CN106777976A - Radiotherapy machine human tumour motion estimation prediction system and method based on particle filter - Google Patents

Abstract

The invention discloses a system and method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering. The method includes: S1, using a breath tracking unit and an image positioning unit to respectively collect three-dimensional motion data of body surface marker points and tumors in the body; S2 , According to the three-dimensional motion data, establish the motion relationship model between the tumor at the current moment and the tumor at the historical moment, and use this model as the state transition equation of the particle filter to establish the motion relationship model between the body surface markers and the tumor in the body within a period of time As the observation equation of the particle filter; S3, based on the state transition equation and the observation equation, use the particle filter algorithm to estimate the movement position of the tumor in the body according to the movement data of the body surface marker points at the current moment. The invention can be used in any form of state space model, has stronger modeling ability for the nonlinear characteristics of variable parameters, and has higher prediction accuracy.

Description

Tumor Motion Estimation and Prediction System and Method for Radiotherapy Robot Based on Particle Filter

technical field

The invention relates to the field of medical technology, in particular to a system and method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering.

Background technique

At present, the main method for the treatment of lung cancer is stereotactic radiotherapy, but the respiratory movement of the human body seriously affects the accuracy of radiotherapy. In order to reduce the influence of respiratory motion, the most effective method is the real-time tracking technology of respiratory motion. This technology establishes an association model between the tumor in the body and the marker points on the body surface, and uses a prediction algorithm to obtain the future movement of the tumor according to the movement of the marker points. Information, so as to adjust the radiation beam in real time to ensure the relative stillness of the radiation and the tumor, so as to achieve real-time tracking of tumor radiation therapy.

The prediction algorithm is used to compensate for the time delay of the stereoradiotherapy system and to predict in advance where the tumor will reach in the future. Some of the predictive algorithms currently studied include artificial neural network, Kalman filter, fuzzy control, etc. But the above algorithm has the following disadvantages:

The learning efficiency of artificial neural network is low, the convergence speed is slow, and it takes a long time for network training, which is not suitable for real-time prediction;

The Kalman filter requires the system to be a Gaussian white noise system, and it is only suitable for linear systems, but the actual respiratory movement process is a complex nonlinear process, so the Kalman filter is not suitable for the prediction of tumor movement;

The fuzzy control prediction effect is better, but this method has limited adaptive ability and high complexity.

The particle filter solves the constraint that the system must satisfy the Gaussian distribution, and is suitable for any nonlinear system that can be expressed by a state-space model, and has high precision. This method is applied to the real-time tracking technology of radiotherapy robots, which is very important for improving the patient's health. It is of great significance to improve the survival rate and promote the development of medical technology in our country.

Therefore, in view of the above technical problems, it is necessary to provide a system and method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering.

Contents of the invention

In view of this, the object of the present invention is to provide a system and method for estimating and predicting the tumor motion of a radiotherapy robot based on particle filtering, so as to solve the above-mentioned problem of predicting the tumor position in the precise treatment of a radiotherapy robot.

In order to achieve the above object, the technical solutions provided by the embodiments of the present invention are as follows:

A particle filter-based radiation therapy robot tumor motion estimation and prediction system, the system includes: a breathing tracking unit, an image positioning unit, an image fusion unit, and a treatment planning unit, and the breathing tracking unit is used to track the movement of marker points on the patient's body surface data, the image positioning unit is used to track the motion data of the tumor in the patient, the image fusion unit is used to fuse the collected motion data based on particle filtering and predict the motion data of the tumor in the future, and the treatment planning unit uses Adjust the treatment beam in real time for precise irradiation.

As a further improvement of the present invention, the system also includes a patient fixation and automatic positioning unit and a robot projection unit. The patient fixation and automatic positioning unit is used to fix the patient and perform positioning. The robot projection unit is used to control the treatment beam to reach the target. Area.

As a further improvement of the present invention, the breathing tracking unit is an infrared tracking unit, which includes a synchronous tracking camera and several light-emitting diodes fixed on the abdomen of the patient.

As a further improvement of the present invention, the image positioning unit is an X-ray image positioning unit, which includes an X-ray source and an amorphous silicon image receiver.

As a further improvement of the present invention, the system further includes a respiratory motion simulator for simulating the relationship of motion changes between tumors in the body and marker points on the body surface during radiotherapy.

The technical scheme provided by another embodiment of the present invention is as follows:

A method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering, the method comprising:

S1. Use the breath tracking unit and the image positioning unit to collect the three-dimensional motion data of body surface markers and tumors in the body respectively;

S2. According to the three-dimensional motion data, establish the motion relationship model between the tumor at the current moment and the tumor at the historical moment, and use this model as the state transition equation of the particle filter to establish the motion relationship between the body surface markers and the tumor in the body within a period of time The model is used as the observation equation of the particle filter;

S3. Based on the state transition equation and the observation equation, the particle filter algorithm is used to estimate the movement position of the tumor in the body according to the movement data of the body surface marker points at the current moment.

As a further improvement of the present invention, the step S3 includes:

The motion position of the tumor is estimated by the particle filter, the state transfer equation is established through the kinematic process of the tumor, and the observation equation is obtained by fitting the motion relationship between the tumor in the body and the marker points on the body surface.

As a further improvement of the present invention, the state transition equation in the step S2 is x _k =f(x _k-1 ), and the observation equation is z _k =h(x _k ), where x _k is the movement of the tumor in the body at time k data, z _k is the motion data of body surface marker points measured at time k.

As a further improvement of the present invention, in the step S3, the particle filter is used to estimate the motion position of the tumor specifically as follows:

S31. Generate a set of N random sample points, that is, N particles, according to the tumor position data and its probability distribution at the previous moment;

S32. Substituting the sampled particles into the state transition equation to obtain the estimated value x _k of the position of N tumors at the next moment;

S33. Substituting the estimated values of the N tumor positions into the observation equation to obtain the measured estimated value z _k of the body surface marker points, and using the radiotherapy robot to obtain the actual measured value z of the marked points;

S34. Assign different weights to particles according to the degree of error between z _k and z, assign larger weights to particles with small errors, and assign smaller weights to particles with larger errors, and resample the particles , after several times of resampling, use the final sampled particles and their weight calculations to obtain the best estimate of the tumor location;

S35 , using the finally sampled particles as a group of random samples for the next round of prediction, repeating S32 to S34 , and performing the next round of particle filtering.

As a further improvement of the present invention, in the step S31, the initial states of the N particles are evenly distributed in the entire state space.

The beneficial effects of the present invention are:

It solves the constraint that the system must satisfy the Gaussian distribution. The particle filter uses the particle set to represent the probability and can be used in any form of state space model. It has stronger modeling ability for the nonlinear characteristics of variable parameters and higher prediction accuracy. ;

Using the respiratory motion simulator as an experimental platform to study the synchronous respiration tracking algorithm avoids the human body being exposed to X-rays for a long time, and can effectively integrate real-time image guidance and synchronous respiration tracking technology in actual treatment to ensure that the lesion receives the maximum radiation dose at the same time Avoid damage to surrounding normal tissues and reduce the incidence of complications.

Description of drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings that need to be used in the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are only These are some embodiments described in the present invention. Those skilled in the art can also obtain other drawings based on these drawings without creative work.

1 is a block diagram of a particle filter-based radiotherapy robot tumor motion estimation and prediction system in the present invention;

FIG. 2 is a schematic flow diagram of a method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering in the present invention;

Figures 3a-3c are comparison results of experimentally verified prediction errors between the particle filter-based tumor motion estimation and prediction method for radiotherapy robots and the linear estimation method.

detailed description

In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described The embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

As shown in FIG. 1 , an embodiment of the present invention discloses a particle filter-based radiotherapy robot tumor motion estimation and prediction system, which includes a breathing tracking unit 10 , an image positioning unit 20 , an image fusion unit 30 , and a treatment planning unit 40 . Among them, the breathing tracking unit 10 is used to track the movement data of the marker points on the patient's body surface, the image positioning unit 20 is used to track the movement data of the tumor in the patient's body, and the image fusion unit 30 is used to perform particle filter-based fusion and fusion on the collected movement data. Predicting the future movement data of the tumor, the treatment planning unit 40 is used to adjust the treatment beam in real time for precise irradiation.

Further, the system in this embodiment also includes a patient fixation and automatic positioning unit and a robot projection unit, the patient fixation and automatic positioning unit is used to fix the patient and perform positioning, and the robot projection unit is used to control the treatment beam to reach the target area .

Preferably, the breathing tracking unit 10 in this embodiment is an infrared tracking unit, which includes a synchronous tracking camera and several light-emitting diodes fixed on the abdomen of the patient. The synchronous tracking camera and three light-emitting diodes fixed on the patient's abdomen form an external respiratory tracking unit, which is used to track the movement data of body surface markers.

Preferably, the image positioning unit 20 in this embodiment is an X-ray image positioning unit, which includes an X-ray source and an amorphous silicon image receiver, and is used for tracking movement data of tumors in the body.

A correlation model between two sets of motion data is established, and the particle filter algorithm is used to predict the future position of the tumor. The image fusion unit 30 and the treatment planning unit 40 adjust the treatment beam in real time according to the predicted tumor data for precise irradiation.

Specifically, during the treatment process, the patient lies on the operating table with the parallel structure of the radiotherapy robot, the synchronous tracking camera in the respiratory tracking unit is used to track the motion data of the body surface markers, and the X-ray source and amorphous silicon image receiver are used to track the movement data of the body surface markers. For the motion data of the tumor, by establishing a correlation model between the marker points and the tumor, the particle filter algorithm is used to predict the motion data of the tumor at the future moment, and the treatment planning unit and the image fusion unit adjust the radiotherapy robot to drive the linear accelerator according to the obtained tumor position prediction data. Move to the part that needs to be treated for irradiation, so as to realize the real-time tracking and treatment of the tumor by the treatment beam.

Since continuous irradiation is harmful to the human body, in order to avoid patients being exposed to X-rays for a long time, the movement data of the tumor in the body is collected intermittently. In order to obtain continuous breathing movement data, the breathing tracking unit can be replaced by a breathing movement simulator The patient is used to simulate the movement change relationship between the tumor in the body and the body surface markers during radiotherapy. The simulator can not only reproduce the real breathing movement, but also measure the continuous simulated breathing movement data in real time with the NDI Polaris optical positioning system, and can also Use the data it provides to verify the effectiveness of the forecasting algorithm. For example, a respiratory motion simulator disclosed in the patent No. ZL201420836958.X is used to simulate the patient's respiratory motion. The respiratory motion simulator consists of a three-degree-of-freedom linear slide, three stepper motors, a simulated tumor and two Composed of springs, it is mainly used to simulate the movement change relationship between the tumor in the body and the body surface markers in radiotherapy. The simulator can not only reproduce the real human breathing movement, but also use the data provided to verify the effectiveness of the prediction algorithm. Among them, the motion data of simulated markers and simulated tumors can be measured in real time by NDI Polaris optical positioning system.

Some prediction algorithms currently studied include artificial neural network, Kalman filter, fuzzy control, etc., but these algorithms have more or less deficiencies for complex nonlinear respiratory motion processes. The particle filter solves the constraint that the system must satisfy the Gaussian distribution, is suitable for any nonlinear system that can be expressed by a state space model, and has high precision, and is suitable for the prediction of the position of the tumor in respiratory motion.

Referring to Fig. 2, another embodiment of the present invention discloses a method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering, which includes:

S1. Using the breath tracking unit and the image positioning unit to collect the three-dimensional motion data of body surface markers and tumors in the body respectively;

S2. According to the three-dimensional motion data, establish the motion relationship model between the tumor at the current moment and the tumor at the historical moment, and use this model as the state transition equation of the particle filter to establish the motion relationship between the body surface markers and the tumor in the body within a period of time The model is used as the observation equation of the particle filter;

S3. Based on the state transition equation and the observation equation, the particle filter algorithm is used to estimate the movement position of the tumor in the body according to the movement data of the body surface marker points at the current moment.

Preferably, in step S3 of this embodiment, the particle filter is used to estimate the motion position of the tumor, the state transition equation is established through the kinematic process of the tumor, and the observation equation is obtained by fitting the motion relationship between the tumor in the body and the marker points on the body surface.

Specifically, the state transition equation in step S2 is x _k =f(x _k-1 ), and the observation equation is z _k =h(x _k ), where x _k is the movement data of the tumor in the body at time k, and z _k is k The motion data of body surface marker points measured at all times.

The motion position of the tumor is estimated by the particle filter as follows:

S31. Generate a set of random sample points or particles according to the tumor position data x _k-1 and x _k-2 at the previous two moments and their probability distribution. Assuming that there are N particles, since the initial state is unknown, it is considered that x ₀ and x ₁ is evenly distributed in the entire state space;

S32. Substituting the sampled particles into the state transition equation x _k = f(x _k-1 ) to obtain the estimated value x _k of the position of N tumors at the next moment;

S33. Substituting the estimated value x _k of N tumor positions into the observation equation z _k =h(x _k ) to obtain the measured estimated value z _k of the body surface marker points, and the actual measured value z of the marked points can be obtained by using the radiotherapy robot system;

S34. Assign different weights to particles according to the degree of error between z _k and z, assign larger weights to particles with small errors, and assign smaller weights to particles with larger errors, and resample the particles , after several times of resampling, use the final sampled particles and their weight calculations to obtain the best estimate of the tumor location;

S35 , using the finally sampled particle points as a group of random samples for the next round of prediction, repeating S32 to S34 , and performing the next round of particle filtering.

In order to verify the effectiveness of the particle filter-based radiotherapy robot tumor motion estimation and prediction system and method, the experimental verification prediction error of the particle filter-based radiotherapy robot tumor motion estimation and prediction method was compared with the linear estimation method. Import 2,000 sets of real respiratory motion data into the respiratory motion simulator for experiments, and apply the collected simulated respiratory motion data to two prediction algorithms. Experimental verification of the prediction error result graph, using the clinically applied linear estimation method and the particle filter algorithm to predict the movement of the tumor. The graph shows the difference between the two prediction algorithms' respiratory cycle prediction errors, that is, the error data of the linear estimation method minus the The result obtained from the error data of the filtering algorithm, so the data greater than zero indicates that the error based on the particle filter algorithm is small, and if the data is less than zero, it indicates that the error of the traditional linear estimation method is small. It can be seen from Figures 3a to 3c that during the entire respiratory movement process, the proportion of data above zero is relatively large, indicating that the particle filter is used to predict tumor movement with high prediction accuracy.

As can be seen from the above technical solutions, compared with the prior art, the present invention has the following beneficial effects:

It solves the constraint that the system must satisfy the Gaussian distribution. The particle filter uses the particle set to represent the probability, which can be used in any form of state space model. It has stronger modeling ability for the nonlinear characteristics of variable parameters and higher prediction accuracy. ;

Using the respiratory motion simulator as an experimental platform to study the synchronous respiration tracking algorithm avoids the human body being exposed to X-rays for a long time, and can effectively integrate real-time image guidance and synchronous respiration tracking technology in actual treatment to ensure that the lesion receives the maximum radiation dose at the same time Avoid damage to surrounding normal tissues and reduce the incidence of complications.

The invention can be applied to the treatment of lung tumors, which is of great significance for improving the survival rate of patients and promoting the development of medical technology in my country.

It will be apparent to those skilled in the art that the invention is not limited to the details of the above-described exemplary embodiments, but that the invention can be embodied in other specific forms without departing from the spirit or essential characteristics of the invention. Accordingly, the embodiments should be regarded in all points of view as exemplary and not restrictive, the scope of the invention being defined by the appended claims rather than the foregoing description, and it is therefore intended that the scope of the invention be defined by the appended claims rather than by the foregoing description. All changes within the meaning and range of equivalents of the elements are embraced in the present invention. Any reference sign in a claim should not be construed as limiting the claim concerned.

In addition, it should be understood that although this specification is described according to implementation modes, not each implementation mode only contains an independent technical solution, and this description in the specification is only for clarity, and those skilled in the art should take the specification as a whole , the technical solutions in the various embodiments can also be properly combined to form other implementations that can be understood by those skilled in the art.

Claims (10)

1. A radiotherapy robot tumor motion estimation and prediction system based on particle filtering, characterized in that the system includes: a breathing tracking unit, an image positioning unit, an image fusion unit and a treatment planning unit, and the breathing tracking unit is used to track patients motion data of body surface markers, the image positioning unit is used to track the motion data of the tumor in the patient, and the image fusion unit is used to fuse the collected motion data based on particle filtering and predict the motion data of the tumor at a future moment, The treatment planning unit is used for real-time adjustment of treatment beams for precise irradiation. 2. The radiotherapy robot tumor motion estimation and prediction system based on particle filter according to claim 1, characterized in that, the system also includes a patient fixation and automatic positioning unit and a robot projection unit, and the patient fixation and automatic positioning unit is used for To immobilize and position the patient, the robotic projection unit is used to control the treatment beam to reach the target area. 3. The particle filter-based radiation therapy robot tumor motion estimation and prediction system according to claim 1, wherein the breathing tracking unit is an infrared tracking unit, which includes a synchronous tracking camera and several light-emitting diodes fixed on the patient's abdomen. 4. The radiotherapy robot tumor motion estimation and prediction system based on particle filtering according to claim 1, wherein the image positioning unit is an X-ray image positioning unit, which includes an X-ray source and an amorphous silicon image receiver . 5. The tumor motion estimation and prediction system for radiotherapy robots based on particle filtering according to claim 1, wherein the system also includes a respiratory motion simulator for simulating the distance between tumors in vivo and body surface markers in radiotherapy. Movement change relationship. 6. A method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering, characterized in that the method comprises: S1. Use the breath tracking unit and the image positioning unit to collect the three-dimensional motion data of body surface markers and tumors in the body respectively; S2. According to the three-dimensional motion data, establish the motion relationship model between the tumor at the current moment and the tumor at the historical moment, and use this model as the state transition equation of the particle filter to establish the motion relationship between the body surface markers and the tumor in the body within a period of time The model is used as the observation equation of the particle filter; S3. Based on the state transition equation and the observation equation, the particle filter algorithm is used to estimate the movement position of the tumor in the body according to the movement data of the body surface marker points at the current moment. 7. The particle filter-based radiotherapy robot tumor motion estimation and prediction method according to claim 6, characterized in that the step S3 includes: The motion position of the tumor is estimated by the particle filter, the state transfer equation is established through the kinematic process of the tumor, and the observation equation is obtained by fitting the motion relationship between the tumor in the body and the marker points on the body surface. 8. The particle filter-based radiotherapy robot tumor motion estimation and prediction method according to claim 7, characterized in that, in the step S2, the state transition equation is x _k =f(x _k-1 ), and the observation equation is z _k =h(x _k ), where x _k is the motion data of the tumor in the body at time k, and z _k is the motion data of body surface marker points measured at time k. 9. The method for estimating and predicting tumor motion of a radiotherapy robot based on particle filter according to claim 8, characterized in that, in the step S3, estimating the motion position of the tumor through the particle filter is specifically: S31. Generate a set of N random sample points, that is, N particles, according to the tumor position data and its probability distribution at the previous moment; S32. Substituting the sampled particles into the state transition equation to obtain the estimated value x _k of the position of N tumors at the next moment; S33. Substituting the estimated values of the N tumor positions into the observation equation to obtain the measured estimated value z _k of the body surface marker points, and using the radiotherapy robot to obtain the actual measured value z of the marked points; S34. Assign different weights to particles according to the degree of error between z _k and z, assign larger weights to particles with small errors, and assign smaller weights to particles with larger errors, and resample the particles , after several times of resampling, use the final sampled particles and their weight calculations to obtain the best estimate of the tumor location; S35 , using the finally sampled particles as a group of random samples for the next round of prediction, repeating S32 to S34 , and performing the next round of particle filtering. 10. The method for estimating and predicting tumor motion of a radiotherapy robot based on particle filtering according to claim 9, characterized in that, in the step S31, the initial states of the N particles are evenly distributed in the entire state space.