Performance test method for radiotherapy laser positioning system

Abstract

The invention relates to the technical field of optical equipment testing, in particular to a radiation therapy laser positioning system performance testing method, which comprises the following steps of: testing first positioning data of optical positioning equipment in a load state, constructing a body weight-optical response correlation model, and analyzing the body weight-optical response correlation model to obtain static optical equipment testing data; a tester slightly moves along a preset track, a response signal of the laser photosensitive sensor and space coordinates of the laser tracker are collected to generate second optical positioning data, and a motion-optical response coupling model is constructed to analyze the second optical positioning data to obtain dynamic optical equipment test data; a multi-factor deviation prediction model is constructed to analyze optical interference signals, phase detection data and environment temperature and humidity parameters to generate optical equipment compensation parameters, and static optical equipment test data, dynamic optical equipment test data and the optical equipment compensation parameters are analyzed; and calculating and generating an optical calibration scene fitness score, a positioning precision quantization matrix and an optical equipment acceptance reference value comprehensive evaluation report.

Description

Technical Field

[0001] The present invention relates to the technical field of optical equipment testing, and particularly to a performance testing method for a radiotherapy laser positioning system. Background Technique

[0002] In the clinical performance testing of a laser positioning system, the ability of the system to maintain optical characteristics under dynamic conditions directly affects the positioning accuracy of medical equipment. Currently, the optical property testing methods mainly focus on the static working state, and the system performance is evaluated by analyzing the geometric parameters and intensity distribution of optical markers. However, in the actual clinical environment, dynamic movements such as the respiratory rhythm and micro-position changes of the subject will cause time-varying disturbances in the optical transmission path. Such dynamic interferences will cause changes in the transmission characteristics of the optical system, including but not limited to problems such as light intensity distribution distortion, marker projection offset, and deterioration of optical imaging quality. Existing testing means are still unable to effectively simulate and quantify the impact of dynamic interferences on the optical performance of the system.

[0003] In the current optical testing standards, there is no established testing method for dynamic working conditions, especially the lack of quantitative detection methods for key performance indicators such as light transmission stability and dynamic distortion suppression ability. Therefore, there is an urgent need to construct a testing platform with multi-degree-of-freedom dynamic simulation function, which should integrate a programmable control module for motion parameters and a synchronous acquisition system for optical characteristics to realize the analysis of the transmission characteristics of the optical system, the detection of real-time imaging resolution, and the evaluation of marker projection stability under dynamic conditions.

[0004] Therefore, a performance testing method for a radiotherapy laser positioning system is proposed. Summary of the Invention

[0005] The purpose of the present invention is to provide a performance testing method for a radiotherapy laser positioning system. By testing the first positioning data of the optical positioning device under a load state, constructing a body weight-optical response correlation model to analyze the first positioning data, and obtaining static optical device test data; obtaining the second optical positioning data under the limb micro-motion trajectory, constructing a motion-optical response coupling model to analyze the second optical positioning data, and obtaining dynamic optical device test data; constructing a multi-factor deviation prediction model to analyze the optical interference signal, phase detection data, and environmental temperature and humidity parameters, generating optical device compensation parameters; analyzing the static optical device test data, dynamic optical device test data, and optical device compensation parameters, calculating the interaction coefficient of the optical device impact factor, and generating a comprehensive evaluation report including the optical calibration scene fitness score, positioning accuracy quantization matrix, and optical device acceptance reference value.

[0006] To achieve the above purpose, the present invention provides the following technical solutions:

[0007] A performance testing method for a radiotherapy laser positioning system, including:

[0008] The testers are divided into different groups according to their weights, and the first positioning data of the optical positioning device under load conditions is tested. A weight-optical response correlation model is constructed to analyze the first positioning data, and static optical device test data is obtained;

[0009] The testers are controlled to move slightly along a preset trajectory, and the photosensitive sensor response signals of the laser beam and the spatial coordinates of the laser tracker are synchronously collected to generate second optical positioning data. A motion-optical response coupling model is constructed to analyze the second optical positioning data, and dynamic optical device test data is obtained;

[0010] A multi-factor deviation prediction model is constructed to analyze the optical interference signal, phase detection data of the laser beam, and environmental temperature and humidity parameters collected synchronously, and optical device compensation parameters are generated;

[0011] The static optical device test data, dynamic optical device test data, and optical device compensation parameters are analyzed, the interaction coefficients of the optical device influence factors are calculated, and a comprehensive evaluation report of the optical calibration scene fitness score, positioning accuracy quantization matrix, and optical device acceptance reference value is generated.

[0012] Preferably, the first positioning data is static position information data collected under different load conditions, including coordinate data of the laser beam center position, laser beam slope data, and the deviation distance between the laser beam and the preset theoretical coordinate system;

[0013] The second optical positioning data is dynamic position information data continuously collected under the micro-motion trajectory of the tester's limbs, and also includes motion trajectory, displacement change, speed, acceleration data, and continuous optical signal change data corresponding to the dynamic state of the tester and its time stamp; the range of limb micro-motion is within the set threshold of the optical positioning device.

[0014] Preferably, the weight-optical response correlation model includes a first positioning data processing layer, a correlation feature extraction layer, a correlation feature analysis layer, and a result verification layer;

[0015] The first positioning data processing layer performs preliminary processing on the first positioning data, including filtering, denoising, and normalization operations;

[0016] The correlation feature extraction layer performs correlation analysis on the processed first positioning data and weight data to generate weight-optical response correlation features;

[0017] The correlation feature analysis layer analyzes the weight-optical response correlation features through statistical methods to obtain parameter data of the influence of weight change on the static performance of the optical device, including positioning error, response sensitivity, and stability index;

[0018] The result verification layer obtains the test data of the static optical device by performing statistical verification and error analysis on the parameter data.

[0019] Preferably, the motion-optical response coupling model includes a second optical positioning data processing layer, a dynamic feature extraction layer, a feature coupling layer, and an analysis and prediction layer;

[0020] The second optical positioning data processing layer preprocesses the continuously collected second optical positioning data, including filtering, denoising, and normalization operations;

[0021] The dynamic feature extraction layer processes the dynamic position information data of the limb micro-movement and the optical signal change data, and extracts the motion feature parameters and the optical response feature parameters;

[0022] The feature coupling layer performs coupling analysis on the motion feature parameters and the optical response feature parameters through machine learning to generate the joint motion and optical response features;

[0023] The analysis and prediction layer generates the test data of the dynamic optical device by analyzing the joint motion and optical response features.

[0024] Preferably, the multi-factor deviation prediction model includes a feature extraction layer, a parameter coupling and analysis layer, and a prediction and compensation layer;

[0025] The feature extraction layer extracts features from the optical interference signal of the laser beam, the phase detection data, and the environmental temperature and humidity parameters to obtain the optical interference features, the phase detection features, and the environmental features; the optical interference features include the characteristics of the interference fringes, the phase information, and the signal intensity; the phase detection features include the phase offset, the change trend, and the relevant correction information; the environmental features include the temperature change rate and the humidity fluctuation;

[0026] The feature coupling and analysis layer couples the optical interference features, the phase detection features, and the environmental features to obtain the optical and environmental coupling features; analyzes the optical and environmental coupling features to obtain the deviation amount generated by the optical device due to environmental and signal fluctuations during actual use, and obtains the key error indicators, including phase error, signal drift, and positioning deviation caused by the environment;

[0027] The prediction and compensation layer calculates and generates the compensation parameters of the optical device based on the key error indicators; the compensation parameters of the optical device include the deviation correction coefficient and the temperature and humidity compensation coefficient.

[0028] Preferably, the specific steps for calculating the interaction coefficient of the influencing factor of the optical device include the following:

[0029] Perform feature alignment and standardization processing on the test data of the static optical device, the test data of the dynamic optical device, and the optical device compensation parameters to generate a multi-dimensional fusion data set;

[0030] Construct a multi-factor coupling matrix, and calculate the weight coefficients of each parameter in the multi-dimensional fusion data set through the partial least squares regression algorithm;

[0031] Adopt the particle swarm optimization algorithm to dynamically adjust the weight coefficients, and calculate the non-linear interaction effects between factors by setting a fitness function;

[0032] Conduct a significance test on the optimized weight coefficients to generate a composite interaction coefficient set including linear coupling coefficients and non-linear interference coefficients;

[0033] Verify the composite interaction coefficient set through the holdout validation method. When the root mean square error is lower than the preset threshold, it is determined that the interaction coefficient is valid.

[0034] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0035] 1. The present invention groups testers by weight, conducts systematic preprocessing on the static laser positioning data under different load states, and uses methods such as filtering, denoising, and normalization to eliminate noise interference, thereby accurately extracting core parameters such as the center coordinates, slope, and deviation distance of the laser beam. Construct a correlation model between weight and optical response, conduct statistical analysis and result verification on the positioning error, response sensitivity, and stability indicators, realize refined data processing, and significantly improve the test accuracy and reliability.

[0036] 2. The present invention introduces the micro-motion trajectory data of testers in dynamic testing, continuously collects displacement, velocity, and acceleration information during the movement process, and combines the optical signal time stamps to realize the coupled analysis of movement and optical response. After preprocessing with filtering, denoising, and normalization, through dynamic feature extraction and joint analysis, it effectively captures the change law of laser positioning under the movement state, and quantitatively reflects the system response delay and fluctuation. This method reveals the error generation mechanism under the movement state, provides a scientific basis for the fine testing of laser beam parameters, and improves the dynamic adaptability and practicality of the test results.

[0037] 3. The present invention constructs a multi-factor deviation prediction model, combines the optical interference signal of the laser beam, the phase detection data, and the environmental temperature and humidity parameters to accurately extract the key error indicators. Using the hierarchical structure of feature extraction, parameter coupling, and prediction compensation, calculate the deviation correction coefficient and the temperature and humidity compensation coefficient, and effectively compensate for the positioning deviation caused by environmental changes and signal drift. Fully considering complex environmental factors, significantly reduce error interference, and improve the positioning test accuracy and stability of the device.

[0038] 4. The present invention constructs a multi-dimensional fusion data set by performing feature alignment and standardization processing on static test data, dynamic test data, and optical device compensation parameters, and calculates the weights of each parameter and the non-linear interaction effect based on the partial least squares regression and particle swarm optimization algorithm to generate a composite interaction coefficient set including linear coupling coefficients and non-linear interference coefficients. This method ensures the accuracy and scientific nature of the evaluation model through significance testing and the holdout validation method, and finally forms a comprehensive evaluation report of the scene fitness score, the quality control index quantization matrix, and the acceptance benchmark value, providing a multi-dimensional and systematic decision-making basis for equipment quality control, maintenance optimization, and testing, and significantly improving the overall detection level and data credibility. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 is a schematic flow chart of a method for testing the performance of a radiotherapy laser positioning system provided by the present invention;

[0040] Figure 2 is a schematic structural diagram of a weight-optical response correlation model provided by an embodiment of the present invention;

[0041] Figure 3 is a schematic structural diagram of a motion-optical response coupling model provided by an embodiment of the present invention;

[0042] Figure 4 is a schematic structural diagram of a multi-factor deviation prediction model provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0043] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0044] Embodiment 1:

[0045] Please refer to Figure 1 , the present invention provides a method for testing the performance of a radiotherapy laser positioning system, and the technical solution is as follows:

[0046] The testers are divided into different groups according to their weights, and the first positioning data of the optical positioning device under the load state is tested. A weight-optical response correlation model is constructed to analyze the first positioning data to obtain static optical device test data; the first positioning data is static position information data collected under different load states, including coordinate data of the center position of the laser beam, laser beam slope data, and the deviation distance between the laser beam and the preset theoretical coordinate system;

[0047] In this embodiment, by dividing the test persons into different groups according to their weight and collecting static positioning data of the laser positioning system under different load conditions, the different load effects of the human body on the equipment in actual usage scenarios are effectively simulated.

[0048] The weight-optical response association model includes a first positioning data processing layer, an association feature extraction layer, an association feature analysis layer and a result verification layer. Figure 2 ;

[0049] The first positioning data processing layer performs preliminary processing on the first positioning data, including filtering, denoising and normalization operations;

[0050] The correlation feature extraction layer performs correlation analysis on the processed first positioning data and the weight data to generate a weight-optical response correlation feature;

[0051] The correlation feature analysis layer analyzes the correlation features between the weight and the optical response by using a statistical method to obtain parameter data of the effect of weight change on the static performance of the optical device, including positioning error, response sensitivity and stability index;

[0052] The result verification layer obtains static optical device test data by performing statistical verification and error analysis on parameter data.

[0053] In this embodiment, by constructing a body weight-optical response correlation model and quantitatively analyzing the laser beam center position, slope, and deviation distance from the preset theoretical coordinate system, the effect of weight change on optical positioning performance can be accurately revealed, thereby providing accurate static test data. This method can not only discover potential positioning errors and response instability problems, but also provide a scientific basis for equipment calibration, quality control, and subsequent performance optimization, ultimately helping to improve the reliability and accuracy of optical positioning equipment during radiotherapy, and ensure the stability of treatment effects and patient safety.

[0054] The test person is controlled to slightly move along a preset trajectory, and the response signal of the photosensor of the laser beam and the spatial coordinates of the laser tracker are synchronously collected to generate the second optical positioning data. A motion-optical response coupling model is constructed to analyze the second optical positioning data to obtain the dynamic optical device test data. The second optical positioning data is the dynamic position information data continuously collected under the trajectory of the test person's limb micro-movement, including motion trajectory, displacement change, speed and acceleration data; continuous optical signal change data corresponding to the dynamic state of the test person and its timestamp; the range of the limb micro-movement is within the threshold set by the optical positioning device.

[0055] The motion-optical response coupling model includes a second optical positioning data processing layer, a dynamic feature extraction layer, a feature coupling layer and an analysis and prediction layer. Figure 3 ;

[0056] The second optical positioning data processing layer preprocesses the continuously collected second optical positioning data, including filtering, denoising, and normalization operations;

[0057] The dynamic feature extraction layer processes the dynamic position information data of limb micro-movements and the optical signal change data, and extracts motion feature parameters and optical response feature parameters;

[0058] The feature coupling layer performs coupling analysis on the motion feature parameters and optical response feature parameters through machine learning to generate joint motion and optical response features; the analysis and prediction layer generates dynamic optical device test data by analyzing the joint motion and optical response features.

[0059] In this embodiment, by acquiring the dynamic optical positioning data continuously collected by the tester under the limb micro-movement trajectory and constructing a motion-optical response coupling model for in-depth analysis, this method can truly reflect the response characteristics of the optical positioning device in the dynamic state. Using multi-level processing processes such as data processing, dynamic feature extraction, and machine learning coupling analysis, motion feature parameters and optical response feature parameters can be accurately extracted to generate high-precision dynamic optical device test data. This method significantly improves the detection accuracy of the device's dynamic performance, helps to timely discover the positioning deviation caused by human body micro-movements, and thus provides a scientific basis for device testing and optimization. Dynamic testing relies on manual operation and lacks a synchronization mechanism. The present invention realizes millisecond-level dynamic error capture through high-frequency acquisition and the LSTM model. For details, refer to Table 1.

[0060] Table 1 Comparison table of dynamic test efficiency

[0061] Index Motion-optical response coupling model Manual trajectory recording Data acquisition frequency 1000Hz (real-time synchronous optical signal) 100Hz (manual marking) Trajectory analysis time <5ms (machine learning dimensionality reduction) >50ms (manual fitting) Dynamic positioning error ±0.3mm (LSTM prediction) ±1.2mm (linear interpolation) Motion-optical delay analysis Support timestamp matching (±1ms accuracy) No synchronization mechanism

[0062] Construct a multi-factor deviation prediction model to analyze the optical interference signal, phase detection data, and environmental temperature and humidity parameters of the synchronously collected laser beam, and generate optical device compensation parameters;

[0063] The multi-factor deviation prediction model includes a feature extraction layer, a parameter coupling and analysis layer, and a prediction and compensation layer. For details, refer to Figure 4 ;

[0064] The feature extraction layer extracts features from the optical interference signal, phase detection data, and environmental temperature and humidity parameters of the laser beam to obtain optical interference features, phase detection features, and environmental features; the optical interference features include interference fringe characteristics, phase information, and signal intensity; the phase detection features include phase offset, change trend, and correlation correction information; the environmental features include temperature change rate and humidity fluctuation;

[0065] The feature coupling and analysis layer couples optical interference features, phase detection features, and environmental features to obtain optical and environmental coupling features; analyzes the optical and environmental coupling features to obtain the deviation amount generated by the optical device due to environmental and signal fluctuations during actual use, and obtains key error indicators, including phase error, signal drift, and positioning deviation caused by the environment.

[0066] The prediction and compensation layer calculates and generates optical device compensation parameters based on the key error indicators; the optical device compensation parameters include deviation correction coefficients and temperature and humidity compensation coefficients.

[0067] In this embodiment, after adopting this multi-factor deviation prediction model, various errors caused by environmental fluctuations and signal instability can be accurately identified and quantified before the device runs, so as to generate compensation parameters in advance and achieve automatic correction of the deviation of the optical device. Through in-depth feature extraction and coupling analysis of the optical interference signal of the laser beam, phase detection data, and environmental temperature and humidity parameters, key error indicators such as phase error, signal drift, and positioning deviation caused by the environment can be accurately obtained. This method significantly improves the positioning test accuracy and stability of the optical device in a complex environment.

[0068] Analyze the static optical device test data, dynamic optical device test data, and optical device compensation parameters, calculate the interaction coefficients of the optical device influence factors, and generate a comprehensive evaluation report including the optical calibration scene fitness score, positioning accuracy quantization matrix, and optical device acceptance reference value.

[0069] The specific steps for calculating the interaction coefficients of the optical device influence factors are as follows:

[0070] Perform feature alignment and standardization processing on the static optical device test data, dynamic optical device test data, and optical device compensation parameters to generate a multi-dimensional fusion data set;

[0071] Construct a multi-factor coupling matrix, and calculate the weight coefficients of each parameter in the multi-dimensional fusion data set through the partial least squares regression algorithm;

[0072] Adopt the particle swarm optimization algorithm to dynamically adjust the weight coefficients, and calculate the non-linear interaction effects between factors by setting a fitness function;

[0073] Conduct a significance test on the optimized weight coefficients to generate a composite interaction coefficient set including linear coupling coefficients and non-linear interference coefficients; verify the composite interaction coefficient set through the holdout validation method, and when the root mean square error is lower than the preset threshold, determine that the interaction coefficients are valid.

[0074] In this embodiment, through the deep fusion and comprehensive analysis of the test data and compensation parameters of static and dynamic optical devices, the feature alignment and standardization processing of multi-dimensional data are realized, a multi-factor coupling matrix is constructed, and the partial least squares regression and particle swarm optimization algorithms are used to dynamically adjust the weight coefficients, and the composite interaction coefficient reflecting the linear and non-linear interaction effects between various influencing factors of the device is accurately calculated. After significance testing and holdout validation, the generated evaluation report not only includes the optical calibration scene fitness score, quality control index quantization matrix, and acceptance benchmark value, but also provides a scientific and quantitative basis for device calibration, performance optimization, and clinical application, thus significantly improving the detection accuracy and operation stability of the positioning system during radiotherapy.

[0075] The present invention realizes a comprehensive quantitative evaluation of the performance of an optical positioning device through the integration of static and dynamic test data, environmental compensation parameters, and multi-factor coupling analysis. First, the testers are grouped according to their body weight, and data such as the central position, slope, and deviation distance of the laser beam are collected under different loads, and a body weight-optical response correlation model is constructed to accurately reflect the influence of the human body load on the static performance of the device; secondly, by obtaining the dynamic data and optical signal changes under the micro-motion trajectory of the limb, the motion-optical response coupling model is used to extract the motion and response characteristics, realizing high-precision capture of the device performance in the dynamic state; in addition, a multi-factor deviation prediction model is used to extract key error indicators from laser interference, phase detection, and environmental parameters, and compensation parameters are generated in advance, so as to realize automatic correction of the device error; through data alignment, regression analysis, and particle swarm optimization of the static and dynamic test data and compensation parameters, the composite interaction coefficient reflecting the linear and non-linear interaction effects is calculated, and a comprehensive evaluation report of the scene fitness score, quality control index matrix, and acceptance benchmark value is generated, significantly improving the test accuracy and stability of the device positioning. Traditional linear regression cannot handle multi-factor non-linear relationships. The present invention balances complexity and precision through an optimization algorithm to achieve high-precision prediction. For details, refer to Table 2.

[0076] Table 2 Balance of algorithm complexity and precision

[0077] Algorithm type This invention (PLS + particle swarm optimization) Traditional method (ordinary least squares) Computational complexity <![CDATA[O(n 2 )(parallelizable optimization)]]> O(n) Prediction accuracy (RMSE) 0.12mm 0.35mm Nonlinear effect support Yes (interaction coefficient set) No Weight dynamic adjustment Particle swarm optimization iteration Fixed weight

[0078] Example Two:

[0079] The testers are divided into different groups according to their body weight, and the first positioning data of the optical positioning device in the loaded state is tested. A body weight-optical response correlation model is constructed to analyze the first positioning data to obtain the test data of the static optical device; the first positioning data is static position information data collected under different loaded states, including the coordinate data of the laser beam center position, the laser beam slope data, and the deviation distance between the laser beam and the preset theoretical coordinate system.

[0080] The weight-optical response correlation model includes a first positioning data processing layer, a correlation feature extraction layer, a correlation feature analysis layer, and a result verification layer;

[0081] The first positioning data processing layer preliminarily processes the first positioning data, including filtering, denoising, and normalization operations;

[0082] The correlation feature extraction layer performs correlation analysis on the processed first positioning data and weight data to generate weight-optical response correlation features;

[0083] Calculate the Pearson correlation coefficient r1 of the weight W and the positioning error △d through the correlation analysis method;

[0084]

[0085] where n is the number of testers, W i is the weight of the i-th tester, is the average weight, △d i is the positioning error of the i-th tester, is the average positioning error;

[0086] When r1 > 0.5, it is considered that there is a significant correlation.

[0087] Establish linear relationships for the features with significant correlations in different weight intervals respectively. The specific formula is as follows:

[0088] △d = β0 + β1W + β2W 2 + ε;

[0089] where β0, β1, and β2 are linear regression coefficients obtained through machine learning, and ε is an infinitesimal value.

[0090] Capture the non-linear relationship through polynomial regression and select the optimal model through the F-test to generate correlation features.

[0091] The correlation feature analysis layer analyzes the weight-optical response correlation features through statistical methods to obtain the parameter data of the influence of weight changes on the static performance of the optical device, including positioning error, response sensitivity, and stability index;

[0092] The response sensitivity is obtained through the error change rate S caused by the change in unit weight. The specific calculation formula is:

[0093]

[0094] The stability index is obtained through the variance of the errors within the same weight group.

[0095] The result verification layer obtains the test data of the static optical device by performing statistical verification and error analysis on the parameter data;

[0096] The statistical verification includes F-test and t-test; the error analysis includes systematic error and random error.

[0097] Control the tester to move slightly along the preset trajectory, synchronously collect the response signals of the photosensitive sensor of the laser beam and the spatial coordinates of the laser tracker, generate the second optical positioning data, construct a motion-optical response coupling model to analyze the second optical positioning data, and obtain the test data of the dynamic optical device; the second optical positioning data is the dynamic position information data continuously collected under the micro-motion trajectory of the tester's limb, including motion trajectory, displacement change, speed and acceleration data; the continuous optical signal change data corresponding to the dynamic state of the tester and its time stamp; the range of limb micro-motion is within the set threshold of the optical positioning device.

[0098] The motion-optical response coupling model includes a second optical positioning data processing layer, a dynamic feature extraction layer, a feature coupling layer, and an analysis and prediction layer; the second optical positioning data processing layer preprocesses the continuously collected second optical positioning data, including filtering, denoising, and normalization operations; the dynamic feature extraction layer processes the dynamic position information data and optical signal change data of the limb micro-motion, and extracts motion feature parameters and optical response feature parameters;

[0099] The motion feature parameters include the trajectory curvature κ obtained by analyzing the bending degree of the limb micro-motion trajectory, the displacement fluctuation amplitude A disp , average speed acceleration a v acceleration standard deviation σ a and acceleration peak ratio R peak ;

[0100] The optical response feature parameters include signal delay time T d , signal intensity fluctuation △S, phase change rate k φ , spectral coherence C(f) and dynamic range compression ratio η.

[0101] The feature coupling layer performs coupling analysis on the motion feature parameters and optical response feature parameters F optical =[T d ,△S,k φ ,C(f),η] through machine learning to generate joint features of motion and optical response;

[0102] By combining the motion feature parameters and optical response feature parameters F optical =[T d ,△S,kφ , C(f), η] are concatenated to obtain the combined feature F joint ;

[0103] For the combined feature F joint Dimensionality reduction is performed to generate the combined feature of motion and optical response F PCA , and the specific calculation formula is:

[0104] F PCA = w T F joint ;

[0105] Among them, w T is the transpose matrix of the PCA projection matrix;

[0106] The analysis and prediction layer analyzes the combined feature of motion and optical response to generate dynamic optical device test data;

[0107] The combined motion-optical feature is input into the trained LSTM model to output dynamic optical device test data; the dynamic optical device test data includes dynamic positioning error, response delay, signal distortion degree, and dynamic stability coefficient.

[0108] A multi-factor deviation prediction model is constructed to analyze the optical interference signal, phase detection data, and environmental temperature and humidity parameters of the synchronously collected laser beam to generate optical device compensation parameters;

[0109] The multi-factor deviation prediction model includes a feature extraction layer, a parameter coupling and analysis layer, and a prediction and compensation layer;

[0110] The feature extraction layer extracts features from the optical interference signal, phase detection data, and environmental temperature and humidity parameters of the laser beam to obtain optical interference features, phase detection features, and environmental features; the optical interference features include interference fringe characteristics, phase information, and signal intensity; the phase detection features include phase offset, change trend, and correlation correction information; the environmental features include temperature change rate and humidity fluctuation;

[0111] The feature coupling and analysis layer couples the optical interference features, phase detection features, and environmental features to obtain the optical and environmental coupling features; analyzes the optical and environmental coupling features to obtain the deviation amount generated by the optical device due to environmental and signal fluctuations during actual use, and obtains key error indicators, including phase error, signal drift, and positioning deviation caused by the environment;

[0112] Through the joint covariance matrix of the optical interference feature Fop, the phase detection feature Fph, and the environmental feature Fenv, the principal component direction is extracted by eigenvalue decomposition to determine the key coupling relationship;

[0113] The phase error is obtained from the difference between the actual phase and the theoretical phase; the Hodrick-Prescott filter is used to separate the signal for signal drift detection; the positioning deviation caused by the environment is determined through linear regression.

[0114] Based on the key error metrics, the prediction and compensation layer calculates and generates optical device compensation parameters; the optical device compensation parameters include a deviation correction coefficient and a temperature and humidity compensation coefficient;

[0115] The specific calculation formula for the deviation correction coefficient is:

[0116] φ compensation = φ original - (△φ T +△φ H );

[0117] where, φ 补偿 is the deviation correction compensation coefficient, φ 原始 is the original correction coefficient, △φ T is the temperature compensation coefficient, and △φ H is the humidity compensation coefficient;

[0118] Analyze the static optical device test data, dynamic optical device test data, and optical device compensation parameters, calculate the interaction coefficient of the optical device impact factor, and generate a comprehensive evaluation report including the optical calibration scene fitness score, positioning accuracy quantization matrix, and optical device acceptance benchmark value.

[0119] The specific steps for calculating the interaction coefficient of the optical device impact factor are as follows:

[0120] Perform feature alignment and standardization processing on the static optical device test data, dynamic optical device test data, and optical device compensation parameters to generate a multi-dimensional fusion dataset; feature alignment includes timestamp matching, data sampling rate synchronization, and coordinate system unification; the standardization processing uses the z-score algorithm to normalize the static positioning error, dynamic displacement variance, and compensation coefficient;

[0121] Construct a multi-factor coupling matrix, and calculate the weight coefficients of each parameter in the multi-dimensional fusion dataset through the partial least squares regression algorithm; the dimensions of the multi-factor coupling matrix include:

[0122] Static factor dimension: includes the mean value of the positioning error associated with body weight and the attenuation rate of response sensitivity; dynamic factor dimension: includes the maximum movement trajectory offset and the amplitude of phase jitter caused by acceleration; environmental compensation dimension: includes the temperature and humidity compensation coefficient and the phase drift correction amount;

[0123] The partial least squares regression algorithm includes a dual feature extraction process, specifically including: performing covariance analysis on the static factor dimension and the dynamic factor dimension to extract the principal component projection vector; taking the environmental compensation dimension as a moderating variable and calculating the moderating effect of the compensation factor on the principal component vector through the ridge regression algorithm; iteratively updating the weight coefficients until the coefficient of determination of the model.

[0124] The particle swarm optimization algorithm is used to dynamically adjust the weight coefficients, and the non-linear interaction effect between factors is calculated by setting the fitness function; a significance test is performed on the optimized weight coefficients to generate a composite interaction coefficient set including linear coupling coefficients and non-linear interference coefficients; the composite interaction coefficient set includes three-level influencing factors: primary interaction coefficient: the synergistic influence coefficient of body weight - motion acceleration on positioning accuracy; secondary interaction coefficient: the non-linear interference coefficient of environmental humidity - phase drift on the dynamic trajectory; tertiary interaction coefficient: the time-varying coupling coefficient of the compensation parameter and the static positioning error;

[0125] The composite interaction coefficient set is verified by the holdout validation method. When the root mean square error is lower than the preset threshold, the interaction coefficient is determined to be valid; the verification method is to substitute the composite interaction coefficient set into the laser positioning simulation model and obtain the root mean square error by predicting the positioning accuracy and the actual measurement value;

[0126] The laser positioning simulation model includes: a human tissue optical property database storing light scattering parameters corresponding to different BMI indices; a motion interference simulation module generating random micro-motion patterns based on a finite state machine; an environmental noise injection unit simulating electromagnetic interference and temperature and humidity fluctuations according to the IEEE 1159 standard.

[0127] Among them, the specific calculation formula for the optical calibration scene fitness score A is:

[0128] A = α1A1 + α2A2 + α3A3;

[0129] Among them, A1 is the score of the static optical device test data, α1 is the weight of the static optical device test data score, A2 is the score of the dynamic optical device test data, α2 is the weight of the dynamic optical device test data score, A3 is the score of the optical device compensation parameter, and α3 is the weight of the optical device compensation parameter score;

[0130] Among them, the static optical device test data score is obtained through the positioning error comprehensive factor P e 、response sensitivity R s and stability factor S t and the specific calculation formula is Among them, k1 is the adjustment parameter, and χ1, χ2, χ3 are the corresponding weights respectively; the positioning error comprehensive factor P eObtained by calculating the N - th positioning error through non - linear weighting, and the specific calculation formula is △x i is the positioning error of the i - th measurement, ω i is the corresponding body weight of the i - th measurement; the response sensitivity R s is obtained through the optical response function f(ω) and the standard deviation σ of the body weight ω and the specific calculation formula is ω is the body weight; the stability factor S t is obtained through the average error ε and the measurement period T, and the specific calculation formula is ε(t) is the error at time t.

[0131] The score of the dynamic optical device test data is obtained through the dynamic positioning error E d 、response delay L r 、signal distortion degree D s 、dynamic stability S d and motion complexity V m and the specific calculation formula is where μ1, μ2, μ3, μ4 are the corresponding weights respectively, and λ is the attenuation factor;

[0132] The dynamic positioning error E d is obtained through the predicted position and the actual position x(t), and the specific calculation formula is t d is the dynamic test duration; the response delay L r is obtained through the delay time t, the signal change amplitude △s i and the number of sampling points M, and the specific calculation formula is The signal distortion degree D s is obtained through the spectrum S(w) of the actual signal and the spectrum of the ideal signal w is the frequency, and the specific calculation formula is w0 is the initial frequency, w max is the maximum frequency; the dynamic stability S d is obtained through the trajectory velocity and the variance, and the specific calculation formula is The motion complexity V m is obtained through the acceleration and the specific calculation formula is

[0133] The score of the optical device compensation parameter is obtained through the temperature deviation factor η T 、humidity deviation factor η H and the compensation effect C e and the specific calculation formula is ν is the enhancement factor; the temperature deviation factor is obtained through the temperature influence function and the temperature deviation; the humidity deviation factor is obtained through the humidity influence function and the humidity deviation; the compensation effect is obtained through the difference between the compensated value and the actual output value;

[0134] The positioning accuracy quantization matrix is a 7-row and 6-column matrix, where:

[0135] Rows: represent the main factors affecting the positioning accuracy, covering static factors, dynamic factors, and environmental compensation factors;

[0136] Columns: represent the key indicators for evaluating the performance of optical devices;

[0137] Matrix elements: each element is a numerical value (usually in the range of 0 to 1), indicating the degree of influence of a specific influencing factor on a specific evaluation index; calculated by methods such as normalizing data, correlation coefficient, or regression analysis.

[0138] The rows of the matrix are the following main influencing factors:

[0139] Positioning error associated with weight (static factor): resulting from the linear or non-linear relationship between weight and positioning error in static tests;

[0140] Response sensitivity (static factor): calculated by the rate of change of error caused by a unit change in weight;

[0141] Stability (static factor): obtained through the variance of errors within the same weight group;

[0142] Movement trajectory deviation (dynamic factor): resulting from the displacement change of the limb micro-movement trajectory in dynamic tests;

[0143] Phase jitter caused by acceleration (dynamic factor): calculating the phase change through the acceleration data in dynamic tests;

[0144] Temperature and humidity compensation coefficient (environmental compensation factor): resulting from the environmental feature analysis in the multi-factor deviation prediction model;

[0145] Phase drift correction amount (environmental compensation factor): calculated through the deviation correction of the optical interference signal and the phase detection data;

[0146] The columns of the matrix are performance parameters, which are respectively:

[0147] Positioning error: the deviation distance between the center position of the laser beam and the theoretical coordinate in static and dynamic tests;

[0148] Response delay: the response time in static tests or the signal delay time in dynamic tests;

[0149] Signal distortion: the difference between the actual signal spectrum and the ideal signal spectrum in dynamic tests;

[0150] Dynamic stability: The stability coefficient of the trajectory speed and variance in dynamic testing;

[0151] Motion complexity: The degree of motion complexity reflected by the acceleration and trajectory curvature in dynamic testing;

[0152] Compensation effect: The degree of improvement of the positioning error after environmental compensation.

[0153] The numerical values of the matrix elements are quantified in the following way:

[0154] Static factor: The correlation analysis (such as Pearson correlation coefficient) or regression coefficient based on the weight-optical response correlation model;

[0155] Dynamic factor: Based on the joint feature analysis in the motion-optical response coupling model (such as the output of the LSTM model);

[0156] Environmental compensation factor: Based on the deviation correction coefficient and temperature and humidity compensation coefficient data of the multi-factor deviation prediction model, refer to Table 3.

[0157] Table 3 Positioning accuracy quantization matrix

[0158]

[0159] The comprehensive evaluation report of the acceptance benchmark value of the optical device includes the static test results, dynamic test results, environmental compensation parameters, comprehensive score and quality control index matrix, as well as the acceptance conclusion and suggestions.

[0160] Existing models ignore environmental interference. The present invention significantly improves the stability of the device in complex environments through deviation prediction and compensation parameter generation, specifically refer to Table 4.

[0161] Table 4 Environmental adaptability comparison table

[0162] Environmental conditions Temperature and humidity compensation model Uncompensated model Temperature fluctuation (±5°C) Positioning deviation ≤ 0.1mm Deviation fluctuation ±0.4mm Humidity change (30% - 80%) Signal drift suppression rate 95% Drift amount increase 120% Electromagnetic interference (IEEE1159) Error correction coefficient adaptive Not detected, deviation accumulation Extreme environment stability Acceptance score ≥ 85 points Score < 60 points (failure)

[0163] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for testing the performance of a radiotherapy laser positioning system, characterized in that, include: Divide the test persons into different groups according to their weight, test the first positioning data of the optical positioning device under load, construct a weight-optical response correlation model to analyze the first positioning data, and obtain static optical device test data; Control the tester to move slightly along the preset trajectory, synchronously collect the photosensitive sensor response signal of the laser beam and the spatial coordinates of the laser tracker, generate the second optical positioning data, build a motion-optical response coupling model to analyze the second optical positioning data, and obtain dynamic optical equipment test data; Construct a multi-factor deviation prediction model to analyze the optical interference signal, phase detection data and environmental temperature and humidity parameters of the synchronously collected laser beam to generate compensation parameters for the optical device; Static optical equipment test data, dynamic optical equipment test data and optical equipment compensation parameters are analyzed, the interaction coefficients of optical equipment influencing factors are calculated, and a comprehensive evaluation report of optical calibration scene fitness score, positioning accuracy quantification matrix and optical equipment acceptance benchmark value is generated.

2. The method for testing the performance of a radiotherapy laser positioning system according to claim 1, characterized in that: The first positioning data is static position information data collected under different load conditions, including coordinate data of the center position of the laser beam, slope data of the laser beam, and deviation distance between the laser beam and a preset theoretical coordinate system; The second optical positioning data is dynamic position information data continuously collected under the trajectory of micro-movement of the test person's limbs, and also includes motion trajectory, displacement change, speed, acceleration data, and continuous optical signal change data corresponding to the dynamic state of the test person and its timestamp; the range of micro-movement of the limbs is within the threshold set by the optical positioning device.

3. The method for testing the performance of a radiotherapy laser positioning system according to claim 1, characterized in that: The body weight-optical response association model includes a first positioning data processing layer, an association feature extraction layer, an association feature analysis layer and a result verification layer; The first positioning data processing layer performs preliminary processing on the first positioning data, including filtering, denoising and normalization operations; The correlation feature extraction layer performs correlation analysis on the processed first positioning data and the weight data to generate a weight-optical response correlation feature; The correlation feature analysis layer analyzes the correlation features between the weight and the optical response by using a statistical method to obtain parameter data of the effect of weight change on the static performance of the optical device, including positioning error, response sensitivity and stability index; The result verification layer obtains static optical device test data by performing statistical verification and error analysis on parameter data.

4. The method for testing the performance of a radiotherapy laser positioning system according to claim 1, characterized in that: The motion-optical response coupling model includes a second optical positioning data processing layer, a dynamic feature extraction layer, a feature coupling layer and an analysis and prediction layer; The second optical positioning data processing layer pre-processes the continuously collected second optical positioning data, including filtering, denoising and normalization operations; The dynamic feature extraction layer processes the dynamic position information data of limb micro-movements and the optical signal change data, and extracts motion feature parameters and optical response feature parameters; The feature coupling layer performs coupling analysis on the motion feature parameters and the optical response feature parameters through machine learning to generate joint motion and optical response features; The analysis and prediction layer generates dynamic optical device test data by analyzing the joint motion and optical response features; 5. The performance test method of a radiotherapy laser positioning system according to claim 1, characterized in that: The multi-factor deviation prediction model includes a feature extraction layer, a parameter coupling and analysis layer, and a prediction and compensation layer; The feature extraction layer extracts features from the optical interference signal, phase detection data, and environmental temperature and humidity parameters of the laser beam to obtain optical interference features, phase detection features, and environmental features; the optical interference features include interference fringe characteristics, phase information, and signal intensity; the phase detection features include phase offset, change trend, and correlation correction information; the environmental features include temperature change rate and humidity fluctuation; The feature coupling and analysis layer couples the optical interference features, phase detection features, and environmental features to obtain optical and environmental coupling features; analyzes the optical and environmental coupling features to obtain the deviation amount generated by the optical device due to environmental and signal fluctuations during actual use, and obtains key error indicators, including phase error, signal drift, and positioning deviation caused by the environment; The prediction and compensation layer calculates and generates optical device compensation parameters based on the key error indicators; the optical device compensation parameters include deviation correction coefficients and temperature and humidity compensation coefficients; 6. A performance testing method for a radiotherapy laser positioning system according to claim 1, characterized in that: The specific steps for calculating the interaction coefficient of the optical device influence factor are as follows: Perform feature alignment and standardization processing on the static optical device test data, dynamic optical device test data, and optical device compensation parameters to generate a multi-dimensional fusion data set; Construct a multi-factor coupling matrix, and calculate the weight coefficients of each parameter in the multi-dimensional fusion data set through the partial least squares regression algorithm; Use the particle swarm optimization algorithm to dynamically adjust the weight coefficients, and calculate the non-linear interaction effects between the factors by setting a fitness function; Perform a significance test on the optimized weight coefficients to generate a composite interaction coefficient set including linear coupling coefficients and non-linear interference coefficients; Verify the composite interaction coefficient set through the leave-one-out validation method. When the root mean square error is lower than the preset threshold, it is determined that the interaction coefficient is valid.

Images