Magnetic resonance eddy current prediction method and system

Abstract

The embodiment of the invention provides a magnetic resonance eddy current prediction method, system and device. The method comprises the following steps: acquiring a gradient parameter of a gradient signal of a magnetic resonance system; the gradient parameters are processed through an eddy current prediction model, eddy current field components corresponding to gradient signals are determined, and the gradient signals comprise single gradient signals or gradient signal combinations.

Description

Technical Field

[0001] The present invention relates to the field of medical technology, and in particular to a magnetic resonance eddy current prediction method and system. Background Art

[0002] Magnetic resonance imaging (MRI) is one of the most widely used medical imaging technologies in the medical field. In MRI equipment, the eddy currents generated by the gradient need to be eliminated to ensure the quality of the image. Specifically, the eddy currents need to be predicted and eliminated by compensating for them. With the enhancement of gradient performance and the increase of the main magnetic field, the eddy currents generated by the gradient become more complex, and the nonlinearity, non-superposition, asymmetry and other characteristics become more obvious. At present, the traditional compensation method based on the linear time-invariant system is mainly used to eliminate eddy currents, but this method cannot completely eliminate eddy currents, which may cause a series of image problems, such as artifacts, displacement, deformation, etc.

[0003] Therefore, it is hoped to provide a magnetic resonance eddy current prediction method to improve the effect of eddy current prediction and evaluation. Summary of the invention

[0004] One of the embodiments of this specification provides a method for predicting eddy currents in magnetic resonance, which includes: obtaining gradient parameters of a gradient signal of a magnetic resonance system; processing the gradient parameters through an eddy current prediction model to obtain eddy current field components corresponding to the gradient signal, wherein the gradient signal includes a single gradient signal or a combination of gradient signals.

[0005] In some embodiments, the gradient parameters of the single gradient signal may include at least two of the following: gradient amplitude, gradient polarity and platform time; the gradient parameters of the gradient signal combination may include at least two of the following: the number of gradients, the time interval between gradients, the gradient amplitude of each gradient signal, the gradient polarity of each gradient signal, and the platform time of each gradient signal.

[0006] In some embodiments, the gradient parameter of the single gradient signal may further include a climbing rate; and the gradient parameter of the gradient signal combination may further include at least one of the climbing rates of the respective gradient signals.

[0007] In some embodiments, the eddy current prediction model may include at least one layer of deconvolution network.

[0008] In some embodiments, the eddy current prediction model is obtained through training, and the training may include: obtaining multiple gradient parameter samples corresponding to the gradient signal and eddy current field component samples corresponding to the gradient signal, wherein the number of the multiple gradient parameter samples is determined based on the sum of the number of values ​​of all gradient parameters in the multiple gradient parameter samples, and each gradient parameter in the multiple gradient parameter samples takes at least one value within its corresponding value range; training the eddy current prediction model based on the multiple gradient parameter samples and the eddy current field component samples.

[0009] In some embodiments, when the gradient signal includes the gradient signal combination, the gradient parameter samples corresponding to the gradient signal combination may satisfy a preset parameter constraint condition.

[0010] One of the embodiments of the present specification provides a magnetic resonance eddy current prediction system, comprising an acquisition module and a processing module; the acquisition module is used to acquire gradient parameters of a gradient signal of a magnetic resonance system; the processing module is used to process the gradient parameters through an eddy current prediction model to obtain eddy current field components corresponding to the gradient signal, wherein the gradient signal includes a single gradient signal or a gradient signal combination.

[0011] One of the embodiments of the present specification provides a magnetic resonance eddy current prediction device, comprising a processor, wherein the processor is used to execute the magnetic resonance eddy current prediction method.

[0012] One of the embodiments of the present specification provides a computer-readable storage medium, wherein the storage medium stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the magnetic resonance eddy current prediction method.

[0013] One of the embodiments of the present specification provides a method for driving a magnetic resonance system based on eddy current correction. The method includes: obtaining gradient parameters of a gradient signal of a magnetic resonance system; processing the gradient parameters through an eddy current prediction model to obtain an eddy current field component corresponding to the gradient signal, wherein the gradient signal includes a single gradient signal or a gradient signal combination; pre-modulating the gradient signal based on the eddy current field component to obtain a modulated gradient signal; and driving the gradient coil of the magnetic resonance system based on the modulated gradient signal.

[0014] One of the embodiments of the present specification provides a magnetic resonance system driving device based on eddy current correction, comprising a processor, wherein the processor is used to execute the magnetic resonance system driving method based on eddy current correction.

[0015] One of the embodiments of the present specification provides a magnetic resonance imaging system, comprising a gradient coil, wherein the gradient coil is used to generate a gradient field based on a modulated gradient signal, and the system performs magnetic resonance imaging based on the gradient field, wherein the modulated gradient signal is obtained by pre-modulating a gradient signal of the system based on an eddy current field component, and the eddy current field component is obtained by processing a gradient parameter of the gradient signal using an eddy current prediction model.

[0016] One of the embodiments of this specification provides a magnetic resonance imaging method based on eddy current correction. The method includes: obtaining gradient parameters of a gradient signal of a magnetic resonance system; processing the gradient parameters through an eddy current prediction model to obtain an eddy current field component corresponding to the gradient signal, wherein the gradient signal includes a single gradient signal or a gradient signal combination; obtaining a magnetic resonance image, wherein the magnetic resonance image is obtained by imaging the magnetic resonance system based on the gradient signal; and correcting the magnetic resonance image based on the eddy current field component to obtain a corrected magnetic resonance image.

[0017] One of the embodiments of the present specification provides a magnetic resonance imaging device based on eddy current correction, comprising a processor, wherein the processor is used to execute the magnetic resonance imaging method based on eddy current correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0019] Figure 1 is a schematic diagram of an application scenario of a magnetic resonance eddy current prediction system according to some embodiments of this specification;

[0020] Figure 2 is a schematic diagram of a magnetic resonance eddy current prediction system according to some embodiments of the present specification;

[0021] Figure 3 is an exemplary flow chart of a magnetic resonance eddy current prediction method according to some embodiments of this specification;

[0022] Figure 4 is an exemplary flow chart of a magnetic resonance system driving method based on eddy current correction according to some embodiments of this specification;

[0023] Figure 5 is an exemplary flow chart of a magnetic resonance imaging method based on eddy current correction according to some embodiments of the present specification;

[0024] Figure 6 is a schematic diagram of a gradient waveform of an eddy current according to some embodiments of the present specification;

[0025] Figure 7 is a schematic diagram of an eddy current prediction model according to some embodiments of this specification;

[0026] Figure 8 It is a schematic diagram of a method for training an eddy current prediction model according to some embodiments of the present specification. DETAILED DESCRIPTION

[0027] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0028] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0029] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0030] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0031] Figure 1 1 is a schematic diagram of an application scenario of a magnetic resonance eddy current prediction system according to some embodiments of the present specification. In the present specification, the magnetic resonance eddy current prediction system 100 is referred to as the system 100 for short.

[0032] like Figure 1 As shown, in some embodiments, the system 100 may include a medical imaging device 110 , a first computing device 120 , a second computing device 130 , a user terminal 140 , a storage device 150 , and a network 160 .

[0033] The medical imaging device 110 may refer to a device that uses different media to reproduce the internal structure of a target object (e.g., a human body) as an image. In some embodiments, the medical imaging device 110 may be any device that uses magnetic resonance imaging technology to image or treat a designated body part of a target object (e.g., a human body), such as an MRI device. The medical imaging device 110 provided above is for illustrative purposes only and is not intended to limit its scope. In some embodiments, the medical imaging device 110 may send its device parameters (e.g., gradient parameters of a gradient signal) and / or acquired medical images (e.g., magnetic resonance (MRI) images, etc.) to other components of the system 100 (e.g., a first computing device 120, a second computing device 130, a storage device 150). In some embodiments, the medical imaging device 110 may exchange data and / or information with other components in the system 100 via the network 160.

[0034] The first computing device 120 and the second computing device 130 are systems with computing and processing capabilities, and may include various computers, such as servers, personal computers, or computing platforms composed of multiple computers connected in various structures. In some embodiments, the first computing device 120 and the second computing device 130 may be the same device or different devices.

[0035] The first computing device 120 and the second computing device 130 may include one or more sub-processing devices (e.g., a single-core processing device or a multi-core multi-core processing device), and the processing device may execute program instructions. As an example only, the processing device may include various common general-purpose central processing units (CPUs), graphics processing units (GPUs), microprocessors, application-specific integrated circuits (ASICs), or other types of integrated circuits.

[0036] The first computing device 120 can process information and data related to the medical imaging device and / or the medical image. In some embodiments, the first computing device 120 may include an eddy current prediction model, and the first computing device 120 may execute the magnetic resonance eddy current prediction method as shown in some embodiments of this specification, and process the gradient parameters of the gradient signal of the medical imaging device 110 through the eddy current prediction model to obtain the eddy current result (i.e., eddy current signal data) corresponding to the gradient signal. In some embodiments, the first computing device 120 can obtain the trained eddy current prediction model from the second computing device 130. In some embodiments, the first computing preparation 120 can perform system eddy current correction, eddy current correction of reconstructed medical images, etc. on the medical imaging device 110 based on the eddy current field component corresponding to the gradient signal of the medical imaging device 110. In some embodiments, the first computing device 120 can exchange information and data through the network 160 and / or other components in the system 100 (e.g., the medical imaging device 110, the second computing device 130, the user terminal 140, the storage device 150). In some embodiments, the first computing device 120 can be directly connected to the second computing device 130 and exchange information and / or data.

[0037] The second computing device 130 can be used for model training. In some embodiments, the second computing device 130 can execute the training method of the eddy current prediction model as shown in some embodiments of this specification to obtain a trained eddy current prediction model. In some embodiments, the second computing device 130 can obtain gradient parameter samples and eddy current field component samples corresponding to the gradient signal as training samples for training the eddy current prediction model. For example, the gradient parameters and eddy current field components of the gradient signal are obtained from the medical imaging device 110 as training data for the model. In some embodiments, the first computing device 120 and the second computing device 130 can also be the same computing device.

[0038] The user terminal 140 can receive and / or display the processing results of the medical image. In some embodiments, the user terminal 140 can receive the eddy current corrected medical image of the medical imaging device 110 from the first computing device 120, and diagnose and treat the patient based on the medical image. In some embodiments, the user terminal 140 can instruct the first computing device 120 to execute the magnetic resonance eddy current prediction method as shown in some embodiments of this specification. In some embodiments, the user terminal 140 can control the medical imaging device 110 to obtain the medical image of the patient. In some embodiments, the user terminal 140 can be one of the mobile device 140-1, tablet computer 140-2, laptop computer 140-3, desktop computer and other devices with input and / or output functions or any combination thereof.

[0039] The storage device 150 may store data or information generated by other devices. In some embodiments, the storage device 150 may store medical images collected by the medical imaging device 110. In some embodiments, the storage device 150 may store data and / or information processed by the first computing device 120 and / or the second computing device 130, for example, an eddy current prediction model, gradient parameters of a gradient signal, eddy current field components corresponding to a gradient signal, etc. The storage device 150 may include one or more storage components, each of which may be an independent device or part of another device. The storage device may be local or implemented through the cloud.

[0040] The network 160 can connect the components of the system and / or connect the system with external resources. The network 160 enables communication between the components and with other parts outside the system to facilitate the exchange of data and / or information. In some embodiments, one or more components in the system 100 (e.g., the medical imaging device 110, the first computing device 120, the second computing device 130, the user terminal 140, the storage device 150) can send data and / or information to other components via the network 160. In some embodiments, the network 160 can be any one or more of a wired network or a wireless network.

[0041] It should be noted that the above description is provided for illustrative purposes only and is not intended to limit the scope of this specification. For those of ordinary skill in the art, various changes and modifications may be made under the guidance of the contents of this specification. The features, structures, methods and other features of the exemplary embodiments described in this specification may be combined in various ways to obtain additional and / or alternative exemplary embodiments. For example, the first computing device 120 and / or the second computing device 130 may be based on a cloud computing platform, such as a public cloud, a private cloud, a community and a hybrid cloud. However, these changes and modifications will not deviate from the scope of this specification.

[0042] Figure 2 is a schematic diagram of a magnetic resonance eddy current prediction system according to some embodiments of the present specification.

[0043] like Figure 2 As shown, in some embodiments, the magnetic resonance eddy current prediction system 200 may include an acquisition module 210 and a processing module 220. In some embodiments, each module of the magnetic resonance eddy current prediction system 200 may be implemented by the first computing device 120 and the second computing device 130.

[0044] In some embodiments, the acquisition module 210 may be used to acquire gradient parameters of a gradient signal of a magnetic resonance system.

[0045] In some embodiments, the processing module 220 may be used to process the gradient parameters through the eddy current prediction model to obtain the eddy current field component corresponding to the gradient signal, wherein the gradient signal may include a single gradient signal or a combination of gradient signals.

[0046] In some embodiments, the gradient parameters of a single gradient signal may include at least two of the following: gradient amplitude, gradient polarity, and platform time, etc.; the gradient parameters of a gradient signal combination may include at least two of the following: the number of gradients, the time interval between gradients, the gradient amplitude of each gradient signal, the gradient polarity of each gradient signal, the platform time of each gradient signal, etc.

[0047] In some embodiments, the gradient parameter of a single gradient signal may further include a climbing rate; and the gradient parameter of a gradient signal combination may further include at least one of the climbing rates of each gradient signal.

[0048] In some embodiments, the eddy current prediction model may include at least one layer of a deconvolutional network.

[0049] In some embodiments, the magnetic resonance eddy current prediction system 200 may further include a model training module ( Figure 2 (not shown). The model training module can be used to train the eddy current prediction model, and the training may include: obtaining multiple gradient parameter samples corresponding to the gradient signal and eddy current field component samples corresponding to the gradient signal, wherein the number of the multiple gradient parameter samples is determined based on the sum of the number of values ​​of all gradient parameters in the multiple gradient parameter samples, and each gradient parameter in the multiple gradient parameter samples takes at least one value within its corresponding value range; and training the eddy current prediction model based on the multiple gradient parameter samples and the eddy current field component samples.

[0050] In some embodiments, when the gradient signal includes a gradient signal combination, the gradient parameter samples corresponding to the gradient signal combination may satisfy a preset parameter constraint condition.

[0051] In some embodiments, each module in the magnetic resonance eddy current prediction system 200 can be connected to the pre-modulation module and the driving module ( Figure 2 The pre-modulation module can be used to pre-modulate the gradient signal based on the eddy current field component to obtain the modulated gradient signal; the driving module can be used to drive the gradient coil of the magnetic resonance system based on the modulated gradient signal.

[0052] In some embodiments, each module in the magnetic resonance eddy current prediction system 200 can be connected to the image acquisition module and the image correction module ( Figure 2The image acquisition module can be used to acquire a magnetic resonance image, which is obtained by imaging based on a gradient signal by the magnetic resonance system; the image correction module can correct the magnetic resonance image based on the eddy current field component to obtain a corrected magnetic resonance image.

[0053] Figure 3 is an exemplary flow chart of a magnetic resonance eddy current prediction method according to some embodiments of the present specification.

[0054] In traditional eddy current compensation technology, it is usually assumed that the eddy current generated by the gradient is a linear time-invariant system. Then, by measuring the eddy current generated by a gradient with a fixed amplitude and a fixed climbing speed, a model (for example, a multi-exponential convolution model) is established to fit the eddy current, and the model is used to pre-correct all gradients used. The common eddy current multi-exponential convolution model is shown in the following formula: Wherein, G is the gradient, and in some embodiments, G may also represent the gradient strength; G eddy (t) is the eddy current generated by the gradient G(t); e(t) is the multi-exponential response function; τ is the time constant; α is the coefficient of the time constant; Represents convolution.

[0055] As shown in the above formulas (1) and (2), the eddy current can be the convolution of the gradient change rate and the multi-exponential response function. In some embodiments, the multi-exponential response function can be obtained by collecting and analyzing the eddy current of a gradient waveform with a fixed parameter; the eddy current of a gradient waveform with any parameter can be obtained by the convolution of its gradient change rate and the multi-exponential response function.

[0056] Therefore, the traditional eddy current compensation / prediction method based on the multi-exponential convolution model only needs to give a set of gradient waveform parameters. Among them, the parameters are usually the gradient intensity and the climbing rate, or the gradient intensity and the climbing time. Because the gradient change rate needs to be obtained to solve the multi-exponential response function, there is generally no specific requirement for this parameter.

[0057] In an improvement to the traditional method, a nonlinear function is added to the traditional eddy current correction, that is, α in the above formula (2) is no longer a fixed value, but is related to the gradient intensity, to solve the compensation problem of nonlinear eddy current. However, the form of the nonlinear function is given in advance and relies on manual experience. At the same time, it cannot solve other problems, such as non-superposition and asymmetry.

[0058] In another improvement on the traditional method, magnetic resonance eddy current compensation is performed based on a neural network model. The input of the neural network model is the result of the eddy current, and the output is the time constant and weight of the eddy current. Therefore, it can be regarded as extracting eigenvalues ​​from the data. This method still uses a multi-exponential convolution model, but only improves the calculation method of solving the time constant and amplitude coefficient in the multi-exponential response function from the eddy current data measurement results from the traditional method to deep learning to increase the calculation speed. However, this method does not take into account the nonlinearity and asymmetry of the eddy current with respect to the gradient, and therefore cannot solve the corresponding problem.

[0059] In some embodiments, the method shown in the execution process 300 can be used to measure the eddy currents generated by the gradient or gradient combination of different random parameters (e.g., gradient amplitude, gradient polarity, climbing rate, platform time, time interval between gradients, number of gradients, etc.), and the parameters of the gradient or gradient combination can be used as the input of the eddy current prediction model, and the predicted eddy current field components can be output, so that the eddy current of the gradient or gradient combination under any parameter combination can be predicted. Among them, since the input of the eddy current prediction model is the parameters of the gradient waveform and the output is the result of the eddy current, it can be regarded as converting the characteristic value into the input. In some embodiments, the eddy current prediction model can include a deconvolutional network.

[0060] like Figure 3 As shown, the process 300 includes the following steps: In some embodiments, the process 300 may be executed by the first computing device 120 .

[0061] Step 310 , obtaining gradient parameters of the gradient signal of the magnetic resonance system. In some embodiments, step 310 may be performed by the obtaining module 210 .

[0062] A magnetic resonance system is a medical imaging system using magnetic resonance imaging technology, such as an MRI device. A gradient signal is a magnetic resonance signal generated by a gradient magnetic field change in a magnetic resonance system, i.e., a gradient echo signal, which can be represented by a gradient waveform diagram. In some embodiments, the gradient signal may include a single gradient signal, a gradient signal combination, etc., wherein the gradient signal combination is a set of multiple single gradient signals. A gradient parameter is a parameter that can represent a gradient signal, such as a gradient amplitude, a gradient polarity, a platform time, a climbing rate, the number of gradients, the time interval between gradients, the gradient amplitude of each gradient signal, the gradient polarity of each gradient signal, the platform time of each gradient signal, etc.

[0063] In some embodiments, the gradient parameters may include at least two. For example, for a single gradient signal, its gradient parameters may include at least two of the following: gradient amplitude, gradient polarity, and platform time, etc. For another example, for a gradient signal combination, its gradient parameters may include at least two of the following: number of gradients, time interval between gradients, gradient amplitude of each gradient signal, gradient polarity of each gradient signal, platform time of each gradient signal, etc.

[0064] In some embodiments, the gradient parameters of a single gradient signal may further include a climbing rate, and the gradient parameters of a gradient signal combination may further include at least one of the climbing rates of each gradient signal, that is, each gradient signal in the gradient signal combination has its own climbing rate, and the gradient parameters of the gradient signal combination may include the climbing rates of one or more gradient signals in the combination.

[0065] like Figure 6 In the gradient waveform diagram of the eddy current shown, the gradient waveform includes a positive polarity gradient and a negative polarity line gradient, both of which are gradient signals and can be used as a gradient signal combination. The gradient amplitude, the rising edge climbing time, the falling edge climbing time, and the platform time can be the gradient parameters of the positive polarity gradient. The gradient amplitude, the rising edge climbing time, and the falling edge climbing time can be used to calculate the climbing rate, for example, climbing rate = gradient amplitude / rising edge climbing time. The gradient interval is the interval time between the positive polarity gradient and the negative polarity line gradient, which can be used as a gradient parameter of the gradient signal combination composed of the positive polarity gradient and the negative polarity line gradient.

[0066] In some embodiments, the first computing device 120 may obtain the gradient parameters of the gradient signal of the magnetic resonance system (e.g., the medical imaging device 110), for example, by obtaining the gradient signal when the magnetic resonance system is working, and then extracting the gradient parameters therefrom. In some embodiments, the first computing device 120 may obtain the gradient parameters of the gradient signal of the magnetic resonance system by other means, for example, by obtaining the historical scanning data of the magnetic resonance system from a storage device (e.g., the storage device 150), and obtaining the gradient parameters of the gradient signal therefrom.

[0067] Step 320 , processing the gradient parameters through the eddy current prediction model to obtain the eddy current field component corresponding to the gradient signal. In some embodiments, step 320 may be performed by the processing module 220 .

[0068] In some embodiments, after obtaining the gradient parameters of the gradient signal of the magnetic resonance system, the first computing device 120 can input the gradient parameters into the eddy current prediction model to obtain the eddy current field component corresponding to the output gradient signal, that is, the eddy current result generated by the gradient signal corresponding to the gradient parameter.

[0069] In some embodiments, the eddy current prediction model may include various machine learning models, such as a convolutional neural network (CNN), a recurrent neural network (RNN), etc.

[0070] In some embodiments, the eddy current prediction model may be a CNN, including at least one layer of deconvolution network, wherein the deconvolution network may be used to perform data processing on the input gradient parameters to obtain eddy current field components.

[0071] In some embodiments, the gradient parameters input to the eddy current prediction model may include at least two parameters, and the first computing device 120 may input these parameters into at least one layer of deconvolution network in the eddy current prediction model, and perform data processing in sequence through each layer of the at least one layer of deconvolution network to obtain eddy current field components. The data processing of each layer of the deconvolution network may include: performing eigenvalue conversion processing, eigenvalue de-pooling processing, and de-pooling feature deconvolution processing on the input data in sequence. Among them, if the current deconvolution network is the first layer, the input data is the input gradient parameter; if the current deconvolution network is the Mth layer (M>1), the input data is the feature combination value of multiple features output by the previous layer of deconvolution network.

[0072] The following Figure 7 Take eddy current prediction model as an example to illustrate the data processing flow. Figure 7 is a schematic diagram of an eddy current prediction model according to some embodiments of this specification. Figure 7 As shown, the model 700 includes two layers of deconvolution networks, wherein the first layer of deconvolution networks includes a pooling feature layer 710 and a reverse pooling feature layer 720, and the second layer of deconvolution networks includes a pooling feature layer 730 and a reverse pooling feature layer 740. The first computing device 120 can input the gradient parameters 711-1, 711-2, 711-3, ..., 711-N (N>1) into the pooling feature layer 710 in the first layer of deconvolution networks, and convert the above gradient parameters into feature values ​​in this layer; then input these feature values ​​into the reverse pooling feature layer 720 in the first layer of deconvolution networks, perform reverse pooling processing, and obtain the reverse pooling features corresponding to each gradient parameter, namely, reverse pooling features 721-1, 721-2, 721-3, ..., 721-N; then perform deconvolution processing on these reverse pooling features, and sum the obtained values ​​pairwise, Figure 7In the example, the deconvolution values ​​of the de-pooled features 721-1 and 721-3 are summed, the deconvolution values ​​of the de-pooled features 721-2 and 721-N are summed, and so on, thereby obtaining the feature combination values ​​of the multiple features. After obtaining the above feature combination values, the first computing device 120 can input these feature combination values ​​into the pooling feature layer 730 in the second layer deconvolution network, such as Figure 7 As shown, two feature combination values ​​are outputted in the first layer of deconvolution network, which are processed by the pooling feature layer 730 to obtain corresponding pooling features 731-1 and 731-2, and then 731-1 and 731-2 are input into the anti-pooling feature layer 740 in the second layer of deconvolution network, and corresponding anti-pooling features 741-1 and 741-2 are obtained after anti-pooling processing. After combining 741-1 and 741-2, the eddy current result 750 outputted by the second layer of deconvolution network is obtained, and the eddy current result 750 is the eddy current field component generated by the gradient signal corresponding to the input gradient parameters 711-1, 711-2, 711-3, ..., 711-N predicted by the model 700.

[0073] In some embodiments, at least one of the number of layers of the deconvolution network and the number of summed de-pooling features in the eddy current prediction model can be adjusted. In some embodiments, the eddy current prediction model can be obtained through training, and when training is performed using training data, the optimal deconvolution kernel is actually solved.

[0074] In some embodiments, the second computing device 130 can use multiple gradient parameter samples corresponding to the gradient signal of the magnetic resonance system and eddy current field component samples corresponding to the gradient signal to train the eddy current prediction model, wherein the eddy current field component samples can be used as training labels. The second computing device 130 can input multiple gradient parameter samples into an untrained initial eddy current prediction model to obtain output predicted eddy current field components, compare the predicted eddy current field components with the eddy current field component samples, adjust the parameters of the eddy current prediction model based on the comparison results, and iterate the above operations until the difference between the output predicted eddy current field components and the eddy current field component samples is less than a preset threshold, and use the eddy current prediction model at this time as a trained eddy current prediction model. For more information on how to train the eddy current prediction model, please visit Figure 8 The relevant description will not be repeated here.

[0075] In some embodiments, for a gradient signal combination composed of gradient signals of different gradient axes, the first computing device 120 may separately predict the gradient signal of each gradient axis, and combine the prediction results on all gradient axes to form a final eddy current prediction result, i.e., the eddy current field component corresponding to the above gradient signal combination.

[0076] In some embodiments of the present specification, by inputting the parameters of the gradient or gradient combination into the machine learning model, the output eddy current field component is obtained, so that the eddy current generated by the gradient can be predicted by artificial intelligence, which well solves the problem of prediction and evaluation of eddy currents, especially complex eddy currents (for example, problems such as nonlinearity, non-superposition, and asymmetry of the gradient), and can obtain more accurate eddy current prediction results.

[0077] Figure 4 is an exemplary flow chart of a magnetic resonance system driving method based on eddy current correction according to some embodiments of the present specification.

[0078] like Figure 4 As shown, the process 400 includes the following steps: In some embodiments, the process 400 may be executed by the first computing device 120 .

[0079] Step 410, obtaining the gradient parameters of the gradient signal of the magnetic resonance system. In some embodiments, step 410 may be performed by the acquisition module 210. Step 410 may be the same as step 310, and will not be described in detail here.

[0080] Step 420, the gradient parameters are processed by the eddy current prediction model to obtain the eddy current field component corresponding to the gradient signal. In some embodiments, step 420 may be performed by the processing module 220. Step 420 may be the same as step 320, and will not be described in detail here.

[0081] Step 430, pre-modulate the gradient signal based on the eddy current field component to obtain a modulated gradient signal. In some embodiments, step 430 may be performed by a pre-modulation module.

[0082] In some embodiments, the first computing device 120 performs pre-modulation of the gradient signal of the magnetic resonance system according to the eddy current field component obtained in the above steps to obtain a modulated gradient signal. The gradient field generated by the modulated gradient signal can offset the eddy current generated by it. In some embodiments, the gradient signal of the magnetic resonance system may include an ideal gradient signal of the system, that is, the scanning sequence effect using the gradient signal is optimal in theory or in practice, wherein the ideal gradient signal can be determined by various methods such as calculation or based on manual experience.

[0083] Step 440, driving the gradient coil of the magnetic resonance system based on the modulated gradient signal. In some embodiments, step 440 may be performed by a driving module.

[0084] In some embodiments, the first computing device 120 may send the modulated gradient signal to the magnetic resonance system, so that the magnetic resonance system may amplify the modulated gradient signal through its gradient power amplifier, and then use the amplified gradient signal to drive the gradient coil to generate a corresponding gradient field.

[0085] In some embodiments of the present specification, an eddy current field component corresponding to a gradient signal is generated based on the gradient parameters of the gradient signal of a magnetic resonance system through a machine learning model, and then a premodulated gradient signal is generated according to the eddy current field component. The gradient coil is driven to generate a gradient field based on the premodulated gradient signal. Since the gradient field generated by the premodulated gradient signal can offset the eddy current generated by it, the gradient field finally generated is basically consistent with the desired ideal gradient field, thereby greatly improving and ensuring the effect of the system eddy current correction.

[0086] Figure 5 is an exemplary flow chart of a magnetic resonance imaging method based on eddy current correction according to some embodiments of the present specification.

[0087] like Figure 5 As shown, the process 500 includes the following steps. In some embodiments, the process 500 may be executed by the first computing device 120 .

[0088] Step 510, obtaining the gradient parameters of the gradient signal of the magnetic resonance system. In some embodiments, step 510 may be performed by the acquisition module 210. Step 510 may be the same as step 310, and will not be described in detail here.

[0089] Step 520, the gradient parameters are processed by the eddy current prediction model to obtain the eddy current field component corresponding to the gradient signal. In some embodiments, step 520 may be performed by the processing module 220. Step 520 may be the same as step 320, and will not be described in detail here.

[0090] Step 530, acquiring a magnetic resonance image. The magnetic resonance image may be acquired by a magnetic resonance system based on a gradient signal. In some embodiments, step 530 may be performed by an image acquisition module.

[0091] In some embodiments, the first computing device 120 may generate a gradient field using the gradient signal in step 510 to scan the scan object (e.g., a biological body, a phantom, air, etc.), and then use the scan data to perform image reconstruction to obtain an original magnetic resonance image without eddy current compensation. In some embodiments, the first computing device 120 may also obtain the magnetic resonance image without eddy current compensation in other ways, for example, from a storage device (e.g., storage device 150).

[0092] Step 540, correcting the magnetic resonance image based on the eddy current field component to obtain a corrected magnetic resonance image. In some embodiments, step 540 may be performed by an image correction module.

[0093] In some embodiments, the first computing device 120 can calculate the impact result on the reconstructed magnetic resonance image (for example, the position offset of the pixel point, the deviation of the pixel value, etc.) based on the eddy current field component, and then reversely compensate the impact result (for example, position correction, pixel value correction, etc.) to the magnetic resonance image without eddy current compensation obtained in step 630, thereby obtaining a corrected magnetic resonance image.

[0094] In some embodiments of the present specification, an eddy current field component corresponding to a gradient signal is generated based on the gradient parameters of the gradient signal of a magnetic resonance system through a machine learning model, the influence on the reconstructed original magnetic resonance image is determined based on the eddy current field component, and eddy current correction is performed on the original magnetic resonance image based on the influence, thereby eliminating the influence of the eddy current on the reconstructed magnetic resonance image and greatly improving the quality of the reconstructed magnetic resonance image.

[0095] Figure 8 It is a schematic diagram of a method for training an eddy current prediction model according to some embodiments of the present specification.

[0096] In some embodiments, process 800 may be executed by the second computing device 130 or the model training module. By executing the steps shown in process 800, the training process of the eddy current prediction model described in processes 300, 400, and 500 may be implemented. The following description will be made by taking the second computing device 130 executing process 800 as an example.

[0097] In some embodiments, the second computing device 130 may acquire multiple gradient parameter samples corresponding to the gradient signal of the magnetic resonance system and eddy current field component samples corresponding to the gradient signal. The number of the multiple gradient parameter samples may be determined based on the sum of the number of values ​​of all gradient parameters in the multiple gradient parameter samples, and each gradient parameter in the multiple gradient parameter samples may take at least one value within its corresponding value range. In some embodiments, the magnetic resonance system here may be the same as the magnetic resonance system in processes 300, 400, and 500.

[0098] like Figure 8 As shown, in some embodiments, the second computing device 130 may determine the gradient parameters of the gradient signal that need to be used as the input parameters of the eddy current prediction model, that is, the gradient parameters 810. The gradient signal may include a single gradient signal or a combination of gradient signals, and the gradient parameters may include at least two of the gradient amplitude, gradient polarity, climbing rate, platform time, number of gradients, time interval between gradients, gradient amplitude of each gradient signal, gradient polarity of each gradient signal, platform time of each gradient signal, etc.

[0099] like Figure 8As shown, in some embodiments, the second computing device 130 may determine a range of each gradient parameter, ie, a parameter range 820 , based on the gradient parameter 810 .

[0100] In some embodiments, the second computing device 130 may determine the range of the gradient parameter according to at least one of the performance limitation of the magnetic resonance system and the range of use of the sequence gradient. For example, the minimum and maximum values ​​of the gradient amplitude may be determined according to the negative and positive values ​​of the maximum gradient intensity that can be achieved when the gradient system of the magnetic resonance system is at maximum performance. For another example, since the number of bipolar diffusion gradients is 4, the maximum value of the number of gradients may be set to 4.

[0101] like Figure 8 As shown, in some embodiments, the second computing device 130 may determine the number of training data of the eddy current prediction model based on the gradient parameter 810 and the parameter range 820, that is, the number of training data 830. In some embodiments, the more parameters the gradient parameter 810 includes and the larger the parameter range 820 is, the larger the number of training data 830 may be.

[0102] In some embodiments, assuming that the number of training data of the eddy current prediction model is recorded as N, then N=∑Ni, where i represents the i-th gradient parameter, and Ni represents that the i-th gradient waveform parameter selects Ni values ​​as parameters of the training data within its value range. In some embodiments, the second computing device 130 can determine Ni according to the range of the i-th gradient waveform parameter. Ni cannot be too small, because it is necessary to cover the range of the i-th gradient waveform parameter to avoid overfitting caused by too little data; Ni cannot be too large to improve data acquisition efficiency. For example, when the gradient intensity range is -40mT / m to 40mT / m, if the average division method is used, it is relatively appropriate to use 5mT to 10mT as a step, 20mT / m as a step, then Ni is too small, and 1mT / m is too large. In some embodiments, the second computing device 130 can determine the Ni of all gradient parameters, and then sum all Ni to obtain the number of training data N.

[0103] like Figure 8 As shown, in some embodiments, the second computing device 130 can select training data according to the gradient parameter 810, the parameter range 820 and the number of training data 830 to obtain a gradient parameter combination 840, wherein the number of gradient parameter combinations included in the gradient parameter combination 840 can be the same as the number of training data 830.

[0104] In some embodiments, the second computing device 130 may select training data for each gradient parameter included in the gradient parameter 810 within its value range (obtained according to the parameter range 802) in various ways (e.g., random, equal division, etc.), i.e., select Ni values, thereby obtaining N groups of gradient parameter combinations, i.e., gradient parameter combination 840.

[0105] like Figure 8 As shown, in some embodiments, the second computing device 130 can measure the eddy current result of the gradient combination corresponding to the gradient parameters of each group in the gradient parameter combination 840 through the magnetic resonance system, thereby obtaining the eddy current result 850. The eddy current result is the eddy current field component, which has two dimensions of space and time, that is, the eddy current result has a distribution in space, and attenuates or / and oscillates with time.

[0106] In some embodiments, the second computing device 130 may use multiple gradient magnetic field signal samples and eddy current field component samples as training samples, and train the eddy current prediction model based on the multiple gradient magnetic field signal samples and eddy current field component samples. Specifically, the second computing device 130 may use multiple gradient magnetic field signal samples as model inputs, and use eddy current field component sample training as training labels to train the initial eddy current prediction model, thereby obtaining a trained eddy current prediction model. For more information on how to use training data to train the eddy current prediction model, please refer to the relevant description in step 320, which will not be repeated here.

[0107] like Figure 8 As shown, in some embodiments, the second computing device 130 can use the gradient parameter combination 840 and its corresponding eddy current result 850 as training samples, input them into the initial eddy current prediction model 860, and obtain a trained eddy current prediction model 870 after training.

[0108] For some special scanning sequences, in addition to the conventional gradients used for imaging, there are also diffusion gradients for diffusing water molecules, such as diffusion-weighted imaging (DWI) and diffusion tensor imaging (DTI). For the special gradients in these special scanning sequences, some parameters in the gradient combination are special. For example, three gradients are combined to form a gradient combination that does not affect imaging but can produce a diffusion effect. For this gradient combination, the amplitudes of the three gradients are usually the same, and they are all the maximum values ​​within the range of gradient amplitude values. For another example, for the aforementioned gradient combination, in order not to affect the imaging effect, the sum of the first-order moments of the gradients must be 0. The first-order moment of the gradient is defined as the integral of the gradient intensity over time, as shown in the following formula: M(G)=∫Gdt (3) Where G is the gradient. Since the gradient can usually be expressed by the gradient strength, the gradient here can be equivalent to the gradient strength; M(G) is the first-order moment of the gradient G; and t is the time.

[0109] In some embodiments, when the gradient signal includes a gradient signal combination, the gradient parameter sample corresponding to the gradient signal combination (e.g., gradient parameter combination 840) may satisfy a preset parameter constraint. The preset constraint may include at least one, for example, the sum of the first-order moments of the gradient is 0, the amplitude of each gradient is the maximum value within the value range, etc. In some embodiments, the second computing device 130 may constrain some or all parameters using the preset constraint; measure the eddy current of the gradient combination using the constrained gradient parameter samples; and train the eddy current prediction model using these gradient parameter samples and eddy current data. The trained eddy current prediction model may be applicable to the eddy current prediction of any gradient combination that satisfies the preset constraint.

[0110] In some embodiments, after obtaining the trained eddy current prediction model 870, the first computing device 120 can input the gradient parameters corresponding to the gradient signal when the magnetic resonance system is working normally (for example, scanning a patient) into the eddy current prediction model 870 to obtain the predicted eddy current field components corresponding to the gradient signal, and perform system eddy current correction, eddy current correction of the reconstructed original magnetic resonance image, and other operations based on the eddy current field components.

[0111] It should be noted that the above descriptions of processes 300, 400, 500 and 800 are only for illustration and explanation, and do not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to processes 300, 400, 500 and 800 under the guidance of this specification. However, these modifications and changes are still within the scope of this specification. For example, an image can be reconstructed based on the scan data obtained after step 440, and then the reconstructed image and the corrected magnetic resonance image obtained after step 540 are fused to obtain the final reconstructed image.

[0112] The beneficial effects that may be brought about by the embodiments of the present specification include but are not limited to: (1) by inputting the parameters of the gradient or gradient combination into the machine learning model, the output eddy current field component is obtained, so that the eddy current generated by the gradient can be predicted by artificial intelligence, which well solves the problem of prediction and evaluation of eddy currents, especially complex eddy currents (for example, the nonlinearity, non-superposition, asymmetry and other problems of the gradient), and can obtain more accurate eddy current prediction results, and improve the effects of subsequent system eddy current correction, eddy current correction of the reconstructed original magnetic resonance image and other operations, thereby improving and ensuring the quality of the reconstructed image; (2) by training the machine learning model with a variety of gradient parameters, the adaptability of the machine learning model to different gradient parameters is improved, and more accurate and comprehensive prediction results of the eddy current field component can be obtained; (3) by constraining the gradient parameter samples with preset constraints, and then using the constrained gradient parameter samples and their corresponding eddy current data to train the machine learning model, for some special gradients in special scanning sequences, the trained machine learning model can be applied to the eddy current prediction of any gradient combination that meets the preset constraints, thereby improving the adaptability of the trained machine learning model. It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects that may be produced may be any one or a combination of the above, or any other beneficial effects that may be obtained.

[0113] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0114] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0115] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0116] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0117] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0118] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0119] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A magnetic resonance eddy current prediction method, comprising: Acquiring gradient parameters of a gradient signal of a magnetic resonance system; The gradient parameters are processed by an eddy current prediction model to determine an eddy current field component corresponding to the gradient signal, wherein the gradient signal includes a single gradient signal or a combination of gradient signals.

2. The method as claimed in claim 1, wherein the gradient parameters of the single gradient signal include at least two of the following: gradient amplitude, gradient polarity and platform time; and the gradient parameters of the gradient signal combination include at least two of the following: number of gradients, time interval between gradients, gradient amplitude of each gradient signal, gradient polarity of each gradient signal, and platform time of each gradient signal.

3. The method according to claim 2, wherein the gradient parameters of the single gradient signal further include a climbing rate; and the gradient parameters of the gradient signal combination further include a climbing rate of at least one gradient signal.

4. The method according to claim 1, wherein the eddy current prediction model comprises at least one layer of deconvolution network.

5. The method according to claim 1, wherein the eddy current prediction model is obtained by training, and the training comprises: Acquire a plurality of gradient parameter samples corresponding to the gradient signal and eddy current field component samples corresponding to the gradient signal, wherein the number of the plurality of gradient parameter samples is determined based on the sum of the number of values ​​of all gradient parameters in the plurality of gradient parameter samples, and each gradient parameter in the plurality of gradient parameter samples takes at least one value within its corresponding value range; The eddy current prediction model is trained based on the plurality of gradient parameter samples and the eddy current field component samples. 6 . The method according to claim 5 , wherein when the gradient signal comprises the gradient signal combination, the gradient parameter samples corresponding to the gradient signal combination satisfy preset parameter constraints.

7. A magnetic resonance eddy current prediction system, comprising an acquisition module and a processing module; The acquisition module is used to acquire the gradient parameters of the gradient signal of the magnetic resonance system; The processing module is used to process the gradient parameter through the eddy current prediction model to obtain the eddy current field component corresponding to the gradient signal, wherein: The gradient signal includes a single gradient signal or a gradient signal combination.

8. A method for driving a magnetic resonance system based on eddy current correction, comprising: Acquiring gradient parameters of a gradient signal of a magnetic resonance system; Processing the gradient parameter through an eddy current prediction model to obtain an eddy current field component corresponding to the gradient signal, wherein the gradient signal includes a single gradient signal or a combination of gradient signals; Pre-modulating the gradient signal based on the eddy current field component to obtain a modulated gradient signal; The gradient coil of the magnetic resonance system is driven based on the modulated gradient signal.

9. A magnetic resonance imaging system, comprising a gradient coil, wherein the gradient coil is used to generate a gradient field based on a modulated gradient signal, and the system performs magnetic resonance imaging based on the gradient field, wherein the modulated gradient signal is obtained by pre-modulating a gradient signal of the system based on an eddy current field component, and the eddy current field component is obtained by processing a gradient parameter of the gradient signal using an eddy current prediction model.

10. A magnetic resonance imaging method based on eddy current correction, comprising: Acquiring gradient parameters of a gradient signal of a magnetic resonance system; Processing the gradient parameter through an eddy current prediction model to obtain an eddy current field component corresponding to the gradient signal, wherein the gradient signal includes a single gradient signal or a combination of gradient signals; Acquiring a magnetic resonance image, wherein the magnetic resonance image is obtained by imaging by the magnetic resonance system based on the gradient signal; The magnetic resonance image is corrected based on the eddy current field component to obtain a corrected magnetic resonance image.

Images