Method, device, computer equipment and storage medium for determining tissue parameters

Abstract

The present application relates to a method, apparatus, computer equipment, storage medium and computer program product for determining tissue parameters. The method comprises: based on a magnetic resonance fingerprint imaging (MRF) sequence and an imaging object, adopting a target downsampling mode to obtain a downsampling time series corresponding to each voxel on the image of the imaging object; inputting the target parameter into a pre-trained artifact filtering model to obtain a full-sampling time series corresponding to the target voxel; and determining the tissue parameter corresponding to the target voxel based on the full-sampling time series corresponding to the target voxel and a pre-constructed dictionary. Since the artifact filtering model can remove the aliased downsampling artifacts in the downsampling time series to obtain a full-sampling time series, the tissue parameters obtained by matching the full-sampling time series with the time series in the dictionary are more accurate.

Description

Technical Field

[0001] The present application relates to the field of magnetic resonance imaging technology, and in particular to a method, apparatus, computer equipment, and storage medium for determining tissue parameters. Background Art

[0002] Magnetic resonance fingerprinting (MRF) technology uses theoretical estimates of all possible magnetic resonance signals generated by biological tissues as prior information to establish a correspondence between observed signals and quantitative parameters. By using certain information processing techniques to find this object relationship, it achieves the quantitative inversion of physiological and pathological parameters.

[0003] In traditional schemes, based on the MRF sequence and the imaging object, the downsampling time series corresponding to each voxel on the image of the imaging object is obtained, and then the downsampling time series is matched with the entries in the dictionary to determine the tissue parameters.

[0004] However, downsampling artifacts are aliased in the downsampling time series, and the tissue parameters determined by the above method are not accurate. Summary of the Invention

[0005] Based on this, it is necessary to provide a tissue parameter determination method, apparatus, computer equipment and storage medium that can improve the accuracy of tissue parameters in order to address the above technical problems.

[0006] In a first aspect, the present application provides a method for determining tissue parameters. The method comprises:

[0007] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object;

[0008] Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0009] The tissue parameters corresponding to the target voxel are determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0010] In a second aspect, the present application further provides a device for determining tissue parameters, the device comprising:

[0011] An acquisition module is used to acquire a downsampling time series corresponding to each voxel on an image of the imaging object using a target downsampling mode based on a magnetic resonance fingerprint imaging MRF sequence and an imaging object;

[0012] The acquisition module is further configured to input target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0013] The determination module is used to determine the tissue parameters corresponding to the target voxel according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0014] In one embodiment, the acquisition module is further configured to:

[0015] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are used to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel; based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, a sample set is constructed. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set; based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0016] In one embodiment, the acquisition module is further configured to:

[0017] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are used to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel; based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, a sample set is constructed. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure diagram corresponding to the imaging object sample constitute a sample in the sample set; based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0018] In one embodiment, the acquisition module is specifically configured to:

[0019] Based on the MRI signal and imaging object at each time point, the target downsampling mode is adopted to obtain the K-space data of the imaging object corresponding to the corresponding time point; and the K-space data of the imaging object is inverse Fourier transformed to obtain the image of the imaging object corresponding to the corresponding time point, thereby obtaining N images; for voxels at the same position on the N images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0020] In one embodiment, the determination module is specifically configured to:

[0021] The full sampling time series corresponding to the target voxel is matched with entries in a pre-built dictionary to obtain a target entry in the pre-built dictionary that matches the full sampling time series corresponding to the target voxel, and the tissue parameter corresponding to the target entry is used as the tissue parameter corresponding to the target voxel.

[0022] In one embodiment, the acquisition module is specifically configured to:

[0023] Based on the MRI signals and imaging object samples at each time point in the MRF sequence samples, the downsampling mode samples are used to obtain the K-space data of the imaging object samples corresponding to the corresponding time point; and the K-space data of the imaging object samples are inverse Fourier transformed to obtain the images of the imaging object samples corresponding to the corresponding time point, thereby obtaining N sample images; for voxels at the same position on the N sample images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0024] In one embodiment, the tissue parameter includes at least one of the following: longitudinal relaxation time, transverse relaxation time, and density.

[0025] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program. When the processor executes the computer program, the following steps are performed:

[0026] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object;

[0027] Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0028] The tissue parameters corresponding to the target voxel are determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0029] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps:

[0030] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object;

[0031] Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0032] The tissue parameters corresponding to the target voxel are determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0033] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the following steps:

[0034] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object;

[0035] Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0036] The tissue parameters corresponding to the target voxel are determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0037] The above-mentioned tissue parameter determination method, apparatus, computer equipment and storage medium first adopt a target downsampling mode based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object to obtain the downsampling time series corresponding to each voxel on the image of the imaging object. Then, the target parameter is input into a pre-trained artifact filtering model to obtain the full-sampling time series corresponding to the target voxel; finally, the tissue parameter corresponding to the target voxel is determined based on the full-sampling time series corresponding to the target voxel and a pre-constructed dictionary. Since the artifact filtering model can remove the aliased downsampling artifacts in the downsampling time series to obtain the full-sampling time series, the full-sampling time series is matched with the time series in the dictionary, and the obtained tissue parameters are more accurate. Compared with the method in some embodiments in which the number of scanning points is increased by increasing the time frames, thereby making the time series approach the full-sampling time series, since the present application is based on the pre-trained artifact filtering model to obtain the corresponding full-sampling time series, it can rely on fewer time frames to achieve the same accuracy, thereby shortening the scanning time. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A schematic diagram of the system structure provided in an embodiment of the present application;

[0039] Figure 2 Schematic diagram of a flow chart of a method for determining tissue parameters in one embodiment;

[0040] Figure 3 A schematic diagram of the principle of obtaining a full sampling time series corresponding to a target voxel in one embodiment;

[0041] Figure 4 A schematic diagram of the principle of obtaining a full sampling time series corresponding to a target voxel in another embodiment;

[0042] Figure 5 1 is a flow chart of the training steps of an artifact filtering model in one embodiment;

[0043] Figure 6 is a flowchart of the training steps of the artifact filtering model in another embodiment;

[0044] Figure 7 A schematic flow chart of a method for determining tissue parameters in another embodiment;

[0045] Figure 8 is a structural block diagram of a device for determining tissue parameters in one embodiment;

[0046] Figure 9 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0048] The tissue parameter determination method provided in the embodiment of the present application can be applied to Figure 1 In the system shown, Figure 1 The illustrated system includes a terminal 102 and a server 104. Terminal 102 communicates with server 104 via a network. A data storage system can store data that server 104 needs to process. The data storage system can be integrated with server 104 or located in the cloud or on another network server. Server 104 is equipped with a pre-trained artifact filtering model.

[0049] The terminal 102 is configured to, based on a magnetic resonance fingerprinting (MR fingerprinting, MRF) sequence and an imaging object, employ a target downsampling mode to obtain a downsampling time series corresponding to each voxel on an image of the imaging object, and transmit the downsampling time series corresponding to each voxel to the server 104. The server 104 is configured to, for each voxel, input a target parameter into a pre-trained artifact filtering model to obtain a full-sampling time series corresponding to the voxel, and then determine the tissue parameter corresponding to the voxel based on the full-sampling time series corresponding to the voxel and a pre-built dictionary. After obtaining the tissue parameter corresponding to each voxel, the server 104 may return the tissue parameter corresponding to each voxel to the terminal 102.

[0050] The terminal 102 is a device with a magnetic resonance fingerprint imaging function. The server 104 can be implemented as an independent server or a server cluster composed of multiple servers.

[0051] It should be noted that the above system is only one possible way to implement the tissue parameter determination method provided in the embodiment of the present application. The tissue parameter determination method provided in the embodiment of the present application can also be implemented only through the terminal, or only through the server. In the case of being implemented only through the terminal, the pre-trained artifact filter model is installed on the terminal, and the entire process of determining the tissue parameters is executed by the terminal. In the case of being implemented only through the server, the pre-trained artifact filter model is installed on the server, and the entire process of determining the tissue parameters is executed by the server. The embodiment of the present application is based on Figure 1 The process of determining tissue parameters is explained using the system shown as an example.

[0052] In some embodiments, as Figure 2 As shown, a method for determining tissue parameters is provided, comprising the following steps:

[0053] S202 : Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, a target downsampling mode is adopted to obtain a downsampling time series corresponding to each voxel on the image of the imaging object.

[0054] The Magnetic Resonance Fingerprinting (MRF) sequence is used to excite the imaging subject and generate K-space data. The MRF sequence consists of Magnetic Resonance Imaging (MRI) signals at N time points, also known as a radiofrequency pulse sequence. The pulse sequence parameters of the MRI signals at these N time points vary, including but not limited to the radiofrequency flip angle (FA), repetition time (TR), and echo time (TE).

[0055] Among them, the imaging object can be the surface part corresponding to various organs of the human body, such as the stomach, liver, abdomen, etc. The embodiment of the present application does not limit the imaging object.

[0056] Among them, the target reduction sampling mode is a sampling method used when scanning the imaging object. The target reduction sampling mode can be, for example, a non-uniform spiral sampling mode, a radiation sampling mode, an echo plane sampling mode, etc. The embodiment of the present application does not limit the target reduction sampling mode.

[0057] Specifically, the terminal excites the imaging object based on the MRI signal at each time point in the MRF sequence. During the scanning process of the imaging object, the terminal adopts the target downsampling mode for sampling to obtain the K-space data corresponding to the corresponding time point. Based on the K-space data corresponding to each of the N time points, the terminal determines the images corresponding to each of the N time points. Based on the images corresponding to each of the N time points, the terminal determines the downsampling time series corresponding to each voxel.

[0058] S204 , inputting the target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to the target voxel.

[0059] Among them, the artifact filtering model can remove the aliased downsampling artifacts in the downsampling time series to obtain a full-sampling time series. The full-sampling time series is matched with the time series in the dictionary to accurately improve the obtained tissue parameters.

[0060] Specifically, different samples are used when training the artifact filtering model, and different target parameters are input when the artifact filtering model is actually used. In one embodiment, see Figure 3As shown, the artifact filtering model can be trained based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel. In this case, the target parameters input by the server include: MRF sequence, target downsampling mode, and downsampling time series corresponding to the target voxel. After the target parameters are input into the artifact filtering model, the artifact filtering model will output the full-sampling time series corresponding to the target voxel. In another embodiment, see Figure 4 As shown, the artifact filtering model can be trained based on the MRF sequence samples, the downsampling mode samples, the full-sampling static structure diagram corresponding to the imaging object samples, the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel. In this case, the target parameters input by the server include: MRF sequence, target downsampling mode, downsampling time series corresponding to the target voxel, and the full-sampling static structure diagram corresponding to the imaging object. After the target parameters are input into the artifact filtering model, the artifact filtering model will output the full-sampling time series corresponding to the target voxel.

[0061] The target voxel may be any voxel on the image of the imaging object.

[0062] It should be noted that the training process of the artifact filtering model can be carried out on a server, or on other processing devices, and then loaded onto the server after the training is completed. This embodiment of the present application does not limit this.

[0063] S206 : Determine the tissue parameter corresponding to the target voxel according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0064] The pre-built dictionary includes multiple entries, each of which is a time series and corresponds to a set of tissue parameters. The server can match the fully sampled time series corresponding to the target voxel with the entries in the dictionary and use the tissue parameters corresponding to the successfully matched entries as the tissue parameters corresponding to the target voxel. Exemplarily, the tissue parameters include at least one of the following: longitudinal relaxation time, transverse relaxation time, and density.

[0065] In the above embodiment, first, based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, a target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object, and then the target parameter is input into a pre-trained artifact filtering model to obtain the full-sampling time series corresponding to the target voxel; finally, according to the full-sampling time series corresponding to the target voxel and a pre-constructed dictionary, the tissue parameter corresponding to the target voxel is determined. Since the artifact filtering model can remove the aliased downsampling artifacts in the downsampling time series to obtain the full-sampling time series, the full-sampling time series is matched with the time series in the dictionary, and the obtained tissue parameters are more accurate. Compared with the method in some embodiments in which the number of scanning points is increased by increasing the time frames, so that the time series approaches the full-sampling time series, since the present application is based on a pre-trained artifact filtering model to obtain the corresponding full-sampling time series, it can rely on fewer time frames to achieve the same accuracy, thereby shortening the scanning time.

[0066] In some embodiments, see Figure 5 As shown in FIG, the steps of training an artifact filtering model based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel include:

[0067] S501 , based on the MRF sequence samples and the imaging object samples, using the downsampling mode samples, obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel.

[0068] The imaging object samples may be body surface parts corresponding to various organs of the human body, and the downsampling mode samples are sampling methods used when scanning the imaging object samples.

[0069] Specifically, the downsampling time series can be obtained in the following manner: based on the MRI signal at each time point in the MRF sequence sample, the imaging object sample is excited, and in the process of scanning the imaging object sample, the downsampling mode sample is used for sampling to obtain the downsampling K-space data corresponding to the corresponding time point, and based on the downsampling K-space data corresponding to each of the N time points in the MRF sequence sample, the downsampling images corresponding to each of the N time points are determined, and on the N downsampling images, the voxels at the same position are locked, and the N image values ​​of the voxels are obtained, and the downsampling time series of the voxels is determined based on the sequence of the N time points and the N image values. For example, if there are 256*256 voxels on each image, a downsampling time series of 256*256 voxels can be obtained.

[0070] Specifically, the method for obtaining a full-sampling time series is similar to the downsampling time series, except that when obtaining a full-sampling time series, sampling is performed in a full-sampling manner. The specific process is as follows: based on the MRI signal at each time point in the MRF sequence sample, the imaging object sample is excited. During the scanning process of the imaging object sample, sampling is performed in a full-sampling manner to obtain the full-sampling K-space data corresponding to the corresponding time point. Based on the full-sampling K-space data corresponding to each of the N time points in the MRF sequence sample, the full-sampling image corresponding to each of the N time points is determined. On the N full-sampling images, the voxel at the same position is locked, and N image values ​​of the voxel are obtained. Based on the order of the N time points and the N image values, the full-sampling time series of the voxel is determined. For example, if there are 256*256 voxels on each image, a full-sampling time series of 256*256 voxels can be obtained.

[0071] It should be noted that the above method of obtaining the full-sampling time series is only an example. In other embodiments, the full-sampling time series can be obtained through simulation. This application does not limit the method of obtaining the full-sampling time series.

[0072] S502. Construct a sample set based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set.

[0073] S503: Based on the sample set, the initial model is trained to obtain an artifact filtering model.

[0074] Here are some examples:

[0075] Assume that the imaging object sample is the abdomen, and the downsampling mode sample is a non-uniform spiral sampling mode. On the one hand, based on the MRI signal at each time point in the MRF sequence sample, the abdomen is excited. During the scanning of the abdomen, the non-uniform spiral sampling mode is used for sampling to obtain the downsampling K-space data corresponding to the corresponding time point. Based on the downsampling K-space data corresponding to each of the N time points in the MRF sequence sample, the downsampling images corresponding to each of the N time points are determined. On the N downsampling images, the voxels at the same position are locked, and the N image values ​​of the voxels are obtained. Based on the sequence of the N time points and the N image values, the downsampling time series of the voxel is determined, thereby obtaining the downsampling time series of all voxels on the downsampling image. On the other hand, based on the MRI signal at each time point in the MRF sequence sample, the abdomen is excited. During the scanning of the abdomen, sampling is performed in a full-sampling manner to obtain the full-sampling K-space data corresponding to the corresponding time point. Based on the full-sampling K-space data corresponding to each of the N time points in the MRF sequence sample, the full-sampling image corresponding to each of the N time points is determined. On the N full-sampling images, the voxels at the same position are locked, and the N image values ​​of the voxels are obtained. Based on the sequence of the N time points and the N image values, the full-sampling time series of the voxel is determined, thereby obtaining the full-sampling time series of all voxels on the full-sampling image. For the voxels at the same position on the downsampled image and the full-sampled image, the downsampled time series and the full-sampled time series of the voxel are extracted. The downsampled time series, the full-sampled time series, the MRF sequence samples and the non-uniform spiral sampling pattern can constitute a sample. Assuming that there are 256*256 voxels on the downsampled image and the full-sampled image, 256*256 samples can be obtained. A sample set can be constructed based on the 256*256 samples. In the training stage, the initial model can be trained based on the sample set. After reaching the convergence condition, the artifact filtering model is obtained.

[0076] In the above embodiment, an artifact filtering model is trained based on the downsampling time series corresponding to each voxel on the image of the MRF sequence samples, the downsampling mode samples, and the imaging object samples, as well as the full-sampling time series corresponding to each voxel. In actual application, after the MRF sequence, the target downsampling mode, and the downsampling time series corresponding to the target voxel are input into the artifact filtering model, the artifact filtering model will output the full-sampling time series corresponding to the target voxel. Since the artifact filtering model can remove the aliased downsampling artifacts in the downsampling time series to obtain the full-sampling time series, the full-sampling time series is matched with the time series in the dictionary, and the obtained tissue parameters are more accurate.

[0077] In some embodiments, see Figure 6As shown, the steps of training an artifact filtering model based on the MRF sequence samples, the downsampling mode samples, the full-sampling static structure diagram corresponding to the imaging object samples, the downsampling time series corresponding to each voxel on the image of the imaging object samples, and the full-sampling time series corresponding to each voxel include:

[0078] S601 , based on the MRF sequence samples and the imaging object samples, using the downsampling mode samples, obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel.

[0079] S602. Construct a sample set based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure diagram corresponding to the imaging object sample constitute a sample in the sample set.

[0080] S603: Based on the sample set, the initial model is trained to obtain an artifact filtering model.

[0081] Specifically, based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are used to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel. Figure 5 The description of the corresponding embodiments will not be repeated here.

[0082] It should be noted that: this embodiment and Figure 5 The difference between the corresponding embodiments is that different samples are used when training the artifact filtering model. Figure 5 The examples in the corresponding embodiments are as follows: Figure 5 In the corresponding embodiment, after obtaining the downsampling time series of all voxels on the downsampling image and the downsampling time series of all voxels on the full-sampling image, the downsampling time series of the voxel and the full-sampling time series of the voxel at the same position on the downsampling image and the full-sampling image are extracted. The downsampling time series, the full-sampling time series, the MRF sequence sample, and the non-uniform spiral sampling pattern can constitute a sample. In this embodiment, the downsampling time series, the full-sampling time series, the MRF sequence sample, the non-uniform spiral sampling pattern, and the full-sampling static structure diagram corresponding to the abdomen constitute a sample. It can be seen that the full-sampling static structure diagram is added to the sample in this embodiment. The full-sampling static structure diagram can provide additional information about the tissue and improve the accuracy of the full-sampling time series output by the artifact filtering model.

[0083] In the above embodiment, an artifact filtering model is trained based on the MRF sequence samples, the downsampling mode samples, the full-sampling static structure diagram corresponding to the imaging object samples, the downsampling time series corresponding to each voxel on the image of the imaging object samples, and the full-sampling time series corresponding to each voxel. In actual application, after the MRF sequence, the target downsampling mode, the downsampling time series corresponding to the target voxel and the full-sampling static structure diagram corresponding to the imaging object are input into the artifact filtering model, the artifact filtering model will output the full-sampling time series corresponding to the target voxel. Since the full-sampling static structure diagram can provide additional information about the tissue, the full-sampling time series output by the artifact filtering model is more accurate. By matching the full-sampling time series with the time series in the dictionary, the obtained tissue parameters are more accurate.

[0084] In some embodiments, the MRF sequence includes MRI signals at N time points. Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, a target downsampling mode is adopted to obtain a downsampling time series corresponding to each voxel on the image of the imaging object, including:

[0085] Based on the MRI signal and imaging object at each time point, the target downsampling mode is adopted to obtain the K-space data of the imaging object corresponding to the corresponding time point; and the K-space data of the imaging object is inverse Fourier transformed to obtain the image of the imaging object corresponding to the corresponding time point, thereby obtaining N images; for voxels at the same position on the N images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0086] Specifically, the imaging object is excited based on the MRI signal at each time point in the MRF sequence. During the scanning process of the imaging object, the target downsampling mode is used for sampling to obtain the K-space data of the imaging object corresponding to the corresponding time point. The downsampling K-space data is inverse Fourier transformed to obtain the image of the imaging object corresponding to the corresponding time point, thereby obtaining the image corresponding to each of the N time points. For the voxels at the same position on the N images, the N image values ​​of the voxels are extracted, and the downsampling time series of the voxels is determined based on the sequence of the N time points and the N image values.

[0087] It should be noted that the image obtained by performing inverse Fourier transform on the K-space data of the imaging object corresponding to a certain time point can also be called a downsampled image. The embodiment of the present application does not limit the name of the image.

[0088] In the above embodiment, a specific implementation method for obtaining a down-sampling time series corresponding to each voxel on an image of an imaging object is provided. The down-sampling time series can be used for subsequent input into an artifact filtering model. Since the artifact filtering model can remove the aliased down-sampling artifacts in the down-sampling time series to obtain a full-sampling time series, the full-sampling time series is matched with the time series in the dictionary, and the obtained tissue parameters are more accurate.

[0089] Accordingly, during the training phase, the downsampling time series can also be obtained by the method of the aforementioned embodiment. Specifically, based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are used to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, including:

[0090] Based on the MRI signals and imaging object samples at each time point in the MRF sequence samples, the downsampling mode samples are used to obtain the K-space data of the imaging object samples corresponding to the corresponding time point; and the K-space data of the imaging object samples are inverse Fourier transformed to obtain the images of the imaging object samples corresponding to the corresponding time point, thereby obtaining N sample images; for voxels at the same position on the N sample images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0091] Specifically, based on the MRI signal at each time point in the MRF sequence sample, the imaging object sample is excited, and in the process of scanning the imaging object sample, the downsampling mode sample is used for sampling to obtain the K-space data of the imaging object sample corresponding to the corresponding time point, and the downsampling K-space data is inverse Fourier transformed to obtain the image of the imaging object sample corresponding to the corresponding time point, thereby obtaining the image corresponding to each of the N time points (N sample images), and for the voxels at the same position on the N sample images, the N image values ​​of the voxels are extracted, and the downsampling time series of the voxels is determined based on the sequence of the N time points and the N image values. It should be noted that the image obtained after the inverse Fourier transform of the K-space data of the imaging object sample corresponding to a certain time point can also be called a downsampling image, and the embodiment of the present application does not limit the name of the image.

[0092] In some embodiments, determining the tissue parameter corresponding to the target voxel based on the full sampling time series corresponding to the target voxel and a pre-built dictionary includes:

[0093] The full sampling time series corresponding to the target voxel is matched with entries in a pre-built dictionary to obtain a target entry in the pre-built dictionary that matches the full sampling time series corresponding to the target voxel, and the tissue parameter corresponding to the target entry is used as the tissue parameter corresponding to the target voxel.

[0094] Specifically, a dictionary can be constructed in advance, which includes several entries, each entry being a time series. After obtaining the full-sampled time series corresponding to the target voxel, the full-sampled time series corresponding to the target voxel can be matched with the entries in the dictionary through information processing technologies such as pattern recognition and data mining. The entry with the highest matching degree is used as the target entry, and the tissue parameters corresponding to the target entry are indexed, and the tissue parameters are used as the tissue parameters corresponding to the target voxel.

[0095] In the above embodiment, after obtaining the full sampling time series corresponding to the target voxel, the full sampling time series corresponding to the target voxel is matched with the entries in the dictionary. Compared with the method of directly using the reduced sampling time series and the entries in the dictionary for matching in some implementations, the obtained tissue parameters are more accurate.

[0096] In one embodiment, see Figure 7 As shown, a method for determining tissue parameters is provided, comprising:

[0097] S701. Based on the MRI signals and imaging object samples at each time point in the MRF sequence samples, a downsampling mode sample is used to obtain K-space data of the imaging object samples corresponding to the corresponding time point; and an inverse Fourier transform is performed on the K-space data of the imaging object samples to obtain an image of the imaging object samples corresponding to the corresponding time point, thereby obtaining N sample images; for voxels at the same position on the N sample images, N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, a downsampling time series corresponding to the corresponding voxels is determined.

[0098] S702. Construct a sample set based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure diagram corresponding to the imaging object sample constitute a sample in the sample set.

[0099] It should be noted that the sample in S702 is only one possible way. It can also be that, for each voxel, the MRF sequence sample, the downsampling mode sample, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set.

[0100] S703: Based on the sample set, the initial model is trained to obtain an artifact filtering model.

[0101] S704. Based on the MRI signal and imaging object at each time point, a target downsampling mode is used to obtain K-space data of the imaging object corresponding to the corresponding time point; and an inverse Fourier transform is performed on the K-space data of the imaging object to obtain an image of the imaging object corresponding to the corresponding time point, thereby obtaining N images; for voxels at the same position on the N images, N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, a downsampling time series corresponding to the corresponding voxels is determined.

[0102] S705 , inputting the target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to the target voxel.

[0103] It should be noted that when the samples include MRF sequence samples, downsampling mode samples, downsampling time series corresponding to the corresponding voxels, full-sampling time series corresponding to the corresponding voxels, and full-sampling static structure diagram corresponding to the imaging object samples, the target parameters include: MRF sequence, target downsampling mode, downsampling time series corresponding to the target voxels, and full-sampling static structure diagram corresponding to the imaging object. When the samples include MRF sequence samples, downsampling mode samples, downsampling time series corresponding to the corresponding voxels, and full-sampling time series corresponding to the corresponding voxels, the target parameters include: MRF sequence, target downsampling mode, and downsampling time series corresponding to the target voxels.

[0104] S706. Match the full-sample time series corresponding to the target voxel with entries in a pre-constructed dictionary to obtain a target entry in the pre-constructed dictionary that matches the full-sample time series corresponding to the target voxel, and use the tissue parameter corresponding to the target entry as the tissue parameter corresponding to the target voxel.

[0105] In the above embodiment, first, based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, a target downsampling mode is adopted to obtain a downsampling time series corresponding to each voxel on the image of the imaging object. Then, the target parameter is input into a pre-trained artifact filtering model to obtain a full-sampling time series corresponding to the target voxel. Finally, the tissue parameter corresponding to the target voxel is determined based on the full-sampling time series corresponding to the target voxel and a pre-constructed dictionary. Since the artifact filtering model can remove the aliased downsampling artifacts in the downsampling time series to obtain a full-sampling time series, the full-sampling time series is matched with the time series in the dictionary, and the obtained tissue parameters are more accurate.

[0106] It should be understood that, although the various steps in the flowcharts involved in the various embodiments described above are displayed in sequence according to the instructions of the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the flowcharts involved in the various embodiments described above can include multiple steps or multiple stages, and these steps or stages are not necessarily executed and completed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of steps or stages in other steps.

[0107] Based on the same inventive concept, embodiments of the present application further provide a tissue parameter determination device for implementing the tissue parameter determination method described above. The solution provided by this device is similar to the solution described in the method described above. Therefore, the specific limitations of one or more tissue parameter determination device embodiments provided below can be found in the limitations of the tissue parameter determination method described above and will not be further elaborated here.

[0108] In one embodiment, Figure 8 As shown, a tissue parameter determination device is provided, comprising:

[0109] An acquisition module 801 is configured to acquire a downsampling time series corresponding to each voxel on an image of the imaging object using a target downsampling mode based on an MRF sequence and an imaging object;

[0110] The acquisition module 801 is further configured to input target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object; the target voxel is any voxel in an image of the imaging object;

[0111] The determination module 802 is configured to determine the tissue parameter corresponding to the target voxel according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0112] In some embodiments, the acquisition module 801 is further configured to:

[0113] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0114] Based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, a sample set is constructed. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set;

[0115] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0116] In some embodiments, the acquisition module 801 is further configured to:

[0117] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0118] A sample set is constructed based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure diagram corresponding to the imaging object sample constitute a sample in the sample set.

[0119] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0120] In some embodiments, the acquisition module 801 is specifically configured to:

[0121] Based on the MRI signal and imaging object at each time point, a target downsampling mode is used to obtain K-space data of the imaging object corresponding to the corresponding time point; and an inverse Fourier transform is performed on the K-space data of the imaging object to obtain an image of the imaging object corresponding to the corresponding time point, thereby obtaining N images;

[0122] For voxels at the same position on the N images, N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, a downsampling time series corresponding to the corresponding voxels is determined.

[0123] In some embodiments, the determination module 802 is specifically configured to:

[0124] The full sampling time series corresponding to the target voxel is matched with entries in a pre-built dictionary to obtain a target entry in the pre-built dictionary that matches the full sampling time series corresponding to the target voxel, and the tissue parameter corresponding to the target entry is used as the tissue parameter corresponding to the target voxel.

[0125] In some embodiments, the acquisition module 801 is specifically configured to:

[0126] Based on the MRI signals and imaging object samples at each time point in the MRF sequence samples, the downsampling mode samples are used to obtain the K-space data of the imaging object samples corresponding to the corresponding time point; and the K-space data of the imaging object samples are inverse Fourier transformed to obtain the images of the imaging object samples corresponding to the corresponding time point, thereby obtaining N sample images; for voxels at the same position on the N sample images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0127] In some embodiments, the tissue parameter includes at least one of: longitudinal relaxation time, transverse relaxation time, and density.

[0128] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 9 As shown. The computer device includes a processor, a memory, and a network interface connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The database of the computer device is used to store data such as a dictionary. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for determining tissue parameters is implemented.

[0129] Those skilled in the art will understand that Figure 9 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0130] In one embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the following steps are implemented:

[0131] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object;

[0132] Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0133] The tissue parameters corresponding to the target voxel are determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0134] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0135] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0136] Based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, a sample set is constructed. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set;

[0137] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0138] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0139] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0140] A sample set is constructed based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure diagram corresponding to the imaging object sample constitute a sample in the sample set.

[0141] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0142] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0143] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object, including:

[0144] Based on the MRI signal and imaging object at each time point, a target downsampling mode is used to obtain K-space data of the imaging object corresponding to the corresponding time point; and an inverse Fourier transform is performed on the K-space data of the imaging object to obtain an image of the imaging object corresponding to the corresponding time point, thereby obtaining N images;

[0145] For voxels at the same position on the N images, N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, a downsampling time series corresponding to the corresponding voxels is determined.

[0146] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0147] The full sampling time series corresponding to the target voxel is matched with entries in a pre-built dictionary to obtain a target entry in the pre-built dictionary that matches the full sampling time series corresponding to the target voxel, and the tissue parameter corresponding to the target entry is used as the tissue parameter corresponding to the target voxel.

[0148] In one embodiment, when the processor executes the computer program, the processor further implements the following steps:

[0149] Based on the MRI signals and imaging object samples at each time point in the MRF sequence samples, the downsampling mode samples are used to obtain the K-space data of the imaging object samples corresponding to the corresponding time point; and the K-space data of the imaging object samples are inverse Fourier transformed to obtain the images of the imaging object samples corresponding to the corresponding time point, thereby obtaining N sample images; for voxels at the same position on the N sample images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0150] In one embodiment, the tissue parameter includes at least one of the following: longitudinal relaxation time, transverse relaxation time, and density.

[0151] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0152] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object;

[0153] Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0154] The tissue parameters corresponding to the target voxel are determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0155] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0156] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0157] Based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, a sample set is constructed. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set;

[0158] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0159] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0160] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0161] A sample set is constructed based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure diagram corresponding to the imaging object sample constitute a sample in the sample set.

[0162] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0163] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0164] Based on the MRI signal and imaging object at each time point, a target downsampling mode is used to obtain K-space data of the imaging object corresponding to the corresponding time point; and an inverse Fourier transform is performed on the K-space data of the imaging object to obtain an image of the imaging object corresponding to the corresponding time point, thereby obtaining N images;

[0165] For voxels at the same position on the N images, N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, a downsampling time series corresponding to the corresponding voxels is determined.

[0166] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0167] The full sampling time series corresponding to the target voxel is matched with entries in a pre-built dictionary to obtain a target entry in the pre-built dictionary that matches the full sampling time series corresponding to the target voxel, and the tissue parameter corresponding to the target entry is used as the tissue parameter corresponding to the target voxel.

[0168] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0169] Based on the MRI signals and imaging object samples at each time point in the MRF sequence samples, the downsampling mode samples are used to obtain the K-space data of the imaging object samples corresponding to the corresponding time point; and the K-space data of the imaging object samples are inverse Fourier transformed to obtain the images of the imaging object samples corresponding to the corresponding time point, thereby obtaining N sample images; for voxels at the same position on the N sample images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0170] In one embodiment, the tissue parameter includes at least one of the following: longitudinal relaxation time, transverse relaxation time, and density.

[0171] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the following steps:

[0172] Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, the target downsampling mode is adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object;

[0173] Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: an MRF sequence, a target downsampling mode, and a downsampling time series corresponding to the target voxel; or, the target parameters include: an MRF sequence, a target downsampling mode, a downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, where the target voxel is any voxel on an image of the imaging object;

[0174] The tissue parameters corresponding to the target voxel are determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

[0175] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0176] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0177] Based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, a sample set is constructed. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set;

[0178] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0179] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0180] Based on the MRF sequence samples and the imaging object samples, the downsampling mode samples are adopted to obtain the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel;

[0181] A sample set is constructed based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel. For each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure diagram corresponding to the imaging object sample constitute a sample in the sample set.

[0182] Based on the sample set, the initial model is trained to obtain the artifact filtering model.

[0183] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0184] Based on the MRI signal and imaging object at each time point, a target downsampling mode is used to obtain K-space data of the imaging object corresponding to the corresponding time point; and an inverse Fourier transform is performed on the K-space data of the imaging object to obtain an image of the imaging object corresponding to the corresponding time point, thereby obtaining N images;

[0185] For voxels at the same position on the N images, N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, a downsampling time series corresponding to the corresponding voxels is determined.

[0186] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0187] The full sampling time series corresponding to the target voxel is matched with entries in a pre-built dictionary to obtain a target entry in the pre-built dictionary that matches the full sampling time series corresponding to the target voxel, and the tissue parameter corresponding to the target entry is used as the tissue parameter corresponding to the target voxel.

[0188] In one embodiment, when the computer program is executed by a processor, the following steps are further implemented:

[0189] Based on the MRI signals and imaging object samples at each time point in the MRF sequence samples, the downsampling mode samples are used to obtain the K-space data of the imaging object samples corresponding to the corresponding time point; and the K-space data of the imaging object samples are inverse Fourier transformed to obtain the images of the imaging object samples corresponding to the corresponding time point, thereby obtaining N sample images; for voxels at the same position on the N sample images, the N image values ​​of the corresponding voxels are extracted, and based on the N image values ​​of the corresponding voxels, the downsampling time series corresponding to the corresponding voxels is determined.

[0190] In one embodiment, the tissue parameter includes at least one of the following: longitudinal relaxation time, transverse relaxation time, and density.

[0191] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, database or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory may include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in the various embodiments provided herein may include at least one of a relational database and a non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchains. The processor involved in the various embodiments provided herein may be, but are not limited to, a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic unit, a data processing logic unit based on quantum computing, and the like.

[0192] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0193] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present application. It should be noted that a person of ordinary skill in the art may make various modifications and improvements without departing from the spirit of the present application, and these modifications and improvements fall within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be determined by the appended claims.

Claims

1. A method for determining tissue parameters, characterized in that: The method comprises: Based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, a target downsampling mode is adopted to obtain a downsampling time series corresponding to each voxel on the image of the imaging object; the MRF sequence includes magnetic resonance imaging MRI signals at N time points, and the pulse sequence parameters corresponding to the MRI signals at the N time points are different; based on the magnetic resonance fingerprint imaging MRF sequence and the imaging object, a target downsampling mode is adopted to obtain a downsampling time series corresponding to each voxel on the image of the imaging object, including: based on the MRI signal at each time point and the imaging object, the target downsampling mode is adopted to obtain the K-space data of the imaging object corresponding to the corresponding time point, and perform inverse Fourier transform on the K-space data of the imaging object to obtain the image of the imaging object corresponding to the corresponding time point, thereby obtaining N images, and for voxels at the same position on the N images, extracting N image values ​​of the corresponding voxels, and determining the downsampling time series corresponding to the corresponding voxels based on the N image values ​​of the corresponding voxels and the order of the N time points; Inputting target parameters into a pre-trained artifact filtering model to obtain a full-sampled time series corresponding to a target voxel; the target parameters include: the MRF sequence, the target downsampling mode, and the downsampling time series corresponding to the target voxel, or the target parameters include: the MRF sequence, the target downsampling mode, the downsampling time series corresponding to the target voxel, and a full-sampled static structure image corresponding to the imaging object, wherein the target voxel is any voxel on the image of the imaging object; the full-sampled time series is a sequence in which aliased downsampling artifacts in the downsampling time series are removed; The tissue parameter corresponding to the target voxel is determined according to the full sampling time series corresponding to the target voxel and a pre-built dictionary.

2. The method according to claim 1, characterized in that The target parameters include: the MRF sequence, the target downsampling mode, and the downsampling time series corresponding to the target voxels; Before inputting the target parameters into the pre-trained artifact filtering model, the method further includes: Based on the MRF sequence samples and the imaging object samples, using the downsampling mode samples, obtaining the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel; Constructing a sample set based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, wherein for each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, and the full-sampling time series corresponding to the corresponding voxel constitute a sample in the sample set; Based on the sample set, an initial model is trained to obtain the artifact filtering model.

3. The method according to claim 1, characterized in that The target parameters include: the MRF sequence, the target downsampling mode, the downsampling time series corresponding to the target voxels, and the full-sampling static structure diagram corresponding to the imaging object; Before inputting the target parameters into the pre-trained artifact filtering model, the method further includes: Based on the MRF sequence samples and the imaging object samples, using the downsampling mode samples, obtaining the downsampling time series corresponding to each voxel on the image of the imaging object sample, and the full-sampling time series corresponding to each voxel; constructing a sample set based on the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to each voxel, and the full-sampling time series corresponding to each voxel, wherein for each voxel, the MRF sequence samples, the downsampling mode samples, the downsampling time series corresponding to the corresponding voxel, the full-sampling time series corresponding to the corresponding voxel, and the full-sampling static structure image corresponding to the imaging object sample constitute a sample in the sample set; Based on the sample set, an initial model is trained to obtain the artifact filtering model.

4. The method according to any one of claims 1 to 3, characterized in that Determining the tissue parameter corresponding to the target voxel according to the full sampling time series corresponding to the target voxel and a pre-built dictionary includes: The full-sample time series corresponding to the target voxel is matched with the entries in the pre-constructed dictionary to obtain a target entry in the pre-constructed dictionary that matches the full-sample time series corresponding to the target voxel, and the tissue parameter corresponding to the target entry is used as the tissue parameter corresponding to the target voxel.

5. The method according to any one of claims 1 to 3, characterized in that The tissue parameter includes at least one of the following: longitudinal relaxation time, transverse relaxation time and density.

6. A tissue parameter determination device, characterized in that: The device comprises: An acquisition module is configured to acquire a downsampling time series corresponding to each voxel on an image of the imaging object using a target downsampling mode based on an MRF sequence and an imaging object; The acquisition module is further configured to input target parameters into a pre-trained artifact filtering model to obtain a full-sample time series corresponding to a target voxel; the target parameters include: the MRF sequence, the target downsampling mode, and the downsampling time series corresponding to the target voxel, or the target parameters include: the MRF sequence, the target downsampling mode, the downsampling time series corresponding to the target voxel, and a full-sample static structure image corresponding to the imaging object, wherein the target voxel is any voxel on the image of the imaging object; the full-sample time series is a sequence in which aliased downsampling artifacts in the downsampling time series are removed; a determination module, configured to determine the tissue parameter corresponding to the target voxel according to the full sampling time series corresponding to the target voxel and a pre-built dictionary; The MRF sequence includes magnetic resonance imaging (MRI) signals at N time points, and the pulse sequence parameters corresponding to the MRI signals at the N time points are different. The acquisition module is further used to adopt the target downsampling mode to acquire the K-space data of the imaging object corresponding to the corresponding time point based on the MRI signal at each time point and the imaging object, and perform an inverse Fourier transform on the K-space data of the imaging object to obtain the image of the imaging object corresponding to the corresponding time point, thereby obtaining N images. For voxels at the same position on the N images, N image values ​​of the corresponding voxels are extracted, and the downsampling time sequence corresponding to the corresponding voxels is determined based on the N image values ​​of the corresponding voxels and the sequence of the N time points.

7. The device according to claim 6, characterized in that The determination module is used to match the full-sampled time series corresponding to the target voxel with the entries in the pre-constructed dictionary to obtain the target entry in the pre-constructed dictionary that matches the full-sampled time series corresponding to the target voxel, and use the tissue parameter corresponding to the target entry as the tissue parameter corresponding to the target voxel.

8. The device according to claim 6, characterized in that The tissue parameter includes at least one of the following: longitudinal relaxation time, transverse relaxation time and density.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images