial position and rotational state of the screw by referring to Table 1 below (Figs. 9 and 10).

Screw angle 60° How to display Marked with pitch size Pitch Water outer diameter cancer endothelium Recommended punching dimensions #16(M16) 1.5 16 14.4 17 #20(M20) 1.5 20 18.4 21 #25(M25) 1.5 25 23.4 26 #32(M32) 1.5 32 30.4 33 #40(M40) 1.5 40 38.4 42 #50(M50) 1.5 50 48.4 52 #63(M63) 1.5 63 61.4 65 #75(M75) 1.5 75 73.4 77

Based on the fluoroscopic image of the screw thread, the virtual patient plane of the screw or needle and the angle of the inserted structure can be estimated, and the error in the calculated angle can be corrected through the changed position during 360° rotation.

Although the present invention has been described with reference to the above-described embodiments, these are merely exemplary, and those skilled in the art will understand that various modifications and equivalent other embodiments are possible therefrom. Therefore, the true technical protection scope of the present invention should be determined by the technical idea of the appended claims.

(National Research and Development Project that supported this invention)

(Task number) 23041

(Task number) 2022R1C1C2008148

(Ministry Name) Ministry of Science, ICT and Future Planning

(Name of the task management (specialized) organization) National Research Foundation of Korea

(Research Project Name) Sejong Science Fellowship

(Research Project Name) Optimization Study of Vagus Nerve Stimulation to Promote Neuroplasticity

(Contribution rate) 1/1

(Name of the organization performing the project) Eulji University Industry-Academic Cooperation Group

(Research Period) 2022.03.01 ~ 2024.02.29

Claims (8)

A step of acquiring a one-dimensional spinal image of a patient requiring spinal surgery;
A feature point designation step of identifying the location of the vertebral body in the spinal image and designating feature points of the vertebral body using a pre-learned U-net structure;
Step of setting a virtual coordinate space for the target area based on the above-mentioned specified feature points:
A step of deriving a triangulation point by applying the coordinate values of the above virtual coordinate space to a trigonometric function;
A step of calculating the insertion length and angle of the needle from the entry point to the target area from the above triangulation point; and
A method for planning a spinal surgery, comprising a planning completion step of determining the length, diameter, etc. of a pedicle screw and determining a plan for the insertion location and path. In the first paragraph,
The above feature point designation step is
A step of training the model using spinal image data with specified feature points marked after preparing the UNET model;
A step of extracting feature point location information from the model by the above training;
A step of representing the above feature point location information as a specific pixel location in a spine image; and
A method comprising a step of identifying and analyzing a structure of a vertebral body by interpreting positional information of the above-mentioned feature points. In the second paragraph,
The above vertebral structure is a spinous process, a transverse process or a pedicle. In the first paragraph,
The above characteristic points are the position values of the terminal end of the spinous process of the vertebral body, the first pedicle, or the second pedicle. In the first paragraph,
The above planning completion step is a method including the shortest path from the entry point to the target area. In the first paragraph,
The above one-dimensional image is an X-ray (C-arm) or a spine CT. In the first paragraph,
The above spinal surgery is a method of spinal fusion, endoscopic spinal surgery, or minimally invasive spinal surgery. In Article 7,
A method wherein the spinal fusion is selected from the group consisting of transforaminal lumbar interbody fusion (TLIF), posterior lumbar interbody fusion (PLIF), anterior lumbar interbody fusion (ALIF), and direct lumbar interbody fusion (DLIF).