CN113850883A - Magnetic particle imaging reconstruction method based on attention mechanism - Google Patents
Magnetic particle imaging reconstruction method based on attention mechanism Download PDFInfo
- Publication number
- CN113850883A CN113850883A CN202111198399.5A CN202111198399A CN113850883A CN 113850883 A CN113850883 A CN 113850883A CN 202111198399 A CN202111198399 A CN 202111198399A CN 113850883 A CN113850883 A CN 113850883A
- Authority
- CN
- China
- Prior art keywords
- magnetic particle
- network model
- neural network
- particle imaging
- image
- 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.)
- Pending
Links
- 239000006249 magnetic particle Substances 0.000 title claims abstract description 55
- 238000003384 imaging method Methods 0.000 title claims abstract description 47
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000007246 mechanism Effects 0.000 title claims abstract description 17
- 238000003062 neural network model Methods 0.000 claims abstract description 37
- 230000006870 function Effects 0.000 claims description 36
- 230000004913 activation Effects 0.000 claims description 23
- 238000004088 simulation Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 16
- 238000003491 array Methods 0.000 claims description 6
- 238000011176 pooling Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 2
- 238000006243 chemical reaction Methods 0.000 claims description 2
- 238000005070 sampling Methods 0.000 abstract description 4
- 239000000284 extract Substances 0.000 abstract description 2
- 239000013598 vector Substances 0.000 description 7
- 238000013135 deep learning Methods 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 239000000700 radioactive tracer Substances 0.000 description 2
- 206010028980 Neoplasm Diseases 0.000 description 1
- 230000004931 aggregating effect Effects 0.000 description 1
- 210000003484 anatomy Anatomy 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 238000003759 clinical diagnosis Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000002059 diagnostic imaging Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- WTFXARWRTYJXII-UHFFFAOYSA-N iron(2+);iron(3+);oxygen(2-) Chemical compound [O-2].[O-2].[O-2].[O-2].[Fe+2].[Fe+3].[Fe+3] WTFXARWRTYJXII-UHFFFAOYSA-N 0.000 description 1
- 230000003902 lesion Effects 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 229940031182 nanoparticles iron oxide Drugs 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000035945 sensitivity Effects 0.000 description 1
- 238000002603 single-photon emission computed tomography Methods 0.000 description 1
- 230000002123 temporal effect Effects 0.000 description 1
- 238000003325 tomography Methods 0.000 description 1
Images
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
- G06N3/045—Combinations of networks
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The invention provides a magnetic particle imaging reconstruction method based on an attention mechanism, which utilizes the strong computing power of a neural network model to match the acquired large data volume, and extracts effective information in signals by integrating a self-attention mechanism learning, thereby reducing the signal information loss caused by down-sampling or truncation, and realizing the end-to-end reconstruction from one-dimensional frequency domain signals to two-dimensional images.
Description
Technical Field
The invention belongs to the field of magnetic particle imaging, and particularly relates to a magnetic particle imaging reconstruction method from a one-dimensional frequency domain signal to a two-dimensional image based on a deep learning self-attention mechanism.
Background
How to accurately and objectively locate tumors and other lesions in clinical diagnosis and detection has been an international research hotspot and challenging problem. The traditional medical imaging technologies such as CT, MRI, SPECT and the like have the problems of great harm, poor positioning, low precision and the like. In recent years, a brand new tracer-based imaging mode, namely, Magnetic Particle Imaging (MPI), has been proposed, which utilizes a tomography technology to accurately locate a target by detecting the spatial concentration distribution of superparamagnetic iron oxide nanoparticles (SPIOs) harmless to a human body, and has the characteristics of three-dimensional imaging, high spatial and temporal resolution and high sensitivity. In addition, MPI does not show anatomical structures and is free from background signal interference, so that the intensity of the signal is directly proportional to the concentration of the tracer, which is a detection method with potential for medical applications.
At present, MPI mostly adopts original signals of magnetic particle imaging, including one-dimensional time domain signals and one-dimensional frequency domain signals, to carry out imaging reconstruction through a traditional algorithm, such as a system matrix-based or X-space-based reproduction method. The magnetic particle imaging reconstruction method based on deep learning mainly uses a multilayer perceptron structure.
However, the conventional method or the deep learning method has obvious disadvantages, which are mainly reflected in the following aspects: (1) because the amount of originally acquired data is large, the existing method, especially the traditional method, needs to preprocess the data, and increases the reconstruction complexity; (2) the existing preprocessing methods include down sampling or truncation and the like, which all belong to artificial preprocessing methods, can cause the loss of sample information in data, and finally cause the loss of reconstructed image quality; (3) because the lengths of the acquired signals are different under different conditions, a general processing method for extracting key information in different signals is lacked.
Disclosure of Invention
Aiming at the defects in the prior art, the invention provides a magnetic particle imaging reconstruction method based on an attention mechanism, which utilizes the strong computing power of a neural network model to match the acquired large data volume and extracts effective information in signals by integrating a self-attention mechanism learning, thereby reducing the signal information loss caused by down-sampling or truncation and realizing the end-to-end reconstruction from one-dimensional frequency domain signals to two-dimensional images.
The technical scheme of the invention is as follows:
a magnetic particle imaging reconstruction method based on an attention mechanism comprises the following steps:
s1, acquiring a data set, and acquiring a simulation image and a corresponding one-dimensional frequency domain signal based on real magnetic particle imaging;
s2, building a neural network model, and building a neural network model fused with a self-attention mechanism;
s3 training the neural network model, dividing the data generated in the step S1 into a training set and a verification set, training the neural network model, updating parameters of the neural network model and storing the optimal neural network model;
and S4, reconstructing an image, inputting any real magnetic particle imaging one-dimensional frequency domain signal into the optimal neural network model obtained in S3, and obtaining a reconstructed image result from end to end.
Preferably, in S1, the simulation image is a black-and-white binary image, white represents a signal, and black represents a background, the simulation image is input as an actual magnetic particle distribution by using a reconstruction algorithm, and the FFP scanning process is simulated by using a simulation code, so as to obtain a one-dimensional time domain signal and a one-dimensional frequency domain signal corresponding to the simulation image.
Preferably, in S1, the one-dimensional frequency domain signal is preprocessed to obtain an array with a uniform length of 1 × 10000, where each value is a complex number.
Preferably, in S1, the preprocessing method includes: and performing truncation and non-signal section operation on overlong signals, and performing zero filling operation on overlong signals.
Preferably, in S2, the neural network model has a structure: the signal that magnetic particle produced passes through full articulamentum in proper order, activation function, full articulamentum, activation function, and in an energy attention module, then through a reshape layer with one-dimensional signal conversion two-dimensional image, again in proper order through the convolutional layer, activation function, convolutional layer, activation function obtains the output and rebuilds the image.
Preferably, the signals input into the energy attention module are firstly processed by a one-dimensional convolutional layer to obtain multi-channel signals, then information of different channels is aggregated by a global average pooling layer and a global maximum pooling layer, two pooled feature maps are processed by a multi-layer perceptron consisting of three full-connection layers and three activation functions, two corresponding output feature maps obtained by the multi-layer perceptron are connected, and corresponding energy attention results are calculated by a Sigmoid activation function.
Preferably, the S3 further includes, before training the neural network model, processing the input data as follows: extracting the real part and the imaginary part of each complex number in the original one-dimensional frequency domain signal of the magnetic particle imaging 1 x 10000 to obtain two real number arrays with the length of 1 x 10000, wherein the real part and the imaginary part of the magnetic particle imaging are respectively of a structure which is symmetrical about the center, taking the first half of the real part and the imaginary part to obtain two arrays with the length of 1 x 5000, and connecting the normalized real part signal and the imaginary part signal to obtain the array with the length of 1 x 10000.
Preferably, in S3, the processed data is used as input, the simulation image is used as a label image, the neural network model is trained, a mean square error is used as a loss function, the loss function is calculated according to the output result and the label image, parameters of the neural network model are updated, the loss function calculation is performed on the verification set, parameters of the network model that minimizes the loss function are updated and stored, and the optimal neural network model is stored after the iterative training is completed.
Preferably, the loss function is as follows:
wherein I is the result of the image output,
is a label image, and H, W and C represent the height, width and number of channels, respectively, of the feature map.
Preferably, in S4, preprocessing any acquired real magnetic particle imaging one-dimensional frequency domain signal, converting the preprocessed real magnetic particle imaging one-dimensional frequency domain signal into an array with a length of 1 × 10000, and inputting the array into the optimal neural network model obtained in S3 to obtain an accurate magnetic particle distribution image.
Compared with the prior art, the invention has the following beneficial effects:
1. the magnetic particle imaging reconstruction method based on the attention mechanism adopts a deep learning method to complete the magnetic particle imaging reconstruction process, fully utilizes the high-efficiency computing capability of a neural network model, and matches the computing requirements of a large amount of data in the reconstruction process;
2. the constructed convolutional neural network model comprises an energy attention module, and the neural network model fused with a self attention mechanism is beneficial to extracting effective information in signals, reducing the influence of noise signals, reducing the signal information loss caused by down-sampling or truncation, and improving the resolution and accuracy of reconstructed images;
3. the magnetic particle imaging reconstruction method based on the attention mechanism has certain generalization performance on magnetic particle imaging one-dimensional frequency domain signals with different lengths.
Drawings
The invention may be better understood by reference to the following drawings. The components in the figures are not to be considered as drawn to scale, emphasis instead being placed upon illustrating the principles of the invention.
FIG. 1 is a flow chart of a magnetic particle imaging reconstruction method based on an attention mechanism according to the present invention;
FIG. 2 is a schematic diagram of a real part of a magnetic particle imaging one-dimensional frequency domain signal;
FIG. 3 is a schematic diagram of an imaginary part of a magnetic particle imaging one-dimensional frequency domain signal;
FIG. 4 is a schematic structural diagram of a neural network model constructed by the present invention.
Detailed Description
For the purpose of facilitating an understanding and practicing the invention by those of ordinary skill in the art, the invention is described in further detail below with reference to the following detailed description of illustrative embodiments and drawings:
the technical scheme of the magnetic particle imaging reconstruction method based on the attention mechanism is shown in fig. 1, and mainly comprises the steps of data set acquisition, neural network model construction, model training and real magnetic particle one-dimensional frequency domain signal test, wherein the specific implementation scheme is as follows:
acquisition of the S1 dataset: and generating a simulation image, and simulating the magnetic particle imaging result in the actual situation. The simulation image is a black-white binary image, white represents a signal, black represents a background, and simple shapes such as an ellipse and a rectangle are used for simulating the shape of a sample collected in an actual situation. And inputting the generated simulation image as actual magnetic particle distribution by using the existing reconstruction algorithm, and simulating the FFP scanning process by using a simulation code to finally obtain a one-dimensional time domain signal and a one-dimensional frequency domain signal corresponding to the simulation image. The method comprises the following steps of simply preprocessing one-dimensional frequency domain signals to obtain an array with the length being 1 x 10000 uniformly, and comprises the following specific steps: and performing truncation and back non-signal section operation on the overlong signal, and performing zero filling operation on the overlong signal, wherein each value of the signal is a complex number. 10000 groups of the obtained simulation images and the corresponding one-dimensional frequency domain signals are used as training set and verification set data of the neural network.
S2 building a neural network model: building a neural network model fused with a self attention mechanism, wherein the model has the structure as follows: the signal generated by the magnetic particles first passes through the full-link layer, the activation function, the full-link layer, and the activation function in sequence. The number of neurons of the two full-junction layers is 20000 and 10000 respectively, and the two activation functions are both ReLU activation functions; the output from the second activation function is a 1 x 10000 vector, which is the input to the energy attention module. The energy attention module suppresses a noise signal by learning an energy concentrated portion (effective information portion) in the attention signal. Specifically, in the module, firstly, a multi-channel signal is obtained through a one-dimensional convolution layer, the convolution layer comprises 32 convolution kernels with 1 × 5, and after processing, a multi-channel feature vector with 1 × 10000 × 32 is obtained. And then aggregating information of different channels by adopting a global average pooling layer and a global maximum pooling layer to obtain two characteristic vectors of 1 x 10000, and enabling the two characteristic vectors to pass through a multilayer sensing machine, wherein the multilayer sensing machine has the structure that the number of neurons of a full connection layer, an activation function, a full connection layer, an activation function and three full connection layers which are sequentially connected is 10000, 1250 and 10000 respectively. The activation function is ReLU. And adding the output of the multilayer perceptron, namely two 1 x 10000 feature maps, and calculating the output result of the corresponding energy attention module through a Sigmoid activation function. The output of the energy attention module is multiplied by the input of the energy attention module in element level to obtain a feature vector with the length of 1 x 10000 processed by the attention module, then the feature vector with the length of 1 x 10000 is converted into a two-dimensional feature map with the size of 100 x 100 through a reshape layer, and then the feature map passes through a convolution layer which comprises 16 convolution kernels with the size of 3 x 3 and then passes through a ReLU activation function. Then, inputting a convolution layer containing 1 convolution kernel with the size of 1 × 1, and finally obtaining an output two-dimensional reconstruction image through a ReLU activation function.
Training of the S3 neural network model: the data generated in S1 is divided into training and verification sets by 9:1, with the one-dimensional frequency domain signal as input and the simulated image as the label image. Because the one-dimensional frequency domain signal is complex, the signal is simply preprocessed, and the specific method comprises the following steps: and extracting the real part and the imaginary part of each complex number in the original one-dimensional frequency domain signal of the magnetic particle imaging 1 x 10000 to obtain two real number arrays with the length of 1 x 10000. Because the real part and the imaginary part of the magnetic particle imaging are respectively of a structure which is symmetrical about the center, in order to reduce data redundancy and more efficiently train, only the first half of the real part and the imaginary part is taken to obtain two arrays of 1 x 5000, and the normalized real part signal and the normalized imaginary part signal are connected to obtain an array with the length of 1 x 10000, namely the final input vector;
the mean square error is used as a loss function:
wherein I is the result of the image output,
is a label image, and H, W and C represent the height, width and number of channels, respectively, of the feature map.
Calculating a loss function according to the output result of the neural network model and the label image, updating parameters of the neural network model, storing the network model, reducing the loss function through iterative training, iteratively training 300 epochs, and finally storing the optimal model.
S4 testing of the one-dimensional frequency domain signal of the real magnetic particle: firstly, a simulated body is manufactured, then the conventional commercial magnetic particle imaging instrument is utilized to acquire an image of the simulated body, an original one-dimensional frequency domain signal acquired from the instrument is used as a test input signal, and a real image is used as a true value image. The input signal is simply preprocessed and converted into an array with the length of 1 x 10000, a final result is obtained by testing a stored optimal model, and a reconstruction result is evaluated after the final result is compared with a true value image. After the model is verified, the acquired real magnetic particle imaging one-dimensional frequency domain signal is simply preprocessed, and an accurate distribution image of the magnetic particles can be reconstructed through the stored optimal model.
In the present invention, unless otherwise expressly stated or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally formed; can be mechanically or electrically connected; either directly or indirectly through intervening media, either internally or in any other relationship. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
In the present invention, unless otherwise expressly stated or limited, "above" or "below" a first feature means that the first and second features are in direct contact, or that the first and second features are not in direct contact but are in contact with each other via another feature therebetween. Also, a first feature being "on," "above" or "over" a second feature includes the first feature being directly on or obliquely above the second feature, or simply indicating that the first feature is at a higher level than the second feature. A first feature being "under", beneath and "under" a second feature includes the first feature being directly under and obliquely under the second feature, or simply means that the first feature is at a lesser elevation than the second feature.
In the present invention, the terms "first", "second", third "and" fourth "are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. The term "plurality" means two or more unless expressly limited otherwise.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A magnetic particle imaging reconstruction method based on an attention mechanism is characterized by comprising the following steps:
s1, acquiring a data set, and acquiring a simulation image and a corresponding one-dimensional frequency domain signal based on real magnetic particle imaging;
s2, building a neural network model, and building a neural network model fused with a self-attention mechanism;
s3 training the neural network model, dividing the data generated in the step S1 into a training set and a verification set, training the neural network model, updating parameters of the neural network model and storing the optimal neural network model;
and S4, reconstructing an image, inputting any real magnetic particle imaging one-dimensional frequency domain signal into the optimal neural network model obtained in S3, and obtaining a reconstructed image result from end to end.
2. The magnetic particle imaging reconstruction method of claim 1, wherein in S1, the simulation image is a black-and-white binary image, white represents a signal, and black represents a background, and the simulation image is input as an actual magnetic particle distribution by using a reconstruction algorithm, and a one-dimensional time domain signal and a one-dimensional frequency domain signal corresponding to the simulation image are obtained by simulating an FFP scanning process with a simulation code.
3. The magnetic particle imaging reconstruction method of claim 2, wherein in S1, the one-dimensional frequency domain signal is preprocessed to obtain an array with a uniform length of 1 × 10000, each value being a complex number.
4. The magnetic particle imaging reconstruction method according to claim 3, wherein in S1, the preprocessing method is: and performing truncation and non-signal section operation on overlong signals, and performing zero filling operation on overlong signals.
5. The magnetic particle imaging reconstruction method of claim 1, wherein in S2, the structure of the neural network model is: the signal that magnetic particle produced passes through full articulamentum in proper order, activation function, full articulamentum, activation function, and in an energy attention module, then through a reshape layer with one-dimensional signal conversion two-dimensional image, again in proper order through the convolutional layer, activation function, convolutional layer, activation function obtains the output and rebuilds the image.
6. The magnetic particle imaging reconstruction method of claim 5, wherein the signals inputted into the energy attention module are first passed through a one-dimensional convolutional layer to obtain multi-channel signals, then information of different channels is aggregated by using a global average pooling layer and a global maximum pooling layer, two pooled feature maps are passed through a multi-layer perceptron composed of three full-connection layers and three activation functions, two corresponding output feature maps obtained by the multi-layer perceptron are connected, and a corresponding energy attention result is calculated by a Sigmoid activation function.
7. The magnetic particle imaging reconstruction method of claim 1, wherein the step S3 further comprises, before training the neural network model, processing the input data as follows: extracting the real part and the imaginary part of each complex number in the original one-dimensional frequency domain signal of the magnetic particle imaging 1 x 10000 to obtain two real number arrays with the length of 1 x 10000, wherein the real part and the imaginary part of the magnetic particle imaging are respectively of a structure which is symmetrical about the center, taking the first half of the real part and the imaginary part to obtain two arrays with the length of 1 x 5000, and connecting the normalized real part signal and the imaginary part signal to obtain the array with the length of 1 x 10000.
8. The magnetic particle imaging reconstruction method according to claim 7, wherein in S3, the processed data is used as input, the simulated image is used as a label image, the neural network model is trained, a mean square error is used as a loss function, the loss function is calculated according to the output result and the label image, parameters of the neural network model are updated, the loss function calculation is performed on the verification set, the network model parameters which minimize the loss function are updated and stored, and the optimal neural network model is stored after the iterative training is completed.
9. The magnetic particle imaging reconstruction method of claim 8, wherein the loss function is as follows:
wherein I is the result of the image output,
is a label image, and H, W and C represent the height, width and number of channels, respectively, of the feature map.
10. The magnetic particle imaging reconstruction method of claim 1, wherein in S4, any collected real magnetic particle imaging one-dimensional frequency domain signals are preprocessed and converted into an array with a length of 1 × 10000, and then input into the optimal neural network model obtained in S3 to obtain an accurate distribution image of the magnetic particles.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111198399.5A CN113850883A (en) | 2021-10-14 | 2021-10-14 | Magnetic particle imaging reconstruction method based on attention mechanism |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111198399.5A CN113850883A (en) | 2021-10-14 | 2021-10-14 | Magnetic particle imaging reconstruction method based on attention mechanism |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113850883A true CN113850883A (en) | 2021-12-28 |
Family
ID=78978304
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111198399.5A Pending CN113850883A (en) | 2021-10-14 | 2021-10-14 | Magnetic particle imaging reconstruction method based on attention mechanism |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113850883A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114998471A (en) * | 2022-06-22 | 2022-09-02 | 中国科学院自动化研究所 | Magnetic particle imaging reconstruction method based on RecNet model |
| CN115063500A (en) * | 2022-06-10 | 2022-09-16 | 北京航空航天大学 | Magnetic nanoparticle imaging reconstruction method based on generation of countermeasure network |
| CN115640506A (en) * | 2022-11-18 | 2023-01-24 | 中国科学院自动化研究所 | Magnetic particle distribution model reconstruction method and system based on time-frequency spectrum signal enhancement |
| CN115880440A (en) * | 2023-01-31 | 2023-03-31 | 中国科学院自动化研究所 | Magnetic particle three-dimensional reconstruction imaging method based on generation of countermeasure network |
| CN116503507A (en) * | 2023-06-26 | 2023-07-28 | 中国科学院自动化研究所 | Magnetic particle image reconstruction method based on pre-training model |
| CN115063500B (en) * | 2022-06-10 | 2024-06-07 | 北京航空航天大学 | Magnetic nanoparticle imaging reconstruction method based on generation countermeasure network |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020134826A1 (en) * | 2018-12-24 | 2020-07-02 | 深圳先进技术研究院 | Parallel magnetic resonance imaging method and related equipment |
| CN111696168A (en) * | 2020-06-13 | 2020-09-22 | 中北大学 | High-speed MRI reconstruction method based on residual self-attention image enhancement |
| CN112200161A (en) * | 2020-12-03 | 2021-01-08 | 北京电信易通信息技术股份有限公司 | Face recognition detection method based on mixed attention mechanism |
| CN113255738A (en) * | 2021-05-06 | 2021-08-13 | 武汉象点科技有限公司 | Abnormal image detection method based on self-attention generation countermeasure network |
-
2021
- 2021-10-14 CN CN202111198399.5A patent/CN113850883A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020134826A1 (en) * | 2018-12-24 | 2020-07-02 | 深圳先进技术研究院 | Parallel magnetic resonance imaging method and related equipment |
| CN111696168A (en) * | 2020-06-13 | 2020-09-22 | 中北大学 | High-speed MRI reconstruction method based on residual self-attention image enhancement |
| CN112200161A (en) * | 2020-12-03 | 2021-01-08 | 北京电信易通信息技术股份有限公司 | Face recognition detection method based on mixed attention mechanism |
| CN113255738A (en) * | 2021-05-06 | 2021-08-13 | 武汉象点科技有限公司 | Abnormal image detection method based on self-attention generation countermeasure network |
Non-Patent Citations (2)
| Title |
|---|
| DONGDONG YU 等: "CONVOLUTIONAL NEURAL NETWORKS FOR PREDICTING MOLECULAR PROFILES OF NON-SMALL CELL LUNG CANCER", 2017 IEEE 14TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (ISBI 2017), 19 June 2017 (2017-06-19), pages 569 - 572 * |
| 黄浦: "基于注意力机制的神经网络技术研究及应用", 中国博士学位论文全文数据库基础科学辑, 15 August 2020 (2020-08-15), pages 37 - 56 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115063500A (en) * | 2022-06-10 | 2022-09-16 | 北京航空航天大学 | Magnetic nanoparticle imaging reconstruction method based on generation of countermeasure network |
| CN115063500B (en) * | 2022-06-10 | 2024-06-07 | 北京航空航天大学 | Magnetic nanoparticle imaging reconstruction method based on generation countermeasure network |
| CN114998471A (en) * | 2022-06-22 | 2022-09-02 | 中国科学院自动化研究所 | Magnetic particle imaging reconstruction method based on RecNet model |
| CN114998471B (en) * | 2022-06-22 | 2022-12-16 | 中国科学院自动化研究所 | Magnetic particle imaging reconstruction method based on RecNet model |
| US11835602B1 (en) * | 2022-06-22 | 2023-12-05 | Institute Of Automation, Chinese Academy Of Sciences | Magnetic particle imaging (MPI) reconstruction method based on RecNet model |
| US20230417847A1 (en) * | 2022-06-22 | 2023-12-28 | Institute Of Automation, Chinese Academy Of Sciences | Magnetic particle imaging (mpi) reconstruction method based on recnet model |
| CN115640506A (en) * | 2022-11-18 | 2023-01-24 | 中国科学院自动化研究所 | Magnetic particle distribution model reconstruction method and system based on time-frequency spectrum signal enhancement |
| CN115640506B (en) * | 2022-11-18 | 2023-03-21 | 中国科学院自动化研究所 | Magnetic particle distribution model reconstruction method and system based on time-frequency spectrum signal enhancement |
| CN115880440A (en) * | 2023-01-31 | 2023-03-31 | 中国科学院自动化研究所 | Magnetic particle three-dimensional reconstruction imaging method based on generation of countermeasure network |
| CN115880440B (en) * | 2023-01-31 | 2023-04-28 | 中国科学院自动化研究所 | Magnetic particle three-dimensional reconstruction imaging method based on generation countermeasure network |
| CN116503507A (en) * | 2023-06-26 | 2023-07-28 | 中国科学院自动化研究所 | Magnetic particle image reconstruction method based on pre-training model |
| CN116503507B (en) * | 2023-06-26 | 2023-08-29 | 中国科学院自动化研究所 | Magnetic particle image reconstruction method based on pre-training model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113850883A (en) | Magnetic particle imaging reconstruction method based on attention mechanism | |
| CN110458813B (en) | Image area positioning method and device and medical image processing equipment | |
| WO2020206755A1 (en) | Ray theory-based method and system for ultrasound ct image reconstruction | |
| CN111553892B (en) | Lung nodule segmentation calculation method, device and system based on deep learning | |
| CN110731773B (en) | Abnormal electrocardiogram screening method based on fusion of global and local depth features of electrocardiogram | |
| CN113628296B (en) | Magnetic particle imaging reconstruction method from time-frequency domain signal to two-dimensional image | |
| CN115496771A (en) | Brain tumor segmentation method based on brain three-dimensional MRI image design | |
| CN112488976B (en) | Multi-modal medical image fusion method based on DARTS network | |
| CN114037714B (en) | 3D MR and TRUS image segmentation method for prostate system puncture | |
| CN110297041A (en) | A kind of 3D woven composite defect inspection method based on FCN and GRU | |
| JP7492640B1 (en) | Brain network data feature extraction method and device | |
| CN112364920A (en) | Thyroid cancer pathological image classification method based on deep learning | |
| US11816767B1 (en) | Method and system for reconstructing magnetic particle distribution model based on time-frequency spectrum enhancement | |
| CN115496720A (en) | Gastrointestinal cancer pathological image segmentation method based on ViT mechanism model and related equipment | |
| CN112150564A (en) | Medical image fusion algorithm based on deep convolutional neural network | |
| CN112465924B (en) | Rapid medical image reconstruction method based on multi-feature fusion | |
| CN113935957A (en) | Medical image comparison method and device, electronic equipment and storage medium | |
| CN116152235A (en) | Cross-modal synthesis method for medical image from CT (computed tomography) to PET (positron emission tomography) of lung cancer | |
| CN116228520A (en) | Image compressed sensing reconstruction method and system based on transform generation countermeasure network | |
| CN115294300A (en) | Multi-branch attention prior parameterized finite projection fast fluorescence tomography reconstruction method | |
| CN114494952B (en) | Mammary gland MRI image time sequence generation method based on perception loss | |
| CN114529519A (en) | Image compressed sensing reconstruction method and system based on multi-scale depth cavity residual error network | |
| CN114569116A (en) | Three-channel image and transfer learning-based ballistocardiogram ventricular fibrillation auxiliary diagnosis system | |
| CN114119558A (en) | Method for automatically generating nasopharyngeal carcinoma image diagnosis structured report | |
| Yang et al. | An EIT image reconstruction method based on DenseNet with multi-scale convolution |
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 |