WO2022198655A1 - 磁共振氧十七代谢成像方法、装置、存储介质及终端设备 - Google Patents

磁共振氧十七代谢成像方法、装置、存储介质及终端设备 Download PDF

Info

Publication number
WO2022198655A1
WO2022198655A1 PCT/CN2021/083359 CN2021083359W WO2022198655A1 WO 2022198655 A1 WO2022198655 A1 WO 2022198655A1 CN 2021083359 W CN2021083359 W CN 2021083359W WO 2022198655 A1 WO2022198655 A1 WO 2022198655A1
Authority
WO
WIPO (PCT)
Prior art keywords
data
radial
matrix data
sub
sampling data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/CN2021/083359
Other languages
English (en)
French (fr)
Inventor
王珊珊
荣楚誉
郑海荣
刘新
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Publication of WO2022198655A1 publication Critical patent/WO2022198655A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/11—Region-based segmentation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T3/00—Geometric image transformations in the plane of the image
    • G06T3/40—Scaling of whole images or parts thereof, e.g. expanding or contracting
    • G06T3/4007—Scaling of whole images or parts thereof, e.g. expanding or contracting based on interpolation, e.g. bilinear interpolation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/90—Dynamic range modification of images or parts thereof
    • G06T5/94—Dynamic range modification of images or parts thereof based on local image properties, e.g. for local contrast enhancement
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10072—Tomographic images
    • G06T2207/10088—Magnetic resonance imaging [MRI]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning

Definitions

  • the present application belongs to the field of computer technology, and in particular relates to a magnetic resonance oxygen seventeen metabolic imaging method, device, computer-readable storage medium and terminal device.
  • the regulation of cerebral fluid plays an important role in the brain.
  • the dynamic image of cerebral fluid can be obtained by magnetic resonance oxygen seventeen (oxygen-17) metabolic imaging, which can be used as a basis for judging whether the brain is in a normal state or a diseased state.
  • an imaging method based on a gridding algorithm can be used. In this method, after density compensation is performed on the sampled data, the data is resampled to a large grid matrix by interpolation, and then the magnetic resonance image is reconstructed. Although this method can achieve fast imaging, the imaging accuracy is low.
  • embodiments of the present application provide a magnetic resonance oxygen seventeen metabolic imaging method, device, computer-readable storage medium, and terminal device, so as to solve the problem of low accuracy of the existing imaging method.
  • a first aspect of the embodiments of the present application provides a magnetic resonance oxygen seventeen metabolic imaging method, which may include:
  • Different radial interpolation networks are used to process the radial sampling data of each sub-region respectively to obtain the filled radial sampling data; wherein, each radial interpolation network corresponds to a frequency partition;
  • the use of different radial interpolation networks to respectively process the radial sampling data of each sub-region to obtain padded radial sampling data may include:
  • the augmented matrix data of each sub-region is restored to the form of radial spokes, and the filled radial sampling data is obtained.
  • the method may further include:
  • the training sample set of the gth radial interpolation network 1 ⁇ g ⁇ G, G is the number of frequency partitions;
  • the training sample set includes several training samples, and each training sample includes input matrix data and expected output matrix data, the input matrix data is consistent with the size of the gth subregion matrix data, and the expected output matrix data is consistent with the size of the gth subregion augmented matrix data;
  • the g-th radial interpolation network is trained by using the training sample set to obtain the g-th radial interpolation network after training.
  • the construction of the training sample set of the gth radial interpolation network may include:
  • a training sample consisting of the input matrix data and the expected output matrix data is added to the training sample set.
  • the process of using different radial interpolation networks to process the matrix data of each sub-region to obtain the augmented matrix data of each sub-region may include:
  • the process of using different radial interpolation networks to process the matrix data of each sub-region to obtain the augmented matrix data of each sub-region may include:
  • the process of using different radial interpolation networks to process the matrix data of each sub-region to obtain the augmented matrix data of each sub-region may include:
  • the number of rows of each subregional matrix data is enlarged to ⁇ g times, and the number of columns is enlarged to ⁇ g times, and the augmented matrix data of each subregion is obtained.
  • performing imaging processing on the padded radial sampling data to obtain a target image may include:
  • a second aspect of the embodiments of the present application provides a magnetic resonance oxygen seventeen metabolic imaging device, which may include:
  • the sampling data division module is used to obtain the golden angle radial sampling data of magnetic resonance oxygen seventeen metabolism, and divide the radial sampling data into several sub-regional radial sampling data according to the preset frequency division;
  • the radial interpolation processing module is used to process the radial sampling data of each sub-region by using different radial interpolation networks to obtain the filled radial sampling data; wherein, each radial interpolation network corresponds to a frequency partition;
  • the imaging processing module is configured to perform imaging processing on the filled radial sampling data to obtain a target image.
  • the radial interpolation processing module may include:
  • the data rearrangement unit is used to rearrange the radial sampling data of each sub-region into a matrix form respectively, and obtain the matrix data of each sub-region;
  • the radial interpolation processing unit is used to process the matrix data of each sub-region by using different radial interpolation networks to obtain the augmented matrix data of each sub-region;
  • a data restoration unit configured to restore the augmented matrix data of each sub-region to the form of radial spokes, and obtain the filled radial sampling data.
  • the magnetic resonance oxygen seventeen metabolic imaging device may further include:
  • the training sample set building module is used to construct the training sample set of the gth radial interpolation network, 1 ⁇ g ⁇ G, G is the number of frequency partitions; the training sample set includes several training samples, each training sample Including input matrix data and expected output matrix data, the input matrix data is consistent with the size of the gth subregional matrix data, and the expected output matrix data is consistent with the size of the gth subregion augmented matrix data;
  • the radial interpolation network training module is used for training the g th radial interpolation network by using the training sample set to obtain the g th radial interpolation network after training.
  • the training sample set building module may include:
  • the full sampling data acquisition unit is used to obtain fully sampled full sampling data
  • an under-sampling data processing unit configured to perform data deletion on the fully-sampled data to obtain under-sampled data corresponding to the fully-sampled data
  • a first rearranging unit used for rearranging the fully collected data to obtain the input matrix data
  • a second rearranging unit configured to rearrange the undersampled data to obtain the expected output matrix data
  • a training sample adding unit configured to add a training sample composed of the input matrix data and the expected output matrix data to the training sample set.
  • the radial interpolation processing unit is specifically configured to: use different radial interpolation networks to expand the number of rows of each subregional matrix data to ⁇ g times, respectively, to obtain each subregional matrix data.
  • Region augmentation matrix data where ⁇ g is the augmentation coefficient corresponding to the g-th frequency partition, and ⁇ g is an integer greater than 1.
  • the radial interpolation processing unit is specifically configured to: use different radial interpolation networks to expand the number of columns of each subregional matrix data to ⁇ g times, respectively, to obtain each subregional matrix data.
  • Region augmentation matrix data where ⁇ g is the augmentation coefficient corresponding to the g-th frequency partition, and ⁇ g is an integer greater than 1.
  • the radial interpolation processing unit is specifically configured to: use different radial interpolation networks to expand the number of rows and columns of each subregional matrix data to ⁇ g times, respectively. to ⁇ g times to obtain the augmented matrix data of each sub-region.
  • the imaging processing module is specifically configured to perform inverse fast Fourier transform processing on the padded radial sampling data to obtain the target image.
  • a third aspect of the embodiments of the present application provides a computer-readable storage medium, where a computer program is stored in the computer-readable storage medium, and when the computer program is executed by a processor, any one of the above-mentioned magnetic resonance oxygen seventeen metabolism is realized The steps of the imaging method.
  • a fourth aspect of the embodiments of the present application provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, when the processor executes the computer program The steps of realizing any one of the above-mentioned magnetic resonance oxygen seventeen metabolic imaging methods.
  • a fifth aspect of the embodiments of the present application provides a computer program product, which, when the computer program product runs on a terminal device, enables the terminal device to perform any of the steps of the above-mentioned magnetic resonance oxygen seventeen metabolic imaging method.
  • the embodiment of the present application has the following beneficial effects: the embodiment of the present application obtains the golden angle radial sampling data of magnetic resonance oxygen seventeen metabolism, and divides the radial sampling data according to preset frequency divisions is the radial sampling data of several sub-regions; different radial interpolation networks are used to process the radial sampling data of each sub-region respectively to obtain the filled radial sampling data; wherein, each radial interpolation network corresponds to a frequency Partition; perform imaging processing on the filled radial sampling data to obtain a target image.
  • different radial interpolation networks are introduced to fill in the sampled data in each frequency partition respectively, so that the sampled data in each different frequency partition has a stronger pertinence, and the filled data is more accurate, thereby effectively improving the the final imaging accuracy.
  • FIG. 1 is a flowchart of an embodiment of a magnetic resonance oxygen seventeen metabolic imaging method in an embodiment of the present application
  • Fig. 2 is the schematic diagram of golden angle radial sampling trajectory
  • Fig. 3 is the schematic flow chart of using different radial interpolation networks to process the radial sampling data of each sub-region respectively to obtain the filled radial sampling data;
  • Fig. 4 is the schematic diagram of frequency domain data filling process
  • FIG. 5 is a structural diagram of an embodiment of a magnetic resonance oxygen seventeen metabolic imaging device in an embodiment of the present application.
  • FIG. 6 is a schematic block diagram of a terminal device in an embodiment of the present application.
  • the term “if” may be contextually interpreted as “when” or “once” or “in response to determining” or “in response to detecting” .
  • the phrases “if it is determined” or “if the [described condition or event] is detected” may be interpreted, depending on the context, to mean “once it is determined” or “in response to the determination” or “once the [described condition or event] is detected. ]” or “in response to detection of the [described condition or event]”.
  • an embodiment of a magnetic resonance oxygen seventeen metabolic imaging method in the embodiment of the present application may include:
  • Step 101 Acquire the golden angle radial sampling data of magnetic resonance oxygen seventeen metabolism, and divide the radial sampling data into several sub-regional radial sampling data according to preset frequency partitions.
  • Magnetic Resonance Imaging Magnetic Resonance Imaging
  • MRI Magnetic Resonance Imaging
  • X-CT X-ray computed tomography
  • MRI does not produce harmful radiation to the human body, and can image human soft tissue at the same time.
  • X-CT X-ray computed tomography
  • MRI does not produce harmful radiation to the human body, and can image human soft tissue at the same time.
  • dynamic imaging which requires high temporal resolution, causes motion artifacts due to the slow imaging speed of MRI, which limits its wide application.
  • the original data collected by MRI is frequency domain data (that is, k-space data).
  • k-space data frequency domain data
  • researchers have done a lot of research on undersampling, non-uniform sampling, and sampling trajectory design.
  • the golden angle radial sampling targeted by the embodiments of the present application is different from the ordinary radial sampling. Because the ordinary radial sampling is densely sampled in the low-frequency region in the center, and sparsely sampled in the high-frequency region, this causes certain difficulties for reconstruction.
  • the golden angle radial sampling adopts different sampling densities for different annular areas according to the distance from the center, that is, reducing the sampling spokes in the low frequency area and increasing the sampling spokes in the high frequency area, which improves the sampling efficiency.
  • Figure 2 shows a schematic diagram of the golden angle radial sampling trajectory.
  • the golden angle radial sampling rotates a fixed angle of 111.25 degrees each time. This value is calculated based on the golden ratio, so it is called the golden angle.
  • the advantage of the golden angle radial sampling data is that the k-space at any time in the sampling process presents a state of relatively uniform sampling, which is conducive to image reconstruction based on the data of any number of sampling lines.
  • the terminal device that executes the embodiments of the present application may be a magnetic resonance scanner, or may be other terminal devices that can acquire radial sampling data through wired communication, wireless communication, transfer storage media, or the like.
  • the executive body can be an image workstation, and the image workstation can communicate and connect with the magnetic resonance scanner through data lines, wifi, bluetooth, etc., to obtain radial sampling data, and perform magnetic resonance imaging work.
  • the radial sampling data may be divided into several sub-regional radial sampling data according to preset frequency partitions.
  • the specific number of frequency partitions and the frequency range of each frequency partition may be set according to actual conditions, which are not specifically limited in this embodiment of the present application.
  • three frequency partitions can be set in the low frequency area, the intermediate frequency area and the high frequency area, and the radial sampling data can be divided into the low frequency area radial sampling data, the intermediate frequency area radial sampling data and the high frequency area radial sampling data correspondingly.
  • Step 102 using different radial interpolation networks to process the radial sampling data of each sub-region respectively to obtain the filled radial sampling data.
  • step 102 may include the process shown in FIG. 3 :
  • Step 1021 Rearrange the radial sampling data of each sub-region into a matrix form to obtain matrix data of each sub-region.
  • the number of radial spokes is denoted as N
  • the number of sampling points on each spoke is denoted as M.
  • the sampling on each spoke is If the data is taken as a row of the matrix, a matrix data with a size of N ⁇ M (N rows and M columns) can be obtained, that is, subregional matrix data.
  • Step 1022 Use different radial interpolation networks to process the matrix data of each sub-region respectively to obtain the augmented matrix data of each sub-region.
  • radial interpolation networks corresponding to each frequency partition may be preset. For example, as shown in Figure 4, for the three frequency partitions of the low frequency region, the intermediate frequency region and the high frequency region, corresponding radial interpolation networks can be set respectively (that is, the radial interpolation module 1 shown in the figure, the radial interpolation module 2 and radial interpolation module 3).
  • any neural network in the prior art may be selected and used as the radial interpolation network according to the actual situation, which is not specifically limited here.
  • Size expansion can specifically include the following three situations:
  • Case 1 Use different radial interpolation networks to expand the number of rows of each sub-regional matrix data to ⁇ g times, that is, increase the number of spokes, and obtain the augmented matrix data of each sub-region ;
  • the augmentation coefficient corresponding to the frequency partition, and ⁇ g is an integer greater than 1, and its specific value can be set according to the actual situation, which is not specifically limited in this embodiment of the present application, 1 ⁇ g ⁇ G, G is the number of frequency partitions .
  • Case 2 Use different radial interpolation networks to expand the number of columns of each subregional matrix data to ⁇ g times, that is, increase the number of sampling points on each spoke, and obtain the augmented matrix data of each subregion; among them, ⁇ g is the augmentation coefficient corresponding to the g-th frequency partition, and ⁇ g is an integer greater than 1, and its specific value can be set according to the actual situation, which is not specifically limited in this embodiment of the present application.
  • Case 3 Using different radial interpolation networks to expand the number of rows of each subregional matrix data to ⁇ g times, and the number of columns to ⁇ g times, that is to say, the number of spokes and the number of sampling points on each spoke are increased at the same time.
  • Each subregion augments the matrix data.
  • any one of the above-mentioned situations may be selected according to the actual situation to process the matrix data of each subregion.
  • the radial interpolation network used in the embodiments of the present application has undergone the training process of deep learning in advance. Processing Radial Interpolation Network) as an example to describe its training process in detail.
  • the training sample set includes several training samples.
  • Each training sample includes input matrix data and expected output matrix data.
  • the size of the input matrix data is the same as that of the gth subregional matrix data, and the expected output matrix data is increased with the gth subregional data.
  • the size of the wide matrix data is consistent.
  • the fully sampled data of the corresponding frequency partition may be obtained first, and then the fully sampled data is deleted according to the preset undersampling ratio to obtain the same data as the fully sampled data.
  • the undersampled data corresponding to the data.
  • rearrange the fully sampled data to obtain input matrix data; and rearrange the undersampled data to obtain expected output matrix data.
  • the training samples consisting of the input matrix data and the expected output matrix data are added to the training sample set.
  • the training sample set is empty, that is, there is no training sample.
  • the g th radial interpolation network can be trained by using the training sample set to obtain the g th radial interpolation network after training.
  • the radial interpolation network can be used to process the input matrix data in the training sample to obtain the actual output matrix data, and then according to the expectations in the training sample
  • the output matrix data and the actual output matrix data calculate the training loss value.
  • the specific calculation method of the training loss value can be set according to the actual situation.
  • the squared error between the expected output matrix data and the actual output matrix data can be calculated, and the squared error can be determined. is the training loss value.
  • the model parameters of the radial interpolation network can be adjusted according to the training loss value.
  • the model parameter of the radial interpolation network is W1
  • the training loss value is back-propagated to modify the model parameter W1 of the radial interpolation network to obtain the modified model parameter W2.
  • the training loss value is back-propagated to modify the model parameter W1 of the radial interpolation network to obtain the modified model parameter W2.
  • the training conditions can be that the number of training times reaches the preset number of times threshold, the number of times
  • the threshold can be set according to the actual situation, for example, it can be set to thousands, tens of thousands, hundreds of thousands or even larger values; the training condition can also be that the radial interpolation network converges; it may occur that the number of training has not yet reached the number of times threshold value, but the radial interpolation network has converged, which may lead to unnecessary repetition of work; or the radial interpolation network can never converge, which may lead to an infinite loop and cannot end the training process.
  • the training condition can also be training. The degree reaches the degree threshold or the radial interpolation network converges. When the training conditions are met, the trained radial interpolation network can be obtained.
  • Step 1023 restore the augmented matrix data of each sub-region to the form of radial spokes, and obtain the filled radial sampling data.
  • step 1023 is the inverse process of step 1021.
  • N′ N′ rows and M′ columns
  • the process of data restoration Taking the data of each row of the matrix as the sampling data on one spoke, the radial sampling data after corresponding sub-region filling can be obtained, including N′ spokes, and the number of sampling points on each spoke is M′. After restoring and obtaining the filled radial sampling data of each sub-region, these data can be combined into complete filled radial sampling data.
  • Figure 4 shows a schematic diagram of the data rearrangement, radial interpolation network processing and data restoration process. After such a process, the data filling in the frequency domain is effectively realized, and the finally obtained radial sampling data after filling is compared with the original. The radially sampled data has higher data precision.
  • Step 103 Perform imaging processing on the padded radial sampling data to obtain a target image.
  • inverse fast Fourier transform processing may be performed on the padded radial sampling data, so as to obtain the final imaging result, that is, the target image.
  • the embodiment of the present application acquires the golden angle radial sampling data of magnetic resonance oxygen seventeen metabolism, and divides the radial sampling data into several sub-regional radial sampling data according to preset frequency partitions; using Different radial interpolation networks respectively process the radial sampling data of each subregion to obtain filled radial sampling data; wherein, each radial interpolation network corresponds to a frequency partition; The data is subjected to imaging processing to obtain the target image.
  • different radial interpolation networks are introduced to fill in the sampled data in each frequency partition respectively, so that the sampled data in each different frequency partition has a stronger pertinence, and the filled data is more accurate, thereby effectively improving the the final imaging accuracy.
  • FIG. 5 shows a structural diagram of an embodiment of a magnetic resonance oxygen seventeen metabolic imaging device provided in an embodiment of the present application.
  • a magnetic resonance oxygen seventeen metabolic imaging device may include:
  • the sampling data division module 501 is used to obtain the golden angle radial sampling data of magnetic resonance oxygen seventeen metabolism, and divide the radial sampling data into several sub-regional radial sampling data according to preset frequency divisions;
  • the radial interpolation processing module 502 is used to process the radial sampling data of each sub-region by using different radial interpolation networks to obtain the filled radial sampling data; wherein, each radial interpolation network corresponds to a frequency partition ;
  • the imaging processing module 503 is configured to perform imaging processing on the filled radial sampling data to obtain a target image.
  • the radial interpolation processing module may include:
  • the data rearrangement unit is used to rearrange the radial sampling data of each sub-region into a matrix form respectively, and obtain the matrix data of each sub-region;
  • the radial interpolation processing unit is used to process the matrix data of each sub-region by using different radial interpolation networks to obtain the augmented matrix data of each sub-region;
  • a data restoration unit configured to restore the augmented matrix data of each sub-region to the form of radial spokes, and obtain the filled radial sampling data.
  • the magnetic resonance oxygen seventeen metabolic imaging device may further include:
  • the training sample set building module is used to construct the training sample set of the gth radial interpolation network, 1 ⁇ g ⁇ G, G is the number of frequency partitions; the training sample set includes several training samples, each training sample Including input matrix data and expected output matrix data, the input matrix data is consistent with the size of the gth subregional matrix data, and the expected output matrix data is consistent with the size of the gth subregion augmented matrix data;
  • the radial interpolation network training module is used for training the g th radial interpolation network by using the training sample set to obtain the g th radial interpolation network after training.
  • the training sample set building module may include:
  • the full sampling data acquisition unit is used to obtain fully sampled full sampling data
  • an under-sampling data processing unit configured to perform data deletion on the fully-sampled data to obtain under-sampled data corresponding to the fully-sampled data
  • a first rearranging unit used for rearranging the fully collected data to obtain the input matrix data
  • a second rearranging unit configured to rearrange the undersampled data to obtain the expected output matrix data
  • a training sample adding unit configured to add a training sample composed of the input matrix data and the expected output matrix data to the training sample set.
  • the radial interpolation processing unit is specifically configured to: use different radial interpolation networks to expand the number of rows of each subregional matrix data to ⁇ g times, respectively, to obtain each subregional matrix data.
  • Region augmentation matrix data where ⁇ g is the augmentation coefficient corresponding to the g-th frequency partition, and ⁇ g is an integer greater than 1.
  • the radial interpolation processing unit is specifically configured to: use different radial interpolation networks to expand the number of columns of each subregional matrix data to ⁇ g times, respectively, to obtain each subregional matrix data.
  • Region augmentation matrix data where ⁇ g is the augmentation coefficient corresponding to the g-th frequency partition, and ⁇ g is an integer greater than 1.
  • the radial interpolation processing unit is specifically configured to: use different radial interpolation networks to expand the number of rows and columns of each subregional matrix data to ⁇ g times, respectively. to ⁇ g times to obtain the augmented matrix data of each sub-region.
  • the imaging processing module is specifically configured to perform inverse fast Fourier transform processing on the padded radial sampling data to obtain the target image.
  • FIG. 6 shows a schematic block diagram of a terminal device provided by an embodiment of the present application. For convenience of description, only parts related to the embodiment of the present application are shown.
  • the terminal device 6 in this embodiment includes: a processor 60 , a memory 61 , and a computer program 62 stored in the memory 61 and running on the processor 60 .
  • the processor 60 executes the computer program 62
  • the steps in each of the above embodiments of the magnetic resonance oxygen seventeen metabolic imaging method are implemented, for example, steps 101 to 103 shown in FIG. 1 .
  • the processor 60 executes the computer program 62
  • the computer program 62 may be divided into one or more modules/units, and the one or more modules/units are stored in the memory 61 and executed by the processor 60 to complete the this application.
  • the one or more modules/units may be a series of computer program instruction segments capable of performing specific functions, and the instruction segments are used to describe the execution process of the computer program 62 in the terminal device 6 .
  • FIG. 6 is only an example of the terminal device 6, and does not constitute a limitation on the terminal device 6, and may include more or less components than the one shown, or combine some components, or different components
  • the terminal device 6 may further include an input and output device, a network access device, a bus, and the like.
  • the processor 60 may be a central processing unit (Central Processing Unit, CPU), or other general-purpose processors, digital signal processors (Digital Signal Processors). Signal Processor, DSP), application specific integrated circuit (Application Specific Integrated Circuit, ASIC), Field Programmable Gate Array (Field-Programmable Gate Array, FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • a general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
  • the memory 61 may be an internal storage unit of the terminal device 6 , such as a hard disk or a memory of the terminal device 6 .
  • the memory 61 may also be an external storage device of the terminal device 6, such as a plug-in hard disk equipped on the terminal device 6, a smart memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash memory card (Flash Card), etc.
  • the memory 61 may also include both an internal storage unit of the terminal device 6 and an external storage device.
  • the memory 61 is used to store the computer program and other programs and data required by the terminal device 6 .
  • the memory 61 can also be used to temporarily store data that has been output or will be output.
  • the disclosed apparatus/terminal device and method may be implemented in other manners.
  • the apparatus/terminal device embodiments described above are only illustrative.
  • the division of the modules or units is only a logical function division. In actual implementation, there may be other division methods, such as multiple units. Or components may be combined or may be integrated into another system, or some features may be omitted, or not implemented.
  • the shown or discussed mutual coupling or direct coupling or communication connection may be through some interfaces, indirect coupling or communication connection of devices or units, and may be in electrical, mechanical or other forms.
  • the units described as separate components may or may not be physically separated, and components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution in this embodiment.
  • each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically alone, or two or more units may be integrated into one unit.
  • the above-mentioned integrated units may be implemented in the form of hardware, or may be implemented in the form of software functional units.
  • the integrated modules/units if implemented in the form of software functional units and sold or used as independent products, may be stored in a computer-readable storage medium.
  • the present application can implement all or part of the processes in the methods of the above embodiments, and can also be completed by instructing the relevant hardware through a computer program.
  • the computer program can be stored in a computer-readable storage medium, and the computer When the program is executed by the processor, the steps of the foregoing method embodiments can be implemented.
  • the computer program includes computer program code, and the computer program code may be in the form of source code, object code, executable file or some intermediate form, and the like.
  • the computer-readable storage medium may include: any entity or device capable of carrying the computer program code, a recording medium, a U disk, a removable hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM, Read-Only). Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the content contained in the computer-readable storage medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, computer-readable Storage media exclude electrical carrier signals and telecommunications signals.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Magnetic Resonance Imaging Apparatus (AREA)

Abstract

一种磁共振氧十七代谢成像方法、装置、计算机可读存储介质及终端设备。该方法包括：获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将径向采样数据划分为若干个分区域径向采样数据（S101）；使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据（S102）；对填补后的径向采样数据进行成像处理，得到目标图像（S103）。由于引入不同的径向差值网络分别在各个频率分区上对采样数据进行填补，有效提高最终的成像精度。

Description

磁共振氧十七代谢成像方法、装置、存储介质及终端设备 技术领域
本申请属于计算机技术领域，尤其涉及一种磁共振氧十七代谢成像方法、装置、计算机可读存储介质及终端设备。
 背景技术
脑液的调节在大脑中起着重要作用，通过磁共振氧十七（氧-17）代谢成像可以得到脑液的动态图像，从而作为判别大脑处于正常情况或是患病情况的依据。现有技术中，可以采用基于网格化算法的成像方法，这种方法在对采样数据进行密度补偿后，通过插值将数据重采样至一个大的网格矩阵下，再进行磁共振图像重建，这种方法虽然可以实现快速成像，但成像精度较低。
 技术问题
有鉴于此，本申请实施例提供了一种磁共振氧十七代谢成像方法、装置、计算机可读存储介质及终端设备，以解决现有的成像方法精度较低的问题。
 技术解决方案
本申请实施例的第一方面提供了一种磁共振氧十七代谢成像方法，可以包括：
获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将所述径向采样数据划分为若干个分区域径向采样数据；
使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据；其中，每个径向插值网络均对应一个频率分区；
对所述填补后的径向采样数据进行成像处理，得到目标图像。
在第一方面的一种可能的实现方式中，所述使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据，可以包括：
将各个分区域径向采样数据分别重排列为矩阵形式，得到各个分区域矩阵数据；
使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据；
将各个分区域增广矩阵数据还原为径向辐条形式，得到所述填补后的径向采样数据。
在第一方面的一种可能的实现方式中，在使用不同的径向插值网络分别对各个分区域矩阵数据进行处理之前，所述方法还可以包括：
构建第g个径向插值网络的训练样本集合，1≤g≤G，G为频率分区的数目；所述训练样本集合中包括若干个训练样本，每个训练样本包括输入矩阵数据和预期输出矩阵数据，所述输入矩阵数据与第g个分区域矩阵数据的尺寸一致，所述预期输出矩阵数据与第g个分区域增广矩阵数据的尺寸一致；
使用所述训练样本集合对第g个径向插值网络进行训练，得到训练后的第g个径向插值网络。
在第一方面的一种可能的实现方式中，所述构建第g个径向插值网络的训练样本集合，可以包括：
获取充分采样的全采数据；
对所述全采数据进行数据删除，得到与所述全采数据对应的欠采样数据；
对所述全采数据进行重排列，得到所述输入矩阵数据；
对所述欠采样数据进行重排列，得到所述预期输出矩阵数据；
将所述输入矩阵数据和所述预期输出矩阵数据组成的训练样本添加入所述训练样本集合中。
在本申请实施例的第一种具体实现中，所述使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据，可以包括：
使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，得到各个分区域增广矩阵数据；其中， α g 为与第g个频率分区对应的增广系数，且 α g 为大于1的整数。
在本申请实施例的第二种具体实现中，所述使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据，可以包括：
使用不同的径向插值网络分别将各个分区域矩阵数据的列数扩大至β g 倍，得到各个分区域增广矩阵数据；其中，β g 为与第g个频率分区对应的增广系数，且β g 为大于1的整数。
在本申请实施例的第三种具体实现中，所述使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据，可以包括：
使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，列数扩大至β g 倍，得到各个分区域增广矩阵数据。
在第一方面的一种可能的实现方式中，所述对所述填补后的径向采样数据进行成像处理，得到目标图像，可以包括：
对所述填补后的径向采样数据进行快速傅里叶逆变换处理，得到所述目标图像。
本申请实施例的第二方面提供了一种磁共振氧十七代谢成像装置，可以包括：
采样数据划分模块，用于获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将所述径向采样数据划分为若干个分区域径向采样数据；
径向插值处理模块，用于使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据；其中，每个径向插值网络均对应一个频率分区；
成像处理模块，用于对所述填补后的径向采样数据进行成像处理，得到目标图像。
在第二方面的一种可能的实现方式中，所述径向插值处理模块可以包括：
数据重排列单元，用于将各个分区域径向采样数据分别重排列为矩阵形式，得到各个分区域矩阵数据；
径向插值处理单元，用于使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据；
数据还原单元，用于将各个分区域增广矩阵数据还原为径向辐条形式，得到所述填补后的径向采样数据。
在第二方面的一种可能的实现方式中，所述磁共振氧十七代谢成像装置还可以包括：
训练样本集合构建模块，用于构建第g个径向插值网络的训练样本集合，1≤g≤G，G为频率分区的数目；所述训练样本集合中包括若干个训练样本，每个训练样本包括输入矩阵数据和预期输出矩阵数据，所述输入矩阵数据与第g个分区域矩阵数据的尺寸一致，所述预期输出矩阵数据与第g个分区域增广矩阵数据的尺寸一致；
径向插值网络训练模块，用于使用所述训练样本集合对第g个径向插值网络进行训练，得到训练后的第g个径向插值网络。
在第二方面的一种可能的实现方式中，所述训练样本集合构建模块可以包括：
全采数据获取单元，用于获取充分采样的全采数据；
欠采样数据处理单元，用于对所述全采数据进行数据删除，得到与所述全采数据对应的欠采样数据；
第一重排列单元，用于对所述全采数据进行重排列，得到所述输入矩阵数据；
第二重排列单元，用于对所述欠采样数据进行重排列，得到所述预期输出矩阵数据；
训练样本添加单元，用于将所述输入矩阵数据和所述预期输出矩阵数据组成的训练样本添加入所述训练样本集合中。
在本申请实施例的第一种具体实现中，所述径向插值处理单元具体用于：使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，得到各个分区域增广矩阵数据；其中， α g 为与第g个频率分区对应的增广系数，且 α g 为大于1的整数。
在本申请实施例的第二种具体实现中，所述径向插值处理单元具体用于：使用不同的径向插值网络分别将各个分区域矩阵数据的列数扩大至β g 倍，得到各个分区域增广矩阵数据；其中，β g 为与第g个频率分区对应的增广系数，且β g 为大于1的整数。
在本申请实施例的第三种具体实现中，所述径向插值处理单元具体用于：使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，列数扩大至β g 倍，得到各个分区域增广矩阵数据。
在第二方面的一种可能的实现方式中，所述成像处理模块具体用于对所述填补后的径向采样数据进行快速傅里叶逆变换处理，得到所述目标图像。
本申请实施例的第三方面提供了一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，所述计算机程序被处理器执行时实现上述任一种磁共振氧十七代谢成像方法的步骤。
本申请实施例的第四方面提供了一种终端设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现上述任一种磁共振氧十七代谢成像方法的步骤。
本申请实施例的第五方面提供了一种计算机程序产品，当计算机程序产品在终端设备上运行时，使得终端设备执行上述任一种磁共振氧十七代谢成像方法的步骤。
 有益效果
本申请实施例与现有技术相比存在的有益效果是：本申请实施例获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将所述径向采样数据划分为若干个分区域径向采样数据；使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据；其中，每个径向插值网络均对应一个频率分区；对所述填补后的径向采样数据进行成像处理，得到目标图像。通过本申请实施例，引入不同的径向插值网络分别在各个频率分区上对采样数据进行填补，对于各个不同频率分区上的采样数据具有更强的针对性，填补的数据更加精准，从而有效提高了最终的成像精度。
 附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其它的附图。
图1为本申请实施例中一种磁共振氧十七代谢成像方法的一个实施例流程图；
图2为黄金角径向采样轨迹的示意图；
图3为使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据的示意流程图；
图4为频率域数据填补过程的示意图；
图5为本申请实施例中一种磁共振氧十七代谢成像装置的一个实施例结构图；
图6为本申请实施例中一种终端设备的示意框图。
 本发明的实施方式
为使得本申请的发明目的、特征、优点能够更加的明显和易懂，下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，下面所描述的实施例仅仅是本申请一部分实施例，而非全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其它实施例，都属于本申请保护的范围。
应当理解，当在本说明书和所附权利要求书中使用时，术语“包括”指示所描述特征、整体、步骤、操作、元素和/或组件的存在，但并不排除一个或多个其它特征、整体、步骤、操作、元素、组件和/或其集合的存在或添加。
还应当理解，在此本申请说明书中所使用的术语仅仅是出于描述特定实施例的目的而并不意在限制本申请。如在本申请说明书和所附权利要求书中所使用的那样，除非上下文清楚地指明其它情况，否则单数形式的“一”、“一个”及“该”意在包括复数形式。
还应当进一步理解，在本申请说明书和所附权利要求书中使用的术语“和/或”是指相关联列出的项中的一个或多个的任何组合以及所有可能组合，并且包括这些组合。
如在本说明书和所附权利要求书中所使用的那样，术语“如果”可以依据上下文被解释为“当...时”或“一旦”或“响应于确定”或“响应于检测到”。类似地，短语“如果确定”或“如果检测到[所描述条件或事件]”可以依据上下文被解释为意指“一旦确定”或“响应于确定”或“一旦检测到[所描述条件或事件]”或“响应于检测到[所描述条件或事件]”。
另外，在本申请的描述中，术语“第一”、“第二”、“第三”等仅用于区分描述，而不能理解为指示或暗示相对重要性。
在本申请实施例中，针对现有的成像方法精度较低的问题，引入不同的径向插值网络分别在各个频率分区上对采样数据进行填补，对于各个不同频率分区上的采样数据具有更强的针对性，填补的数据更加精准，从而有效提高了最终的成像精度。
请参阅图1，本申请实施例中一种磁共振氧十七代谢成像方法的一个实施例可以包括：
步骤101、获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将径向采样数据划分为若干个分区域径向采样数据。
磁共振成像（Magnetic Resonance Imaging，MRI）是当前最先进的医学诊疗手段之一，与X光和X射线计算机断层扫描成像（X-CT）相比，MRI不会产生对人体有害的辐射，同时可以对人体软组织成像，能对多种病变提供早期诊断。其中对于时间分辨率要求较高的动态成像由于MRI的成像速度慢，造成运动伪影，这限制了其广泛应用。MRI采集到的原始数据是频率域数据（即k空间数据），在k空间采样技术层面，为了提高采样速度，研究人员在欠采样、非均匀采样、采样轨迹设计方面做了很多研究。与此相对应的，对于不同采样模式对应的图像域重建研究同样重要。本申请实施例针对的黄金角径向采样与普通径向采样存在区别，由于普通的径向采样在中心的低频区域采样密集，在高频区域采样稀疏，这对重建造成一定的困难。而黄金角径向采样则按到中心的距离对于不同的环状区域采取了不同的采样密度，即在低频区域减少采样辐条，在高频区域增多采样辐条，提高了采样效率。
图2所示为黄金角径向采样轨迹的示意图，黄金角径向采样每次均旋转固定角度111.25度，这个数值是基于黄金分割比例计算得到的，因此被称为黄金角。黄金角径向采样数据的好处是使得采样过程中的任意时刻的k空间都呈现出采样比较均匀的状态，有利于基于任意数量采样线的数据进行图像重建。
执行本申请实施例的终端设备可以是磁共振扫描仪，也可以是其他可以通过有线通信、无线通信、转移存储介质等方式获取到径向采样数据的终端设备。例如，执行主体可以是图像工作站，图像工作站可以通过数据线、wifi、蓝牙等方式与磁共振扫描仪通信连接，获取径向采样数据，执行磁共振成像工作。
在获取到磁共振氧十七代谢的黄金角径向采样数据之后，则可以按照预设的频率分区将径向采样数据划分为若干个分区域径向采样数据。具体的频率分区数目，以及每个频率分区的频率范围均可以根据实际情况进行设置，本申请实施例对此不作具体限定。例如，可以设置低频区、中频区和高频区共三个频率分区，则径向采样数据可对应划分为低频区域径向采样数据、中频区域径向采样数据和高频区域径向采样数据。
步骤102、使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据。
在本申请实施例的一种具体实现中，步骤102可以包括如图3所示的过程：
步骤1021、将各个分区域径向采样数据分别重排列为矩阵形式，得到各个分区域矩阵数据。
以任意一个分区域径向采样数据为例，将其径向辐条数目记为N，将其每根辐条上的采样点数记为M，在数据重排列的过程中，将每根辐条上的采样数据作为矩阵的一行，则可以得到一个尺寸为N×M（N行M列）的矩阵数据，即分区域矩阵数据。
步骤1022、使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据。
由于k空间数据具有区域差异，低频区和高频区对于最终重建出的图像的重要性并不相同。在本申请实施例中，可以预先设置与各个频率分区分别对应的径向插值网络。例如，如图4所示，对于低频区、中频区和高频区这三个频率分区，可以分别设置相对应的径向插值网络（即图中所示的径向插值模块1、径向插值模块2和径向插值模块3）。在本申请实施例中，可以根据实际情况选择使用现有技术中的任意一种神经网络来作为径向插值网络，此处对其不作具体限定。
当径向插值网络对分区域矩阵数据进行处理时，可以将分区域矩阵数据的尺寸扩大。尺寸扩大具体可以包括以下三种情况：
情况一、使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，即增加了辐条数，得到各个分区域增广矩阵数据；其中， α g 为与第g个频率分区对应的增广系数，且 α g 为大于1的整数，其具体取值可以根据实际情况进行设置，本申请实施例对此不作具体限定，1≤g≤G，G为频率分区的数目。
情况二、使用不同的径向插值网络分别将各个分区域矩阵数据的列数扩大至β g 倍，即增加了每根辐条上的采样点数，得到各个分区域增广矩阵数据；其中，β g 为与第g个频率分区对应的增广系数，且β g 为大于1的整数，其具体取值可以根据实际情况进行设置，本申请实施例对此不作具体限定。
情况三、使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，列数扩大至β g 倍，即同时增加了辐条数以及每根辐条上的采样点数，得到各个分区域增广矩阵数据。
在本申请实施例中，可以根据实际情况选择上述的任意一种情况来对各个分区域矩阵数据进行处理。
本申请实施例中所使用的径向插值网络预先经过深度学习的训练过程，此处以其中的第g个径向插值网络（即对应于第g个频率分区，对第g个分区域矩阵数据进行处理的径向插值网络）为例对其训练过程进行详细说明。
首先，构建第g个径向插值网络的训练样本集合。训练样本集合中包括若干个训练样本，每个训练样本包括输入矩阵数据和预期输出矩阵数据，输入矩阵数据与第g个分区域矩阵数据的尺寸一致，预期输出矩阵数据与第g个分区域增广矩阵数据的尺寸一致。
在本申请实施例的一种具体实现中，可以首先获取对应的频率分区的充分采样的数据，即全采数据，然后按照预设的欠采样倍率对全采数据进行数据删除，得到与全采数据对应的欠采样数据。接着，对全采数据进行重排列，得到输入矩阵数据；并对欠采样数据进行重排列，得到预期输出矩阵数据。最后将输入矩阵数据和预期输出矩阵数据组成的训练样本添加入训练样本集合中。在初始状态下，训练样本集合为空，即不存在任何的训练样本，重复上述过程，则可以不断构建出新的训练样本添加入训练样本集合中，直至达到预定的训练样本数目为止。
在完成训练样本集合的构建之后，则可以使用训练样本集合对第g个径向插值网络进行训练，得到训练后的第g个径向插值网络。
在训练的过程中，可以针对训练数据集合中的每个训练样本，使用径向插值网络对该训练样本中的输入矩阵数据进行处理，得到实际输出的矩阵数据，然后根据该训练样本中的预期输出矩阵数据和实际输出的矩阵数据计算训练损失值。训练损失值的具体计算方式可以根据实际情况进行设置，在本申请实施例的一种具体实现中，可以计算预期输出矩阵数据和实际输出的矩阵数据之间的平方误差，并将该平方误差确定为训练损失值。
在计算得到训练损失值之后，则可以根据训练损失值对径向插值网络的模型参数进行调整。在本申请实施例中，假设径向插值网络的模型参数为W1，将训练损失值反向传播修改径向插值网络的模型参数W1，得到修改后的模型参数W2。修改参数之后再继续执行下一次的训练过程，在该次训练过程中，重新计算得到训练损失值，将该训练损失值反向传播修改径向插值网络的模型参数W2，得到修改后的模型参数W3，……，以此类推，不断重复以上过程，每次训练过程均可对模型参数进行修改，直至满足预设的训练条件，其中，训练条件可以是训练次数达到预设的次数阈值，次数阈值可以根据实际情况进行设置，例如，可以将其设置为数千、数万、数十万甚至更大的数值；训练条件也可以是径向插值网络收敛；由于可能出现训练次数还未达到次数阈值，但径向插值网络已经收敛，可能导致重复不必要的工作；或者径向插值网络始终无法收敛，可能导致无限循环，无法结束训练的过程，基于上述两种情况，训练条件还可以是训练次数达到次数阈值或径向插值网络收敛。当满足训练条件，即可得到已训练的径向插值网络。
步骤1023、将各个分区域增广矩阵数据还原为径向辐条形式，得到填补后的径向采样数据。
容易理解地，步骤1023为步骤1021的逆过程，以任意一个分区域增广矩阵数据为例，假设其尺寸为N′× M′（N′行M′列），在数据还原的过程中，将矩阵每一行的数据作为一根辐条上的采样数据，则可以得到对应的分区域填补后的径向采样数据，其中包括N′根辐条，每根辐条上的采样点数为M′。在还原得到各个分区域填补后的径向采样数据之后，则可以将这些数据组合为完整的填补后的径向采样数据。
图4所示即为数据重排列、径向插值网络处理以及数据还原过程的示意图，经过这样的过程，有效实现了频率域的数据填补，最终得到的填补后的径向采样数据较之于原始的径向采样数据具有更高的数据精度。
步骤103、对填补后的径向采样数据进行成像处理，得到目标图像。
具体地，可以对填补后的径向采样数据进行快速傅里叶逆变换处理，从而得到最终的成像结果，也即目标图像。
综上所述，本申请实施例获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将所述径向采样数据划分为若干个分区域径向采样数据；使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据；其中，每个径向插值网络均对应一个频率分区；对所述填补后的径向采样数据进行成像处理，得到目标图像。通过本申请实施例，引入不同的径向插值网络分别在各个频率分区上对采样数据进行填补，对于各个不同频率分区上的采样数据具有更强的针对性，填补的数据更加精准，从而有效提高了最终的成像精度。
应理解，上述实施例中各步骤的序号的大小并不意味着执行顺序的先后，各过程的执行顺序应以其功能和内在逻辑确定，而不应对本申请实施例的实施过程构成任何限定。
对应于上文实施例所述的一种磁共振氧十七代谢成像方法，图5示出了本申请实施例提供的一种磁共振氧十七代谢成像装置的一个实施例结构图。
本实施例中，一种磁共振氧十七代谢成像装置可以包括：
采样数据划分模块501，用于获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将所述径向采样数据划分为若干个分区域径向采样数据；
径向插值处理模块502，用于使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据；其中，每个径向插值网络均对应一个频率分区；
成像处理模块503，用于对所述填补后的径向采样数据进行成像处理，得到目标图像。
在本实施例的一种可能的实现方式中，所述径向插值处理模块可以包括：
数据重排列单元，用于将各个分区域径向采样数据分别重排列为矩阵形式，得到各个分区域矩阵数据；
径向插值处理单元，用于使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据；
数据还原单元，用于将各个分区域增广矩阵数据还原为径向辐条形式，得到所述填补后的径向采样数据。
在本实施例的一种可能的实现方式中，所述磁共振氧十七代谢成像装置还可以包括：
训练样本集合构建模块，用于构建第g个径向插值网络的训练样本集合，1≤g≤G，G为频率分区的数目；所述训练样本集合中包括若干个训练样本，每个训练样本包括输入矩阵数据和预期输出矩阵数据，所述输入矩阵数据与第g个分区域矩阵数据的尺寸一致，所述预期输出矩阵数据与第g个分区域增广矩阵数据的尺寸一致；
径向插值网络训练模块，用于使用所述训练样本集合对第g个径向插值网络进行训练，得到训练后的第g个径向插值网络。
在本实施例的一种可能的实现方式中，所述训练样本集合构建模块可以包括：
全采数据获取单元，用于获取充分采样的全采数据；
欠采样数据处理单元，用于对所述全采数据进行数据删除，得到与所述全采数据对应的欠采样数据；
第一重排列单元，用于对所述全采数据进行重排列，得到所述输入矩阵数据；
第二重排列单元，用于对所述欠采样数据进行重排列，得到所述预期输出矩阵数据；
训练样本添加单元，用于将所述输入矩阵数据和所述预期输出矩阵数据组成的训练样本添加入所述训练样本集合中。
在本申请实施例的第一种具体实现中，所述径向插值处理单元具体用于：使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，得到各个分区域增广矩阵数据；其中， α g 为与第g个频率分区对应的增广系数，且 α g 为大于1的整数。
在本申请实施例的第二种具体实现中，所述径向插值处理单元具体用于：使用不同的径向插值网络分别将各个分区域矩阵数据的列数扩大至β g 倍，得到各个分区域增广矩阵数据；其中，β g 为与第g个频率分区对应的增广系数，且β g 为大于1的整数。
在本申请实施例的第三种具体实现中，所述径向插值处理单元具体用于：使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，列数扩大至β g 倍，得到各个分区域增广矩阵数据。
在本实施例的一种可能的实现方式中，所述成像处理模块具体用于对所述填补后的径向采样数据进行快速傅里叶逆变换处理，得到所述目标图像。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的装置，模块和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述或记载的部分，可以参见其它实施例的相关描述。
图6示出了本申请实施例提供的一种终端设备的示意框图，为了便于说明，仅示出了与本申请实施例相关的部分。
如图6所示，该实施例的终端设备6包括：处理器60、存储器61以及存储在所述存储器61中并可在所述处理器60上运行的计算机程序62。所述处理器60执行所述计算机程序62时实现上述各个磁共振氧十七代谢成像方法实施例中的步骤，例如图1所示的步骤101至步骤103。或者，所述处理器60执行所述计算机程序62时实现上述各装置实施例中各模块/单元的功能，例如图5所示模块501至模块503的功能。
示例性的，所述计算机程序62可以被分割成一个或多个模块/单元，所述一个或者多个模块/单元被存储在所述存储器61中，并由所述处理器60执行，以完成本申请。所述一个或多个模块/单元可以是能够完成特定功能的一系列计算机程序指令段，该指令段用于描述所述计算机程序62在所述终端设备6中的执行过程。
本领域技术人员可以理解，图6仅仅是终端设备6的示例，并不构成对终端设备6的限定，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件，例如所述终端设备6还可以包括输入输出设备、网络接入设备、总线等。
所述处理器60可以是中央处理单元（Central Processing Unit，CPU），还可以是其它通用处理器、数字信号处理器（Digital Signal Processor，DSP）、专用集成电路（Application Specific Integrated Circuit，ASIC）、现场可编程门阵列（Field-Programmable Gate Array，FPGA）或者其它可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
所述存储器61可以是所述终端设备6的内部存储单元，例如终端设备6的硬盘或内存。所述存储器61也可以是所述终端设备6的外部存储设备，例如所述终端设备6上配备的插接式硬盘，智能存储卡（Smart Media Card，SMC），安全数字（Secure Digital，SD）卡，闪存卡（Flash Card）等。进一步地，所述存储器61还可以既包括所述终端设备6的内部存储单元也包括外部存储设备。所述存储器61用于存储所述计算机程序以及所述终端设备6所需的其它程序和数据。所述存储器61还可以用于暂时地存储已经输出或者将要输出的数据。
所属领域的技术人员可以清楚地了解到，为了描述的方便和简洁，仅以上述各功能单元、模块的划分进行举例说明，实际应用中，可以根据需要而将上述功能分配由不同的功能单元、模块完成，即将所述装置的内部结构划分成不同的功能单元或模块，以完成以上描述的全部或者部分功能。实施例中的各功能单元、模块可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中，上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。另外，各功能单元、模块的具体名称也只是为了便于相互区分，并不用于限制本申请的保护范围。上述系统中单元、模块的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在上述实施例中，对各个实施例的描述都各有侧重，某个实施例中没有详述或记载的部分，可以参见其它实施例的相关描述。
本领域普通技术人员可以意识到，结合本文中所公开的实施例描述的各示例的单元及算法步骤，能够以电子硬件、或者计算机软件和电子硬件的结合来实现。这些功能究竟以硬件还是软件方式来执行，取决于技术方案的特定应用和设计约束条件。专业技术人员可以对每个特定的应用来使用不同方法来实现所描述的功能，但是这种实现不应认为超出本申请的范围。
在本申请所提供的实施例中，应该理解到，所揭露的装置/终端设备和方法，可以通过其它的方式实现。例如，以上所描述的装置/终端设备实施例仅仅是示意性的，例如，所述模块或单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通讯连接可以是通过一些接口，装置或单元的间接耦合或通讯连接，可以是电性，机械或其它的形式。
所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部单元来实现本实施例方案的目的。
另外，在本申请各个实施例中的各功能单元可以集成在一个处理单元中，也可以是各个单元单独物理存在，也可以两个或两个以上单元集成在一个单元中。上述集成的单元既可以采用硬件的形式实现，也可以采用软件功能单元的形式实现。
所述集成的模块/单元如果以软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读存储介质中。基于这样的理解，本申请实现上述实施例方法中的全部或部分流程，也可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一计算机可读存储介质中，该计算机程序在被处理器执行时，可实现上述各个方法实施例的步骤。其中，所述计算机程序包括计算机程序代码，所述计算机程序代码可以为源代码形式、对象代码形式、可执行文件或某些中间形式等。所述计算机可读存储介质可以包括：能够携带所述计算机程序代码的任何实体或装置、记录介质、U盘、移动硬盘、磁碟、光盘、计算机存储器、只读存储器（ROM，Read-Only Memory）、随机存取存储器（RAM，Random Access Memory）、电载波信号、电信信号以及软件分发介质等。需要说明的是，所述计算机可读存储介质包含的内容可以根据司法管辖区内立法和专利实践的要求进行适当的增减，例如在某些司法管辖区，根据立法和专利实践，计算机可读存储介质不包括电载波信号和电信信号。
以上所述实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围，均应包含在本申请的保护范围之内。

Claims (14)

  1. 一种磁共振氧十七代谢成像方法，其特征在于，包括：
    获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将所述径向采样数据划分为若干个分区域径向采样数据；
    使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据；其中，每个径向插值网络均对应一个频率分区；
    对所述填补后的径向采样数据进行成像处理，得到目标图像。
  2. 根据权利要求1所述的磁共振氧十七代谢成像方法，其特征在于，所述使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据，包括：
    将各个分区域径向采样数据分别重排列为矩阵形式，得到各个分区域矩阵数据；
    使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据；
    将各个分区域增广矩阵数据还原为径向辐条形式，得到所述填补后的径向采样数据。
  3. 根据权利要求2所述的磁共振氧十七代谢成像方法，其特征在于，在使用不同的径向插值网络分别对各个分区域矩阵数据进行处理之前，还包括：
    构建第g个径向插值网络的训练样本集合，1≤g≤G，G为频率分区的数目；所述训练样本集合中包括若干个训练样本，每个训练样本包括输入矩阵数据和预期输出矩阵数据，所述输入矩阵数据与第g个分区域矩阵数据的尺寸一致，所述预期输出矩阵数据与第g个分区域增广矩阵数据的尺寸一致；
    使用所述训练样本集合对第g个径向插值网络进行训练，得到训练后的第g个径向插值网络。
  4. 根据权利要求3所述的磁共振氧十七代谢成像方法，其特征在于，所述构建第g个径向插值网络的训练样本集合，包括：
    获取充分采样的全采数据；
    对所述全采数据进行数据删除，得到与所述全采数据对应的欠采样数据；
    对所述全采数据进行重排列，得到所述输入矩阵数据；
    对所述欠采样数据进行重排列，得到所述预期输出矩阵数据；
    将所述输入矩阵数据和所述预期输出矩阵数据组成的训练样本添加入所述训练样本集合中。
  5. 根据权利要求2所述的磁共振氧十七代谢成像方法，其特征在于，所述使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据，包括：
    使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，得到各个分区域增广矩阵数据；其中， α g 为与第g个频率分区对应的增广系数，且 α g 为大于1的整数，1≤g≤G，G为频率分区的数目。
  6. 根据权利要求2所述的磁共振氧十七代谢成像方法，其特征在于，所述使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据，包括：
    使用不同的径向插值网络分别将各个分区域矩阵数据的列数扩大至β g 倍，得到各个分区域增广矩阵数据；其中，β g 为与第g个频率分区对应的增广系数，且β g 为大于1的整数，1≤g≤G，G为频率分区的数目。
  7. 根据权利要求2所述的磁共振氧十七代谢成像方法，其特征在于，所述使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据，包括：
    使用不同的径向插值网络分别将各个分区域矩阵数据的行数扩大至 α g 倍，列数扩大至β g 倍，得到各个分区域增广矩阵数据；其中， α g 和β g 为与第g个频率分区对应的增广系数，且 α g 和β g 均为大于1的整数，1≤g≤G，G为频率分区的数目。
  8. 根据权利要求1至7中任一项所述的磁共振氧十七代谢成像方法，其特征在于，所述对所述填补后的径向采样数据进行成像处理，得到目标图像，包括：
    对所述填补后的径向采样数据进行快速傅里叶逆变换处理，得到所述目标图像。
  9. 一种磁共振氧十七代谢成像装置，其特征在于，包括：
    采样数据划分模块，用于获取磁共振氧十七代谢的黄金角径向采样数据，并按照预设的频率分区将所述径向采样数据划分为若干个分区域径向采样数据；
    径向插值处理模块，用于使用不同的径向插值网络分别对各个分区域径向采样数据进行处理，得到填补后的径向采样数据；其中，每个径向插值网络均对应一个频率分区；
    成像处理模块，用于对所述填补后的径向采样数据进行成像处理，得到目标图像。
  10. 根据权利要求9所述的磁共振氧十七代谢成像装置，其特征在于，所述径向插值处理模块包括：
    数据重排列单元，用于将各个分区域径向采样数据分别重排列为矩阵形式，得到各个分区域矩阵数据；
    径向插值处理单元，用于使用不同的径向插值网络分别对各个分区域矩阵数据进行处理，得到各个分区域增广矩阵数据；
    数据还原单元，用于将各个分区域增广矩阵数据还原为径向辐条形式，得到所述填补后的径向采样数据。
  11. 根据权利要求10所述的磁共振氧十七代谢成像装置，其特征在于，还包括：
    训练样本集合构建模块，用于构建第g个径向插值网络的训练样本集合，1≤g≤G，G为频率分区的数目；所述训练样本集合中包括若干个训练样本，每个训练样本包括输入矩阵数据和预期输出矩阵数据，所述输入矩阵数据与第g个分区域矩阵数据的尺寸一致，所述预期输出矩阵数据与第g个分区域增广矩阵数据的尺寸一致；
    径向插值网络训练模块，用于使用所述训练样本集合对第g个径向插值网络进行训练，得到训练后的第g个径向插值网络。
  12. 根据权利要求11所述的磁共振氧十七代谢成像装置，其特征在于，所述训练样本集合构建模块包括：
    全采数据获取单元，用于获取充分采样的全采数据；
    欠采样数据处理单元，用于对所述全采数据进行数据删除，得到与所述全采数据对应的欠采样数据；
    第一重排列单元，用于对所述全采数据进行重排列，得到所述输入矩阵数据；
    第二重排列单元，用于对所述欠采样数据进行重排列，得到所述预期输出矩阵数据；
    训练样本添加单元，用于将所述输入矩阵数据和所述预期输出矩阵数据组成的训练样本添加入所述训练样本集合中。
  13. 一种计算机可读存储介质，所述计算机可读存储介质存储有计算机程序，其特征在于，所述计算机程序被处理器执行时实现如权利要求1至8中任一项所述的磁共振氧十七代谢成像方法的步骤。
  14. 一种终端设备，包括存储器、处理器以及存储在所述存储器中并可在所述处理器上运行的计算机程序，其特征在于，所述处理器执行所述计算机程序时实现如权利要求1至8中任一项所述的磁共振氧十七代谢成像方法的步骤。
PCT/CN2021/083359 2021-03-23 2021-03-26 磁共振氧十七代谢成像方法、装置、存储介质及终端设备 Ceased WO2022198655A1 (zh)

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN2021082440 2021-03-23
CNPCT/CN2021/082440 2021-03-23
CN202110310782.9 2021-03-23
CN202110310782 2021-03-23

Publications (1)

Publication Number Publication Date
WO2022198655A1 true WO2022198655A1 (zh) 2022-09-29

Family

ID=76278820

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/083359 Ceased WO2022198655A1 (zh) 2021-03-23 2021-03-26 磁共振氧十七代谢成像方法、装置、存储介质及终端设备

Country Status (2)

Country Link
CN (1) CN112967297B (zh)
WO (1) WO2022198655A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPWO2005023108A1 (ja) * 2003-09-05 2007-11-01 株式会社日立メディコ 磁気共鳴イメージング装置
US20150077112A1 (en) * 2012-04-19 2015-03-19 New York University System, method and computer-accessible medium for highly-accelerated dynamic magnetic resonance imaging using golden-angle radial sampling and compressed sensing
US20190054194A1 (en) * 2017-08-17 2019-02-21 Case Western Reserve University System and Method For Evaluation of Subjects Using Magnetic Resonance Imaging and Oxygen-17
WO2020033566A1 (en) * 2018-08-07 2020-02-13 Human Longevity, Inc. Neural networks for volumetric segmentation and parcellated surface representations
CN111492256A (zh) * 2017-12-18 2020-08-04 皇家飞利浦有限公司 运动补偿的磁共振成像

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102017202535A1 (de) * 2017-02-16 2018-08-16 Siemens Healthcare Gmbh Erstellen von MR-Bildern
US11633123B2 (en) * 2017-10-31 2023-04-25 Koninklijke Philips N.V. Motion artifact prediction during data acquisition
CN110133558B (zh) * 2018-02-09 2021-04-30 深圳先进技术研究院 一种磁共振动态成像方法、装置及可读介质
CN111292388A (zh) * 2020-01-22 2020-06-16 上海东软医疗科技有限公司 磁共振成像方法、装置、电子设备、存储介质
CN111445547B (zh) * 2020-03-20 2023-05-09 中国科学院深圳先进技术研究院 一种基于神经网络的网格化磁共振图像重建方法和装置

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPWO2005023108A1 (ja) * 2003-09-05 2007-11-01 株式会社日立メディコ 磁気共鳴イメージング装置
US20150077112A1 (en) * 2012-04-19 2015-03-19 New York University System, method and computer-accessible medium for highly-accelerated dynamic magnetic resonance imaging using golden-angle radial sampling and compressed sensing
US20190054194A1 (en) * 2017-08-17 2019-02-21 Case Western Reserve University System and Method For Evaluation of Subjects Using Magnetic Resonance Imaging and Oxygen-17
CN111492256A (zh) * 2017-12-18 2020-08-04 皇家飞利浦有限公司 运动补偿的磁共振成像
WO2020033566A1 (en) * 2018-08-07 2020-02-13 Human Longevity, Inc. Neural networks for volumetric segmentation and parcellated surface representations

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
WU ZHANXIONG; LI XUN: "Construction of Brain Anatomical Connectivity Network Based on Q-ball Imaging", HANGTIAN YIXUE YU YIXUE GONGCHENG - SPACE MEDICINE AND MEDICALENGINEERING, GAI KAN BIANJIBU, BEIJING, CN, vol. 29, no. 6, 15 December 2016 (2016-12-15), CN , pages 428 - 433, XP009539802, ISSN: 1002-0837, DOI: 10.16289/j.cnki.1002-0837.2016.06.008 *

Also Published As

Publication number Publication date
CN112967297A (zh) 2021-06-15
CN112967297B (zh) 2024-05-03

Similar Documents

Publication Publication Date Title
Lv et al. Transfer learning enhanced generative adversarial networks for multi-channel MRI reconstruction
CN111461983B (zh) 一种基于不同频度信息的图像超分辨率重建模型及方法
CN109581253B (zh) 用于磁共振成像的方法和系统
CN111951344B (zh) 一种基于级联并行卷积网络的磁共振图像重建方法
CN113077527A (zh) 一种基于欠采样的快速磁共振图像重建方法
CN111047660B (zh) 图像重建方法、装置、设备及存储介质
CN112014782B (zh) 磁共振成像方法、装置及计算机存储介质
CN108896943A (zh) 一种磁共振定量成像方法和装置
CN101975936A (zh) 一种基于cs压缩感知技术的快速磁共振成像方法
CN110264538B (zh) 用于核磁共振图像重建的方法、系统和装置
CN111784792A (zh) 基于双域卷积神经网络的快速磁共振重建系统及其训练方法与应用
CN114255291B (zh) 用于磁共振参数定量成像的重建方法、系统
CN106680748A (zh) 磁共振图像获取方法及磁共振系统
CN108010094A (zh) 一种磁共振图像重建方法和装置
CN102096055B (zh) 一种用于磁共振成像非均匀采样数据的快速精确重建方法
CN114913262A (zh) 采样模式与重建算法联合优化的核磁共振成像方法及系统
CN112213673B (zh) 动态磁共振成像方法、装置、重建计算机及磁共振系统
CN114782567B (zh) 基于深度学习的磁共振参数成像方法、装置及智能终端
CN114926559B (zh) 一种基于字典学习思想无衰减校正的pet重建方法
WO2022198655A1 (zh) 磁共振氧十七代谢成像方法、装置、存储介质及终端设备
CN106093814A (zh) 一种基于多尺度低秩模型的心脏磁共振成像方法
Sabbagh et al. Cardiac MRI compressed sensing image reconstruction with a graphics processing unit
CN107703467A (zh) 三维动态磁共振成像的采集方法、装置、设备及存储介质
CN117830373B (zh) 基于加速预条件邻近梯度算法的pet图像重建方法
US12148071B2 (en) Methods and apparatus for MRI reconstruction and data acquisition

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21932276

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21932276

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 21932276

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 28/05/2024)

122 Ep: pct application non-entry in european phase

Ref document number: 21932276

Country of ref document: EP

Kind code of ref document: A1