Contactless physiological signal extraction and look-ahead motion correction device

Abstract

The invention discloses a non-contact physiological signal extraction and look-ahead motion correction device, which comprises a signal separation module, a parameter calculation module, a real-time separation and filtering module and an adaptive threshold module, and is characterized in that the signal separation module obtains a feature vector and a separation matrix of a motion signal; the parameter calculation module sets a trigger threshold value of the motion signal; the real-time separation and filtering module performs operation on the multi-channel BPT signal, the feature vector and the separation matrix to obtain a real-time motion signal; and the self-adaptive threshold module responds to the change trend of the real-time motion signal and updates the trigger threshold. The invention further discloses the rapid compressor and a compression ratio adjusting method. According to the method, the prospective motion correction of the motion signal is realized, and the magnetic resonance image which is less influenced by motion artifacts can be quickly obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of motion correction, in particular to a non-contact physiological signal extraction and prospective motion correction device. BACKGROUND

[0002] Magnetic Resonance Imaging (MRI) is a non-invasive diagnostic medical imaging technology. Due to its high spatial resolution, excellent soft tissue contrast and no ionizing radiation, MRI is widely used in imaging examination of brain, heart, abdomen and other organ regions, and can provide accurate image data for clinical diagnosis of cerebral vascular diseases, tumors and other diseases, and plays a key role in developing treatment plans, guiding surgical operations and evaluating treatment effects.

[0003] Compared with other imaging technologies such as computed tomography, MRI is more susceptible to motion artifacts. The main reason is that the subject is prone to voluntary or involuntary body movement during imaging, and the imaging mechanism of MRI itself is particularly sensitive to disturbances caused by motion. Therefore, exploring effective motion artifact suppression and correction methods is of great significance to improve image quality and diagnostic accuracy.

[0004] To avoid or alleviate motion artifacts in MRI, there are mainly three methods: motion suppression, artifact mitigation sequence design, and motion tracking correction. Among them, motion suppression is the most commonly used method in current clinical practice, including installing bite bars on head coils, using fixation devices or guiding subjects to hold their breath, etc. However, these methods often cause significant discomfort to the subjects, and limit the available scan time and imaging sequences, thereby affecting image quality. Artifact mitigation sequences reduce the sensitivity of images to motion by shortening the imaging time, optimizing the phase encoding direction and scanning order, etc., but often at the expense of reducing spatial resolution, and may introduce new artifact types. Traditional motion tracking and correction methods usually require the installation of multiple contact-type complex hardware devices, such as electrode patches, on the subject's body to detect the subject's motion state synchronously during imaging, and use the collected motion information for image correction. However, such devices are not only cumbersome to install, but also susceptible to magnetic field interference, and may cause skin burns and other safety hazards.

[0005] In recent years, a new type of pilot tone (PT) non-contact physiological motion monitoring technology has emerged. By placing a single-frequency radio frequency signal source near the subject's body surface, the PT signal generated after receiving the physiological motion modulation is received by the coil of the MRI system, which can effectively obtain the motion information of the subject, and is easy to install and does not cause the patient to feel obvious discomfort. However, due to the limitation of the frequency of the PT signal to the Larmor frequency of the MRI system, its sensitivity to physiological motion is limited. Therefore, the Beat Pilot Tone (BPT) technology has emerged, which overcomes the sensitivity limitation by the second-order intermodulation characteristics in the preamplifier in the magnetic resonance receiving chain and improves the sensitivity to motion.

[0006] The motion tracking correction technology is mainly forward trigger gating technology and retrospective motion correction method, and the trigger gating technology is more commonly used in clinical practice. Since the physiological effects such as heartbeat and respiration cause periodic patient motion, the magnetic resonance system is used to detect the physiological motion signal collected in real time, and when it is determined that a certain threshold is reached, the sequence scanning is started, the radio frequency pulse is emitted and the signal is collected. This method can quickly reconstruct a magnetic resonance image without motion artifacts after scanning, which is suitable for clinical application scenarios. The retrospective motion correction technology reorders the magnetic resonance data obtained by scanning through the motion signal synchronized with the magnetic resonance acquisition, and classifies it into the same motion state to effectively reduce the influence of motion artifacts.

[0007] However, the existing motion correction based on BPT technology is limited to retrospective motion correction, and the reconstruction time is long, the lag is large, and the magnetic resonance image cannot be obtained in real time after scanning, which reduces the clinical application value. The current PT technology only has obvious effect on respiratory motion detection in medium and high fields, and has limited detection capability for cardiac motion; at the same time, the existing technology mainly estimates the heartbeat motion of the subject by detecting the low-frequency component of the heartbeat signal, which is easily disturbed by respiratory motion, and it is difficult to synchronize the correction of the heartbeat motion in the free breathing state. The frequency of the existing PT technology is limited by the Larmor frequency of the magnetic resonance system, and the accuracy of the detection of human motion is limited, especially in medium and low field MRI, the technology cannot sensitively detect body motion.

[0008] Therefore, the skilled in the art is committed to providing a multi-pilot tone based non-contact physiological signal extraction and forward motion correction device under magnetic resonance, which realizes non-contact respiratory and heartbeat signal extraction and forward motion correction in magnetic resonance scanning. SUMMARY

[0009] In view of the defects in the prior art, the technical problem to be solved by the present application is how to provide a device capable of realizing motion signal extraction and forward motion correction.

[0010] In order to achieve the above object, the application provides a device for contactless physiological signal extraction and prospective motion correction, comprising a signal separation module, a parameter calculation module, a real-time separation and filtering module, and an adaptive threshold module, wherein the signal separation module acquires a feature vector and a separation matrix of a motion signal; the parameter calculation module sets a trigger threshold of the motion signal; the real-time separation and filtering module operates a multi-channel BPT signal with the feature vector and the separation matrix to obtain a real-time motion signal; and the adaptive threshold module updates the trigger threshold in response to a change trend of the real-time motion signal.

[0011] Further, the device further comprises a TTL level output circuit which sends a control signal to a magnetic resonance system when a trigger event occurs.

[0012] Further, the motion signal is from a subject, and the motion signal comprises a breathing signal and a heartbeat signal.

[0013] Further, the signal separation module adopts principal component analysis for dimension compression on a multi-dimensional signal, extracts a signal component with independent statistical characteristics through fast independent component analysis, and acquires the breathing signal and the heartbeat signal.

[0014] Preferably, the signal separation module selects an energy maximum in a 0.2-0.7 Hz frequency band as the breathing signal, compares the breathing signal with a channel with a maximum modulation ratio to unify the polarity, acquires a breathing feature vector and a separation matrix, selects a target component in a 3-7 Hz frequency band as the heartbeat signal, and acquires a heartbeat feature vector and a separation matrix through polarity adjustment.

[0015] Preferably, the parameter calculation module adjusts the polarity of the breathing signal so that an ascending waveform represents an exhalation direction, and a breathing trigger threshold is set as , wherein is a mean value of the breathing signal in a learning stage, is a standard deviation of the breathing signal in the learning stage, k is a threshold ratio design coefficient.

[0016] Preferably, the parameter calculation module sets a sine wave positive half-cycle structure element with a corresponding length to perform opening and closing operations to enhance a J wave based on a one-dimensional mathematical morphological feature enhancement algorithm and a time scale feature of the J wave in the heartbeat waveform, calculates an average heart rate and a waveform amplitude based on the enhanced heartbeat signal.

[0017] Preferably, the real-time separation and filtering module multiplies the breathing signal with the breathing feature vector to obtain a BPT signal after dimension compression, and then multiplies the BPT signal with a breathing separation matrix to obtain a real-time breathing signal.

[0018] Preferably, the real-time separation and filtering module performs third-order Butterworth high-pass filtering on the heartbeat signal to remove the low-frequency respiratory component, and then multiplies the filtered heartbeat signal with a heartbeat feature vector and a heartbeat separation matrix to obtain a real-time heartbeat signal.

[0019] Preferably, the adaptive threshold module calculates the mean and variance of the real-time respiratory signal in real time to adjust the trigger threshold, and detects the change of the real-time heartbeat signal to adjust the trigger delay and the waiting period setting.

[0020] The present application has at least the following beneficial technical effects: 1. The present application uses BPT technology to extract the motion information of the subject in real time, detects the characteristics of the end-expiratory period and the heartbeat mechanical motion in advance, and combines the delay compensation and prediction algorithm to output a trigger signal before the motion moment, thereby realizing forward-looking motion correction, quickly obtaining a magnetic resonance image less affected by motion artifacts, shortening the time required for magnetic resonance image reconstruction, reducing the consumption of computing resources, effectively improving the imaging efficiency, and expanding the applicability of the method in various magnetic resonance application scenarios.

[0021] 2. The present application separates the low-frequency and high-frequency components in the BPT signal in a low-delay manner by combining a linear filter and a Kalman filter, and estimates the breathing and heartbeat motion of the subject, thereby realizing synchronous detection and correction of the two types of physiological motion; at the same time, a feature enhancement algorithm based on mathematical morphology is introduced to effectively improve the robustness and stability of heartbeat motion detection.

[0022] 3. The present application uses two high-frequency signals to detect the physiological motion and rigid body motion of the subject, and uses the second-order intermodulation property of the high-frequency signal at the receiving end to obtain the BPT signal within the MR receiving bandwidth, breaking through the limitation that the detection sensitivity of the traditional PT method depends on the Larmor frequency of the magnetic resonance system, increasing the signal frequency by nearly 20 times, realizing high-sensitivity motion detection at low and medium field strengths, and improving the applicability of the method.

[0023] The concept, specific structure and technical effects of the present application will be further described below with reference to the accompanying drawings to fully understand the purpose, features and effects of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 is a non-contact physiological signal extraction and forward-looking motion correction device schematic diagram of an embodiment of the present application; Figure 2 is a flowchart of a feature enhancement algorithm based on mathematical morphology of an embodiment of the present application; Figure 3 is a TTL level output module circuit diagram of an embodiment of the present application; Figure 4is a schematic diagram of the effect of contactless physiological signal extraction and prospective motion correction of the embodiment of the present application; Figure 5 is a comparison chart of the effect of respiratory motion artifact correction of the embodiment of the present application. DETAILED DESCRIPTION

[0025] The preferred embodiments of the present application are described below to make the technical content of the present application clearer and easier to understand. The present application can be embodied in many different forms and the protection scope of the present application is not limited to the embodiments described herein.

[0026] In the drawings, components of the same structure are denoted by the same reference numerals, and components having similar structures or functions are denoted by similar reference numerals. The size and thickness of each component shown in the drawings are arbitrarily shown, and the present application is not limited to the size and thickness of each component. In order to make the drawing clearer, the thickness of some components is appropriately exaggerated in some places in the drawing.

[0027] The present application provides a contactless physiological signal extraction and prospective motion correction device, which is based on BPT technology to realize contactless respiratory and heartbeat signal extraction and prospective motion correction in magnetic resonance scanning. The present application can stably obtain physiological motion information of a subject without additional electrodes and contact type sensors such as respiration, avoiding problems such as cumbersome electrode attachment, signal distortion, and subject discomfort in traditional methods.

[0028] As shown in Figure 1 The contactless physiological signal extraction and prospective motion correction device of the present embodiment includes a signal separation module, a parameter calculation module, a real-time separation and filtering module, and an adaptive threshold module. The signal separation module obtains a feature vector and a separation matrix of a motion signal, the parameter calculation module sets a trigger threshold of the motion signal, the real-time separation and filtering module operates the multi-channel BPT signal with the feature vector and the separation matrix obtained by the signal separation module to obtain a real-time motion signal, and the adaptive threshold module updates the trigger threshold in response to the change trend of the real-time motion signal. The signal separation module and the parameter calculation module are applied to a learning stage, and the real-time separation and filtering module and the adaptive threshold module are applied to a trigger stage.

[0029] The motion signal comes from the motion information of the subject. A signal emitting device can be added on or near the subject to convert the signal containing the motion information of the subject into a BPT signal after Fourier transform, and then send it to the signal separation module to obtain the motion signal. The motion signal is divided into a respiratory signal and a heartbeat signal, which respectively represent respiratory motion and heartbeat motion.

[0030] The signal separation module uses principal component analysis for dimension compression on the multi-dimensional signal, and extracts a signal component with independent statistical characteristics through fast independent component analysis, to obtain a stable motion signal. Specifically, the signal separation module selects the one with the maximum energy in the 0.2-0.7 Hz frequency band as the respiratory signal, and compares it with the channel with the maximum modulation ratio to unify the polarity, to obtain a respiratory feature vector and a separation matrix. For the heartbeat signal, to reduce the interference of the respiratory signal, the signal separation module first applies a third-order Butterworth high-pass filter with a cutoff frequency of 3 Hz before PCA, and then performs PCA and FastICA analysis, to screen the target component in the 3-7 Hz frequency band, and adjust the polarity, to obtain a heartbeat feature vector and a separation matrix.

[0031] The parameter calculation module adjusts the polarity of the respiratory signal, so that the rising waveform represents the exhalation direction, and the respiratory trigger threshold is set as , wherein is the mean value of the respiratory signal in the learning stage, is the standard deviation of the respiratory signal in the learning stage, k is the threshold ratio design coefficient. For the heartbeat signal, the parameter calculation module sets a sinusoidal positive half-cycle structure element with a length corresponding to the time scale feature of the J wave in the heartbeat waveform of the subject to perform opening and closing operations to enhance the J wave and suppress other components, based on the enhanced heartbeat signal, calculates the average heart rate and waveform amplitude, and sets the sampling delay and window length.

[0032] In one specific embodiment, k The value can be set to 0.3, to facilitate identification of the plateau stage at the end of exhalation. The flowchart of the feature enhancement algorithm based on mathematical morphology is shown in Figure 2 , in which a sinusoidal positive half-cycle with a length corresponding to the J wave is used as a structure element to perform opening and closing operations on the one-dimensional signal according to the flowchart.

[0033] In the real-time processing stage, the real-time separation and filtering module obtains the multi-channel BPT signal in real time after each K-line data is acquired in the magnetic resonance acquisition sequence, and reconstructs the respiratory signal and the heartbeat signal according to the feature vector and the separation matrix saved in the learning stage, respectively. For the respiratory signal, the real-time acquired signal is first multiplied by the respiratory feature vector obtained in the learning stage to obtain the BPT signal after dimension compression, and then multiplied by the corresponding separation matrix, to obtain the real-time respiratory signal with the same polarity as in the learning stage. For the heartbeat signal, the BPT signal is first subjected to a third-order Butterworth high-pass filter, for example, with a cutoff frequency of 3 Hz, to remove the low-frequency respiratory component interference; then, it is also multiplied by the heartbeat feature vector and the separation matrix, to obtain the real-time heartbeat signal required.

[0034] To further improve the signal-to-noise ratio and control the processing delay, the real-time separation and filtering module introduces an extended Kalman filter modeling process to the reconstructed signal, wherein the respiratory signal is modeled as a sinusoidal signal varying with time:

[0035] wherein represents the estimated value of the amplitude of the respiratory signal at time t, is a direct current component, is the respiratory frequency of the subject, and are the phase and amplitude of the sinusoidal signal. The heartbeat signal is modeled as a superposition of 4th and 5th harmonics:

[0036] wherein represents the estimated value of the amplitude of the heartbeat signal at time t, is a direct current component, is the estimated heartbeat frequency of the subject by the model, and are the amplitude and phase of the 4th harmonic of the heartbeat signal, and are the amplitude and phase of the 5th harmonic of the heartbeat signal.

[0037] In the initial stage of the trigger phase, the trigger parameters are obtained from the learning phase; subsequently in the data collection, the real-time signal is continuously written into a fixed-length buffer for updating the judgment threshold and responding to the signal change trend. For the respiratory signal, the adaptive threshold module calculates the mean and variance of the data in the buffer in real time to adjust the trigger threshold. For the heartbeat signal, the adaptive threshold module also automatically detects the change of the R-R interval and adjusts the trigger delay and the waiting period setting accordingly to cope with unstable heart rate, abnormal heartbeat and the like, thereby reducing the risk of false triggering and missed triggering and ensuring that the imaging timing matches the heart motion state.

[0038] Other embodiments of the present application also have a TTL level output circuit, as shown in Figure 3 The TTL level output circuit is based on a CH340 serial communication chip and communicates with the upper computer through a USB interface. When a trigger event is recognized, the upper computer sends a control signal to the lower computer, and the polarity inversion circuit generates a standard TTL level pulse with an output amplitude of 5V and a duration of about 7.5ms. The pulse can be used as an effective signal for the external trigger interface of the magnetic resonance system to achieve precise control of the scanning timing and meet the requirements of the respiratory or heartbeat prospective motion correction.

[0039] The effect diagram of the physiological signal extraction and prospective correction technology of one specific embodiment of the present application is as shown in Figure 4As shown, the left figure is a schematic diagram of extraction and forward correction based on a respiratory signal, and the right figure is a schematic diagram of extraction and forward correction based on a heartbeat signal.

[0040] A specific embodiment of the present application is a comparison chart of correction effects of artifacts caused by respiratory motion, as shown in the figure. Figure 5 As shown, the left figure is a liver imaging result in a free breathing state without using the present method, and the right figure is a liver imaging result in a free breathing state using the present method. By comparing the regions indicated by the arrows, the left image presents ghosting caused by respiratory motion, and the right image has more sharp tissue edges and clearer contours.

[0041] The contactless physiological signal extraction and forward motion correction device of the present application can be directly embedded in the receiving chain and trigger module of an existing magnetic resonance system, only needs to add a signal source and software modules for pilot signal demodulation and real-time processing, and has good compatibility and easy integration. The correction device of the present application uses two high-frequency signals to improve the working frequency and effectively enhance the detection sensitivity of motion signals, so that it also has good applicability in low-field magnetic resonance systems, thereby significantly expanding the application scenarios. At the same time, the present application estimates the respiratory motion of the subject using the low-frequency component of the BPT signal and estimates the heartbeat motion using the high-frequency component, thereby achieving effective separation and mutual interference of the two types of signals; under the condition of free breathing, it can simultaneously suppress the motion artifacts caused by breathing and heartbeat; experimental results show that its triggering performance is highly consistent with the commonly used respiratory and electrocardiogram triggering methods, and it exhibits good stability and reliability.

[0042] The above describes the preferred embodiments of the present application in detail. It should be understood that those skilled in the art can make many modifications and changes without creative labor based on the concept of the present application. Therefore, any technical solution obtained by logical analysis, reasoning or limited experiments based on the prior art according to the concept of the present application shall be within the protection scope defined by the claims.

Claims

1. A non-contact physiological signal extraction and prospective motion correction device, characterized in that, The system includes a signal separation module, a parameter calculation module, a real-time separation and filtering module, and an adaptive threshold module. The signal separation module acquires the feature vector and separation matrix of the motion signal; the parameter calculation module sets the trigger threshold of the motion signal; the real-time separation and filtering module performs calculations on the multi-channel BPT signal with the feature vector and separation matrix to obtain the real-time motion signal; and the adaptive threshold module updates the trigger threshold in response to the changing trend of the real-time motion signal.

2. The contactless physiological signal extraction and prospective motion correction device as described in claim 1, characterized in that, It also includes a TTL level output circuit, which sends a control signal to the magnetic resonance system when a trigger event occurs.

3. The contactless physiological signal extraction and prospective motion correction device as described in claim 1, characterized in that, The motion signals are from the subject and include respiratory and heart rate signals.

4. The contactless physiological signal extraction and prospective motion correction device as described in claim 3, characterized in that, The signal separation module performs dimensionality compression on the multidimensional signal using principal component analysis and extracts signal components with independent statistical characteristics through rapid independent component analysis to obtain the respiratory signal and heartbeat signal.

5. The contactless physiological signal extraction and prospective motion correction device as described in claim 4, characterized in that, The signal separation module selects the highest energy signal in the 0.2-0.7Hz frequency band as the respiratory signal and compares it with the channel with the highest modulation ratio to unify the polarity, thereby obtaining the respiratory feature vector and separation matrix; it also selects the target component in the 3-7Hz frequency band as the heartbeat signal and obtains the heartbeat feature vector and separation matrix by adjusting the polarity.

6. The contactless physiological signal extraction and prospective motion correction device as described in claim 5, characterized in that, The parameter calculation module adjusts the polarity of the respiratory signal so that the rising waveform represents the direction of exhalation, and the respiratory trigger threshold is set to... ,in To learn how to calculate the mean of the respiratory signal, The standard deviation of the respiratory signal during the learning phase. k The threshold ratio is designed with a coefficient.

7. The contactless physiological signal extraction and prospective motion correction device as described in claim 6, characterized in that, The parameter calculation module is based on a one-dimensional mathematical morphology feature enhancement algorithm. It combines the time scale characteristics of the J wave in the heartbeat waveform, sets a sine wave positive half-cycle structural element of corresponding length to perform opening and closing operations to enhance the J wave, and calculates the average heart rate and waveform amplitude based on the enhanced heartbeat signal.

8. The contactless physiological signal extraction and prospective motion correction device as described in claim 7, characterized in that, The real-time separation and filtering module multiplies the respiratory signal with the respiratory feature vector to obtain the BPT signal after compression, and then multiplies it with the respiratory separation matrix to obtain the real-time respiratory signal.

9. The contactless physiological signal extraction and prospective motion correction device as described in claim 8, characterized in that, The real-time separation and filtering module performs a third-order Butterworth high-pass filter on the heartbeat signal to remove low-frequency breathing components, and then multiplies the filtered heartbeat signal with the heartbeat feature vector and the heartbeat separation matrix to obtain the real-time heartbeat signal.

10. The contactless physiological signal extraction and prospective motion correction device as described in claim 9, characterized in that, The adaptive threshold module calculates the mean and variance of the real-time respiratory signal in real time and adjusts the trigger threshold; the adaptive threshold module detects changes in the real-time heartbeat signal and adjusts the trigger delay and waiting period settings.

Images