CN113124744B - Intelligent intervertebral trial mold, implant and control method - Google Patents

Abstract

A new type of intelligent intervertebral trial mold, implant and control method, after completing the discectomy and decompression in medicine, it is necessary to use the trial mold to determine the appropriate size of the intervertebral disc space, and then select the appropriate prosthesis to fill in When using a mold to select a prosthesis, the selection method is more subjective, and the size of the intervertebral space is often determined according to the doctor's experience. In order to overcome the deficiencies of the prior art, the present invention provides a novel intelligent intervertebral test mold, which can directly read the size of the intervertebral disc space when the test mold is used. The technical scheme of the present invention to solve the technical problem is as follows: an intelligent intervertebral trial mold. The trial mold is composed of an upper cover, a lower cover and a support rod. The connection between the upper and lower covers ensures that the upper cover can be displaced in a small part, and the lower cover has 4 Strain gauges.

Description

Intelligent intervertebral trial mold, implant and control method

technical field

The present invention generally relates to an intelligent intervertebral trial mold, an implant and a control method for assisting the treatment of the spine, and more particularly relates to an intelligent intervertebral trial mold, an implant that can be fixed to the patient's spine and fill the intervertebral space. entry and control methods.

Background technique

Now after discectomy and decompression are completed in medicine, it is necessary to use a test mold to determine the appropriate size of the intervertebral disc space, and then choose a suitable prosthesis to fill the gap. When using a mold to select a prosthesis, the selection method is more subjective. The size of the intervertebral space is often determined based on the doctor's experience.

Pain and pathology of the spine can affect individuals of all ages and can cause great distress to the victim. Pain may be caused by a variety of factors, such as congenital deformities, traumatic injuries, degenerative changes in the spine, etc. This change causes painful hypermotion or collapse of the motion segment, causing the spinal canal to constrict and compress the nerve structures, causing debilitating pain, paralysis, or both, which in turn causes nerve root compression or spinal stenosis.

Intervertebral discs, which are disposed between the endplates of adjacent vertebrae, both stabilize the spine and cushion the vertebral body. However, diseased, degenerated, replaced or otherwise damaged discs (such as herniated or ruptured discs) exhibit a number of undesirable symptoms, including nerve damage, pain, spinal instability, numbness and reduced mobility.

To correct the aforementioned symptoms, surgical intervention, which is of the nature of a discectomy, is usually required. During this procedure, the vertebrae are exposed and the intervertebral disc is removed, either to remove the invading tissue or to provide access for removal.

During spinal implantation after discectomy, a prosthetic implant, or spinal implant, is inserted into the intervertebral space. The implant may be a bone graft removed from another part of the patient's body. Bone grafting has the important advantage of avoiding implant rejection, but it also has some disadvantages. There is always a risk of opening up the second surgical site to obtain the implant, which can lead to infection or pain for the patient, and weaken the site of the implant by removing bone material. Bone implants may not be perfectly shaped and placed, leading to slippage or resorption of the implant or failure of the implant to fit in the vertebrae.

SUMMARY OF THE INVENTION

In order to overcome the deficiencies of the existing technology, the invention provides a novel intelligent intervertebral trial mold, which can directly read the size of the intervertebral disc space when using the trial mold.

A gap forms when a surgeon performs a cervical or lumbar spondylectomy by removing part of the spine. In one embodiment of the invention, an appropriately sized replacement implant can be selected for attachment to the spine on each side of the gap.

An intelligent intervertebral test mold. The test mold includes an upper cover 1, a lower cover 2, and a support rod 3. The connection between the upper and lower covers ensures that the upper cover can be displaced in a small part, and the lower cover has 4 sets of strain sensors.

The surface of the test mold shell is curved, and there is a cylinder 4 inside. After bonding with the lower cover, the cylinder 4 of the upper cover is in contact with the beam 5 of the lower cover. When the upper cover is deformed by force, the cylinder 4 transmits the force to the beam. On the center 5, the beam is bent, and four sets of strain sensors are arranged on the beam, which are the first set of strain sensors 6, the second set of strain sensors 7, the third set of strain sensors 8, and the fourth set of strain sensors 9. The beam is deformed so that the first group of strain sensors 6 and the second group of strain sensors 7 are under the same force, the third group of strain sensors 8 and the fourth group of strain sensors 9 are under the same force, and the first and second groups of strain The force sensors 6 and 7 are opposite to the third and fourth groups of strain sensors 8 and 9, the first group of strain sensors 6, the second group of strain sensors 7, the third group of strain sensors 8, and the fourth group of strain sensors 9 As shown in Figure 4, the intervertebral space can be obtained by converting the output voltage signal into a displacement signal.

Preferably, the cylinder 4 is a telescopic structure, including a driving motor, a lead screw and a nut.

Alternatively, the implant can have the same structure as the intervertebral trial mold, that is, the implant includes an upper cover 1, a lower cover 2, and a support rod 3. The surface of the outer shell is curved, and there is a cylinder 4 inside, which is bonded to the lower cover The cylinder 4 of the rear upper cover is in contact with the beam 5 of the lower cover. When the upper cover is deformed by force, the cylinder 4 transmits the force to the center of the beam 5 to make the beam bend. Four sets of strain sensors are arranged on the beam. They are the first group of strain sensors 6, the second group of strain sensors 7, the third group of strain sensors 8, and the fourth group of strain sensors 9. The beam deforms so that the first group of strain sensors 6 and the second group of strain sensors The force sensors 7 are subjected to the same force, the third group of strain sensors 8 and the fourth group of strain sensors 9 are subjected to the same force, and the first and second groups of strain sensors 6 and 7 are in the same direction as the third and fourth groups of strain sensors 8 and 9. on the contrary. The cylinder 4 includes a driving motor, a lead screw and a nut. The control module controls the telescopic movement of the column body 4 . The lead screw includes a layered structure, which is a protective layer 13, a sealing layer, a phase change layer 14, a sealing layer, and a central layer 15 from outside to inside; the protective layer can be titanium, titanium alloy, stainless steel, cobalt chromium or any combination thereof; The sealing layer is usually a thin film material; the phase change layer consists of a phase change material that switches between solid, liquid/glass, preferably solid below 20°C, liquid/glass above 30°C, and remains solid during the surgical phase , to provide sufficient rigidity support for the patient's spine, and gradually turn into a liquid/glass state after the operation, providing better comfort for the patient.

Preferably, the cylinder 4 has a top seat 10, a telescopic motion pair 11 and a base 12, and the telescopic motion pair 11 includes a drive motor, a lead screw and a nut, and the lead screw includes a layered structure, followed by a protective layer 13, a sealing layer from the outside to the inside. Layer, phase change layer 14, sealing layer, center layer 15; protective layer 13 and center layer 15 can be titanium, titanium alloy, stainless steel, cobalt chromium or any combination thereof; sealing layer is usually a thin film material; phase change layer includes phase change The material is switched between solid state and liquid/glass state. Preferably, it is solid below 20°C and liquid/glass state above 30°C. It remains solid during the operation stage to provide sufficient rigid support for the patient's spine. After the operation is over Slowly turns into a liquid/glass state for better patient comfort.

Preferably, the main body of the implant also includes a control module and a communication device. The control module includes a self-learning function, and the self-learning function includes three levels. First, through self-learning of the historical medical record data of multiple patients, the preset stretching control and pressure The corresponding relationship between the sensors; secondly, the communication device receives the current temperature, humidity, and weather data (whether there is rain, snow, wind, etc.) in real time, and the control module adjusts the length of the implant in real time according to the above data; thirdly, the patient can manually Feedback on the feeling of the spine, the control module learns independently from the patient's feedback data, and readjusts the control strategy of the telescopic control.

Alternatively, the implant may include a surface material and/or a textured solid tip that mates with the inner surface of the void in the spine and facilitates bone growth or solidification, such surface material may include nanostructured regions, including Nanostructures and nanopore structures. The solid end may also be a solid block for securing the implant in bone. This solid block is made of titanium, bone graft material, or other biocompatible material.

Description of drawings

Figure 1 is a schematic diagram of the overall structure;

Figure 2 is a bottom view and a front view of the upper cover;

Figure 3 is a top view and a front view of the lower cover;

Figure 4 is a wiring diagram of the strain sensor;

5 is a schematic diagram of the main body of the implant;

Fig. 6 is a schematic diagram of the layered structure of the lead screw.

Detailed ways

The present invention will be further elaborated below in conjunction with embodiment.

In order to make the purpose and technical advantages of the present invention clearer, the present invention will be described in detail in conjunction with the accompanying drawings.

An intelligent intervertebral test mold. The test mold includes an upper cover 1, a lower cover 2, and a support rod 3. The connection between the upper and lower covers ensures that the upper cover can be displaced in a small part, and the lower cover has 4 sets of strain sensors.

The test mold includes an upper cover 1, a lower cover 2, and a support rod 3. The upper cover is shown in Figure 2. The surface of the outer shell is a curved surface, and there is a cylinder 4 inside. The beam 5 is in contact with the lower cover as shown in Figure 3. When the upper cover is deformed by force, the column 4 transmits the force to the center of the beam 5 to make the beam bend. Four sets of strain sensors are arranged on the beam. The first group of strain sensors 6, the second group of strain sensors 7, the third group of strain sensors 8, the fourth group of strain sensors 9, the beam 5 is deformed so that the first group of strain sensors 6, the second group of strain sensors 7 The force is the same, the first group of strain sensors 6, the second group of strain sensors 7 are in opposite directions to the third group of strain sensors 8, and the fourth group of strain sensors 9, the arrangement of the strain sensors is shown in Figure 4, and the output The intervertebral space can be obtained by converting the voltage signal into a displacement signal.

Preferably, the cylinder 4 has a top seat 10, a telescopic motion pair 11 and a base 12, and the telescopic motion pair 11 includes a drive motor, a lead screw and a nut, and the lead screw includes a layered structure, followed by a protective layer 13, a sealing layer from the outside to the inside. Layer, phase change layer 14, sealing layer, central layer 15; protective layer can be titanium, titanium alloy, stainless steel, cobalt chromium or its any combination; sealing layer is usually thin film material; phase change layer comprises phase change material, in solid state, Switch between liquid/glass state, preferably solid state below 20°C, liquid/glass state above 30°C, maintain solid state during operation, provide sufficient rigid support for the patient's spine, and gradually turn into liquid state after operation / Glass state, providing better comfort for patients.

As shown in Figure 4, when the cylinder 4 is under pressure, the resistance of the strain gauge 6 decreases by ΔR, the resistance of the strain gauge 8 increases by μΔR, the resistance of the strain gauge 7 decreases by ΔR, and the resistance of the strain gauge 9 increases by μΔR, the output voltage signal is:

Among them, V _e represents the excitation voltage, V _out2 is the output voltage signal under pressure, and R is the resistance value of the strain gauge when it is not deformed. Due to the different layout, the deformation of the strain gauge 5 and the strain gauge 6 is different, resulting in resistance The changing resistance values are different, and the resistance value changes are ΔR and μΔR respectively, and μ is the ratio of the resistance value change.

When V _out2 exceeds the preset threshold range, the control module controls the lead screw to extend or shorten.

Preferably, the angle of the lead screw threads is chosen to minimize the torque generated by the drive, especially since this torque is reacted by the patient's own anatomy. The extensions are not limited to being formed from titanium, titanium alloys, stainless steel, cobalt chromium, or any combination thereof, or other suitable medical material capable of withstanding spinal loads without excessive bending or twisting. The extensions may have a diameter approximately that of a spine.

When the implant is initially implanted in the patient, the telescopic kinematic pairs 11 are in their fully retracted position, this retracted or telescoping feature allows the device to be significantly smaller in length. When implanting, the smaller length of the telescopic kinematic pair 11 reduces the operation risk and the trauma to the patient.

Preferably, the implant contains necessary power supply and communication equipment to control the expansion and contraction of the implant. The implant of the present invention can be completely independent without physical external connection, and the power supply is a battery pack composed of several small batteries. Alternatively, the battery pack may include one or more rechargeable batteries that may be inductively charged while remaining in place within the patient. The power supply may be an inductive power supply that is only energized when coupled inductively (not physically) to an external device. An inductive power supply can be incorporated into the patient bed so that the power supply is only charged or powered when the patient is lying in bed, keeping the battery small for the processor.

Communication devices may be configured to transmit and receive radio frequency signals. A pair of encoders may be provided to measure the angle and speed of rotation of each drive element in the assembly. The drive element can accept external control commands via a communication device allowing manual adjustment of the implant.

In order to ensure that the implant device is not accidentally shortened or elongated, the implant may include an electromagnetic locking device mounted on the motor and lead screw, allowing the motor and lead screw to rotate only when the electromagnetic locking device is in an unlocked state, the processor Can be configured to automatically lock the electromagnetic locking device at the end of exercise and automatically unlock the electromagnetic locking device at the beginning of exercise.

The base and driver are formed from a medical material that is strong enough to resist the load of the extension. The housing and driver may be formed from a polymer, such as a polymeric resin.

Preferably, the lead screw includes a layered structure, which is followed by a protective layer 13, a sealing layer, a phase change layer 14, a sealing layer, and a central layer 15 from outside to inside; the protective layer 13 and the central layer 15 can be titanium, titanium alloy, stainless steel , cobalt chromium or any combination thereof; the sealing layer is usually a thin film material; the phase change layer includes a phase change material, switching between solid state, liquid state/glass state, preferably solid state below 20°C and liquid state above 30°C /glass state, it remains solid during the operation stage, providing sufficient rigidity support for the patient's spine, and gradually turns into a liquid/glass state after the operation, providing better comfort for the patient.

Preferably, the control module includes a self-learning function, and the self-learning function includes three levels. First, through self-learning of historical medical record data of multiple patients, the correspondence between the preset telescopic control and the strain sensor data is preset; secondly, the communication device Receive the current temperature, humidity, and weather data (whether there is rain, snow, wind, etc.) in real time, and the control module adjusts the length of the implant in real time according to the above data; again, the patient can manually feed back the feeling of the spine, and the control module can give feedback data to the patient Self-learning, readjusting the control strategy of telescoping control.

The autonomous learning function includes the following steps:

1. System modeling

This application regards the control strategy of the column 4 as an intelligent body, which can control the expansion and contraction of the column 4 according to the change of the environment state by interacting with the environment. Assume that the action set for the agent to control stretching is A={c,o,d}, where c,o,d represent the three actions of stretching, constant, and shortening respectively. When the agent chooses to perform action c or d once, the cylinder 4 will be extended or shortened by one unit; when the agent chooses to perform action o, the cylinder 4 will keep the original length unchanged. Set the set of environmental states as S={T, D, L, y, x, f}, where T, D, and L respectively represent the air temperature, relative humidity, and the length of the current column 4, and y, x, and f respectively Indicates whether there are weather data such as rain, snow, wind, etc., which is a Boolean value.

其中Q(c,s_t)表示智能体在状态s_t下选择动作c时预测能够获得的期望回报，以此类推。当智能体选择其中的一个动作a_t执行后，环境状态信息变为s_t+1，并反馈给智能体一个即时奖励r_t。t时刻的环境奖励r_t表示如下：Input the environmental state information s _t ∈ S at time t into the neural network model, and the neural network model outputs the predicted value of the three actions

Where Q(c, s _t ) represents the expected reward that the agent can obtain when it chooses action c in state s _t , and so on. When the agent chooses one of the actions at to execute, the environment state information becomes _{st +1} , and an immediate reward r _t is fed back to the agent. The environmental reward r _t at time t is expressed as follows:

Where L _t represents the actual length of the cylinder 4 to be adjusted at time t, and G _t represents the optimum length of the cylinder 4 under the environment state s _t .

The cylinder 4 telescoping control strategy algorithm adopted in this application includes two feedforward neural network models with the same structure, called the predictive model and the target model. The target model is a regular copy of the prediction model. By using the stochastic gradient descent method to iteratively optimize the prediction model, it can effectively adjust the length of the cylinder 4 in real time through environmental data, and can realize the patient feedback data according to the patient's manual feedback information. Self-learning.

2. Model training

Step 1: Model establishment and initialization. This patent uses a 5-layer feed-forward neural network as a prediction model, and the number of neurons in each layer is: 6 (input layer), 32, 128, 128, 64, 3 (output layer). After each layer of affine layer, ReLU nonlinear mapping is performed, and the output layer uses Softmax activation function. Initialize the weight parameters of the prediction model, and copy a copy of it as the target model.

将s₀输入预测模型，得到模型预测的3个动作的Q值。以90％的概率选择3个Q值中最大值

对应的动作，以10％的概率随机选择一个动作作为对柱体4在t＝0时刻执行的动作a₀。执行动作a₀，柱体4长度变为

对比

和Gⁱ，获得t＝0时刻的即时奖励r₀，环境状态变为

将获得的数据{s₀,a₀,r₀,s₁}保存到预设容量大小的经验池中用于模型训练。再将s₁输入预测模型，按相同的策略执行动作a₁，获得r₁和s₂，并将相应的数据保存到经验池中。迭代上述动作直至r_t＝0时结束一个回合的采样。判断经验池是否已满，若是，进入步骤3；否则，重复本步骤。Step 2: Random sampling. The historical case data include weather T ⁱ , humidity D ⁱ , and whether there is rain y ⁱ , snow ^xi , wind f ⁱ , etc., and the corresponding optimal length G ⁱ of column 4 . Start a round of random sampling: Randomly sample a set of environmental state information {T ⁱ , D ⁱ , y ⁱ , x ⁱ , f ⁱ } and G ⁱ in the historical case data, and obtain the actual length L ⁱ of the current cylinder 4, constitute the initial state

Input s ₀ into the prediction model to get the Q values of the three actions predicted by the model. Choose the largest of the 3 Q values with 90% probability

For the corresponding action, an action is randomly selected with a probability of 10% as the action a _{0 executed on the cylinder 4 at time t=0} . Execute action a ₀ , the length of cylinder 4 becomes

Compared

and G ⁱ , get the immediate reward r ₀ at time t=0, and the environment state becomes

Save the obtained data {s ₀ , a ₀ , r ₀ , s ₁ } to the experience pool with preset capacity for model training. Then input s ₁ into the prediction model, execute action a ₁ according to the same strategy, obtain r ₁ and s ₂ , and save the corresponding data in the experience pool. The above actions are iterated until r _t =0 to end a round of sampling. Determine whether the experience pool is full, if so, go to step 3; otherwise, repeat this step.

计算状态s_t的期望回报为：Step 3: Train historical data until the model converges. According to the preset batch parameter value, randomly sample a batch from the data set {s _t , a _t , r _t , s _t+1 } saved in the experience pool in step 2, input _st into the prediction model, and output the model prediction 3 Q values of , get the value Q _a corresponding to a _t . Input s _t+1 into the target model to obtain the maximum value

Calculate the expected return of state s _t as:

Computing the loss for model training using the mean square error is:

The stochastic gradient descent method is used to iteratively train the model to minimize the model loss Loss. Only the prediction model is trained but not the target model. When the number of training times reaches the preset threshold, the weight parameters of the prediction model are copied to update the target model. After training all the data in the experience pool once, judge whether the number of model trainings has been reached, if so, save the model and end; otherwise, go to step 3.

3. Model prediction:

选择其中的最大值

对应的动作a₀控制柱体4做相应的调整。再次获取实时环境状态s₁，输入预测模型，获得最大Q值对应的动作a₁，并对柱体4执行动作a₁。不断重复上述操作，使柱体4根据外界环境状态的变化而做出实时自适应调整。During the application process, the communication device receives weather data such as the current temperature T, humidity D, whether there is rain y, snow x, wind f, etc. in real time, and forms the initial environmental state s ₀ ={T ₀ , D ₀ , L ₀ , y ₀ , x ₀ , f ₀ }. Input s ₀ into the trained prediction model, and output the Q value of stretching, unchanged, and shortening 3 actions

choose the maximum

The corresponding action a ₀ controls the cylinder 4 to make corresponding adjustments. Obtain the real-time environment state s ₁ again, input it into the prediction model, obtain the action a 1 corresponding to the maximum Q value, and execute the action a ₁ on the column ₄ . The above operations are repeated continuously, so that the column 4 can make real-time self-adaptive adjustments according to changes in the external environment.

4. Self-learning based on patient manual feedback data:

According to the feeling of the spine, the patient manually adjusts the length of the column body 4 to the best state through 3 buttons {extend (c), remain unchanged (o), shorten (d)} and press the constant (o) key to move this The environmental state s at the time and the length G of the column 4 are stored as historical case data. On the basis of the existing model, enter the step 2 of model training for sampling and training, thereby realizing autonomous learning and readjustment of the patient's feedback data. The purpose of the control strategy for scaling control.

Although the present application has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can use the methods and technical contents disclosed above to analyze the present invention without departing from the spirit and scope of the present application. Possible changes and modifications are made in the technical solution. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments according to the technical essence of the application without departing from the content of the technical solution of the application belong to the technical solution of the application. protected range.

Claims (7)

1. an intelligent intervertebral test mold, is characterized in that: comprise loam cake (1), lower cover (2), strut bar (3), the shell surface of loam cake (1) is along the lengthwise direction of strut bar (3) It is a curved surface with a cylinder (4) inside. After connecting with the lower cover (2), the cylinder (4) of the upper cover is in contact with the beam (5) of the lower cover. When the upper cover is deformed by force, the cylinder ( 4) The force is transmitted to the center of the beam (5) to bend the beam. Four sets of strain sensors are arranged on the beam, and the beam is deformed so that the first set of strain sensors (6), the second set of strain sensors (7) The direction of the force is the same, the third group of strain sensors (8) and the fourth group of strain sensors (9) are in the same direction of force, the first and second groups of strain sensors (6, 7) are the same as the third and fourth groups of strain sensors The force direction of the sensors (8, 9) is opposite; the cylinder (4) is a telescopic structure. 2. A kind of intelligent intervertebral trial mold according to claim 1, characterized in that: the cylinder (4) includes a drive motor, a leading screw and a nut; the leading screw includes a layered structure, and the layered The structure includes a phase change layer (14). 3. A control method of the intelligent intervertebral trial mold according to claim 1 or 2, characterized in that: the output voltage signal of the strain gauge group is obtained under pressure, and when the voltage signal exceeds the preset threshold range, the control method The module controls the lead screw to extend or shorten. 4. An intelligent intervertebral implant is characterized in that: comprise upper cover (1), lower cover (2), support rod (3) and control module, the shell surface of upper cover (1) is along the support rod (3) ) is a curved surface, and there is a cylinder (4) inside. After connecting with the lower cover (2), the cylinder (4) of the upper cover is in contact with the beam (5) of the lower cover. When the upper cover is deformed due to force , the column (4) transmits the force to the center of the beam (5) to make the beam bend, and four sets of strain sensors are arranged on the beam, and the beam deforms so that the first set of strain sensors (6), the second set of strain sensors The force direction of the sensors (7) is the same, the force direction of the third group of strain sensors (8) and the fourth group of strain sensors (9) is the same, and the first and second groups of strain sensors (6, 7) are connected with the third and fourth groups of strain sensors (9). The four sets of strain sensors (8, 9) are stressed in opposite directions; the column (4) is a telescopic structure. 5. A kind of intelligent intervertebral implant according to claim 4, characterized in that: the column (4) includes a drive motor, a leading screw and a nut; the leading screw includes a layered structure, and the branch The layer structure includes a phase change layer (14). 6. A control method of the intelligent intervertebral implant according to claim 5, characterized in that: use the intelligent intervertebral trial mold according to claim 1 to measure the intervertebral space data; when the intelligent intervertebral implant When implanted into the patient's body, the column (4) is at the shortest position. After implanting into the intervertebral space, the column (4) is adjusted to the target length according to the previous measured intervertebral space data; the implant includes a power supply and The communication device that controls the expansion and contraction of the implant, the power source is a battery pack consisting of batteries; the battery pack includes one or more rechargeable batteries, and the rechargeable batteries can be inductively charged while in the patient's body. 7. A control method for an intelligent intervertebral implant according to claim 5, characterized in that: the control module includes an autonomous learning function; the autonomous learning function includes three levels, first through multiple patient history Self-learning of medical record data, preset the corresponding relationship between stretch control and strain sensor data; secondly, the communication equipment receives the current temperature, humidity, and weather data in real time, and the control module adjusts the implantation in real time according to the temperature, humidity, and weather data Thirdly, the patient can manually feedback the feeling of the spine, and the control module learns independently from the patient's feedback data, and readjusts the control strategy of the telescopic control.

Images