Human body 5.0 T multi-nuclear magnetic resonance imaging index evaluation device

Abstract

The invention discloses a human body 5.0 T multi-nuclear magnetic resonance imaging index evaluation device, which comprises a clinical index module, a multi-core index module, a main control module, a display module and an index accuracy evaluation module, and is characterized in that a brain multi-core index evaluation module is constructed in the index accuracy evaluation module, and is trained based on a training set; and inputting the to-be-identified sample group into the trained brain multi-core index evaluation module to obtain a prediction score. According to the invention, a dual-channel transformer module is adopted, and self-attention is used in non-overlapping local windows, so that the calculation cost is reduced, and local information is allowed to be better paid attention to; according to the invention, collection of multi-nuclear indexes obtained by human body 5.0 T multi-nuclear magnetic resonance imaging can be realized, and clinical indexes are synchronously collected, so that accuracy evaluation of 5.0 T multi-nuclear magnetic resonance imaging indexes is carried out, and reliability and stability of a human body 5.0 T multi-nuclear magnetic resonance imaging system are ensured.

Description

Technical Field

[0001] The invention belongs to the technical field of magnetic resonance imaging, and in particular relates to an index evaluation device for 5.0T multi-nuclear magnetic resonance imaging of a human body. Background Art

[0002] Magnetic resonance imaging (MRI) can obtain information about the structure of living organisms with high resolution and without loss. Its imaging of blood vessels, nerves, diffusion and other functions as well as high resolution of soft tissues are unmatched by other imaging methods. In addition, MRI has become a powerful tool for clinical applications and life science research due to its advantages of no ionizing radiation, no tissue penetration depth limitation, and revealing dynamic information at atomic resolution.

[0003] At present, existing human MRI equipment is single-core ( 1 H). 1 In addition, the human body also contains 23 Na, 31 P and other elements that can reflect human metabolic information have been used in the study of various diseases such as the brain. Therefore, multi-nuclear MRI is the development direction of the next generation of clinical imaging equipment and has the potential to become a new clinical non-invasive imaging method. In addition, the field strength of clinical 3.0T MRI is relatively low and cannot meet the sensitivity requirements of multi-nuclear nuclide detection and early diagnosis of diseases. Therefore, the higher field strength 5.0T multi-nuclear MRI equipment can further improve its sensitivity, signal-to-noise ratio and spatial resolution, thereby increasing the clinical application scenarios of multi-nuclear MRI.

[0004] At present, the human 5.0T MRI system has been put into use in many hospitals, and further layout is being made to realize clinical human multi-core MRI to occupy the commanding heights of the next generation of MRI equipment. After the successful integration of the human 5.0T multi-core MRI system, it is necessary to establish a complete indicator evaluation method to evaluate the accuracy of the obtained multi-core indicators to ensure the reliability and stability of the system. This will provide a basis for the next step of 5.0T multi-core MRI clinical trials and explore the correlation between 5.0T multi-core MRI indicators and diseases. Summary of the invention

[0005] The purpose of the present invention is to provide an index evaluation device for 5.0T multi-nuclear magnetic resonance imaging of the human body in view of the above-mentioned problems existing in the prior art.

[0006] The technical problem solved by the present invention can be achieved by adopting the following technical solutions:

[0007] A device for evaluating the indexes of 5.0T multi-nuclear magnetic resonance imaging of human body, comprising:

[0008] Clinical indicator module, used to receive clinical indicators of each volunteer;

[0009] The multi-core index module is used to receive 5.0T multi-core MRI indexes of each brain region of each volunteer;

[0010] A main control module, used to transmit the clinical indicators obtained by the clinical indicator module and the 5.0T multi-core magnetic resonance imaging indicators obtained by the multi-core indicator module to the indicator accuracy evaluation module, and also used to receive the prediction score of the indicator accuracy evaluation module;

[0011] The indicator accuracy evaluation module is used to construct a brain multi-core indicator evaluation module, train the brain multi-core indicator evaluation module based on the training set, and input the sample group to be identified into the trained brain multi-core indicator evaluation module to obtain a predicted score. The training set includes multiple sample groups, each sample group includes clinical indicators obtained by the clinical indicator module and 5.0T multi-core magnetic resonance imaging indicators obtained by the multi-core indicator module of the same volunteer, and the 5.0T multi-core magnetic resonance imaging indicators of each sample group correspond to a real score.

[0012] The display module is used to display the input clinical indicators and 5.0T multi-nuclear magnetic resonance imaging indicators in real time, and is also used to display the prediction scores corresponding to the clinical indicators and the 5.0T multi-nuclear magnetic resonance imaging indicators.

[0013] As mentioned above, the brain multi-core index evaluation module is a Transformer model that introduces a dual-channel transformer module in the decoder.

[0014] As described above, the dual-channel transformer module includes a first normalization layer, a window-based multi-head self-attention module, a second normalization layer, a multi-head self-attention module, a third normalization layer, a first feedforward neural layer, a fourth normalization layer, a self-attention module, a fifth normalization layer, and a second feedforward neural layer.

[0015] The feature X1 output by the downsampling module of the Transformer model's encoder is processed by the first normalization layer and the window-based multi-head self-attention module, and then concatenated with the feature X1 as the first concatenated feature.

[0016] The feature X2 finally output by the multi-head self-attention module in the decoder of the Transformer model is concatenated with the feature X2 after being processed by the second normalization layer and the multi-head attention module as the second concatenated feature.

[0017] The first splicing feature and the second splicing feature are spliced ​​to obtain a third splicing feature, the third splicing feature is sequentially processed by the third normalization layer and the first feedforward neural layer, and the feature is spliced ​​with the third splicing feature to obtain a fourth splicing feature, the fourth splicing feature is used as the input feature of the fourth normalization layer, the fourth splicing feature is sequentially passed through the fourth normalization layer and the self-attention module to obtain the feature and is spliced ​​with the fourth splicing feature to obtain the fifth splicing feature, the fifth splicing feature is used as the input feature of the fifth normalization layer, the fifth splicing feature is sequentially passed through the fifth normalization layer and the second feedforward neural layer to obtain the feature and is spliced ​​with the fifth splicing feature to obtain the sixth splicing feature, the sixth splicing feature is input to the output layer of the Transformer model, and the prediction score is output after passing through the softmax function of the output layer.

[0018] As mentioned above, the total loss function Loss of the brain multi-core index evaluation module is:

[0019]

[0020] Where N is the number of sample groups, y i is the true score of the 5.0T multi-nuclear MRI index of the i-th sample group; y , i It indicates the predicted score corresponding to the i-th sample group; the clip function limits the predicted score to the range of 1 to 100; W i represents the weight of the i-th sample group.

[0021] As mentioned above, the training of the brain multi-core index evaluation module is based on minimizing the total loss function Loss.

[0022] The main control module as described above is also used to give suggestions on whether to optimize the sampling method and sequence of the 5.0T multi-nuclear magnetic resonance imaging indicators based on the predicted scores and display the suggestions in the display module.

[0023] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to form a clinical indicator module, a multi-core indicator module, a main control module, a display module and an indicator accuracy evaluation module of the above-mentioned indicator evaluation device.

[0024] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a clinical indicator module, a multi-core indicator module, a main control module, a display module and an indicator accuracy evaluation module of the above-mentioned indicator evaluation device.

[0025] A computer program product comprises a computer program, which, when executed by a processor, implements a clinical indicator module, a multi-core indicator module, a main control module, a display module and an indicator accuracy evaluation module of the above-mentioned indicator evaluation device.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] 1. This paper improves the Transformer model and designs a new dual-channel Transformer module. Unlike the traditional Transformer model that uses global self-attention, this paper uses self-attention in non-overlapping local windows, which reduces the computational cost and allows better attention to local information.

[0028] 2. A loss function is designed to introduce weighting factors to better focus on a certain range of scores. At the same time, normalization is combined to make the output range between 1 and 100.

[0029] 3. The present invention can realize the collection of multi-nuclear indices obtained by 5.0T multi-nuclear MRI of the human body, and simultaneously collect clinical indices, so as to evaluate the accuracy of 5.0T multi-nuclear MRI indices, so as to ensure the reliability and stability of the 5.0T multi-nuclear MRI system of the human body. Ultimately, it provides a basis for carrying out the next step of 5.0T multi-nuclear MRI clinical trials and exploring the correlation between 5.0T multi-nuclear MRI indices and diseases. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a structural schematic diagram of the present invention;

[0031] Figure 2 Schematic diagram of the structure of the dual-channel transformer module. DETAILED DESCRIPTION

[0032] In order to facilitate those skilled in the art to understand and implement the present invention, the present invention is further described in detail below in conjunction with implementation examples. It should be understood that the implementation examples described herein are only used to illustrate and explain the present invention and are not used to limit the present invention.

[0033] Embodiment 1:

[0034] The device for evaluating the indexes of 5.0T multi-core magnetic resonance imaging of the human brain comprises a clinical index module, a multi-core index module, a main control module, a display module, and an index accuracy evaluation module.

[0035] The clinical indicator module is used to receive the clinical indicators of each volunteer, including physiological data and clinical results.

[0036] Among them, physiological data includes gender, family history, symptoms, blood pressure, and blood oxygen;

[0037] Clinical results include clinical diagnosis results, blood routine, blood biochemistry, serum sodium level, serum phosphorus level, electroencephalogram results, and brain disease marker levels. Clinical results also include brain CT results, low field 1 H MRI imaging findings and low field 1 The metabolite concentration ratio of each brain region measured by H MRS. Brain CT results include Hu value measured by CT, low field 1 The imaging results of HMRI include tissue homogeneity, perfusion level and apparent diffusion coefficient (ADC) of each brain region, low field 1 The metabolite concentration ratios of various brain regions measured by HMRS include the choline / N-acetylaspartate (Cho / NAA) concentration ratio and the N-acetylaspartate / creatine-phosphocreatine complex (NAA / Cr) concentration ratio.

[0038] The multi-core index module is used to receive the 5.0T multi-core MRI index of each brain region of each volunteer. The 5.0T multi-core MRI index includes:

[0039] 1 H MRS (magnetic resonance spectroscopy) indicators: choline / N-acetylaspartate (Cho / NAA) concentration ratio and N-acetylaspartate / creatine-phosphocreatine complex (NAA / Cr) concentration ratio, etc.

[0040] 23 Na MRI indicators: TSC range, TSC mean, and TSC uniformity;

[0041] 31 P MRS indicators: ratios of various phosphorus-containing compounds, including creatine phosphate / adenosine triphosphate (PCr / ATP) ratio, inorganic phosphate / adenosine triphosphate (Pi / ATP) ratio, creatine phosphate / inorganic phosphate (PCr / Pi) ratio, phosphomonoester / phosphodiesterase (PME / PDE) ratio, etc.

[0042] And combined 1 H MRS and 23 Na MRI obtains the N-acetylaspartate / tissue sodium concentration (NAA / TSC) ratio, etc.

[0043] The main control module is used to transmit the clinical indicators obtained by the clinical indicator module and the 5.0T multi-core magnetic resonance imaging indicators obtained by the multi-core indicator module to the indicator accuracy evaluation module, and is also used to receive the prediction score of the indicator accuracy evaluation module.

[0044] The display module is used to display the input clinical indicators and 5.0T multi-nuclear magnetic resonance imaging indicators in real time to facilitate technicians to view and modify them. It is also used to display the predicted scores corresponding to the clinical indicators and 5.0T multi-nuclear magnetic resonance imaging indicators.

[0045] The indicator accuracy evaluation module is used to construct a brain multi-core indicator evaluation module, train the brain multi-core indicator evaluation module based on the training set, and input the sample group to be identified into the trained brain multi-core indicator evaluation module to obtain a predicted score. The training set includes multiple sample groups, each of which includes clinical indicators obtained by the clinical indicator module and 5.0T multi-core magnetic resonance imaging indicators obtained by the multi-core indicator module of the same volunteer, and the 5.0T multi-core magnetic resonance imaging indicators of each sample group correspond to a true score.

[0046] The brain multi-core index evaluation module is improved based on the Transformer model. The Transformer model includes an input layer, an encoder, a decoder, and an output layer. A dual-channel transformer module is introduced in the decoder of the Transformer model to improve the performance of the brain multi-core index evaluation model when processing sequence data or image data. The dual-channel transformer module includes a first normalization layer, a window-based multi-head self-attention module, a second normalization layer, a multi-head self-attention module, a third normalization layer, a first feedforward neural layer, a fourth normalization layer, a self-attention module, a fifth normalization layer, and a second feedforward neural layer. The dual-channel transformer module is first divided into two paths. The dual-channel transformer module receives the feature X1 output by the downsampling module of the encoder of the Transformer model and the feature X2 finally output by the multi-head self-attention module in the decoder of the Transformer model. One of them uses traditional multi-head self-attention, and the other uses window-based multi-head self-attention. The feature X1 processed by the first normalization layer and the window-based multi-head self-attention module is concatenated with the feature X1 as the first concatenated feature. The feature X2 processed by the second normalization layer and the multi-head attention module is concatenated with the feature X2 as the second concatenated feature. The first concatenated feature and the second concatenated feature are concatenated to obtain the third concatenated feature. The third concatenated feature processed by the third normalization layer and the first feedforward neural layer is concatenated with the third concatenated feature to obtain the fourth concatenated feature. The fourth concatenated feature is used as the input feature of the fourth normalization layer. The fourth concatenated feature is concatenated with the fourth concatenated feature after passing through the fourth normalization layer and the self-attention module to obtain the fifth concatenated feature. The fifth concatenated feature is used as the input feature of the fifth normalization layer. The fifth concatenated feature is concatenated with the fifth concatenated feature after passing through the fifth normalization layer and the second feedforward neural layer to obtain the sixth concatenated feature. The sixth concatenated feature is used as the output feature of the dual-channel transformer module. The output features of the dual-channel transformer module are input into the output layer of the Transformer model, and the predicted score is output after passing through the softmax function of the output layer.

[0047] The total loss function Loss of the brain multi-core index evaluation module is as shown in the formula:

[0048]

[0049] Where N is the number of sample groups input into the brain multi-core index evaluation module, y i is the true score of the 5.0T multi-nuclear MRI index of the i-th sample group (limited to the range of 1 to 100); y ,i It indicates the predicted score corresponding to the i-th sample group; the clip function limits the predicted score to the range of 1 to 100; W i Represents the weight of the i-th sample group, which can be dynamically adjusted according to the position of the target score, and samples close to the boundary can be given a larger weight.

[0050] Based on minimizing the total loss function, the brain multi-core index evaluation module is end-to-end trained using the training set;

[0051] The network learning rate was initialized to 0.0001, the batch size was set to 8, and the network training was performed on the PyTorch platform using the Adam optimizer. The sample group in the training set was input into the brain multi-core index evaluation module to obtain the prediction results, and the network training was performed according to the total loss function set in step 3. The training was stopped after the total number of iterations of the network training reached 200, and the network parameters corresponding to the brain multi-core index evaluation module were saved.

[0052] The sample group to be identified is input into the trained brain multi-core index evaluation module to obtain the predicted score of the sample group to be identified, and then output to the display module for display through the main control module.

[0053] In the main control module, it is also used to give suggestions on whether to optimize the sampling method and sequence of the 5.0T multi-nuclear MRI indicators according to the predicted score and display the suggestions in the display module. For the predicted score obtained for the sample group to be identified, if the predicted score is ≤60 points, it prompts the technicians to optimize the sampling method and sequence of the 5.0T multi-nuclear MRI indicators of the sample group to be identified. If the predicted score is >60 points, it prompts that the 5.0T multi-nuclear MRI indicators of the sample group to be identified have a certain accuracy, and the next step of 5.0T multi-nuclear MRI clinical trials can be further carried out to explore the correlation between the 5.0T multi-nuclear MRI indicators and brain diseases.

[0054] A person of ordinary skill in the art may understand that all or part of the processes of implementing the clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy assessment module of the above-mentioned embodiments may be completed by instructing related hardware through a computer program, and the computer program may be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it may include processes such as the above-mentioned clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy assessment module.

[0055] Embodiment 2:

[0056] In this embodiment, a computer device is also provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy evaluation module in the above-mentioned embodiment 1 are implemented.

[0057] Embodiment 3:

[0058] In this embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy evaluation module in the above-mentioned embodiment 1 are implemented.

[0059] Embodiment 4:

[0060] In this embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, implements the clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy evaluation module in the above-mentioned embodiment 1.

[0061] It should be noted that the embodiments described in the present invention are merely examples of the spirit of the present invention. Those skilled in the art may make various modifications or additions to the described embodiments or replace them in similar ways, but they will not deviate from the spirit of the present invention or exceed the scope defined by the attached claims.

Claims

1. A human body 5.0T multi-nuclear magnetic resonance imaging index evaluation device, characterized in that: include: Clinical indicator module, used to receive clinical indicators of each volunteer; The multi-core index module is used to receive 5.0T multi-core MRI indexes of each brain region of each volunteer; A main control module, used to transmit the clinical indicators obtained by the clinical indicator module and the 5.0T multi-core magnetic resonance imaging indicators obtained by the multi-core indicator module to the indicator accuracy evaluation module, and also used to receive the prediction score of the indicator accuracy evaluation module; The indicator accuracy evaluation module is used to construct a brain multi-core indicator evaluation module, train the brain multi-core indicator evaluation module based on the training set, and input the sample group to be identified into the trained brain multi-core indicator evaluation module to obtain a predicted score. The training set includes multiple sample groups, each sample group includes clinical indicators obtained by the clinical indicator module and 5.0T multi-core magnetic resonance imaging indicators obtained by the multi-core indicator module of the same volunteer, and the 5.0T multi-core magnetic resonance imaging indicators of each sample group correspond to a real score. The display module is used to display the input clinical indicators and 5.0T multi-nuclear magnetic resonance imaging indicators in real time, and is also used to display the prediction scores corresponding to the clinical indicators and the 5.0T multi-nuclear magnetic resonance imaging indicators.

2. The index evaluation device for 5.0T multi-nuclear magnetic resonance imaging of human body according to claim 1, characterized in that: The brain multi-core index evaluation module is a Transformer model that introduces a dual-channel transformer module in the decoder.

3. The index evaluation device for 5.0T multi-nuclear magnetic resonance imaging of human body according to claim 2, characterized in that: The dual-channel transformer module includes a first normalization layer, a window-based multi-head self-attention module, a second normalization layer, a multi-head self-attention module, a third normalization layer, a first feedforward neural layer, a fourth normalization layer, a self-attention module, a fifth normalization layer, and a second feedforward neural layer, The feature X1 output by the downsampling module of the Transformer model's encoder is processed by the first normalization layer and the window-based multi-head self-attention module, and then concatenated with the feature X1 as the first concatenated feature. The feature X2 finally output by the multi-head self-attention module in the decoder of the Transformer model is concatenated with the feature X2 after being processed by the second normalization layer and the multi-head attention module as the second concatenated feature. The first splicing feature and the second splicing feature are spliced ​​to obtain a third splicing feature, the third splicing feature is sequentially processed by the third normalization layer and the first feedforward neural layer, and the feature is spliced ​​with the third splicing feature to obtain a fourth splicing feature, the fourth splicing feature is used as the input feature of the fourth normalization layer, the fourth splicing feature is sequentially passed through the fourth normalization layer and the self-attention module to obtain the feature and is spliced ​​with the fourth splicing feature to obtain the fifth splicing feature, the fifth splicing feature is used as the input feature of the fifth normalization layer, the fifth splicing feature is sequentially passed through the fifth normalization layer and the second feedforward neural layer to obtain the feature and is spliced ​​with the fifth splicing feature to obtain the sixth splicing feature, the sixth splicing feature is input to the output layer of the Transformer model, and the prediction score is output after passing through the softmax function of the output layer.

4. The index evaluation device for 5.0T multi-nuclear magnetic resonance imaging of human body according to claim 3, characterized in that: The total loss function Loss of the brain multi-core index evaluation module is: Where N is the number of sample groups, y i is the true score of the 5.0T multi-nuclear MRI index of the i-th sample group; y , i It indicates the predicted score corresponding to the i-th sample group; the clip function limits the predicted score to the range of 1 to 100; W i represents the weight of the i-th sample group.

5. The index evaluation device for 5.0T multi-nuclear magnetic resonance imaging of human body according to claim 4, characterized in that: The brain multi-core index evaluation module is trained based on minimizing the total loss function Loss.

6. The index evaluation device for 5.0T multi-nuclear magnetic resonance imaging of human body according to claim 1, characterized in that: The main control module is also used to give suggestions on whether to optimize the sampling method and sequence of the 5.0T multi-nuclear magnetic resonance imaging indicators according to the predicted scores and display the suggestions in the display module.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, the clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy evaluation module of the indicator evaluation device according to any one of claims 1 to 7 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy evaluation module of the indicator evaluation device according to any one of claims 1 to 7 are implemented.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the clinical indicator module, multi-core indicator module, main control module, display module and indicator accuracy evaluation module of the indicator evaluation device according to any one of claims 1 to 7 are implemented.

Images