Pseudo data generating device, pseudo data generating method, and pseudo data generating program

Abstract

To generate data more efficiently.SOLUTION: A pseudo data generation device according to an embodiment includes a collection part and a generation part. The collection part collects a dataset having one or higher dimensional data values. The generation part converts one or higher dimensional data values contained in the data set, and generates a pseudo physical parameter related to each of one or more physical quantities.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The embodiments disclosed in this specification and the drawings relate to a pseudo data generating device, a pseudo data generating method, and a pseudo data generating program. [Background technology]

[0002] Machine learning techniques such as deep neural networks are based on the premise of training using a large amount of data, so if the amount of data is insufficient, the expected performance cannot be achieved. In particular, in the medical field, it is difficult to collect a large amount of diverse medical data, including medical images, due to privacy concerns. Therefore, when designing machine learning models for application in the medical field, the amount of training data tends to be insufficient, making it difficult to achieve high-accuracy training. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2019 / 160850 Summary of the Invention [Problem to be solved by the invention]

[0004] One of the problems to be solved by the embodiments disclosed in this specification and the drawings is to generate data more efficiently. However, the problems to be solved by the embodiments disclosed in this specification and the drawings are not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described below can also be positioned as other problems. [Means for solving the problem]

[0005] The pseudo data generating device according to this embodiment includes a collecting unit and a generating unit. The collecting unit collects a data set having one or more dimensional data values. The generating unit converts the one or more dimensional data values ​​included in the data set to generate pseudo physical parameters for each of one or more physical quantities. [Brief explanation of the drawings]

[0006] [Figure 1] FIG. 1 is a block diagram showing a pseudo data generating device according to this embodiment. [Figure 2] FIG. 2 is a flowchart showing an example of the operation of the pseudo data generating device according to this embodiment. [Figure 3] FIG. 3 is a conceptual diagram showing a first example of a specific process for generating pseudo-physical parameters. [Figure 4] FIG. 4 is a diagram showing an example of the use of the pseudo-physical parameters. [Figure 5] FIG. 5 is a conceptual diagram showing a second example of a specific process for generating pseudo-physical parameters. DETAILED DESCRIPTION OF THE INVENTION

[0007] Hereinafter, a pseudo data generating device, a pseudo data generating method, and a pseudo data generating program according to this embodiment will be described with reference to the drawings. In the following embodiments, parts with the same reference numerals perform similar operations, and redundant explanations will be omitted as appropriate. Hereinafter, one embodiment will be described with reference to the drawings.

[0008] (First embodiment) A pseudo data generating device according to a first embodiment will be described with reference to the block diagram of FIG.

[0009] The pseudo data generating device 1 according to the first embodiment includes a processing circuit 2, an input interface 4, a communication interface 6, and a memory 8. The processing circuit 2 includes an acquisition function 21, a determination function 22, a generation function 23, and a simulation function 24. The processing circuit 2 has a processor (not shown) as a hardware resource.

[0010] The acquisition function 21 collects a data set having one or more dimensional data values. The data set may be, for example, two or more dimensional data such as monochrome or color (RGB) video data, or one or more dimensional time series data. The video data is not limited to medical images, but may be any type of video image, such as images of people, landscapes, or animations. The time series data may be, for example, data that changes over time, such as audio data, seismic waveforms, ECG (Electrocardiogram) waveforms, or stock price charts, or a collection of sampled values ​​of the data. For example, if the data set is image data, the one or more dimensional data values ​​include coordinate information and pixel values ​​for each pixel of the image data, and if the data set is time-series data, they include sampling values ​​(plot data) of the time-series data.

[0011] The determination function 22 determines the type of one or more physical quantities for which the pseudo data generating device 1 generates pseudo physical parameters, which are pseudo physical parameters. The generation function 23 converts one or more dimensional data values ​​included in the data set, and generates pseudo-physical parameters for each of the one or more physical quantities determined by the determination function 22. In other words, the pseudo-physical parameters are values ​​(parameters) of physical quantities artificially generated by data processing or the like.

[0012] The simulation function 24 executes a magnetic resonance simulation using pseudo-physical parameters related to one or more physical quantities, and generates pseudo-acquired data that simulates magnetic resonance signals. The magnetic resonance simulation is a method for simulating MR signals acquired by an input pulse sequence. In this embodiment, a simulation using Bloch equations, which are equations that describe the movement and relaxation phenomenon of macroscopic magnetization, will be described as an example of the magnetic resonance simulation method. However, the present invention is not limited to this, and any method that can simulate MR signals from input data may be used, and a specific description of the simulation method itself will be omitted.

[0013] The input interface 4 has circuits for receiving various instructions and information input from a user. The input interface 4 has circuits related to input devices such as a pointing device such as a mouse or a keyboard. Note that the circuits included in the input interface 4 are not limited to circuits related to physical operation components such as a mouse and a keyboard. For example, the input interface 4 may have an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the pseudo data generation device 1 and outputs the received electrical signal to various circuits within the pseudo data generation device 1. The communication interface 6 exchanges data with external devices via wire or wirelessly.

[0014] The memory 8 stores data sets, pseudo-physical parameters, pseudo-collected data, trained models, etc. The memory 8 may be a semiconductor memory element such as a random access memory (RAM), a flash memory, a hard disk drive (HDD), a solid state drive (SSD), an optical disk, etc. The memory 8 may also be a drive or the like that reads and writes various information from and to a portable storage medium such as a CD-ROM drive, a DVD drive, or a flash memory.

[0015] The magnetic resonance simulation related to the simulation function 24 of the processing circuitry 2 may be executed by an external device. In this case, the pseudo physical parameters may be transmitted from the pseudo data generating device 1 to the external device via the communication interface 6, and the external device may execute the magnetic resonance simulation to generate pseudo acquired data. The pseudo data generating device 1 may receive the generated pseudo acquired data from the external device via the communication interface 6 and store it in the memory 8.

[0016] Furthermore, the various functions of the processing circuit 2 may be stored in the memory 8 in the form of programs executable by a computer. In this case, the processing circuit 2 can be said to be a processor that realizes the functions corresponding to each program by reading the programs corresponding to these various functions from the memory 8 and executing them. In other words, the processing circuit 2 in a state in which each program has been read out will have the multiple functions shown in the processing circuit 2 of FIG. 1.

[0017] 1, it has been explained that these various functions are realized by a single processing circuit 2, but it is also possible to configure the processing circuit 2 by combining multiple independent processors, and have each processor execute a program to realize the function. In other words, it is possible that each of the above functions is configured as a program and one processing circuit executes each program, or that a specific function is implemented in a dedicated, independent program execution circuit.

[0018] Next, an example of the operation of the pseudo data generating device 1 according to the first embodiment will be described with reference to the flowchart in Fig. 2. Note that although the case where processing is performed on one data set will be described here, the pseudo data generating device 1 may acquire multiple data sets, perform similar processing on each data set, and generate one or more pseudo physical parameters for each data set.

[0019] In step S201, the processing circuit 2 acquires a data set using the acquisition function 21. A data set may be a set having one or more dimensions, in other words, a plurality of data values ​​for one or more channels. For example, it may be two-dimensional data having one data for each dimension. Furthermore, when the data set is an image and focusing on pixels, it can be said to be a data set in which three-dimensional data including the pixel value of each pixel and two-dimensional coordinate information (for example, three-dimensional array data of [pixel value, x coordinate, y coordinate]) exists for the number of pixels. Similarly, in the case of an RGB image, it can be said to be a data set in which five-dimensional data ([R value, G value, B value, x coordinate, y coordinate]) including the three RGB pixel values ​​of each pixel and two-dimensional coordinate information (for example, R value, G value, B value, x coordinate, y coordinate) exists for the number of pixels.

[0020] In step S202, the processing circuit 2 determines the type of physical quantity of the pseudo-physical parameter to be generated using the determination function 22. The type of physical quantity may be determined based on a user instruction input via the input interface 4, for example, or may be set as a default. Alternatively, a combination of an application for which a pseudo-physical parameter is to be generated and one or more physical quantities required for that application may be stored in advance in the memory 8, and the corresponding one or more physical quantities may be determined when the user selects an application.

[0021] Specifically, for example, when it is desired to generate pseudo-physical parameters of physical quantities required for magnetic resonance simulation using the Bloch equation, the processing circuit 2 may use the determination function 22 to determine physical quantities such as a value M0 proportional to the number of protons, longitudinal relaxation time T1, transverse relaxation time T2, resonance frequency F0, and diffusion coefficient D. The value M0 proportional to the number of protons may be, for example, the value of thermal equilibrium magnetization including the influence of proton density. The diffusion coefficient D is expressed as D xx , D xy , D yy、 D xz Pseudo-physical parameters may be determined for each component of the diffusion tensor.

[0022] In step S203, the processing circuit 2 generates pseudo-physical parameters of the one or more physical quantities determined in step S202 using the generation function 23, based on one or more data values ​​included in the data set. A method for generating the pseudo-physical parameters may involve, for example, converting one or more data values ​​and setting the converted values ​​as pseudo-physical parameters corresponding to the physical quantities determined in step S202. The conversion process by the generation function 23 may be linear or nonlinear. Linear conversion is, for example, a linear sum of one or more data values. Nonlinear conversion is, for example, a process in which, for a model such as a deep neural network, one or more data items are input and a trained model trained to output one or more pseudo-physical parameters is applied to the one or more data values ​​to generate pseudo-physical parameters corresponding to the one or more physical quantities.

[0023] Furthermore, it is not necessary to calculate all pseudo-physical parameters by conversion processing, and if the pseudo-physical parameters do not need to have variations in value as physical quantities, they may be assigned random values ​​or predetermined values. For example, six physical quantities M0, T1, T2, and D xx , D xy and D yy Among them, for M0, T1 and T2, pseudo-physical parameters are generated by conversion processing, and D xx , D xy and D yy may be assigned random values.

[0024] Note that multiple pseudo-physical parameters for one or more physical quantities may be generated for each physical quantity. For example, for a physical quantity M0 proportional to the number of protons, two linear sums with different coefficients may be performed on data values ​​of one or more dimensions to generate two different pseudo-physical parameters.

[0025] Next, a first example of a specific process for generating pseudo-physical parameters by the generating function 23 will be described with reference to the conceptual diagram of FIG. 3 shows a case where a 256×256 pixel color image 31 is assumed as a data set, and a pseudo-physical parameter set 35 is generated, which is a set of pseudo-physical parameters 351 for each physical quantity. The physical quantities assumed here are five physical quantities: M0, T1, T2, F0, and D.

[0026] Since the color image 31 is an image of three channels of RGB, it is decomposed into images of each component of RGB, that is, an R image 32, a G image 33, and a B image 34. The process of decomposing into images of each component can be performed using a general image processing method, and therefore a description thereof will be omitted here.

[0027] Next, the processing circuit 2 uses the generation function 23 to perform a linear transformation on each pixel of the R image 32, G image 33, and B image 34, calculate pseudo-physical parameters 351 for each physical quantity, and generate one pseudo-physical parameter set 35. That is, pixel values ​​at the same coordinates are extracted from each of the R image 32, G image 33, and B image 34, and a linear sum of the extracted three pixel values ​​is calculated. Specifically, if the horizontal direction of the image is the x-axis and the vertical direction of the image is the y-axis, and the coordinate of the upper left corner is (x, y) = (1, 1), pixel values ​​of the pixel corresponding to the coordinate (1, 1) in the R image 32, G image 33, and B image 34 are extracted, and one or more dimensional data values, here three-dimensional data [(pixel value of the R image), (pixel value of the G image), (pixel value of the B image)], are obtained. A pseudo-physical parameter 351, whose value M0 is proportional to the first physical quantity, the number of protons, is generated from the linear sum of the extracted pixel values. For example, it may be calculated using a linear sum such as F(M0)=0.5×(pixel value of R image)+0.3×(pixel value of G image)+0.2×(pixel value of B image).

[0028] Next, the pseudo-physical parameter 352 of the longitudinal relaxation time T1, which is the second physical quantity, is generated using the above-mentioned three-dimensional data. For example, the pseudo-physical parameter 352 can be calculated by a linear sum with different coefficients, such as F'(T1) = 500 × (pixel value of R image) + 1000 × (pixel value of G image) + 200 × (pixel value of B image). In this way, pseudo-physical parameter set 35 can be obtained by generating pseudo-physical parameters of physical quantities such as the transverse relaxation time T2, the resonance frequency F0, and the diffusion coefficient D using the same three-dimensional data.

[0029] Next, the next three-dimensional data is obtained by extracting the pixel value at the coordinate (2,1), which is immediately to the right of the coordinate (1,1), from each of the R image 32, G image 33, and B image 34. For the next three-dimensional data, pseudo-physical parameters are generated sequentially, such as by linearly adding the function F(M0) for pseudo-physical parameter 351 of the value M0 proportional to the number of protons, and by linearly adding the function F'(T1) for pseudo-physical parameter 352 of the longitudinal relaxation T1, to obtain pseudo-physical parameter set 35. By calculating the pseudo-physical parameters of each physical quantity from the pixel value of each pixel in this way, 256×256=65536 pseudo-physical parameter sets 35 can be generated from a color image of 256×256 pixels.

[0030] Furthermore, pseudo-physical parameters may be calculated not from the pixel values ​​of each pixel but from the pixel values ​​of a partial image (also called a patch) extracted from a partial region of the entire image. The pixel values ​​of a patch may be calculated by using the average pixel value of multiple pixels in the patch as the pixel value of each component image (R image, G image, B image). Alternatively, the pixel values ​​of multiple pixels in the patch may be used to generate pseudo-physical parameters by linearly adding them together using a function in which each pixel value in the patch is multiplied by a coefficient. When generating multiple pseudo-physical parameter sets 35 from an image using patches, more pseudo-physical parameter sets 35 can be generated by appropriately setting the so-called patch stride. For example, data values ​​may be extracted from an entire 256x256 image using a stride of "1" for an 8x8 patch, and then data values ​​may be extracted using a stride of "2." Changing the stride changes the combination of pixel values ​​in the patch, allowing more pseudo-physical parameter sets 35 to be generated.

[0031] The coefficients in the above-described functions may be set in any manner as long as the calculated values ​​satisfy the physical constraints of the pseudo-physical parameters. In other words, the pseudo-physical parameters may be determined in any manner as long as the values ​​satisfy the physical constraints. That is, for example, in the case of the longitudinal relaxation time T1, since a negative value is not possible according to the laws of physics, any value that satisfies the constraint T1>0 can be used as the pseudo-physical parameter 352 for the longitudinal relaxation time T1. The pseudo-physical parameters may have a range of values ​​limited according to the circumstances under which the physical quantity can be observed.

[0032] Furthermore, if the data set is time-series data, pseudo-physical parameters can be generated for data values ​​of one or more dimensions in a similar manner by using, for example, multiple sampled values. That is, the pseudo-physical parameters of physical quantities can be calculated by a linear sum such as F(M0) = 0.5 × (first sampled value) + 0.3 × (second sampled value) + 0.2 × (third sampled value).

[0033] In addition to the linear transformation described above, in the case of nonlinear transformation using a trained model such as a neural network, a trained model is prepared in advance that is trained to input pixel values ​​of each component image and output unknown values. The trained model is then applied to the R image, G image, and B image, and the values ​​output from the trained model are used as pseudo-physical parameters. The network structure of the trained model only needs to be able to output some value for the input; for example, the pseudo-physical parameters can be output by applying a generator of Conditional GAN ​​(Generative Adversarial Networks) to an RGB image.

[0034] The number of values ​​output from the trained model may be trained to be greater than the number of types of physical quantities determined in step S202, and during inference of the trained model, outputs corresponding in number to the number of types of physical quantities determined in step S202 may be extracted as appropriate. Alternatively, if the number of types of physical quantities to be generated is known in advance, the model may be trained to obtain the same number of outputs as the number of types.

[0035] Furthermore, there may be one output from the trained model, and by applying the trained model multiple times, pseudo-physical parameters in the same number as the number of types of physical quantities determined in step S202 may be generated. For example, when using a generator such as a Conditional GAN, inputting a random value as a seed can obtain different outputs even if the input pixel values ​​are the same, so the trained model may be applied multiple times.

[0036] Next, an example of how the generated pseudo-physical parameters are used by the simulation function will be described with reference to FIG. For example, a pseudo-physical parameter set 35 generated by the process shown in FIG. 3 and a pulse sequence 41 are incorporated into a magnetic resonance simulation using the Bloch equations and executed. The pulse sequence and pseudo-physical parameters representing physical quantities such as M0, T1, and T2 are incorporated into the Bloch equations, and the MR signals acquired by the input pulse sequence can be theoretically calculated by solving the Bloch equations. In other words, roughly speaking, protons are arranged in a simulation space according to a field of view (FOV), and the precession of the protons is analyzed. If the FOV is 256 mm, 256 × 256 pseudo-physical parameters generated based on each pixel can be arranged according to the FOV. As a result, simulated values ​​of k-space data (hereinafter simply referred to as k-space data 42) can be generated from the pseudo-physical parameter set 35.

[0037] In addition, parameters related to measurement conditions such as transmission sensitivity and reception sensitivity required for magnetic resonance simulation, which are different from pseudo-physical parameters, may be set to predetermined values ​​or may be set to values ​​measured using an actual device.

[0038] The generated k-space data 42 may be used as training data for various machine learning applications. Specifically, the k-space data 42 may be used to train an image reconstruction model that inputs the k-space data 42 and outputs a reconstructed image. The k-space data 42 may also be used to train a model from which data is transferred in transfer learning, in which a trained model trained in one domain is used to train another domain. Alternatively, the k-space data 42 may be used as test data for predicting learning performance or transfer effectiveness. The k-space data 42 may also be used as test data for product testing to determine whether data can be normally transmitted in DICOM (Digital Imaging and Communications in Medicine) format from an MRI (Magnetic Resonance Imaging) device to, for example, a PACS (Picture Archiving and Communication Systems). The k-space data 42 may also be used for research or educational purposes, such as simulating a reconstruction process using the k-space data. For applications that do not require the accuracy of actual clinical data, such as test data, actual clinical data, for example, k-space data actually acquired by an MRI device, may be used in a state where it is mixed with pseudo-acquired data.

[0039] Next, a second example of the pseudo-physical parameter generation process will be described with reference to FIG. Figure 5 shows an example of generating a pseudo-physical parameter set 51 for MRS (Magnetic Resonance Spectroscopy), a type of chemical shift measurement, using a color image 31 as a data set. MRS is a measurement method that captures biochemical information by utilizing minute differences in the resonance frequencies of molecules that are metabolites in a target living body and signal intensities (peaks) that primarily reflect the number of protons, and can visualize the abundance of metabolites in a living body in a measurement target region (VOI: Voxel of Interest). Examples of metabolite molecules include N-acetylaspartic acid (NAA), choline (Cho), and creatine (Cr).

[0040] The pseudo-physical parameter generation method is the same as in the first example described above, and involves converting the R image 32, G image 33, and B image 34 for each pixel to generate pseudo-physical parameters of physical quantities required for MRS acquisition of molecules. The physical quantities required for MRS acquisition include a value M0 proportional to the number of protons, a chemical shift ΔF0 which is the shift from the reference resonance frequency, the longitudinal relaxation time T1 or longitudinal relaxation rate R1 (= 1 / T1), the transverse relaxation time T2 or transverse relaxation rate R2 (= 1 / T2), and the diffusion coefficient D.

[0041] The chemical shift amount ΔF0 is known to be a value determined for each molecule, for example, the main chemical shift amount of NAA is 2.02 ppm, and the main chemical shift amount of Cho is 3.02 ppm. Therefore, the chemical shift amount ΔF0 may be determined, for example, from a lookup table that stores molecules and corresponding chemical shift amounts ΔF0.

[0042] In addition to the molecule-specific chemical shift amount ΔF0, a global ΔF, which is a chemical shift amount that contributes overall to MRS acquisition, may be set, and a predetermined value or a random value may be set for the global ΔF. Furthermore, the diffusion coefficient D may also be set to a predetermined value or a random value.

[0043] Specifically, in Figure 5, for two molecules, "NAA" and "Cho," the chemical shift amount ΔF0 is determined from a lookup table, and default values ​​are given for the global ΔF and diffusion coefficient D. For other physical quantities, such as the value M0 proportional to the number of protons, the longitudinal relaxation rate R1, and the transverse relaxation rate R2, the processing circuit 2 converts the R image 32, G image 33, and B image 34 pixel by pixel using the generation function 23 to generate a pseudo-physical parameter set 51-1 for "NAA" and a pseudo-physical parameter set 51-2 for "Cho."

[0044] Alternatively, the longitudinal relaxation rate R1 and the transverse relaxation rate R2 may be assigned predetermined or random values ​​for each molecule instead of being generated from a data set. That is, the longitudinal relaxation rate R1 and the transverse relaxation rate R2 may be set for the pseudo-physical parameter set "NAA" and the longitudinal relaxation rate R1 and the transverse relaxation rate R2 may be set for the pseudo-physical parameter set "Cho."

[0045] The generated pseudo-physical parameter set, the pulse sequence for MRS acquisition, and the data acquisition conditions for MRS acquisition are incorporated into the Bloch equations, and the Bloch equations are then solved to generate simulated MR signals. Data acquisition conditions include the repetition time (TR), echo time (TE), number of integrations, spectral width, number of samples, data acquisition method, and region-selective pulses. Known pulse sequences for MRS acquisition include point-resolved spectroscopy (PRESS) and stimulated echo acquisition mode (STEAM). The generated MR signals are subjected to preprocessing such as low-pass filtering and then Fourier transform to generate an MRS spectrum.

[0046] Note that the MRS spectrum obtained using the pseudo-physical parameter set described above is based on the pseudo-physical parameters of a specific molecule, and therefore may lack simulated values ​​of signal intensities corresponding to other chemical shift amounts. Therefore, the MRS spectrum may be generated by applying a trained model that has been trained to fill in the missing simulated values.

[0047] A trained model that fills in missing simulation values ​​can be generated by training the model using training data in which, for example, the measured MRS spectrum is used as the correct data and the spectrum of the correct data is blanked (blanked) for the parts other than the signal intensity corresponding to the chemical shift of the metabolite.

[0048] 5, the pseudo-physical parameters obtained for each molecule may be used not only to obtain an MRS spectrum but also to simulate the physical properties of the molecule. Also, the pseudo-physical parameters may be used to generate CSI (Chemical Shift Imaging) by distributing the spectra obtained by MRS simulation for multiple VOIs.

[0049] Although the pseudo data generating device 1 according to this embodiment is assumed to be connected to an MRI device or a PACS server that handles k-space data, it may also generate pseudo data relating to physical quantities handled by other medical image diagnostic devices. For example, it may be connected to any of a CT (Computed Tomography) device, an X-ray device, a PET (Positron Emission Tomography) device, a SPECT (Single Photon Emission Computed Tomography) device, an ultrasound diagnostic device, etc., and generate pseudo physical parameters of physical quantities relating to medical data acquired by the connected medical image diagnostic device.

[0050] Specifically, physical quantities related to projection data obtained by a CT device include the atomic number (effective atomic number) of atoms constituting body tissues, electron density, incident X-ray dose, tube voltage, etc. As in the above case, pseudo-physical parameters related to the physical quantity of atomic number are generated, and pseudo-projection data can be generated by performing projection processing on the generated pseudo-physical parameters according to the incident X-ray dose and the linear attenuation coefficient of X-rays.

[0051] The pseudo data generating device 1 may be installed in at least one of a server, a workstation, and a medical image diagnostic device.

[0052] According to the embodiment described above, pseudo-physical parameters for one or more physical quantities are generated by converting one or more dimensional data values ​​contained in a wide variety of data sets. This allows a large number of parameters related to the physical quantities to be prepared, and a large number of pseudo-collected data sets can be generated by performing, for example, magnetic resonance simulations using the large number of pseudo-physical parameters. This allows a large number of data sets to be prepared even in the medical field, where there is a shortage of data. Therefore, in applications such as image-based machine learning, product testing, and education, when using actual data is unnecessary or difficult, pseudo-physical parameters can be used, thereby improving the accuracy of machine learning and testing.

[0053] In addition, the learning performance of machine learning models for many applications deteriorates in the following order: (1) when the data directly corresponds to the application of the model, (2) when the data is completely compatible with the data, (3) when the data is not completely compatible with the data but has similar properties, and (4) when the data is completely dissimilar (e.g., random data). In other words, by generating pseudo-physical parameters from data with a meaningful structure, such as image data or time-series data, and learning using pseudo-collected data based on those pseudo-physical parameters, the learning performance of machine learning models in tasks such as medical image classification and segmentation can be improved compared to generating pseudo-physical parameters from random data.

[0054] According to at least one of the embodiments described above, data can be generated more efficiently.

[0055] In addition, each function according to the embodiment can be realized by installing a program that executes the above-described processes in a computer such as a workstation and expanding the program in memory. In this case, the program that causes the computer to execute the above-described methods can be stored and distributed on a storage medium such as a magnetic disk (e.g., a hard disk), an optical disk (e.g., a CD-ROM or a DVD), or a semiconductor memory.

[0056] Although several embodiments have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, modifications, and combinations of embodiments can be made without departing from the spirit of the invention. These embodiments and their modifications are included within the scope and spirit of the invention, as well as within the scope of the invention and its equivalents as defined in the claims. [Explanation of symbols]

[0057] 1. Pseudo data generator 2 Processing circuit 4 Input Interface 6 Communication Interface 8. Memory 21 Acquisition Function 22 Decision Function 23 Generation function 24 Simulation Function 31 color images 32 R images 33G Images 34 B Image 35,51,51-1,51-2 Pseudo-physics parameter set 41 Pulse Sequence 42 k-space data 351 Pseudo-Physical Parameters 352 Pseudo-Physical Parameters

Claims

1. a collection unit for collecting an image data set in which data values ​​in one or more dimensions correspond to one pixel; a generator that generates pseudo-physical parameters to be used in a magnetic resonance simulation, each of which is a value of a plurality of physical quantities, by converting one or more dimensional data values ​​of one pixel included in the image data set through data processing; A pseudo data generating device comprising:

2. 2. The pseudo data generating device according to claim 1, wherein the transformation is a linear sum of the data values ​​of one or more dimensions.

3. The pseudo data generating device according to claim 1 , wherein the conversion is a process of applying a trained model that has been trained to input one or more pieces of data and output one or more pseudo physical parameters to the one or more dimensional data values.

4. 4. The pseudo data generating device according to claim 1, wherein the data values ​​of one or more dimensions include coordinate information and pixel values ​​for each pixel of the image data.

5. one of the plurality of physical quantities is a value proportional to the number of protons, The pseudo data generating device according to claim 4 , wherein the generating unit generates a pseudo physical parameter relating to a value proportional to the number of protons from the pixel value for each pixel.

6. a collection unit that collects time series data that is a set of sampling values; a generating unit that generates pseudo-physical parameters to be used in a magnetic resonance simulation, the pseudo-physical parameters being values ​​of a plurality of physical quantities, by converting a plurality of sampled values ​​of the time-series data through data processing; A pseudo data generating device comprising:

7. The pseudo data generating device according to claim 1 , wherein the generating unit generates the pseudo physical parameters by assigning predetermined values ​​or random values ​​to some of the plurality of physical quantities.

8. 6. The pseudo data generating device according to claim 1, wherein the plurality of physical quantities are physical quantities used in a magnetic resonance simulation using Bloch equations.

9. 5. The pseudo data generating device according to claim 1, wherein the plurality of physical quantities include a value proportional to the number of protons, a longitudinal relaxation time or longitudinal relaxation rate, and a transverse relaxation time or transverse relaxation rate.

10. 5. The pseudo data generating device according to claim 1, wherein the plurality of physical quantities are physical quantities used in a simulation relating to chemical shift measurements of molecules including metabolites in a living body.

11. 11. The pseudo data generating device according to claim 10, wherein the generating unit sets a chemical shift amount specific to the molecule for each of the molecules, and generates the pseudo physical parameters for each of the molecules, including a value proportional to the number of protons, a longitudinal relaxation rate or longitudinal relaxation time, and a transverse relaxation rate or transverse relaxation time.

12. 12. The pseudo data generating device according to claim 1, further comprising a simulation unit that executes a magnetic resonance simulation using the pseudo physical parameters related to each of the plurality of physical quantities, and generates pseudo collected data that simulates a magnetic resonance signal.

13. collecting an image data set in which data values ​​in one or more dimensions correspond to one pixel; generating pseudo-physical parameters to be used in magnetic resonance simulation, which are values ​​of a plurality of physical quantities, by converting one or more dimensional data values ​​of one pixel included in the image data set by data processing; Pseudo data generation method.

14. On the computer, an acquisition function for acquiring an image data set in which data values ​​in one or more dimensions correspond to one pixel; a generation function for generating pseudo-physical parameters to be used in a magnetic resonance simulation, the pseudo-physical parameters being values ​​of a plurality of physical quantities, by converting one or more dimensional data values ​​of one pixel included in the image data set through data processing; A pseudo data generation program that realizes the above.

Images