CN115690253A - Magnetic resonance image reconstruction method and image reconstruction device - Google Patents

Abstract

The present invention relates to the technical field of magnetic resonance imaging, in particular to a magnetic resonance image reconstruction method and an image reconstruction device. The present invention first obtains the deep neural network of the image structure information of the depth image, obtains the underlying image, the background phase, the coil sensitivity of the target object to the magnetic field coil, and the conjugate sensitivity of the coil sensitivity, and then analyzes the information received by the magnetic resonance equipment from the The signal of the target object is sampled to generate a sampling template signal, and then according to the information of each magnetic field applied in the sampling environment, the magnetic field phase difference formed by each magnetic field is obtained, and finally according to the underlying image, coil sensitivity, conjugate sensitivity, and background phase , sampling the template signal and the phase difference of the magnetic field, and reconstructing the magnetic resonance image of the target object. The present invention does not involve low-frequency ACS signals in the process of calculating the coil sensitivity, thus improving the quality of the reconstructed image of the present invention. It also improves the speed of image reconstruction by the present invention.

Description

A magnetic resonance image reconstruction method and image reconstruction device

technical field

The present invention relates to the technical field of magnetic resonance imaging, in particular to a magnetic resonance image reconstruction method and an image reconstruction device.

Background technique

The scanning speed of Magnetic Resonance Imaging (MRI, Magnetic Resonance Imaging) is slow. Excessively long scanning time will not only cause discomfort to the patient, but also easily introduce motion artifacts into the image, thereby affecting the image quality. Magnetic resonance parallel imaging method is a kind of method to accelerate MRI scanning speed, such as sensitivity encoding technology (SENSE, sensitivity encoding) and generalized autocalibrating partially parallel acquisitions technology (GRAPPA, generalized autocalibrating partially parallel acquisitions), etc. This type of method achieves the purpose of fast scanning by reducing the amount of collected data and using the redundant information contained in the multi-channel coil to reconstruct the under-sampled data.

Magnetic resonance imaging includes wave gradient field coding parallel imaging technology, Wave coding imaging technology, virtual conjugate coil (VCC) imaging technology.

Among them, wave gradient field encoding parallel imaging technology (Wave encoding) is a parallel imaging technology used to speed up the scanning speed of magnetic resonance. Coil, VCC) is a parallel imaging technology with similar effects to wave coding, which can provide more channels of spatial coding prior information. The combination of Wave-VCC can play the characteristics of both and provide a higher multiple of acceleration technology, but In the conventional Wave-VCC reconstruction method, only the low-frequency ACS signal in the middle of the k-space is used to estimate the background phase. Due to the lack of surrounding high-frequency information, the estimated background phase is difficult to represent the image of high-frequency phase changes.

的正弦梯度场，并采用可控混叠的快速并行成像技术(2D CAIPIRINHA,two-dimension Controlled Aliasing InParallel Imaging Results In Higher Acceleration)对数据进行欠采，使得欠采样所导致的混叠伪影沿读出、选层和相位方向进行分散，降低各像素点中图像混叠伪影的程度，从而极大的降低了并行成像重建中的几何因子(g-factor,geometry factor)信噪比丢失，达到高倍加速的目的。Wave encoding technology is a parallel imaging technology used to speed up 3D MRI scans. This technology uses MRI gradient coils to apply phases in the layer selection and phase directions while the MRI signal is being acquired (while applying the readout gradient field). difference, using the MRI gradient coil to apply the phase difference in the phase encoding direction as

The sinusoidal gradient field, and the fast parallel imaging technology with controllable aliasing (2D CAIPIRINHA, two-dimension Controlled Aliasing InParallel Imaging Results In Higher Acceleration) is used to undersample the data, so that the aliasing artifacts caused by undersampling are read along The output, layer selection and phase directions are dispersed to reduce the degree of image aliasing artifacts in each pixel, thereby greatly reducing the loss of the signal-to-noise ratio of the geometric factor (g-factor, geometry factor) in parallel imaging reconstruction, reaching The purpose of high-speed acceleration.

Virtual Conjugate Coil (VCC) is another technique to improve the condition of coded matrix systems in parallel imaging. The idea is to incorporate object background and coil phases into the reconstruction process, providing additional encoding capability by adding virtual coils generated from conjugate symmetric k-space signals from real physical coils.

The existing technology needs to collect low-frequency auto-calibration signals (ACS) signals, and then calculate the high-frequency coil sensitivity according to the low-frequency ACS signals, and in the process of collecting low-frequency ACS signals, memory introduces motion errors , resulting in a large error in the calculated coil sensitivity, which in turn leads to poor quality of the reconstructed image based on the coil sensitivity.

To sum up, the quality of the reconstructed image calculated by the prior art is poor.

Therefore, the prior art still needs to be improved and improved.

Contents of the invention

In order to solve the above technical problems, the present invention provides a magnetic resonance image reconstruction method and an image reconstruction device, which solve the problem of poor quality of reconstructed images calculated in the prior art.

To achieve the above object, the present invention adopts the following technical solutions:

In a first aspect, the present invention provides a magnetic resonance image reconstruction method, which includes:

Apply a network structure composed of several neural networks to the image structure information of the depth image, and obtain the underlying image output by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, and the conjugate sensitivity of the coil sensitivity degree, the depth image is used to characterize the depth information of the target object relative to the magnetic resonance equipment, and the magnetic field coil is a coil inside the magnetic resonance equipment;

Sampling the signal received by the magnetic resonance equipment from the target object to generate a sampling template signal;

According to the information of each magnetic field applied in the sampling environment, the phase difference of the magnetic field formed by each magnetic field is obtained;

The magnetic resonance image of the target object is reconstructed according to the underlying image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal and the magnetic field phase difference.

In one implementation, the network structure composed of several neural networks is applied to the image structure information of the depth image to obtain the underlying image output by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, The conjugate sensitivity of the coil sensitivity, the depth image is used to characterize the depth information of the target object relative to the magnetic resonance equipment, and the magnetic field coil is a coil inside the magnetic resonance equipment, including:

Applying a first deep convolutional neural network with a decoding structure to the image structure information of the depth image to obtain an underlying image output by the first deep convolutional neural network;

applying a second deep convolutional neural network to the image structure information of the depth image to obtain a background phase output by the second deep convolutional neural network;

applying a third deep convolutional neural network to the image structure information of the depth image to obtain the coil sensitivity output by the third deep convolutional neural network;

A fourth deep convolutional neural network is applied to the image structure information of the depth image to obtain a conjugate sensitivity output by the fourth deep convolutional neural network.

In an implementation manner, the sampling the signal received by the magnetic resonance equipment from the target object to generate a sampling template signal includes:

Sampling the signal received by the magnetic resonance equipment from the target object using a three-dimensional MRI sequence to obtain a sampling signal in the layer selection direction and a sampling signal in the phase direction;

A sampling template signal is generated according to the sampling signal in the layer selection direction and the sampling signal in the phase direction.

In an implementation manner, while sampling the signal received by the magnetic resonance equipment from the target object using a three-dimensional MRI sequence, a sinusoidal gradient field is applied in the layer selection direction, and a truncation is applied in the phase direction. or a truncated sinusoidal gradient field is applied in the layer selection direction, and a sinusoidal gradient field is applied in the phase direction.

In an implementation manner, the 0th moment of the truncated sinusoidal gradient field is zero.

In an implementation manner, the obtaining the magnetic field phase difference formed by each magnetic field according to the information of each magnetic field applied in the sampling environment includes:

A magnetic field phase difference between the sinusoidal gradient field and the truncated sinusoidal gradient field is calculated according to the magnetic field phase of the sinusoidal gradient field and the magnetic field phase of the truncated sinusoidal gradient field.

In an implementation manner, the reconstruction of the target object is based on the underlying image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal, and the magnetic field phase difference MRI images, including:

multiplying the underlying image by the coil sensitivity to obtain a first result;

applying a Fourier transform to the first result to obtain a second result;

multiplying the second result by the magnetic field phase difference to obtain a third result;

applying an inverse Fourier transform to the third result to obtain a fourth result;

multiplying the fourth result by the sampling template signal to obtain a target signal;

multiplying the background phase, the conjugate sensitivity, and the underlying image to obtain a fifth result;

applying a Fourier transform to said fifth result to obtain a sixth result;

multiplying the sixth result by the magnetic field phase difference to obtain a seventh result;

applying an inverse Fourier transform to the seventh result to obtain an eighth result;

multiplying the eighth result by the sampling template signal to obtain a conjugate symmetric signal of the target signal;

A magnetic resonance image of the target object is reconstructed according to the target signal and the conjugate symmetry signal.

In the second aspect, the embodiment of the present invention also provides a magnetic resonance image reconstruction device, wherein the device includes the following components:

The information analysis module is used to apply a network structure composed of several neural networks to the image structure information of the depth image, and obtain the underlying image output by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, the coil The conjugate sensitivity of the sensitivity, the depth image is used to represent the depth information of the target object relative to the magnetic resonance equipment, and the magnetic field coil is a coil inside the magnetic resonance equipment;

The signal adopting module is used to sample the signal from the target object received by the magnetic resonance equipment, and generate a sampling template signal;

The phase difference calculation module is used to obtain the magnetic field phase difference formed by each magnetic field according to the information of each magnetic field applied in the sampling environment;

An image reconstruction module, configured to reconstruct the magnetic resonance of the target object according to the underlying image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal, and the magnetic field phase difference image.

In the third aspect, an embodiment of the present invention further provides a terminal device, wherein the terminal device includes a memory, a processor, and a magnetic resonance image reconstruction program stored in the memory and operable on the processor. When the processor executes the magnetic resonance image reconstruction program, the steps of the magnetic resonance image reconstruction method described above are implemented.

In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, where a magnetic resonance image reconstruction program is stored on the computer-readable storage medium, and when the magnetic resonance image reconstruction program is executed by a processor, the above-mentioned The steps of the magnetic resonance image reconstruction method described above.

Beneficial effects: the underlying image, background phase, coil sensitivity, and conjugate sensitivity required for image reconstruction in the present invention all come from the neural network. Since the present invention does not involve low-frequency ACS signals in the process of calculating the coil sensitivity, it improves The quality of the reconstructed image of the present invention is improved. In addition, since the underlying image of the present invention is also calculated by the neural network, the speed of image reconstruction in the present invention is improved.

Description of drawings

Fig. 1 is the overall flowchart of the present invention;

Fig. 2 is a structural diagram of a deep convolutional neural network in an embodiment of the present invention;

FIG. 3 is a schematic diagram of a data undersampling strategy in an embodiment of the present invention;

4 is a schematic diagram of a truncated sinusoidal gradient field in an embodiment of the present invention;

Fig. 5 is the schematic diagram of the sinusoidal gradient field in the embodiment of the present invention;

6 is a schematic diagram of a truncated gradient field used in a bSSFP sequence in an embodiment of the present invention;

FIG. 7 is a functional block diagram of an internal structure of a terminal device provided by an embodiment of the present invention.

Detailed ways

The technical solutions in the present invention are clearly and completely described below in conjunction with the embodiments and the accompanying drawings. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts belong to the protection scope of the present invention.

After research, it is found that the existing technology needs to collect low-frequency ACS signals, and then calculate the high-frequency coil sensitivity based on the low-frequency ACS signals. In the process of collecting low-frequency ACS signals, memory introduces motion errors, which leads to the calculated There is a large error in the coil sensitivity, which leads to poor quality of the reconstructed image based on the coil sensitivity.

In order to solve the above technical problems, the present invention provides a magnetic resonance image reconstruction method and an image reconstruction device, which solve the problem of poor quality of reconstructed images calculated in the prior art. In the specific implementation, first apply a network structure composed of several neural networks to the image structure information of the depth image, and obtain the underlying image output by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, and the conjugate of the coil sensitivity Sensitivity: Sampling the signal received by the magnetic resonance equipment from the target object to generate a sampling template signal; then according to the information of each magnetic field applied in the sampling environment, the phase difference of the magnetic field formed by each magnetic field is obtained; finally, according to the underlying image, coil Sensitivity, conjugate sensitivity, background phase, sampling template signal and magnetic field phase difference to reconstruct the magnetic resonance image of the target object. The invention improves the quality of reconstructed images.

For example, it is necessary to use magnetic resonance imaging technology to collect images of the patient's lesion, the magnetic resonance equipment is fixed at one position, and the depth image of the patient's lesion is collected through the magnetic resonance equipment (used to characterize the distance between each point of the lesion and the magnetic resonance equipment). distance). Input the image information of the depth image into the four neural networks respectively to obtain the underlying image, background phase, and coil sensitivity (the sensitivity of the lesion to the internal coil of the magnetic resonance equipment, that is, the change of the coil will cause the image of the lesion to be collected how much the information changes), conjugate sensitivity. In addition, after sampling the signal (the signal is the signal formed after the magnetic resonance equipment sends the original signal to the lesion, and the lesion acts on the original signal) to obtain the sampling template signal, various magnetic fields are applied to the environment where the signal is located . A magnetic field phase difference occurs between various magnetic fields. Finally, the magnetic resonance image of the lesion (target object) can be reconstructed by combining the underlying image, coil sensitivity, conjugate sensitivity, background phase, sampling template signal and magnetic field phase difference.

exemplary method

The magnetic resonance image reconstruction method of this embodiment can be applied to a terminal device, and the terminal device can be a terminal product with an image acquisition function, such as a magnetic resonance device or the like. In this embodiment, as shown in Figure 1, the magnetic resonance image reconstruction method specifically includes the following steps:

S100. Apply a network structure composed of several neural networks to the image structure information of the depth image, and obtain the underlying image output by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, and the common value of the coil sensitivity. Yoke sensitivity, the depth image is used to characterize the depth information of the target object relative to the magnetic resonance equipment, and the magnetic field coil is a coil inside the magnetic resonance equipment.

In this embodiment, the image structure information is utilized, and such image structure information is generally expressed as the sparse property of the image itself in the MRI image, and this property can be represented by an optimized network structure.

The present embodiment includes four neural networks as shown in Figure 2, in Figure 2 from top to bottom are the first deep convolutional neural network CNN _k , the second deep convolutional neural network CNN, the third deep convolutional neural network CNN _C , the fourth deep convolutional neural network CNN _C* . The four deep convolutional neural networks in this embodiment are similar in structure, the difference is that the number of intermediate channels is different. In this embodiment, the Adam optimizer is used to iteratively optimize the parameters of the network. When the parameters of the network are optimized, random Inputting a small amount of noise can generate a complete image.

The optimization parameters are based on the following formula:

为二范数，C为深度卷积神经网络的输入，

为深度卷积神经网络的输出，

为深度卷积神经网络的损失函数，

分别为上述四个深度卷积神经网络的参数，通过上述公式计算损失函数的二范数取最小值时所对应的网络参数的值，以完成对网络参数的优化。In the formula,

is the two-norm, C is the input of the deep convolutional neural network,

is the output of the deep convolutional neural network,

is the loss function of the deep convolutional neural network,

They are the parameters of the above four deep convolutional neural networks, and the values of the corresponding network parameters when the two norms of the loss function take the minimum value are calculated by the above formula, so as to complete the optimization of the network parameters.

第二深度卷积神经网络

输出的背景相位

将深度图像的图像结构信息输入到第三深度卷积神经网络CNN_C，第三深度卷积神经网络CNN_C输出的线圈敏感度CSM。将深度图像的图像结构信息输入到第四深度卷积神经网络CNN_C*，第四深度卷积神经网络CNN_C*输出的共轭敏感度CSM*。其中，CNN_k的输入为

CNN_c、

和CNN_φ的输入为

After optimizing the network, the image structure information of the depth image is input to the first deep convolutional neural network CNN _k of the decoding structure, and the bottom layer image m output by the first deep convolutional neural network CNN _k . Input the image structure information of the depth image into the second deep convolutional neural network

The second deep convolutional neural network

output background phase

The image structure information of the depth image is input to the third deep convolutional neural network CNN _C , and the third deep convolutional neural network CNN _C outputs the coil sensitivity CSM. The image structure information of the depth image is input to the fourth deep convolutional neural network CNN _C* , and the conjugate sensitivity CSM* output by the fourth deep convolutional neural network CNN _C* . Among them, the input of CNN _k is

CNN _c ,

and the input of CNN _φ is

In this embodiment, the ACS signal containing only low-frequency phase information is not directly used to estimate the background phase, but the image and its phase information are characterized according to the image degradation process and the final collected signal, so the image can be generated more accurately and contain background phase information.

S200. Sampling the signal received by the magnetic resonance equipment from the target object to generate a sampling template signal.

For example, the magnetic resonance equipment sends the original signal to the human lesion (target object), and the human lesion interacts with the original signal, so that the original signal becomes a new signal. The new signal is sampled and generated as shown in Figure 3 The sampling template signal of .

Step S200 needs to apply a sinusoidal gradient field and a truncated sinusoidal gradient field before sampling.

In one embodiment, before signal sampling, the MRI gradient field coil is used to apply a sinusoidal gradient field in the slice selection direction, while the MRI gradient field coil is used to apply a truncated sinusoidal gradient field as shown in FIG. 4 in the phase direction. Alternatively, the truncated sinusoidal gradient field shown in FIG. 4 is applied in the slice selection direction by using the MRI gradient field coil, and the sinusoidal gradient field shown in FIG. 5 is applied in the phase direction by the MRI gradient field coil. And the 0th moment of the truncated sinusoidal gradient field is zero.

In this embodiment, the 0th moment of the truncated sinusoidal gradient field (in the phase direction and the layer selection direction) is zero and then the truncated sinusoidal gradient field adopts the direction shown in Figure 4 in the phase direction, which can not only effectively disperse the The aliasing artifacts caused by sampling can reduce the signal-to-noise ratio loss caused by g-factor to achieve high speedup, and at the same time avoid the imaging artifacts caused by the gradient field whose 0th order moment is not zero.

In one embodiment, the magnetic phase difference Psf between the sinusoidal gradient field and the truncated sinusoidal gradient field is

In one embodiment, the expression for the sinusoidal gradient field is as follows:

和

分别为wave-CAIPI技术在相位和选层方向所施加的梯度场；A为正弦梯度场幅值；D_C为一个正弦梯度场周期的持续时间(如图5中所标注)；D_R为读出梯度场的平台持续时间(如图5中所标注)。Wherein, t is time, and t=0 time point has been marked in Fig. 5;

and

are the gradient fields applied by the wave-CAIPI technology in the phase and layer selection directions; A is the amplitude of the sinusoidal gradient field; D _C is the duration of a sinusoidal gradient field period (marked in Figure 5); D _R is the reading The plateau duration of the gradient field (marked in Figure 5).

After applying the above-mentioned sinusoidal gradient field and truncated sinusoidal gradient field, step S200 starts to sample the signal, and step S200 includes the following steps S201 and S202:

S201. Sampling the signal received by the magnetic resonance equipment from the target object using a three-dimensional MRI sequence to obtain a sampling signal in a layer selection direction and a sampling signal in a phase direction.

In this embodiment, the three-dimensional MRI sequence includes GRE sequence, SE sequence, bSSFP sequence and the like.

S202. Generate a sampling template signal according to the sampling signal in the layer selection direction and the sampling signal in the phase direction.

In this embodiment, the three-dimensional MRI sequence is used to under-sample the signal to achieve the purpose of accelerating scanning. The truncated gradient field is applied to the bSSFP sequence, and the signal acquisition strategy of 2D CAIPIRINHA technology is used to reduce the amount of data acquisition. Through the combination of the truncated gradient field and the 2D CAIPIRINHA sampling strategy, the aliasing caused by undersampling can be dispersed to the readout, phase and layer selection directions at the same time, and the background area in the FOV is more effectively used to increase the number of different pixels. The sensitivity difference between points can further reduce the loss of g-factor signal-to-noise ratio.

The truncated gradient field is applied in the bSSFP sequence as shown in Figure 6, and the 2D CAIPIRINHA data acquisition strategy is shown in Figure 3. In Fig. 3, the readout direction is perpendicular to the phase direction and the layer selection direction at the same time, the dotted line intersection is the readout line required for full sampling, and the readout line required for the undersampling strategy adopted in the present invention is represented by a solid origin. Figure 3 shows 3×3 times undersampling (3 times undersampling in the phase direction, 3 times undersampling in the layer selection direction), the total acceleration factor is 9, and the required acquisition time is repetition time (TR, repetition time)×phase encoding Number of lines (Np)×Number of coded lines for layer selection (Ns)/9.

S300. Obtain a magnetic field phase difference formed by each magnetic field according to the information of each magnetic field applied in the sampling environment.

In this embodiment, each magnetic field information is the magnetic field phase of the sinusoidal _gradient field and the magnetic field phase of the truncated sinusoidal gradient field in step S200. In the two-dimensional case, _the The value of the point spread function is Psf _Y (k _x ,y)=waveP _y (k _x ,y)/P _y (k _x ,y), if it is expressed in z direction in another frequency coding direction, then the above formula will be becomes Psf _z (k _x ,y)=waveP _z (k _x ,z)/P _z (k _x ,z), in the three-dimensional case, both phase encoding directions are added with gradient sinusoidal gradient magnetic field and truncated sinusoidal gradient magnetic field, then the three-dimensional point spread function is Psf _yz (k _x ,y,z)=Psf _z (k _x ,z) Psf _y (k _x ,y), that is, any point in the three-dimensional point spread function Psf _yz (k _x ,y,z) is Psf _yz (k _x ,y,z).

S400. Reconstruct a magnetic resonance image of the target object according to the underlying image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal, and the magnetic field phase difference.

Step S400 includes the following steps S401 to S4011:

S401. Multiply the underlying image by the coil sensitivity to obtain a first result.

S402. Apply Fourier transform to the first result to obtain a second result.

S403. Multiply the second result by the magnetic field phase difference Psf to obtain a third result.

S404. Apply an inverse Fourier transform to the third result to obtain a fourth result.

S405. Multiply the fourth result by the sampling template signal to obtain a target signal b:

对应第一深度卷积神经网络输出的底层图像m，

对应第三深度卷积神经网络CNN_C输出的线圈敏感度CSM，

对应第一结果，

为对第一结果应用傅里叶变换所得的第二结果，

为第三结果，

对第三结果应用傅里叶逆变换，M为图3中的采样模板信号。

Corresponding to the underlying image m output by the first deep convolutional neural network,

Corresponding to the coil sensitivity CSM output by the third deep convolutional neural network CNN _C ,

Corresponding to the first result,

is the second result obtained by applying the Fourier transform to the first result,

for the third result,

Apply an inverse Fourier transform to the third result, and M is the sampled template signal in FIG. 3 .

S406. Multiply the background phase, the conjugate sensitivity, and the underlying image to obtain a fifth result;

S407. Apply Fourier transform to the fifth result to obtain a sixth result;

S408. Multiply the sixth result by the magnetic field phase difference Psf to obtain a seventh result;

S409. Apply an inverse Fourier transform to the seventh result to obtain an eighth result;

S4010. Multiply the eighth result by the sampling template signal to obtain the conjugate symmetric signal b ^H of the target signal:

对应第一深度卷积神经网络输出的底层图像m，

对应第四深度卷积神经网络输出的共轭敏感度CSM*，

对应第二深度卷积神经网络输出的背景相位，

为第五结果，

为第六结果，

为第七结果,

为傅里叶逆变换。

Corresponding to the underlying image m output by the first deep convolutional neural network,

The conjugate sensitivity CSM* corresponding to the output of the fourth deep convolutional neural network,

Corresponding to the background phase of the output of the second deep convolutional neural network,

For the fifth result,

For the sixth result,

For the seventh result,

is the inverse Fourier transform.

替代b^H：In one embodiment, substituting _bi for b, replaces b with

Substitute b ^H :

b _i =MF _y,z Psf(k _x ,y,z)F _x C _i m

是wave编码的前向模型，其中M为CAIPI采样模板，如图3所示，在实际的磁共振成像过程中，

为背景相位，D_i为接收线圈固有的线圈敏感度，在一般的模型中并没有单独考虑背景相位，而是将其包含在C_i中进行后续重建工作，通过VCC进行扩展后的数据为，b _i and

is the forward model of wave encoding, where M is the CAIPI sampling template, as shown in Figure 3, in the actual magnetic resonance imaging process,

is the background phase, and D _i is the inherent coil sensitivity of the receiving coil. In the general model, the background phase is not considered separately, but it is included in C _i for subsequent reconstruction work. The expanded data through VCC is,

为

的共轭对称，Psf^*为Psf的共轭对称，接收的信号

为原始信号b_i的共轭对称，通过这样虚拟共轭对称后，相当于提供了另外一组的相位信息，也就是说对wave编码框架中提供了一组额外的接收线圈的额外编码先验信息，进一步提供加速倍数。

for

The conjugate symmetry of Psf ^* is the conjugate symmetry of Psf, the received signal

is the conjugate symmetry of the original signal _bi , through such a virtual conjugate symmetry, it is equivalent to providing another set of phase information, that is to say, a set of additional encoding priors of an additional receiving coil is provided in the wave encoding framework information to further provide the acceleration factor.

S4011. Reconstruct the magnetic resonance image of the target object according to the target signal b and the conjugate symmetry signal b ^H .

The real magnetic resonance image of the target object can be reconstructed according to the target signal b and the conjugate symmetry signal b ^H , and obtaining the magnetic resonance image through b and b ^H is a prior art.

In summary, the underlying image, background phase, coil sensitivity, and conjugate sensitivity required for image reconstruction in the present invention all come from the neural network. Since the present invention does not involve low-frequency ACS signals in the process of calculating the coil sensitivity, it improves The quality of the reconstructed image of the present invention is improved. In addition, since the underlying image of the present invention is also calculated by the neural network, the speed of image reconstruction in the present invention is improved.

In addition, the present invention uses a deep convolutional neural network (Decoder) that does not require training to represent the underlying image after Wave-VCC encoding and expansion, CSM, and the high-frequency changing background phase that cannot be estimated using only the ACS of the low-frequency part. The convolutional neural network first indirectly generates the above three variables. In the present invention, the decoding convolutional neural network without training is used. Compared with the traditional supervised neural network or unsupervised neural network, there is no need to collect training data. The network The update of the parameters is optimized through an optimization algorithm, which is more in line with the difficult collection of fully sampled data in clinical magnetic resonance imaging.

Exemplary device

This embodiment also provides a magnetic resonance image reconstruction device, which includes the following components:

The information analysis module is used to apply a network structure composed of several neural networks to the image structure information of the depth image, and obtain the underlying image output by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, the coil The conjugate sensitivity of the sensitivity, the depth image is used to represent the depth information of the target object relative to the magnetic resonance equipment, and the magnetic field coil is a coil inside the magnetic resonance equipment;

The signal adopting module is used to sample the signal from the target object received by the magnetic resonance equipment, and generate a sampling template signal;

The phase difference calculation module is used to obtain the magnetic field phase difference formed by each magnetic field according to the information of each magnetic field applied in the sampling environment;

An image reconstruction module, configured to reconstruct the magnetic resonance of the target object according to the underlying image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal, and the magnetic field phase difference image.

Based on the foregoing embodiments, the present invention further provides a terminal device, the functional block diagram of which may be shown in FIG. 7 . The terminal equipment includes a processor, a memory, a network interface, a display screen, and a temperature sensor connected through a system bus. Wherein, the processor of the terminal device is used to provide calculation and control capabilities. The memory of the terminal device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The network interface of the terminal device is used to communicate with external terminals via a network connection. When the computer program is executed by the processor, a magnetic resonance image reconstruction method is realized. The display screen of the terminal device may be a liquid crystal display screen or an electronic ink display screen, and the temperature sensor of the terminal device is pre-set inside the terminal device for detecting the operating temperature of the internal device.

Those skilled in the art can understand that the functional block diagram shown in Figure 7 is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the terminal equipment to which the solution of the present invention is applied. The specific terminal equipment It is possible to include more or fewer components than shown in the figures, or to combine certain components, or to have a different arrangement of components.

In one embodiment, a terminal device is provided. The terminal device includes a memory, a processor, and a magnetic resonance image reconstruction program stored in the memory and operable on the processor. When the processor executes the magnetic resonance image reconstruction program, the The following operation instructions:

Apply a network structure composed of several neural networks to the image structure information of the depth image, and obtain the underlying image output by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, and the conjugate sensitivity of the coil sensitivity degree, the depth image is used to characterize the depth information of the target object relative to the magnetic resonance equipment, and the magnetic field coil is a coil inside the magnetic resonance equipment;

Sampling the signal received by the magnetic resonance equipment from the target object to generate a sampling template signal;

According to the information of each magnetic field applied in the sampling environment, the phase difference of the magnetic field formed by each magnetic field is obtained;

The magnetic resonance image of the target object is reconstructed according to the underlying image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal and the magnetic field phase difference.

Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be implemented through computer programs to instruct related hardware, and the computer programs can be stored in a non-volatile computer-readable memory In the medium, when the computer program is executed, it may include the processes of the embodiments of the above-mentioned methods. Wherein, any references to memory, storage, database or other media used in the various embodiments provided by the present invention may include non-volatile and/or volatile memory. Nonvolatile memory can include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: it can still be Modifications are made to the technical solutions described in the foregoing embodiments, or equivalent replacements are made to some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims (10)

1. A magnetic resonance image reconstruction method, comprising:

applying a network structure consisting of a plurality of neural networks to image structure information of a depth image to obtain a bottom layer image, a background phase, a coil sensitivity of a target object to a magnetic field coil and a conjugate sensitivity of the coil sensitivity, wherein the bottom layer image, the background phase, the coil sensitivity of the target object to the magnetic field coil and the conjugate sensitivity of the coil sensitivity are output by the network structure, the depth image is used for representing the depth information of the target object relative to a magnetic resonance device, and the magnetic field coil is a coil in the magnetic resonance device;

sampling a signal from the target object received by the magnetic resonance equipment to generate a sampling template signal;

obtaining magnetic field phase differences formed by the magnetic fields according to the information of the magnetic fields applied in the sampling environment;

and reconstructing a magnetic resonance image of the target object according to the bottom layer image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal and the magnetic field phase difference.

2. The method as claimed in claim 1, wherein the applying a network structure composed of several neural networks to the image structure information of the depth image to obtain the bottom layer image outputted by the network structure, the background phase, the coil sensitivity of the target object to the magnetic field coil, and the conjugate sensitivity of the coil sensitivity, the depth image is used to characterize the depth information of the target object relative to the magnetic resonance device, and the magnetic field coil is a coil inside the magnetic resonance device, and the method includes:

applying a first depth convolution neural network with a decoding structure to the image structure information of the depth image to obtain a bottom layer image output by the first depth convolution neural network;

applying a second depth convolution neural network to the image structure information of the depth image to obtain a background phase output by the second depth convolution neural network;

applying a third depth convolution neural network to the image structure information of the depth image to obtain the coil sensitivity output by the third depth convolution neural network;

and applying a fourth depth convolution neural network to the image structure information of the depth image to obtain the conjugate sensitivity output by the fourth depth convolution neural network.

3. The method of claim 1, wherein the sampling the signal from the target object received by the magnetic resonance apparatus to generate a sampled template signal comprises:

sampling a signal from the target object received by the magnetic resonance equipment by using a three-dimensional MRI sequence to obtain a sampling signal in a layer selection direction and a sampling signal in a phase direction;

and generating a sampling template signal according to the sampling signal in the layer selection direction and the sampling signal in the phase direction.

4. A magnetic resonance image reconstruction method as set forth in claim 3, characterized in that while the signals from the target object received by the magnetic resonance apparatus are sampled using a three-dimensional MRI sequence, a sinusoidal gradient field is applied in the slice selection direction and a truncated sinusoidal gradient field is applied in the phase direction; or applying a truncated sine gradient field in the layer selection direction and applying a sine gradient field in the phase direction.

5. A magnetic resonance image reconstruction method as claimed in claim 4, characterized in that the 0-th order moment of the truncated sinusoidal gradient field is zero.

6. The method as claimed in claim 4, wherein said obtaining the magnetic field phase difference formed by each magnetic field according to each magnetic field information applied in the sampling environment comprises:

and calculating the magnetic field phase difference between the sinusoidal gradient field and the truncated sinusoidal gradient field according to the magnetic field phase of the sinusoidal gradient field and the magnetic field phase of the truncated sinusoidal gradient field.

7. The method of claim 2, wherein the reconstructing the magnetic resonance image of the target object from the underlayer image, the coil sensitivities, the conjugate sensitivities, the background phases, the sampled template signals, and the magnetic field phase differences comprises:

multiplying the bottom layer image by the coil sensitivity to obtain a first result;

applying Fourier transform to the first result to obtain a second result;

multiplying the second result by the magnetic field phase difference to obtain a third result;

applying inverse Fourier transform to the third result to obtain a fourth result;

multiplying the fourth result by the sampling template signal to obtain a target signal;

multiplying the background phase, the conjugate sensitivity and the bottom layer image to obtain a fifth result;

applying Fourier transform to the fifth result to obtain a sixth result;

multiplying the sixth result by the magnetic field phase difference to obtain a seventh result;

applying inverse Fourier transform to the seventh result to obtain an eighth result;

multiplying the eighth result by the sampling template signal to obtain a conjugate symmetric signal of the target signal;

and reconstructing a magnetic resonance image of the target object according to the target signal and the conjugate symmetric signal.

8. A magnetic resonance image reconstruction apparatus, characterized in that the apparatus comprises the following components:

the information analysis module is used for applying a network structure consisting of a plurality of neural networks to image structure information of a depth image to obtain a bottom layer image, a background phase, coil sensitivity of a target object to a magnetic field coil and conjugate sensitivity of the coil sensitivity, wherein the bottom layer image, the background phase, the coil sensitivity of the target object to the magnetic field coil and the conjugate sensitivity of the coil sensitivity are output by the network structure, the depth image is used for representing the depth information of the target object relative to magnetic resonance equipment, and the magnetic field coil is a coil in the magnetic resonance equipment;

the signal application module is used for sampling the signal from the target object received by the magnetic resonance equipment to generate a sampling template signal;

the phase difference calculation module is used for obtaining the magnetic field phase difference formed by each magnetic field according to each magnetic field information applied in the sampling environment;

and the image reconstruction module is used for reconstructing a magnetic resonance image of the target object according to the bottom layer image, the coil sensitivity, the conjugate sensitivity, the background phase, the sampling template signal and the magnetic field phase difference.

9. A terminal device, characterized in that the terminal device comprises a memory, a processor and a magnetic resonance image reconstruction program stored in the memory and executable on the processor, and the processor implements the steps of the magnetic resonance image reconstruction method according to any one of claims 1 to 7 when executing the magnetic resonance image reconstruction program.

10. A computer-readable storage medium, on which a magnetic resonance image reconstruction program is stored which, when executed by a processor, carries out the steps of the magnetic resonance image reconstruction method according to any one of claims 1 to 7.

Images