Electromyographic signal-based tumor motion prediction method and system

Abstract

The present invention relates to the field of medical signal processing technology, and specifically, to an electromyographic signal-based tumor motion prediction method and system. The system of the present invention comprises: an image guidance module, configured to: acquire electromyographic signals and tumor motion of a patient, extract a respiratory-related signal from the electromyographic signals by sequentially using gating, template subtraction, envelope calculation, and offline filtering, and calculate a correlation coefficient and a lookahead time between the respiratory-related signal and the tumor motion by using a cross-correlation function; and a treatment delivery module, configured to: acquire electromyographic signals of the patient, extract a respiratory-related signal from the electromyographic signals by sequentially using gating, template subtraction, envelope calculation, and online filtering, and acquire a prediction result of the tumor motion on the basis of the lookahead time calculated by the image guidance module. The present invention further provides a method for predicting tumor motion by using the system. The present invention can compensate for the system delay of a radiotherapy device in tracking tumor motion, and has good application prospects.

Description

A method and system for predicting tumor movement based on electromyographic signals Technical Field

[0001] This invention belongs to the field of medical signal processing technology, specifically relating to a method and system for predicting tumor movement based on electromyographic signals. Background Technology

[0002] Tumor radiotherapy is a local treatment method that uses radiation to treat tumors. Radiation includes alpha, beta, and gamma rays produced by radioactive isotopes, as well as X-rays, electron beams, proton beams, and other particle beams produced by various X-ray therapy machines or accelerators. Radiotherapy also has a killing effect on normal tissues; therefore, the precision of radiotherapy is a very important issue in this field. Tumors in the chest and abdomen move with the patient's breathing, making tumor localization difficult. Furthermore, similar problems exist for non-tumor lesions; lesion movement during respiration can adversely affect the precision of robotic surgery. Therefore, respiratory motion is a common problem affecting the precision of radiotherapy and robotic surgery for chest and abdominal tumors.

[0003] In existing technologies, tumor motion tracking technology is mainly based on traditional imaging systems such as anatomical imaging systems, optical imaging systems, and electromagnetic imaging systems (e.g., Chinese patent application CN110168657A - Tumor tracking using intelligent tumor size change notifications). This method is limited to tracking historical tumor motion locations or tumor motion reference locations. However, after determining the tumor location, radiotherapy accelerators or surgical robots still need to determine the tumor position and adjust the treatment plan in real time. This means that there may be a significant discrepancy between the tumor location tracked by traditional imaging and the tumor location treated by the accelerator or surgical robot.

[0004] Therefore, how to predict tumor movement more quickly and eliminate this bias is an important issue. Electromyography (EMG) signals have a 320ms look-ahead time relative to respiratory movement; the clinically reported accelerometer response time is mostly around 300ms. This means that EMG signals have the potential to compensate for the accelerometer response time, predicting the future location of lesions in advance, thereby enabling more accurate lesion movement tracking. However, current technologies lack relevant research and have not yet achieved the conversion of EMG signals into predictive information about lesion location. Summary of the Invention

[0005] To address the problems of existing technologies, this invention provides a method and system for predicting tumor movement based on electromyographic signals.

[0006] A tumor movement prediction system based on electromyographic signals, comprising:

[0007] The image-guided module is configured to: acquire the patient's electromyography (EMG) signals and tumor movement; sequentially extract respiratory-related signals from the EMG signals using gating, template subtraction, envelope calculation, and offline filtering; and calculate the correlation coefficient and look-ahead time between the respiratory-related signals and tumor movement using a cross-correlation function.

[0008] The treatment delivery module is configured to: acquire the patient's electromyography (EMG) signals, and sequentially extract respiratory-related signals from the EMG signals using gating, template subtraction, envelope calculation, and online filtering; if the correlation coefficient is higher than a preset value, based on the look-ahead time calculated by the image guidance module, obtain a real-time prediction result of the imminent location and movement trajectory of the tumor by sliding the respiratory-related signals on the time axis.

[0009] Preferably, the gating method and template subtraction are used to clean the electromyographic signals and remove artifacts from the electrocardiogram signals.

[0010] Preferably, the envelope calculation is used to transform the electromyographic signal after removing ECG signal artifacts to obtain the envelope of the electromyographic signal, so as to represent the amplitude changes in the electromyographic signal.

[0011] Preferably, the offline filtering and online filtering smooth the envelope of the electromyographic signal to obtain the respiratory-related signal in the electromyographic signal.

[0012] Preferably, the offline filtering method is selected from moving average smoothing.

[0013] And / or, the online filtering method is selected from online smoothing based on recurrent neural networks.

[0014] Preferably, the recurrent neural network is selected from LSTM or GRU.

[0015] Preferably, during the online smoothing process, the output of the recurrent neural network is restricted by the constraints of spline interpolation.

[0016] Preferably, the preset value of the correlation coefficient is 0.9.

[0017] The present invention also provides a method for predicting tumor movement using the above-mentioned tumor movement prediction system based on electromyography signals, comprising the following steps:

[0018] Step 1: In the image-guided stage, the patient's electromyography (EMG) signals and tumor movement are acquired. Gating, template subtraction, envelope calculation, and offline filtering are used sequentially to extract respiratory-related signals from the EMG signals. The correlation coefficient and look-ahead time between the respiratory-related signals and tumor movement are calculated using a cross-correlation function.

[0019] Step 2: During the treatment delivery phase, the patient's electromyography (EMG) signals are collected, and respiratory-related signals are extracted from the EMG signals by sequentially employing gating, template subtraction, envelope calculation, and online filtering. If the correlation coefficient is higher than a preset value, based on the look-ahead time calculated by the image guidance module, the respiratory-related signals are slid along the time axis to obtain real-time prediction results of the tumor's impending location and movement trajectory.

[0020] The present invention also provides a computer-readable storage medium having stored thereon a computer program for implementing the above-described tumor motion prediction system based on electromyography signals, or a computer program for implementing the above-described method.

[0021] This invention aims to predict tumor movement using electromyography (EMG) signals. It designs a data processing workflow for EMG signals and ultimately predicts tumor movement by correlating respiratory-related signals within the EMG signals with the anticipatory time of tumor movement. This invention is the first to realize the operation of predicting tumor movement using EMG signals. Compared with tumor movement tracking technologies based on traditional imaging (anatomical imaging, electromagnetic imaging, optical imaging), EMG-based tumor movement tracking technology can anticipate the future location of the tumor, which helps to solve the problem of reduced tumor movement tracking accuracy caused by delays in radiotherapy equipment systems. Therefore, this invention has excellent application prospects.

[0022] Obviously, based on the above description of the present invention, and according to common technical knowledge and conventional methods in the field, various other modifications, substitutions or alterations can be made without departing from the basic technical concept of the present invention.

[0023] The following detailed embodiments further illustrate the above-described content of the present invention. However, this should not be construed as limiting the scope of the present invention to the following examples. All technologies implemented based on the above-described content of the present invention fall within the scope of the present invention. Attached Figure Description

[0024] Figure 1 is a flowchart of Embodiment 1 of the present invention;

[0025] Figure 2 is an example diagram of the data processing results of each step in Embodiment 1 of the present invention, wherein (a) is the original data of electromyography (EMG) signal on the surface of the diaphragm, (b) is the EMG signal after removing ECG interference, (c) is the EMG envelope signal, and (d) is the comparison between the EMG-predicted respiratory signal and the actual respiratory signal. Detailed Implementation

[0026] It should be noted that the algorithms for data acquisition, transmission, storage and processing steps not specifically described in the embodiments, as well as the hardware structures and circuit connections not specifically described, can all be implemented using content already disclosed in the prior art.

[0027] Example 1: Tumor Motility Prediction Method and System Based on Electromyographic Signals

[0028] The system in this embodiment includes:

[0029] The image-guided module is configured to: acquire the patient's electromyography (EMG) signals and tumor movement; sequentially extract respiratory-related signals from the EMG signals using gating, template subtraction, envelope calculation, and offline filtering; and calculate the correlation coefficient and look-ahead time between the respiratory-related signals and tumor movement using a cross-correlation function.

[0030] The treatment delivery module is configured to: acquire the patient's electromyography (EMG) signals, and sequentially extract respiratory-related signals from the EMG signals using gating, template subtraction, envelope calculation, and online filtering; if the correlation coefficient is higher than a preset value, based on the look-ahead time calculated by the image guidance module, obtain a real-time prediction result of the imminent location and movement trajectory of the tumor by sliding the respiratory-related signals on the time axis.

[0031] The workflow for tumor motion prediction using the above system is shown in Figure 1, and includes the following steps:

[0032] Step 1: In the image-guided stage, the patient's electromyography (EMG) signals and tumor movement are acquired. Gating, template subtraction, envelope calculation, and offline filtering are used sequentially to extract respiratory-related signals from the EMG signals. The correlation coefficient and look-ahead time between the respiratory-related signals and tumor movement are calculated using a cross-correlation function.

[0033] Step 2: During the treatment delivery phase, the patient's electromyography (EMG) signals are acquired. Gating, template subtraction, envelope calculation, and online filtering are sequentially used to extract respiratory-related signals from the EMG signals. If the correlation coefficient is higher than a preset value, based on the look-ahead time calculated by the image-guided module, the respiratory-related signals are slid along the time axis to obtain a real-time prediction of the tumor's impending location and trajectory. If the correlation coefficient is lower than the preset value, the measurement is repeated after the patient's breathing stabilizes. Preferably, the preset value for the correlation coefficient is 0.9.

[0034] The above sliding operation is performed according to the following formula: S sEMG [t] = S resp [t+t precursor ]

[0035] That is, the electromyographic signal S at the current time t. sEMG Calculate t+t precursor Breathing movements S resp , t precursor It is the look-ahead time calculated in step 1.

[0036] In the above-described electromyography (EMG) signal processing flow, the gating method and template subtraction are used to clean the EMG signal and remove ECG signal artifacts. The envelope calculation is used to transform the EMG signal after removing ECG signal artifacts, obtaining the envelope of the EMG signal to represent the amplitude changes in the EMG signal. Offline filtering and online filtering smooth the envelope of the EMG signal, respectively, to obtain the respiratory-related signals within the EMG signal.

[0037] in,

[0038] Gating methods were used to remove the PR interval component from the electromyographic signal, while template subtraction was used to remove the QT interval component, T wave, and U wave from the electromyographic signal (Figure 1c): S sEMG [t R -Δt g :t R ] = S sEMG [t R -2Δt g :t R -Δt g #(1)

[0039] Gating methods refer to replacing a signal interfered with by the electrocardiogram (ECG) signal with a short, stable signal. Wherein, S... sEMG Represents electromyographic signals, t R Δt represents the time point corresponding to the R-wave peak. g This is a gated window. In this embodiment, Δt before the PR interval is used. g The data of the length window is used to replace the PR interval data to remove ECG signal interference, Δt g The time delay was set to 50 milliseconds, which exceeds the 40 millisecond time delay for R-wave peak detection summarized in the review literature. sEMG [t R :t R +Δt m ] = S sEMG [t R :t R +Δt m ]-ηS ECG [t R :t R +Δt m #(2)

[0040] Among them, S ECG Δt represents the average electrocardiogram signal obtained from the 10-second breath-holding task data. m This refers to the template subtraction window, and η corresponds to S in each ECG cycle. sEMG [t R ] and S ECG [t R The ratio of Δt. In this embodiment, Δtm It was set to 500 milliseconds.

[0041] To eliminate polarity changes in sEMG, this embodiment calculates the envelope of the sEMG signal after removing ECG. i :

[0042] Here, L refers to the envelope calculation window. In this study, L is set to 25 milliseconds, which does not exceed the gating window Δt1, thus avoiding the introduction of additional time delay. i represents the current time point, and j represents the last time point of the envelope calculation window.

[0043] Based on the above data processing, in step 1, the offline filtering method is selected from moving average smoothing to obtain a stable look-ahead time. Offline filtering can provide more accurate noise removal and facilitate accurate calculation of correlation coefficients and look-ahead time. Step 2 is an online intervention in the treatment process, which requires the use of an online filtering algorithm to reflect real-time characteristics. The online filtering method is selected from online smoothing based on recurrent neural networks (RNNs) to meet the real-time prediction requirements. The recurrent neural network is selected from LSTM or GRU, and in this embodiment, GRU with relatively fast computation speed is preferred. During the online smoothing process, since the RNN predicted value is a discrete variable, the predicted value will have large fluctuations. Therefore, the output of the recurrent neural network is restricted by the constraint condition of spline interpolation to achieve the purpose of smoothing the output data.

[0044] Figure 2 shows an example of the results of the above data processing.

[0045] The technical solution of the present invention will be further illustrated by the following experiments.

[0046] Experiment Example 1: Comparison of differences in the results of anticipatory time calculations of electromyographic signals and respiratory movements at different time points. I. Experimental Methods

[0047] Electromyographic signals and respiratory movements were acquired from 15 healthy adults using paired Ag-AgCl electrodes (Noraxon, Scottsdale, AZ, USA) at a sampling rate of 2000 Hz and a pressure sensor integrated into an elastic waistband (Anzai Medical, Tokyo, Japan) at a sampling rate of 40 Hz, respectively. For further analysis, respiratory cycle curves were interpolated to 2000 Hz. Paired measuring electrodes were placed in a bilateral long configuration along the subcostal region, with a reference electrode placed on the sternum, as shown in Figure 1(a). Subjects were instructed to perform a breath-holding task for approximately 10 seconds to capture a template of the electrocardiogram signal, followed by a 10-minute supine free breathing task.

[0048] Electromyography (EMG) signals were processed using MATLAB code to extract respiratory-related signals. The raw diaphragmatic EMG signals were preprocessed using a 1-500Hz bandpass filter and a 50Hz notch filter to obtain pre-processed EMG signals.

[0049] Other experimental steps and settings shall be performed in accordance with the method described in Example 1.

[0050] II. Experimental Results

[0051] The results are shown in Table 1, where r is the correlation coefficient between electromyography (EMG) signals and respiratory movements after a full 10-minute preprocessing period. Δt0 is the look-ahead time for EMG signals and respiratory movements after a full 10-minute preprocessing period. Δt1 and Δt2 are the look-ahead times calculated from the first 1 minute (corresponding to step 1 of Example 1 during radiotherapy) and the last 9 minutes (corresponding to step 2 of Example 1 during radiotherapy), respectively. There was no significant statistical difference between Δt1 and Δt2, indicating that the look-ahead time calculated during the image-guided phase of radiotherapy in Example 1 can be applied during the treatment delivery phase of radiotherapy.

[0052] Table 1

[0053] As can be seen from the above embodiments and experimental examples, the present invention constructs a method and system for predicting tumor movement using electromyographic signals, which can compensate for the system delay when radiotherapy equipment tracks tumor movement in the prior art, and has good application prospects.

Claims

1. A tumor movement prediction system based on electromyographic signals, characterized in that, include: The image-guided module is configured to: acquire the patient's electromyography (EMG) signals and tumor movement; sequentially extract respiratory-related signals from the EMG signals using gating, template subtraction, envelope calculation, and offline filtering; and calculate the correlation coefficient and look-ahead time between the respiratory-related signals and tumor movement using a cross-correlation function. The treatment delivery module is configured to: acquire the patient's electromyography (EMG) signals, and sequentially extract respiratory-related signals from the EMG signals using gating, template subtraction, envelope calculation, and online filtering; if the correlation coefficient is higher than a preset value, based on the look-ahead time calculated by the image guidance module, obtain a real-time prediction result of the imminent location and movement trajectory of the tumor by sliding the respiratory-related signals on the time axis.

2. The tumor motion prediction system based on electromyography signals according to claim 1, characterized in that: The gating method and template subtraction are used to clean electromyographic signals and remove artifacts from electrocardiogram signals.

3. The tumor motion prediction system based on electromyography signals according to claim 1, characterized in that: The envelope calculation is used to transform the electromyographic signal after removing ECG signal artifacts to obtain the envelope of the electromyographic signal, which represents the amplitude changes in the electromyographic signal.

4. The tumor motion prediction system based on electromyography signals according to claim 1, characterized in that: The offline filtering and online filtering respectively smooth the envelope of the electromyographic signal to obtain the respiratory-related signal in the electromyographic signal.

5. The tumor motion prediction system based on electromyography signals according to claim 1 or 4, characterized in that: The offline filtering method is selected from moving average smoothing; And / or, the online filtering method is selected from online smoothing based on recurrent neural networks.

6. The tumor motion prediction system based on electromyography signals according to claim 5, characterized in that: The recurrent neural network is selected from LSTM or GRU.

7. The tumor motion prediction system based on electromyography signals according to claim 5, characterized in that: During the online smoothing process, the output of the recurrent neural network is restricted by the constraints of spline interpolation.

8. The tumor motion prediction system based on electromyography signals according to claim 1, characterized in that: The preset value for the correlation coefficient is 0.

9.

9. A method for predicting tumor movement using the tumor movement prediction system based on electromyographic signals as described in any one of claims 1-8, characterized in that, Includes the following steps: Step 1: In the image-guided stage, the patient's electromyography (EMG) signals and tumor movement are acquired, and respiratory-related signals are extracted from the EMG signals by gating, template subtraction, envelope calculation, and offline filtering in sequence. The correlation coefficient and prospective time between the respiratory-related signals and tumor motion were calculated using a cross-correlation function; Step 2: During the treatment delivery phase, the patient's electromyography (EMG) signals are collected, and respiratory-related signals are extracted from the EMG signals by sequentially employing gating, template subtraction, envelope calculation, and online filtering. If the correlation coefficient is higher than a preset value, based on the look-ahead time calculated by the image guidance module, the respiratory-related signals are slid along the time axis to obtain real-time prediction results of the tumor's impending location and movement trajectory.

10. A computer-readable storage medium, characterized in that, It stores: a computer program for implementing the tumor motion prediction system based on electromyography signals as described in any one of claims 1-8, or a computer program for implementing the method described in claim 9.

Images