CN113763499B - Nuclear magnetic resonance image reconstruction method, device, equipment and medium - Google Patents
Nuclear magnetic resonance image reconstruction method, device, equipment and medium Download PDFInfo
- Publication number
- CN113763499B CN113763499B CN202110995948.5A CN202110995948A CN113763499B CN 113763499 B CN113763499 B CN 113763499B CN 202110995948 A CN202110995948 A CN 202110995948A CN 113763499 B CN113763499 B CN 113763499B
- Authority
- CN
- China
- Prior art keywords
- space data
- frequency
- image
- network
- magnetic resonance
- 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.)
- Active
Links
- 238000005481 NMR spectroscopy Methods 0.000 title claims abstract description 126
- 238000000034 method Methods 0.000 title claims abstract description 118
- 230000006870 function Effects 0.000 claims description 39
- 238000012545 processing Methods 0.000 claims description 16
- 238000004590 computer program Methods 0.000 claims description 13
- 230000000750 progressive effect Effects 0.000 claims description 10
- 230000001360 synchronised effect Effects 0.000 claims description 7
- 238000003384 imaging method Methods 0.000 description 7
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000004891 communication Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012804 iterative process Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000009977 dual effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
- 238000009827 uniform distribution Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T11/00—2D [Two Dimensional] image generation
- G06T11/003—Reconstruction from projections, e.g. tomography
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2211/00—Image generation
- G06T2211/40—Computed tomography
- G06T2211/424—Iterative
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/30—Assessment of water resources
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Magnetic Resonance Imaging Apparatus (AREA)
Abstract
The invention provides a nuclear magnetic resonance image reconstruction method, a device, equipment, a medium and a program product, wherein the method comprises the following steps: acquiring K space data corresponding to a measured target; firstly, acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step; in the process of acquiring K space data, after acquiring low-frequency K space data and acquiring high-frequency K space data in each step, inputting the acquired K space data into a depth image priori network with image priori capability, calculating a loss function by using all the K space data and the K space data of images generated by the depth image priori network, iteratively updating network parameters of the depth image priori network through continuously converging the loss function, and taking a generated image of the depth image priori network based on the network parameters obtained by final iteration updating as a reconstructed image. The nuclear magnetic resonance image reconstruction method, the device, the equipment, the medium and the program product provided by the invention improve the nuclear magnetic resonance speed and the reconstructed image quality.
Description
Technical Field
The present invention relates to the field of computer technologies, and in particular, to a method, an apparatus, a device, a medium, and a program product for reconstructing a nuclear magnetic resonance image.
Background
Nuclear magnetic resonance is to detect electromagnetic waves emitted by a detected target through an externally applied gradient magnetic field and a pulse magnetic field, and draw a structural image of the interior of an object according to different attenuations of released energy in different structural environments of the interior of the object. Nuclear magnetic resonance includes the steps of object scanning and image reconstruction, and the scanning process can be completed by using nuclear magnetic resonance equipment.
In the existing nuclear magnetic resonance, a detected target is scanned to obtain original K space data, and then an image reconstruction is carried out by using a traditional method or a deep learning method. The K-space is the dual space of the ordinary space under fourier transform. Due to the principle of scanning imaging under nuclear magnetic resonance, the object scanning process takes a lot of time, and then the original K space data obtained by scanning is subjected to image reconstruction. The reconstruction process using the conventional method or the deep neural network takes a certain time, which eventually results in long overall nuclear magnetic resonance time, and the reconstructed image cannot be immediately checked.
Disclosure of Invention
In order to solve the problems in the prior art, the invention provides a nuclear magnetic resonance image reconstruction method, a device, equipment, a medium and a program product.
The invention provides a nuclear magnetic resonance image reconstruction method, which comprises the following steps: acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step; in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value; and acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the step-by-step acquisition of high-frequency K space data comprises the following steps: and acquiring high-frequency K space data step by step according to the frequency from low to high.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the step-by-step acquisition of high-frequency K space data comprises the following steps: and acquiring high-frequency K space data step by step according to the uniformity of the frequency in the non-scanning frequency region.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the frequency range corresponding to the low-frequency K space data and the high-frequency K space data is smaller than the scanning frequency range of the nuclear magnetic resonance equipment.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the frequency range occupies the range of 1/4 to 1/3 of the scanning frequency range.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the iterative updating of the network parameters of the depth image prior network comprises the following steps: determining the preset number of steps of iteration based on the quality requirement of the reconstructed image and the synchronous requirement of the reconstruction process and the scanning process; calculating a loss function of the depth image prior network based on the true value and the predicted value; and continuously converging the loss function iteration to update the network parameters of the depth image prior network through the iteration processing of the preset step number.
The invention also provides a nuclear magnetic resonance image reconstruction device, which comprises: the K space data acquisition module is used for: acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step; an image reconstruction progressive processing module for: in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value; a reconstructed image acquisition module for: and acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
The invention also provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the nuclear magnetic resonance image reconstruction method as described in any one of the above when executing the program.
The invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of a method of nuclear magnetic resonance image reconstruction as described in any of the above.
The invention also provides a computer program product comprising a computer program which when executed by a processor carries out the steps of a method of reconstructing a nuclear magnetic resonance image as described in any of the preceding.
According to the nuclear magnetic resonance image reconstruction method, the device, the equipment, the medium and the program product, the image reconstruction process is synchronously started after the scanning is carried out to obtain the low-frequency K space data, then the processing of the image reconstruction process is continued after each step of obtaining the high-frequency K space data, so that the image reconstruction can be completed only by inputting the high-frequency K space data obtained in the last step into the depth image priori network until the scanning is finished, and the network parameter iteration update is completed; and because the depth image prior network with the image prior capability is utilized for image reconstruction, the quality of the reconstructed image is improved; meanwhile, the main contour is acquired by firstly acquiring the low-frequency signal, then acquiring the high-frequency signal, the detail is perfected, the convergence stability of image reconstruction by using the depth image prior network is improved, and the quality of the reconstructed image is further improved.
Drawings
In order to more clearly illustrate the invention or the technical solutions of the prior art, the following description will briefly explain the drawings used in the embodiments or the description of the prior art, and it is obvious that the drawings in the following description are some embodiments of the invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of a nuclear magnetic resonance image reconstruction method provided by the invention;
FIG. 2 is a second flow chart of the method for reconstructing a nuclear magnetic resonance image according to the present invention;
FIG. 3 is a third flow chart of the method for reconstructing a nuclear magnetic resonance image according to the present invention;
FIG. 4 is a graph showing the effect of the present invention;
fig. 5 is a schematic structural diagram of a nmr image reconstruction apparatus provided by the present invention;
fig. 6 is a schematic structural diagram of an electronic device provided by the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The following describes a method, apparatus, device, medium and program product for reconstructing a nuclear magnetic resonance image according to the present invention with reference to fig. 1 to 6.
Fig. 1 is a schematic flow chart of a method for reconstructing a nuclear magnetic resonance image according to the present invention. As shown in fig. 1, the method includes:
step 101, acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step;
102, in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value;
Step 103, obtaining a generated image of the network parameters obtained by updating the depth image prior network based on the last iteration, and taking the generated image as a reconstructed image corresponding to the measured target.
It should be noted that, the step numbers 101, 102, 103 are only for convenience of description, and are not used to strictly limit the execution sequence. For example, the processing steps 101 and 102 are performed synchronously.
The invention adopts progressive learning, namely, the image reconstruction process is started at the same time of acquisition without waiting for the completion of the acquisition of all K space data and entering the image reconstruction process. And in the scanning process, scanning the detected target by using nuclear magnetic resonance equipment to acquire corresponding K space data. In the reconstruction process, progressive reconstruction is performed by using a depth image prior network with image prior capability. The progressive reconstruction process is actually a process in which the loss function of the depth image prior network gradually converges. The depth image prior network is a depth neural network with image prior capability. The depth image prior network can output a certain image based on network parameters thereof, and the loss function is set as the difference between the input K space data and the K space data acquired based on the image output by the depth image prior network. Corresponding K space data can be obtained by performing forward Fourier transform on the image output by the depth image prior network. The network parameters of the depth image prior network are updated by continuously converging the loss function, and the image output by the depth image prior network based on the finally obtained network parameters is used as a reconstruction image. The depth image prior network does not need training in advance.
Because the K space data describes frequency domain information and the low frequency information contains more main contour information, the loss function convergence process of scanning the low frequency information first and then starting image reconstruction is set, then the high frequency information is gradually increased, a section of loss function convergence is continued after a part of high frequency information is added, and finally, network parameter iteration update is basically completed when the information is acquired according to the preset requirement, and the final reconstructed image is obtained after waiting for an iteration moment.
In the present invention, the image reconstruction process can be started synchronously after a short period of time from the start of scanning, without waiting for the complete end of scanning. Thus, after the scanning is finished, the reconstruction can be finished after a small period of time, so that the waiting time of the scanning is utilized, and the integral imaging process is quickened. And meanwhile, the main body contour is obtained by scanning the low-frequency information, and then the refinement treatment is performed by scanning the high-frequency information, so that the convergence stability of image reconstruction by using the depth image prior network is improved.
The specific implementation process of the invention is as follows:
firstly, low-frequency signal acquisition and depth image priori network initial iteration. The low frequency and the high frequency refer to scanning frequencies of the nuclear magnetic resonance equipment, and data of corresponding frequencies or frequency bands, such as low-frequency K space data corresponding to a detected target and high-frequency K space data corresponding to the detected target, are obtained according to different scanning frequencies of the nuclear magnetic resonance equipment.
The nuclear magnetic resonance equipment scans and collects low-frequency partial data first, low-frequency K space data corresponding to a measured object are obtained, the data are input into a depth image prior network for generating an image immediately after the completion of the low-frequency K space data, and a loss function is defined as the difference between the K space data of the generated image of the neural network after forward Fourier transform and the actually input K space data. The network parameters of the depth image prior network have random initial values, and the depth image prior network can generate corresponding images based on the network parameters. Because the obtained low-frequency K space data is input into the depth image prior network at this time, the loss function is calculated by taking the low-frequency K space data as a true value and taking K space data of an image generated by the depth image prior network based on initial network parameters as a predicted value. And (3) converging the loss function through parameter iteration updating, and generating a fuzzy contour form picture of the measured target by the depth image prior network after the iteration updating is completed.
And the nuclear magnetic resonance equipment continuously scans and collects high-frequency partial data, namely, high-frequency K space data, while the nuclear magnetic resonance equipment inputs the acquired low-frequency K space data into the depth image priori network. The invention acquires high-frequency K space data in a step-by-step mode when acquiring high-frequency signals.
In the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring corresponding high-frequency K space data in each step, inputting the high-frequency K space data into a depth image priori network, taking the set of low-frequency K space data and high-frequency K space data which are already input into the depth image priori network as a true value, taking K space data of an image generated by the depth image priori network based on the latest updated network parameters as a predicted value, calculating a loss function of the depth image priori network, and iteratively updating the network parameters of the depth image priori network by continuously converging the loss function. The frequency band of each step of obtaining the high-frequency K space data can be set according to the requirement, or a random mode can be adopted.
For the scanning of the high-frequency part, new data is input into the depth image prior network immediately after the scanning acquisition of the high-frequency K space data of each step is completed. The loss function is updated as the difference between the K-space data of the generated image of the neural network after forward Fourier transform and all the previous input K-space data. The new loss function is converged by parameter iteration update, and after the iteration update is completed, the neural network can generate a new image which becomes clearer. While the nuclear magnetic resonance equipment inputs new data into the depth image prior network, the nuclear magnetic resonance equipment continuously scans and collects the residual high-frequency part data.
The residual high-frequency part is scanned for multiple times according to the sequence of each step and is continuously input into the depth image prior network, and new iteration and parameter update are carried out on the network parameters of the depth image prior network after each input. After all scanning is completed according to the set frequency range, the high-frequency K space data acquired at the last time is input into the depth image priori network, and after the depth image priori network finishes the final iteration update, the depth image priori network generates an image according to the network parameters obtained by the final iteration update, and the generated image is used as a reconstruction image corresponding to the tested object, so that the final image reconstruction is finished.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the image reconstruction process is synchronously started after the scanning is carried out to obtain the low-frequency K space data, then the processing of the image reconstruction process is continued after the high-frequency K space data is obtained in each step, so that the image reconstruction can be completed only by inputting the high-frequency K space data obtained in the last step into a depth image priori network until the scanning is finished and completing the iterative updating of network parameters; and because the depth image prior network with the image prior capability is utilized for image reconstruction, the quality of the reconstructed image is improved; meanwhile, the main contour is acquired by firstly acquiring the low-frequency signal, then acquiring the high-frequency signal, the detail is perfected, the convergence stability of image reconstruction by using the depth image prior network is improved, and the quality of the reconstructed image is further improved.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the step-by-step acquisition of high-frequency K space data comprises the following steps: and acquiring high-frequency K space data step by step according to the frequency from low to high.
FIG. 2 is a second flow chart of the method for reconstructing a nuclear magnetic resonance image according to the present invention. As shown in fig. 2, the low-frequency K-space data of the center position is first obtained by scanning, then the low-frequency K-space data is input into the depth image prior network, the low-frequency K-space data input into the depth image prior network is used as a true value to calculate a loss function of the depth image prior network, and a predicted value of the calculated loss function is K-space data obtained by the depth image prior network according to an image generated by the initial network parameters. The loss function is used to reduce the difference between the predicted and actual values. And continuously converging the loss function through iteration, so as to update the network parameters of the depth image prior network.
In the iterative process, the nuclear magnetic resonance equipment continuously scans to obtain K space data corresponding to the measured target. After the low-frequency K-space data is acquired, the high-frequency K-space data needs to be acquired in a further step. When the high-frequency K space data is acquired, the high-frequency K space data is acquired step by step from low frequency to high frequency. As shown in fig. 2, when K-space data of the second line is acquired, the corresponding frequency range is the higher frequency portion corresponding to the low frequency portion.
After the corresponding high-frequency K space data are obtained in each step, the high-frequency K space data are input into a depth image priori network, a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image priori network is used as a true value, the K space data of an image generated by the depth image priori network based on the latest updated network parameters are used as a predicted value, a loss function of the depth image priori network is calculated, and the network parameters of the depth image priori network are iteratively updated through the continuously converged loss function.
And acquiring a generated image of the network parameters obtained by updating the depth image prior network based on the last iteration, and taking the generated image as a reconstructed image corresponding to the measured target.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the high-frequency K space data is acquired step by step according to the frequency from low to high, so that the regularity of the acquisition of the high-frequency K space data is realized, the iterative update of network parameters is facilitated to be realized steadily, and the quality of the reconstructed image is improved.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the step-by-step acquisition of high-frequency K space data comprises the following steps: and acquiring high-frequency K space data step by step according to the uniformity of the frequency in the non-scanning frequency region.
FIG. 3 is a third flow chart of the method for reconstructing a nuclear magnetic resonance image according to the present invention. As shown in fig. 3, the low-frequency K-space data of the center position is first obtained by scanning, then the low-frequency K-space data is input into the depth image prior network, the low-frequency K-space data input into the depth image prior network is used as a true value to calculate a loss function of the depth image prior network, and a predicted value of the calculated loss function is K-space data obtained by the depth image prior network according to an image generated by the initial network parameters. The loss function is used to reduce the difference between the predicted and actual values. And continuously converging the loss function through iteration, so as to update the network parameters of the depth image prior network.
In the iterative process, the nuclear magnetic resonance equipment continuously scans to obtain K space data corresponding to the measured target. After the low-frequency K-space data is acquired, the high-frequency K-space data needs to be acquired in a further step. And when the high-frequency K space data are acquired, acquiring the high-frequency K space data step by step according to the uniformity of the frequency in the non-scanning frequency area. As shown in fig. 3, when K-space data of the second row is acquired, the corresponding frequency ranges are uniformly distributed over the unscanned area (of course, strictly uniform distribution may not be required).
After the corresponding high-frequency K space data are obtained in each step, the high-frequency K space data are input into a depth image priori network, a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image priori network is used as a true value, the K space data of an image generated by the depth image priori network based on the latest updated network parameters are used as a predicted value, a loss function of the depth image priori network is calculated, and the network parameters of the depth image priori network are iteratively updated through the continuously converged loss function.
And acquiring a generated image of the network parameters obtained by updating the depth image prior network based on the last iteration, and taking the generated image as a reconstructed image corresponding to the measured target.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the high-frequency K space data is acquired step by step according to the uniformity of the frequency in the non-scanning frequency region, so that the regularity of the acquisition of the high-frequency K space data is realized, the iterative update of network parameters is facilitated to be realized steadily, and the quality of the reconstructed image is improved.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the frequency range corresponding to the low-frequency K space data and the high-frequency K space data is smaller than the scanning frequency range of the nuclear magnetic resonance equipment.
Since one nuclear magnetic resonance detection requires a large number of scans of the gradient magnetic field, it takes a long time. Therefore, it is considered that the full scan is not performed, but a part of the frequency segments in the scanning frequency range of the nuclear magnetic resonance equipment are selected for scanning. Since the low frequency portion reflects the subject contour information, a full scan can be performed for the low frequency portion and a partial scan can be performed for the high frequency portion. In the invention, the scanning completion (that is, the high-frequency K space data is acquired in the last step) shown in fig. 2 and 3 refers to the scanning to a set certain proportion of data relative to the complete scanning, and the mode is called downsampling, so that the overall detection speed can be increased, and the waiting time of a patient can be reduced. Therefore, the K-space scan data in the left column of fig. 2 and 3 shows that the uppermost part of the K-space scan data shows that only the center part of the K-space scan data is scanned, and the lowermost part shows that only a part of the scale data of the whole K-space scan data is scanned. In order to reduce the shortage of the acquired data quantity as much as possible, a depth image prior network is introduced for image reconstruction, and as the depth image prior network has a good image prior, the image with still high quality can be recovered according to the downsampled data.
According to the nuclear magnetic resonance image reconstruction method, the frequency range corresponding to the low-frequency K space data and the high-frequency K space data is smaller than the scanning frequency range of nuclear magnetic resonance equipment, so that the integral imaging speed of nuclear magnetic resonance is further improved.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the frequency range occupies the range of 1/4 to 1/3 of the scanning frequency range.
Through verification, when the value range of the ratio of the frequency range corresponding to the low-frequency K space data and the high-frequency K space data to the scanning frequency range of the nuclear magnetic resonance equipment is 1/4 to 1/3, the quality of the reconstructed image can be ensured while the integral imaging speed of nuclear magnetic resonance is improved.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the frequency range corresponding to the low-frequency K space data and the high-frequency K space data occupies the value range of 1/4 to 1/3 of the scanning frequency range of the nuclear magnetic resonance equipment, so that the quality of a reconstructed image is ensured while the integral imaging speed of nuclear magnetic resonance is improved.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the iterative updating of the network parameters of the depth image prior network comprises the following steps: determining the preset number of steps of iteration based on the quality requirement of the reconstructed image and the synchronous requirement of the reconstruction process and the scanning process; calculating a loss function of the depth image prior network based on the true value and the predicted value; and continuously converging the loss function iteration to update the network parameters of the depth image prior network through the iteration processing of the preset step number.
The time required by the nuclear magnetic resonance equipment for scanning the K space data is determined by the nuclear magnetic resonance equipment, and the image reconstruction can be carried out at the same time by utilizing the scanning time as much as possible, so that when the network parameters are updated by inputting new K space data into the depth image prior network, the iteration times can be set according to the acquisition times of the K space data and the time spent by the whole scanning, and the reconstruction process and the scanning process can be synchronized as much as possible on the basis of ensuring the quality of the reconstructed image. Of course, after the high-frequency K space data acquired in the last step is input into the depth image prior network, since the scanning is completed, more steps of iteration can be relatively performed, the last updated network parameters of the depth image prior network are obtained through iteration, and the image generated by the depth image prior network according to the last updated network parameters is used as a reconstruction image.
In the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, calculating a loss function of the depth image priori network, and iteratively updating network parameters of the depth image priori network by continuously converging the loss function; and after the corresponding high-frequency K space data is obtained in each step, the high-frequency K space data is input into a depth image priori network, a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image priori network is used as a true value, the K space data of an image generated by the depth image priori network based on the latest updated network parameters is used as a predicted value, a loss function of the depth image priori network is calculated, and the network parameters of the depth image priori network are iteratively updated through the continuously converged loss function. And acquiring a generated image of the network parameters obtained by updating the depth image prior network based on the last iteration, and taking the generated image as a reconstructed image corresponding to the measured target.
According to the nuclear magnetic resonance image reconstruction method provided by the invention, the preset number of steps of iteration is determined based on the quality requirement of the reconstructed image and the synchronous requirements of the reconstruction process and the scanning process, and the network parameters of the depth image prior network are updated through iteration processing of the preset number of steps by continuously converging the loss function, so that the image reconstruction can be performed by fully utilizing the scanning time on the basis of ensuring the quality of the reconstructed image, and the quality of the reconstructed image is further improved.
Fig. 4 is a graph showing the effect of the nmr image reconstruction method according to the present invention and the nmr image reconstruction method according to the prior art. In the prior art, the reconstructed image is obtained by directly performing inverse Fourier transform on the K space data, as shown in fig. 4, and compared with the nuclear magnetic resonance image reconstruction method provided by the invention, the nuclear magnetic resonance image reconstruction method in the prior art has the advantages of poor definition and lower quality. In the example in fig. 4, in the nuclear magnetic resonance image reconstruction method provided by the invention, only 50 steps are iterated in each intermediate progressive iteration (the iteration process after new K space data is input), 2000 steps are iterated in total, and a reconstructed image with very good quality is generated after 200 seconds. On one hand, the invention can save the scanning time by a downsampling mode, namely, the full data is not scanned; meanwhile, image reconstruction is carried out through a depth priori image network by utilizing the scanning time, and a reconstructed image with higher quality is generated. Therefore, the nuclear magnetic resonance image reconstruction method provided by the invention obviously improves the overall rate of nuclear magnetic resonance, can quickly see the nuclear magnetic resonance result (reconstructed image), avoids long-time waiting, and improves the quality of the reconstructed image.
The following describes a nmr image reconstruction apparatus provided by the present invention, and the nmr image apparatus described below and the nmr image method described above may be referred to correspondingly.
Fig. 5 is a schematic structural diagram of a nmr image reconstruction apparatus according to the present invention. As shown in fig. 5, the apparatus includes a K-space data acquisition module 10, an image reconstruction progressive processing module 20, and a reconstructed image acquisition module 30, wherein:
the K-space data acquisition module 10 is configured to: acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step;
the image reconstruction progressive processing module 20 is configured to: in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value;
The reconstructed image acquisition module 30 is configured to: and acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
According to the nuclear magnetic resonance image reconstruction device, the image reconstruction process is synchronously started after the low-frequency K space data are acquired through scanning, then the processing of the image reconstruction process is continued after the high-frequency K space data are acquired in each step, so that the image reconstruction can be completed only by inputting the high-frequency K space data acquired in the last step into a depth image priori network until the scanning is finished, and the network parameter iteration update is completed; and because the depth image prior network with the image prior capability is utilized for image reconstruction, the quality of the reconstructed image is improved; meanwhile, the main contour is acquired by firstly acquiring the low-frequency signal, then acquiring the high-frequency signal, the detail is perfected, the convergence stability of image reconstruction by using the depth image prior network is improved, and the quality of the reconstructed image is further improved.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the K-space data acquisition module 10 is specifically configured to, when being configured to acquire high-frequency K-space data step by step: and acquiring high-frequency K space data step by step according to the frequency from low to high.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the high-frequency K space data is acquired step by step according to the frequency from low to high, so that the regularity of the acquisition of the high-frequency K space data is realized, the iterative update of network parameters is facilitated to be realized steadily, and the quality of a reconstructed image is improved.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the K-space data acquisition module 10 is specifically configured to, when being configured to acquire high-frequency K-space data step by step: and acquiring high-frequency K space data step by step according to the uniformity of the frequency in the non-scanning frequency region.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the high-frequency K space data is acquired step by step according to the uniformity of the frequency in the non-scanning frequency region, so that the regularity of the acquisition of the high-frequency K space data is realized, the iterative update of network parameters is facilitated to be realized steadily, and the quality of a reconstructed image is improved.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the frequency range corresponding to the low-frequency K space data and the high-frequency K space data is smaller than the scanning frequency range of the nuclear magnetic resonance equipment.
According to the nuclear magnetic resonance image reconstruction device, the frequency range corresponding to the low-frequency K space data and the high-frequency K space data is smaller than the scanning frequency range of nuclear magnetic resonance equipment, so that the integral imaging speed of nuclear magnetic resonance is further improved.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the frequency range occupies the scanning frequency range in a ratio of 1/4 to 1/3.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the frequency range corresponding to the low-frequency K space data and the high-frequency K space data occupies the value range of 1/4 to 1/3 of the scanning frequency range of the nuclear magnetic resonance equipment, so that the quality of a reconstructed image is ensured while the integral imaging speed of nuclear magnetic resonance is improved.
According to the present invention, when the image reconstruction progressive processing module 20 is configured to iteratively update the network parameters of the depth image prior network by continuously converging the loss function, the image reconstruction progressive processing module is specifically configured to: determining the preset number of steps of iteration based on the quality requirement of the reconstructed image and the synchronous requirement of the reconstruction process and the scanning process; calculating a loss function of the depth image prior network based on the true value and the predicted value; and continuously converging the loss function iteration to update the network parameters of the depth image prior network through the iteration processing of the preset step number.
According to the nuclear magnetic resonance image reconstruction device provided by the invention, the preset number of steps of iteration is determined based on the quality requirement of the reconstructed image and the synchronous requirements of the reconstruction process and the scanning process, and the network parameters of the depth image prior network are updated through iteration processing of the preset number of steps by continuously converging the loss function, so that the image reconstruction can be performed by fully utilizing the scanning time on the basis of ensuring the quality of the reconstructed image, and the quality of the reconstructed image is further improved.
Fig. 6 illustrates a physical schematic diagram of an electronic device, as shown in fig. 6, which may include: processor 610, communication interface (Communications Interface) 620, memory 630, and communication bus 640, wherein processor 610, communication interface 620, and memory 630 communicate with each other via communication bus 640. The processor 610 may invoke logic instructions in the memory 630 to perform a method of nuclear magnetic resonance image reconstruction, the method comprising: acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step; in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value; and acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
Further, the logic instructions in the memory 630 may be implemented in the form of software functional units and stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, a server, a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present invention also provides a computer program product, the computer program product comprising a computer program, the computer program being storable on a non-transitory computer readable storage medium, the computer program, when executed by a processor, being capable of performing the nuclear magnetic resonance image reconstruction method provided by the above methods, the method comprising: acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step; in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value; and acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
In yet another aspect, the present invention also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the method for reconstructing a nuclear magnetic resonance image provided by the above methods, the method comprising: acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step; in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value; and acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and are not limiting; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims (9)
1. A method of reconstructing a nuclear magnetic resonance image, comprising:
acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step;
in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value;
And acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
2. A method of reconstructing a nuclear magnetic resonance image according to claim 1, wherein said acquiring high frequency K-space data in steps comprises:
and acquiring high-frequency K space data step by step according to the frequency from low to high.
3. A method of reconstructing a nuclear magnetic resonance image according to claim 1, wherein said acquiring high frequency K-space data in steps comprises:
and acquiring high-frequency K space data step by step according to the uniformity of the frequency in the non-scanning frequency region.
4. The method of reconstructing a nuclear magnetic resonance image according to claim 1, wherein a frequency range corresponding to the low frequency K-space data and the high frequency K-space data is smaller than a scanning frequency range of the nuclear magnetic resonance apparatus.
5. The method of reconstructing a nuclear magnetic resonance image according to claim 4, wherein the ratio of the frequency range to the scanning frequency range is 1/4 to 1/3.
6. The method of nmr image reconstruction according to claim 1, wherein the iteratively updating the network parameters of the depth image prior network comprises:
Determining the preset number of steps of iteration based on the quality requirement of the reconstructed image and the synchronous requirement of the reconstruction process and the scanning process;
calculating a loss function of the depth image prior network based on the true value and the predicted value;
and continuously converging the loss function iteration to update the network parameters of the depth image prior network through the iteration processing of the preset step number.
7. A nuclear magnetic resonance image reconstruction apparatus, comprising:
the K space data acquisition module is used for: acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment; in the process of acquiring K space data corresponding to a detected target by using nuclear magnetic resonance equipment, firstly acquiring low-frequency K space data, and then acquiring high-frequency K space data step by step;
an image reconstruction progressive processing module for: in the process of acquiring K space data corresponding to a measured target by using nuclear magnetic resonance equipment, after acquiring the low-frequency K space data, inputting the low-frequency K space data into a depth image priori network with image priori capability, taking the low-frequency K space data as a true value, and taking K space data of an image generated by the depth image priori network based on network parameters with initial values as a predicted value, and iteratively updating the network parameters of the depth image priori network; after corresponding high-frequency K space data are obtained in each step, inputting the high-frequency K space data into the depth image prior network, taking a set of the low-frequency K space data and the high-frequency K space data which are already input into the depth image prior network as a true value, and iteratively updating the network parameters of the depth image prior network by taking K space data of an image generated by the depth image prior network based on the latest updated network parameters as a predicted value;
A reconstructed image acquisition module for: and acquiring a generated image of the network parameters obtained by the depth image prior network based on the last iteration update, and taking the generated image as a reconstructed image corresponding to the measured target.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the nuclear magnetic resonance image reconstruction method according to any one of claims 1 to 6 when the program is executed.
9. A non-transitory computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor implements the steps of the nuclear magnetic resonance image reconstruction method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110995948.5A CN113763499B (en) | 2021-08-27 | 2021-08-27 | Nuclear magnetic resonance image reconstruction method, device, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110995948.5A CN113763499B (en) | 2021-08-27 | 2021-08-27 | Nuclear magnetic resonance image reconstruction method, device, equipment and medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113763499A CN113763499A (en) | 2021-12-07 |
| CN113763499B true CN113763499B (en) | 2023-08-18 |
Family
ID=78791498
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110995948.5A Active CN113763499B (en) | 2021-08-27 | 2021-08-27 | Nuclear magnetic resonance image reconstruction method, device, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113763499B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8886283B1 (en) * | 2011-06-21 | 2014-11-11 | Stc.Unm | 3D and 4D magnetic susceptibility tomography based on complex MR images |
| KR101886575B1 (en) * | 2017-06-26 | 2018-08-07 | 연세대학교 산학협력단 | Device and Method for Reconstructing Undersampled Magnetic Resonance Image |
| CN110490832A (en) * | 2019-08-23 | 2019-11-22 | 哈尔滨工业大学 | A kind of MR image reconstruction method based on regularization depth image transcendental method |
| CN112927313A (en) * | 2019-12-05 | 2021-06-08 | 上海联影医疗科技股份有限公司 | Magnetic resonance image reconstruction method, magnetic resonance image reconstruction device, computer equipment and readable storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210150783A1 (en) * | 2019-11-19 | 2021-05-20 | Siemens Healthcare Gmbh | Unsupervised learning-based magnetic resonance reconstruction |
-
2021
- 2021-08-27 CN CN202110995948.5A patent/CN113763499B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8886283B1 (en) * | 2011-06-21 | 2014-11-11 | Stc.Unm | 3D and 4D magnetic susceptibility tomography based on complex MR images |
| KR101886575B1 (en) * | 2017-06-26 | 2018-08-07 | 연세대학교 산학협력단 | Device and Method for Reconstructing Undersampled Magnetic Resonance Image |
| CN110490832A (en) * | 2019-08-23 | 2019-11-22 | 哈尔滨工业大学 | A kind of MR image reconstruction method based on regularization depth image transcendental method |
| CN112927313A (en) * | 2019-12-05 | 2021-06-08 | 上海联影医疗科技股份有限公司 | Magnetic resonance image reconstruction method, magnetic resonance image reconstruction device, computer equipment and readable storage medium |
Non-Patent Citations (1)
| Title |
|---|
| Dmitry Ulyanov, Andrea Vedaldi.Deep Image Prior.https://arxiv.org/abs/1711.10925.2020,1-23. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113763499A (en) | 2021-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gözcü et al. | Learning-based compressive MRI | |
| JP5154016B2 (en) | Method for correcting non-uniformity in image and imaging apparatus | |
| CN112001856B (en) | Training method of denoising model, image noise removing method and related device | |
| CN109658468B (en) | Magnetic resonance parameter imaging method, device, equipment and storage medium | |
| CN110095742B (en) | Plane echo imaging method and device based on neural network | |
| AU2015200573B2 (en) | Dynamic cancellation of mri sequencing noise appearing in an ecg signal | |
| US10353038B2 (en) | Magnetic resonance method and apparatus for accelerating magnetic resonance fingerprinting | |
| US9366741B2 (en) | Medical image imaging method, medical diagnostic apparatus using the same, and recording medium therefor | |
| US9507001B2 (en) | Magnetic resonance imaging method | |
| CN113359077A (en) | Magnetic resonance imaging method and related equipment | |
| CN109461192B (en) | Image iterative reconstruction method, device and equipment and storage medium | |
| Ramzi et al. | Density compensated unrolled networks for non-cartesian MRI reconstruction | |
| JP2023155505A (en) | Medical information processing device and medical information processing method | |
| JP7346270B2 (en) | Medical information processing device, medical information processing method, and program | |
| CN113763499B (en) | Nuclear magnetic resonance image reconstruction method, device, equipment and medium | |
| CN109471053B (en) | dielectric characteristic iterative imaging method based on double constraints | |
| CN109633500B (en) | Transverse relaxation map determination method and device and magnetic resonance imaging equipment | |
| CN116466279A (en) | Magnetic resonance imaging method, system, electronic equipment and medium for lung | |
| CN111856364B (en) | Magnetic resonance imaging method, device and system and storage medium | |
| JP2021171490A (en) | Magnetic resonance imaging device and image reconstruction method | |
| US20230142011A1 (en) | Magnetic resonance imaging apparatus, image processing apparatus, and image processing method | |
| US10977838B2 (en) | Method for generating a magnetic resonance image dataset, computer program product, data storage medium, and magnetic resonance system | |
| CN111856365B (en) | Magnetic resonance imaging method, apparatus, system and storage medium | |
| JP6783619B2 (en) | Magnetic resonance imaging device and image analysis method | |
| CN109839608B (en) | Magnetic resonance field map determination method, magnetic resonance field map determination device, electronic apparatus, and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP03 | Change of name, title or address |
Address after: 201114 room 1302, 13 / F, building 16, 2388 Chenhang Road, Minhang District, Shanghai Patentee after: Shanghai Bi Ren Technology Co.,Ltd. Country or region after: China Address before: 201114 room 1302, 13 / F, building 16, 2388 Chenhang Road, Minhang District, Shanghai Patentee before: Shanghai Bilin Intelligent Technology Co.,Ltd. Country or region before: China |
|
| CP03 | Change of name, title or address |