Medical image acquisition device and method

Abstract

The present invention provides a medical image acquisition device and method, comprising: a preprocessing unit that preprocesses raw image signals to obtain a first number of input images; and a determination unit that uses a training model to determine an analytical relationship between pixels at the same position in the first number of input images, and, based on the analytical relationship, determines a second number of medical function parameter maps corresponding to the raw image signals. The above embodiment enables rapid calculation of SyMRI function parameter maps with strong scalability.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of medical equipment technology, and in particular to a medical image acquisition device and method. Background Art

[0002] Medical imaging devices are increasingly used to scan subjects (e.g., the human body) to obtain medical images of specific locations (e.g., entire or partial organs, or specific regions of interest), providing useful information for medical diagnosis. Medical imaging techniques include computed tomography (CT), magnetic resonance imaging (MR), and digital subtraction angiography (DSA).

[0003] MRI is one of the most common diagnostic methods in medicine. It is radiation-free and offers excellent soft tissue imaging. However, MRI examinations are typically time-consuming, averaging 30 minutes per patient. Currently, MRI is primarily used for weighted imaging, but this imaging cannot be used for quantitative component analysis, making it difficult to provide accurate quantitative diagnosis.

[0004] To reduce scanning time, a technology called Synthesized MRI (SyMRI) has been developed. This synthetic MRI technology can determine quantitative parameters based on the signals collected in a single scan and then use magnetic resonance signal formulas to generate multiple medical function parameter maps (also called contrast images). Therefore, the scanning time for routine examinations can be reduced to less than 10 minutes. However, after the scan, the above-mentioned multiple medical function parameter maps need to be calculated using complex analytical formulas. Summary of the Invention

[0005] Traditionally, SyMRI is performed through analytical calculations using complex mathematical formulas. Therefore, the iterative analytical method used in SyMRI to solve multiple medical functional parameter maps is time-consuming and requires powerful hardware processing equipment. This has hindered the advantages of SyMRI's rapid early scanning.

[0006] Deep learning is a hot technology field that can solve many complex computational problems. Convolutional Neural Networks (CNNs) are currently the most popular method in deep learning, but their primary application is extracting regional features from images for image recognition or classification. SyMRI medical functional parameter maps are calculated using specific analytical formulas for each pixel and require summarizing the relationships between adjacent pixels. Therefore, traditional convolutional neural network models cannot be directly used to calculate these maps.

[0007] In addition, the currently popular CNN models are all represented by regional convolution sliders, which collect the relationship between adjacent pixels in the image to extract regional features. This is quite different from the principle of SyMRI calculation of medical functional parameter maps.

[0008] In response to at least one of the above technical problems, the embodiments of the present application provide a medical image acquisition device and method, which can realize the rapid calculation of SyMRI functional parameter maps and has strong scalability.

[0009] According to one aspect of an embodiment of the present application, a medical image acquisition device is provided, wherein the device includes:

[0010] a preprocessing unit, configured to preprocess the original image signal to obtain a first number of input images;

[0011] A determination unit is configured to determine an analytical relationship between pixels at the same position in the first number of input images using a training model, and to determine a second number of medical function parameter maps corresponding to the original image signal based on the analytical relationship.

[0012] In some embodiments, the pre-processing unit acquires the first number of echo signals corresponding to different excitation pulses, and determines the first number of input images according to the first number of echo signals.

[0013] In some embodiments, the pre-processing unit performs a square root on the sum of the square of the real part and the square of the imaginary part corresponding to each echo signal to obtain a corresponding input image.

[0014] In some embodiments, the first number is equal to eight.

[0015] In some embodiments, the training model is a convolutional neural network with an encoder-decoder structure, the encoder module and the decoder module of the convolutional neural network each include at least one convolution layer, and each convolution layer uses a 1×1 convolution kernel.

[0016] In some embodiments, the number of channels of the first convolutional layer of the encoder module is equal to the first number.

[0017] In some embodiments, the convolutional neural network uses a leakage-corrected unit activation function.

[0018] In some embodiments, the medical function parameter map includes a quantitative parameter map and a weighted parameter map.

[0019] According to one aspect of an embodiment of the present application, a medical image acquisition method is provided, wherein the method includes:

[0020] Preprocessing the original image signal to obtain a first number of input images;

[0021] The training model is used to determine an analytical relationship between pixels at the same position in the first number of input images, and a second number of medical function parameter maps corresponding to the original image signal are determined based on the analytical relationship.

[0022] According to one aspect of an embodiment of the present application, a magnetic resonance imaging system is provided, wherein the system includes:

[0023] a gradient coil assembly configured to generate a gradient magnetic field;

[0024] a radio frequency coil assembly configured to transmit radio frequency excitation pulses;

[0025] a controller in communication with the gradient coil assembly and the radio frequency coil assembly, the controller being configured to:

[0026] instructing a gradient coil assembly to generate magnetic field gradients for use during an MRI scan;

[0027] instructing the radio frequency coil assembly to generate and transmit a radio frequency excitation pulse sequence, and to receive MR signals from the scanned object;

[0028] An original image signal is determined based on the MR signal, and the original image signal is preprocessed to obtain a first number of input images; an analytical relationship between pixels at the same position in the first number of input images is determined using a training model, and a second number of medical function parameter maps corresponding to the original image signal are determined based on the analytical relationship.

[0029] According to another aspect of an embodiment of the present application, a medical image acquisition device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to execute the computer program to implement the medical image acquisition method as described above.

[0030] According to another aspect of an embodiment of the present application, a storage medium storing a computer-readable program is provided, wherein the computer-readable program enables a computer to execute the medical image acquisition method described above in a medical image acquisition device.

[0031] One of the beneficial effects of the embodiments of the present application is that a training model is used to determine the analytical relationship between pixels at the same position in multiple input images, and multiple medical function parameter maps are determined based on the analytical relationship, thereby achieving rapid calculation of SyMRI functional parameter maps with strong scalability.

[0032] With reference to the following description and drawings, specific implementations of the embodiments of the present application are disclosed in detail, indicating how the principles of the embodiments of the present application can be employed. It should be understood that the embodiments of the present application are not limited in scope. Within the spirit and scope of the appended claims, the embodiments of the present application include many variations, modifications, and equivalents. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The included drawings are used to provide a further understanding of the embodiments of the present application, which constitute a part of the specification, are used to illustrate the implementation methods of the present application, and together with the text description, explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other implementation methods can be obtained based on these drawings without inventive work. In the drawings:

[0034] Figure 1 is a schematic diagram of a medical image acquisition method according to an embodiment of the present application;

[0035] Figure 2 This is the model of the embodiment of the present application Figure 1 Schematic diagram;

[0036] Figure 3 Schematic diagram of the real and imaginary parts of the SyMRI image in an embodiment of the present application;

[0037] Figure 4 This is an example diagram of the training model structure of an embodiment of the present application;

[0038] Figure 5 is a schematic diagram of a medical image acquisition method according to an embodiment of the present application;

[0039] Figure 6 is a schematic diagram of a medical image acquisition device according to an embodiment of the present application;

[0040] Figure 7 is a schematic diagram of a medical image acquisition device according to an embodiment of the present application;

[0041] Figure 8 is a schematic diagram of a magnetic resonance imaging system according to an embodiment of the present application;

[0042] Figure 9 This is an example diagram of a medical function parameter diagram obtained by the prior art;

[0043] Figure 10 This is an example diagram of the medical function parameter diagram obtained in the embodiment of the present application. DETAILED DESCRIPTION

[0044] The foregoing and other features of the embodiments of the present application will become apparent through the following description with reference to the accompanying drawings. In the description and drawings, specific embodiments of the present application are disclosed in detail, which illustrate some embodiments in which the principles of the embodiments of the present application can be adopted. It should be understood that the present application is not limited to the described embodiments. On the contrary, the embodiments of the present application include all modifications, variations and equivalents that fall within the scope of the appended claims.

[0045] In the embodiments of the present application, the terms "first", "second", etc. are used to distinguish different elements from the name, but do not indicate the spatial arrangement or temporal order of these elements, and these elements should not be limited by these terms. The term "and / or" includes any one and all combinations of one or more of the associated listed terms. The terms "comprising", "including", "having", etc. refer to the presence of the stated features, elements, components or components, but do not exclude the presence or addition of one or more other features, elements, components or components.

[0046] In the embodiments of this application, the singular forms "a," "the," etc. include plural forms and should be broadly understood to mean "a" or "a type" rather than being limited to "one." Furthermore, the term "said" should be understood to include both singular and plural forms, unless the context clearly indicates otherwise. Furthermore, the term "according to" should be understood to mean "at least in part based on...", and the term "based on" should be understood to mean "at least in part based on...", unless the context clearly indicates otherwise.

[0047] Features described and / or illustrated for one embodiment may be used in the same or similar manner in one or more other embodiments, combined with features in other embodiments, or substituted for features in other embodiments. The term "include / comprise" as used herein refers to the presence of a feature, an integral part, a step, or a component, but does not exclude the presence or addition of one or more other features, integral parts, steps, or components.

[0048] The device for obtaining medical imaging data described in this article can be applicable to various medical imaging modalities, including but not limited to computed tomography (CT) equipment, magnetic resonance imaging (MRI) equipment, C-arm imaging equipment, positron emission tomography (PET) equipment, single photon emission computed tomography (SPECT) equipment, or any other suitable medical imaging equipment.

[0049] The system for obtaining medical images may include the aforementioned medical imaging device, a separate computer device connected to the medical imaging device, or a computer device connected to an internet cloud, wherein the computer device is connected to the medical imaging device or a storage device storing medical images via the internet. The imaging method may be performed independently or in combination by the aforementioned medical imaging device, the computer device connected to the medical imaging device, and the computer device connected to the internet cloud.

[0050] For example, the following describes the embodiments of the present application in conjunction with a magnetic resonance imaging device. Those skilled in the art will appreciate that the embodiments of the present application may also be applicable to other medical imaging devices.

[0051] For example, in magnetic resonance imaging, multiple types of images are generated, hereinafter referred to as medical function parameter maps, also referred to as contrast images. Medical function parameter maps include quantitative parameter maps and weighted parameter maps, such as T1 (longitudinal relaxation time) image, T2 (transverse relaxation time) image, proton density (PD) image, T1-weighted image (T1WI), T2-weighted image (T2WI), fluid attenuated inversion recovery (FLAIR) image, short inversion time inversion recovery (STIR), etc. Each type of image can provide tissue information with different contrast to assist doctors in diagnosis. For example, proton density-weighted image emphasizes the difference in spin density of various tissues / fluids being analyzed, T1-weighted image emphasizes the difference in relaxation time of magnetization recovery along the longitudinal direction, T2-weighted image emphasizes the difference in relaxation time of magnetization recovery along the transverse direction, FLAIR suppresses signals from the fluid, etc. Examples are not given one by one here, and reference can be made to the prior art for details.

[0052] Furthermore, the medical imaging workstation can be located locally on the medical imaging device, meaning it can be located near the device. Both can be co-located in the examination room, the imaging department, or within the same hospital. The medical image cloud platform analysis system can be located remotely from the medical imaging device, for example, in a cloud-based environment that communicates with the device.

[0053] For example, after a medical institution completes an imaging scan using medical imaging equipment, the scanned data is stored in a storage device; a medical imaging workstation can directly read the scanned data and perform image processing through its processor. As another example, a medical image cloud platform analysis system can read medical images in a storage device via remote communication to provide "Software as a Service" (SAAS). SAAS can exist between hospitals, between hospitals and imaging centers, or between hospitals and third-party online diagnosis and treatment service providers.

[0054] The following is a detailed description of the embodiments of the present application.

[0055] Embodiments of the first aspect

[0056] An embodiment of the present application provides a method for acquiring medical images. Figure 1 is a schematic diagram of a medical image acquisition method according to an embodiment of the present application. Figure 1 As shown, the method includes:

[0057] 101, preprocessing the original image signal to obtain a first number of input images;

[0058] 102. Use the training model to determine the analytical relationship between pixels at the same position in the first number of input images, and determine a second number of medical function parameter maps corresponding to the original image signals based on the analytical relationship.

[0059] It is worth noting that the above Figure 1 The embodiments of the present application are only schematically described, but the present application is not limited thereto. For example, the execution order of the various operations can be appropriately adjusted, and other operations can be added or some operations can be reduced. Those skilled in the art can make appropriate modifications based on the above content, and are not limited to the above appended examples. Figure 1 Records of.

[0060] In some embodiments, a region of interest of a scanned object is scanned by a magnetic resonance apparatus to obtain the original image signal. For example, during scanning, an excitation pulse signal (scanning sequence) may be emitted via a transmitting coil, and a receiving coil senses and receives a magnetic resonance signal obtained from the scanned object to obtain the original image signal. Optionally, the scanning sequence may be a QRAPMASTER (Quantification of Relaxation Times and Proton Density by Multi-echo Acquisition of a Saturation-recovery using Turbo spin-Echo Readout) sequence, but the present application is not limited thereto. Other types of sequences may also be used. For details, reference may be made to the prior art and will not be repeated here.

[0061] In some embodiments, the signal intensities of multiple echoes and multiple delay times can be directly collected in a sequence to obtain a first number of echo signals corresponding to different excitation pulses, and the first number of input images can be determined based on the first number of echo signals. For example, when the scanning sequence is a QRAPMASTER sequence, multi-stage, multi-echo, and multi-saturation delay signals of saturation recovery can be obtained from Turbo spin echoes, and the first number of input images can be obtained by combining different echo times (TE) and saturation delay times; that is, the first number can be determined by the set TE number and the saturation delay time number, and the first number can be equal to the set The product of the number of TEs and the number of saturation delay times, for example, using 2 TEs and 4 saturation delay times to generate 8 complex images in one section as the first number of input images, the TE value can be set to 16.9ms and 84.5ms, but the embodiment of the present application is not limited to this; wherein, the complex image includes a real part (hereinafter referred to as a real image) and an imaginary part (hereinafter referred to as an imaginary image), that is, using 2 TEs and 4 saturation delay times to generate 8 real images and 8 imaginary images in one section, the above is only an example, but the embodiment of the present application is not limited to this, the number of delay times and the number of echoes can also be other values, which are not given examples here. In addition, the echo signal can be Fourier transformed to obtain the corresponding image data. For details, please refer to the prior art and will not be described here.

[0062] In some embodiments, in order to reduce computational complexity, the square root of the sum of the square of the real part and the square of the imaginary part corresponding to each echo signal can be taken to obtain the corresponding input image (also called a modulus). That is, when acquiring the signal, only the modulus of the received signal is determined (only the signal magnitude is considered, without considering the direction or phase), and the reconstructed image is also a modulus. Figure 2 is a schematic diagram of the calculated model; or, it can be regarded as dividing the aforementioned complex images into real and imaginary parts, Figure 3 This is a diagram of the real and imaginary parts of a SyMRI image (the left side is the real part, the right side is the imaginary part). The square root of the sum of the squares of the real and imaginary parts at the corresponding pixel position is taken. For example, the following formula (1) is used to convert the original echo signal (complex image) into a model image:

[0063]

[0064] Wherein, modular represents a module image, real represents a real part, and imaginary represents an imaginary part. Each echo signal (complex image) is processed using formula (1) to obtain a first number of module images as a first number of input images, which are input into the training model.

[0065] In some embodiments, the training model is a convolutional neural network with an encoder-decoder structure, and the number of channels of the input layer of the training model, that is, the first convolution layer of the encoder module, is the first number. In other words, the input of the training model can be regarded as an image with a first number of channels, and each channel of the image corresponds to an input image.

[0066] In some embodiments, the output layer of the training model, that is, the number of channels of the last convolutional layer of the decoder module, is a second number, and the second number is determined and adjusted based on the number of medical function parameter maps to be obtained. The second number can be 1, 3, or 6, etc., and the embodiments of the present application are not limited thereto. For example, when the second number is 6, the second number of medical function parameter maps can be T1, T2, PD, T1WI, T2WI, and T2WI FLAIR. Examples are not given one by one here.

[0067] In some embodiments, each convolutional layer of the convolutional neural network uses a 1×1 convolution kernel to determine the analytical relationship between pixels at the same position in the first number of input images. Since the convolution kernel size is only 1×1, there is no need to consider the relationship between each pixel and the surrounding adjacent pixels. Instead, it is only necessary to linearly combine the pixel point information at the same position on different channels, which is equivalent to performing convolution on different channels and performing the same processing on each pixel to determine the analytical relationship between pixels at the same position in the first number of input images. In other words, by setting the convolution kernel of each convolutional layer of the convolutional neural network to 1×1, that is, using a single-pixel step-by-step calculation method, the training model is used to automatically find the solution formula for each SyMRI medical function parameter map to output a second number of medical function parameter maps; in addition, by using a 1×1 convolution kernel, nonlinear characteristics can be added while keeping the scale unchanged (that is, without losing resolution).

[0068] In some embodiments, to ensure that the training model can learn and understand complex nonlinear relationships, the convolutional neural network can also use a rectified linear unit (ReLU) as an activation function. In order to retain the learned positive and negative information, the rectified linear unit activation function can be a leaky rectified unit activation function (Leaky ReLU). For example, the rectified linear unit activation function can be used after the convolution operation on each convolutional layer, but the embodiments of the present application are not limited to this.

[0069] In some embodiments, the encoder module and decoder module of the convolutional neural network each include at least one convolutional layer, the first convolutional layer of the encoder module is the input layer of the training model, and the last convolutional layer of the decoder module is the output layer of the training model, wherein the encoder module and the decoder module have the same number of convolutional layers, and the last convolutional layer of the encoder module and the first convolutional layer of the decoder module use the same number of channels. For example, the encoder module and the decoder module each consist of four convolutional layers. Figure 4 This is a schematic diagram of the training model structure, such as Figure 4 As shown, the encoder module 401 includes convolutional layers 4011, 4012, 4013, and 4014, and the decoder module 402 includes convolutional layers 4021, 4022, 4023, and 4024. Each convolutional layer further includes a leakage correction unit activation function. The convolution kernels of each convolutional layer 4011, 4012, 4013, and 4014 and 4021, 4022, 4023, and 4024 are all 1×1. The number of channels of the convolutional layer 4011 is equal to the first number, the number of channels of the convolutional layer 4024 is equal to the second number, and the convolutional layer 401 2 is equal to the number of channels of convolution layer 4023, for example, it is set to 16, the number of channels of convolution layer 4013 is equal to the number of channels of convolution layer 4022, for example, it is set to 32, and the number of channels of convolution layer 4014 is equal to the number of channels of convolution layer 4021, for example, it is set to 64; the above is only an example, the encoder module and the decoder module can also each include 2 convolution layers, 3 convolution layers, or more than 4 convolution layers, and the number of channels on other convolution layers except the input layer and the output layer can be set to other values, and the embodiments of the present application are not limited to this.

[0070] In some embodiments, the method may further include (not shown): obtaining training data, training parameters of the training model using the training data, and inputting the first number of input images into the trained training model at 102. The raw image signals of the training data may be obtained by scanning a scanned object using a magnetic resonance imaging device, and the raw image signals of the training data may be preprocessed to obtain the first number of input images in the training data. The specific implementation of obtaining the raw image signals and preprocessing is as described above and will not be repeated here.

[0071] In some embodiments, the second number of medical function parameter maps required for the training data can be calculated using existing analytical formulas, or the original image signals of the training data can be input into a medical imaging workstation to obtain the second number of medical function parameter maps of the training data, and the second number of medical function parameter maps of the training data can be used as gold standard maps (true value maps).

[0072] In some embodiments, a first number of training input images and their corresponding labeled gold standard images (a second number of medical function parameter images, which can be regarded as the output of the training model) are input into the aforementioned training model as a set of training data, and the parameters of the training model are trained. In particular, multiple sets of training data can be obtained in the same way, and the parameters of the training model can be trained using multiple sets of training data to improve the accuracy of the training model.

[0073] In some embodiments, the second number of medical function parameter maps used in the training phase are of the same type as the second number of medical function parameter maps that need to be acquired in actual application.

[0074] Figure 5 This is a schematic diagram of the method for obtaining medical function parameters of this application, such as Figure 5 As shown, the method includes:

[0075] 501, obtaining an original image signal for training;

[0076] 502, calculating a gold standard image based on the original image signal;

[0077] For example, a second number of medical function parameter maps corresponding to the original image signal may be calculated using an existing formula, or the original image signal may be input into a workstation to obtain a second number of medical function parameter maps for training, i.e., gold standard maps;

[0078] 503, preprocessing the original image signal in 501 to obtain a first number of input images (eg, model images) for training;

[0079] 504, inputting the input image in 503 and the gold standard image in 502 as a set of training data into the training model to train the parameters of the training model;

[0080] For example, the same or different regions of interest of different scanned objects may be scanned to obtain multiple original image signals for training. 502-503 are performed for each original image signal to obtain multiple sets of training data, which are input into the training model to train the parameters of the training model.

[0081] 505 , obtaining an original image signal of a region of interest of a scanned object during an actual application process;

[0082] 506, preprocessing the original image signal in 505 to obtain a first number of input images (eg, model images);

[0083] 507 , input the first number of input images in 506 into the trained training model to obtain a second number of medical function parameter maps output by the training model.

[0084] The implementation methods of the above 501-507 are as described above and will not be repeated here.

[0085] It is worth noting that the above Figure 5 The embodiments of the present application are only schematically described, but the present application is not limited thereto. For example, the execution order of the various operations can be appropriately adjusted, and other operations can be added or some operations can be reduced. Those skilled in the art can make appropriate modifications based on the above content, and are not limited to the above appended examples. Figure 5 Records of.

[0086] It should be noted that the above-mentioned medical function parameter map in the embodiment of the present application is suitable for any region of interest of any scanned object, and the embodiment of the present application is not limited to this.

[0087] The above embodiments are merely exemplary of the present invention, but the present invention is not limited thereto. Appropriate modifications may be made based on the above embodiments. For example, the above embodiments may be used alone, or one or more of the above embodiments may be combined.

[0088] As can be seen from the above embodiments, a training model is used to determine the analytical relationship between pixels at the same position in multiple input images, and multiple medical function parameter maps are determined based on the analytical relationship, thereby achieving fast calculation of SyMRI functional parameter maps with strong scalability.

[0089] Embodiments of the second aspect

[0090] An embodiment of the present application provides a medical image acquisition device, and the contents that are the same as those in the embodiment of the first aspect are not repeated here.

[0091] Figure 6 FIG is a schematic diagram of a medical image acquisition device according to an embodiment of the present application. Figure 6 As shown, the medical image acquisition device 600 includes:

[0092] A preprocessing unit 601 preprocesses the original image signal to obtain a first number of input images;

[0093] The determining unit 602 uses the training model to determine the analytical relationship between pixels at the same position in the first number of input images, and determines a second number of medical function parameter maps corresponding to the original image signal based on the analytical relationship.

[0094] In some embodiments, the implementation of the preprocessing unit 601 and the determination unit 602 can refer to 101-102 of the first aspect embodiment, and the repeated parts will be omitted.

[0095] In some embodiments, the pre-processing unit 601 obtains the first number of echo signals corresponding to different excitation pulses, and determines the first number of input images according to the first number of echo signals.

[0096] In some embodiments, the pre-processing unit 601 performs a square root on the sum of the square of the real part and the square of the imaginary part corresponding to each echo signal to obtain the corresponding input image. For example, the first number is equal to 8.

[0097] In some embodiments, the training model is a convolutional neural network with an encoder-decoder structure, the encoder module and the decoder module of the convolutional neural network each include at least one convolution layer, and each convolution layer uses a 1×1 convolution kernel.

[0098] In some embodiments, the number of channels of the first convolutional layer of the encoder module is equal to the first number, and the number of channels of the last convolutional layer of the decoder module is equal to the second number.

[0099] In some embodiments, the convolutional neural network uses a leakage-corrected unit activation function.

[0100] In some embodiments, the apparatus may further include: (not shown)

[0101] The training unit obtains training data and uses the training data to train the parameters of the training model. The implementation method of the training unit and the structure of the training model can refer to the embodiment of the first aspect and will not be repeated here.

[0102] To keep it simple, Figure 6 The connection relationships and signal paths between various components or modules are shown for illustrative purposes only. However, those skilled in the art will appreciate that various related technologies, such as bus connections, can be employed. The aforementioned components or modules can be implemented using hardware such as processors and memory; this is not a limitation of the present invention.

[0103] The above embodiments are merely exemplary of the present invention, but the present invention is not limited thereto. Appropriate modifications may be made based on the above embodiments. For example, the above embodiments may be used alone, or one or more of the above embodiments may be combined.

[0104] As can be seen from the above embodiments, a training model is used to determine the analytical relationship between pixels at the same position in multiple input images, and multiple medical function parameter maps are determined based on the analytical relationship, thereby achieving fast calculation of SyMRI functional parameter maps with strong scalability.

[0105] Embodiments of the third aspect

[0106] The present embodiment provides a medical image acquisition device, including the medical image acquisition apparatus 600 described in the second embodiment, the contents of which are incorporated herein. The medical image acquisition device can be, for example, a computer, server, workstation, laptop computer, smartphone, etc., but the present embodiment is not limited thereto.

[0107] Figure 7 Schematic diagram of a medical image acquisition device according to an embodiment of the present application. Figure 7 As shown, the medical image acquisition device 700 may include: one or more processors (e.g., central processing units (CPUs)) 710 and one or more memories 720; the memories 720 are coupled to the processors 710. The memories 720 may store various data and training models, etc.; and may also store information processing programs 721, which are executed under the control of the processors 710.

[0108] In some embodiments, the functions of the medical image acquisition apparatus 600 are integrated into the processor 710. The processor 710 is configured to implement the medical image acquisition method as described in the embodiment of the first aspect.

[0109] In some embodiments, the medical image acquisition device 600 is configured separately from the processor 710. For example, the medical image acquisition device 600 can be configured as a chip connected to the processor 710, and the functions of the medical image acquisition device 600 are realized through the control of the processor 710.

[0110] For example, the processor 710 is configured to perform the following control: preprocessing the original image signal to obtain a first number of input images; using a training model to determine the analytical relationship between pixels at the same position in the first number of input images, and determining a second number of medical function parameter maps corresponding to the original image signal based on the analytical relationship.

[0111] For example, the processor 710 is configured to perform the following control: acquiring training data; and training parameters of the training model using the training data.

[0112] In some embodiments, the implementation of the processor 710 and the structure of the training model can refer to the embodiment of the first aspect and will not be repeated here.

[0113] In addition, if Figure 7 As shown, the medical image acquisition device 700 may also include: an input / output (I / O) device 730 and a display 740 (displaying a medical function parameter diagram), etc.; wherein, the functions of the above components are similar to those of the prior art and are not described here in detail. It is worth noting that the medical image acquisition device 700 does not necessarily have to include Figure 7In addition, the medical image acquisition device 700 may also include Figure 7 For components not shown, reference may be made to related art.

[0114] As can be seen from the above embodiments, a training model is used to determine the analytical relationship between pixels at the same position in multiple input images, and multiple medical function parameter maps are determined based on the analytical relationship, thereby achieving fast calculation of SyMRI functional parameter maps with strong scalability.

[0115] Embodiments of the fourth aspect

[0116] An embodiment of the present application provides a magnetic resonance imaging system.

[0117] Figure 8 This is a schematic diagram of the magnetic resonance imaging system. Figure 8 As shown, the system 800 includes:

[0118] a gradient coil assembly 801 configured to generate a gradient magnetic field;

[0119] a radio frequency coil assembly 802 configured to transmit radio frequency excitation pulses;

[0120] A controller 803 in communication with the gradient coil assembly 801 and the radio frequency coil assembly 802 , the controller 803 being configured to:

[0121] instructing the gradient coil assembly 801 to generate magnetic field gradients for use during an MRI scan;

[0122] Instructing the radio frequency coil assembly 802 to generate and transmit a radio frequency excitation pulse sequence, and to receive MR signals from the scanned object;

[0123] An original image signal is determined based on the MR signal, and the original image signal is preprocessed to obtain a first number of input images; an analytical relationship between pixels at the same position in the first number of input images is determined using a training model, and a second number of medical function parameter maps corresponding to the original image signal are determined based on the analytical relationship.

[0124] In some embodiments, the controller 803 is connected to other components of the magnetic resonance imaging system via electrical and / or data connections.

[0125] In some embodiments, the magnetic resonance imaging system optionally further includes a gradient driver 806, to which the controller 803 provides gradient waveforms. The gradient driver includes, among other things, GX, GY, and GZ amplifiers. Each GX, GY, and GZ gradient amplifier excites a corresponding gradient coil in a gradient coil assembly 801 to generate magnetic field gradients for spatially encoding MR signals during an MRI scan. The gradient coil assembly 801 is contained within a resonance assembly (not shown), which further includes a superconducting magnet having superconducting coils 808. During operation, the superconducting coils provide a uniform longitudinal magnetic field throughout an open cylindrical imaging volume enclosed by the resonance assembly. The coil structure of this portion may be described in detail in the prior art and will not be further described herein.

[0126] The resonant assembly also includes a radio frequency coil assembly 802, which can be configured to operate in a transmit and receive mode, a transmit mode, or a receive mode. A subject undergoing an MRI scan can be positioned within the open cylindrical imaging volume of the resonant assembly. The magnetic resonance imaging system also includes an RF generator 807, a memory 804, and a display 805. The RF generator 807 generates RF excitation pulses, which are amplified by an RF amplifier (optional, not shown) and provided to the RF coil assembly 802 via a transmit / receive switch. The RF coil assembly 802 can be used to transmit RF excitation pulses and / or receive resulting MR signals from the patient undergoing the MRI scan. The resulting MR signals emitted by the excited nuclei of the subject undergoing the MRI scan can be sensed and received by the RF coil assembly 802 and sent back to a preamplifier 809 (optional) via the transmit / receive switch. The MR signals are demodulated, filtered, and digitized in a controller 803. The transmit / receive switch electrically connects the RF amplifier to the RF coil assembly 802 during transmit mode and connects the preamplifier to the RF coil assembly 802 during receive mode. The transmit / receive switch may further enable the RF coil assembly 802 to be used in either a transmit mode or a receive mode.The resulting MR signals sensed and received by the RF coil assembly 802 are digitized by the controller 803 and transferred to the memory 804 for storage.

[0127] In some embodiments, the controller 803 can communicate with a medical image acquisition device (not limited to the device of the third aspect embodiment) and receive commands from it to instruct the RF generator 803 to generate an MRI scan sequence to be performed during the MRI scan.

[0128] In some embodiments, the controller 803 acquires an array of raw k-space data corresponding to the received MR signals until the MR scan is completed when the data is subsequently transformed into an image. The raw k-space data is rearranged into separate k-space data arrays, and each of these separate k-space data arrays is processed by the controller 803 to be transformed into an array of image data. For example, the controller 803 uses a known transformation method, such as Fourier transform, to convert the received MR signals into an image. For details, reference may be made to the prior art, and the embodiments of the present application are not limited thereto. After acquiring the raw image signals, the raw image signals are preprocessed to obtain a first number of input images; a training model is used to determine the analytical relationship between pixels at the same position in the first number of input images, and a second number of medical function parameter maps corresponding to the raw image signals are determined based on the analytical relationship. The specific implementation method can refer to the embodiment of the first aspect and will not be repeated here. Optionally, the controller 803 can also instruct the display 805 to display the second number of medical function parameter maps.

[0129] In some embodiments, after the controller 803 obtains the MR signal, it may not process the MR signal, but instead send the MR signal to the medical image acquisition device 700 (such as a workstation or computer system) described in the third embodiment, and the medical image acquisition device 700 determines the original image signal and pre-processes the original image signal to obtain a first number of input images; uses the training model to determine the analytical relationship between pixels at the same position in the first number of input images, and determines a second number of medical function parameter maps corresponding to the original image signal based on the analytical relationship, the controller 803 receives the second number of medical function parameter maps, and instructs the display 805 to display the second number of medical function parameter maps.

[0130] In addition, the magnetic resonance imaging system 800 may further include Figure 8 Components not shown, such as a phase detector physiological acquisition controller, a scanning object positioning device, etc., can be specifically referred to in related technologies and will not be given examples one by one here.

[0131] Figure 9 This is an example diagram of a medical function parameter diagram obtained by prior art. Figure 10 This is an example diagram of a medical function parameter diagram obtained by the medical image acquisition method of this application, such as Figure 9 and 10 As shown, the medical image acquisition method of the present application can reconstruct various medical function parameter maps more accurately, and can realize the rapid calculation of SyMRI function parameter maps, and has strong scalability.

[0132] An embodiment of the present application also provides a computer-readable program, wherein when the program is executed in a medical image acquisition device, the program causes a computer to execute the medical image acquisition method as described in the embodiment of the first aspect in the medical image acquisition device.

[0133] An embodiment of the present application also provides a storage medium storing a computer-readable program, wherein the computer-readable program enables a computer to execute the medical image acquisition method as described in the embodiment of the first aspect in a medical image acquisition device.

[0134] The above devices and methods of the present application can be implemented by hardware or by a combination of hardware and software. The present application relates to such a computer-readable program that, when executed by a logic component, enables the logic component to implement the devices or components described above, or enables the logic component to implement the various methods or steps described above. The present application also relates to a storage medium for storing the above program, such as a hard disk, a magnetic disk, an optical disk, a DVD, a flash memory, etc.

[0135] The method / device described in conjunction with the embodiments of the present application can be directly embodied as hardware, a software module executed by a processor, or a combination of the two. For example, one or more of the functional block diagrams shown in the figure and / or one or more combinations of functional block diagrams can correspond to various software modules of the computer program flow or to various hardware modules. These software modules can respectively correspond to the various steps shown in the figure. These hardware modules can be implemented by solidifying these software modules, for example, using a field programmable gate array (FPGA).

[0136] The software module may be located in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art. A storage medium may be coupled to a processor so that the processor can read information from the storage medium and write information to the storage medium; or the storage medium may be an integral part of the processor. The processor and the storage medium may be located in an ASIC. The software module may be stored in the memory of the mobile terminal or in a memory card that can be inserted into the mobile terminal. For example, if the device (such as a mobile terminal) uses a large-capacity MEGA-SIM card or a large-capacity flash memory device, the software module may be stored in the MEGA-SIM card or the large-capacity flash memory device.

[0137] One or more of the functional blocks and / or one or more combinations of functional blocks described in the accompanying drawings may be implemented as a general-purpose processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, or any appropriate combination thereof for performing the functions described in this application. One or more of the functional blocks and / or one or more combinations of functional blocks described in the accompanying drawings may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors in communication with a DSP, or any other such configuration.

[0138] The present application has been described above with reference to specific embodiments. However, those skilled in the art should understand that these descriptions are merely illustrative and are not intended to limit the scope of protection of the present application. Those skilled in the art may make various modifications and variations to the present application based on the principles of the present application, and such modifications and variations are also within the scope of the present application.

[0139] Regarding the implementation methods including the above embodiments, the following additional notes are also disclosed:

[0140] 1. A medical image acquisition method, characterized in that the method comprises:

[0141] Preprocessing the original image signal to obtain a first number of input images;

[0142] The training model is used to determine an analytical relationship between pixels at the same position in the first number of input images, and a second number of medical function parameter maps corresponding to the original image signals are determined based on the analytical relationship.

[0143] 2. The method according to Supplementary Note 1, wherein preprocessing the original image signal to obtain the first number of input images comprises:

[0144] The first number of echo signals corresponding to different excitation pulses are acquired, and the first number of input images are determined according to the first number of echo signals.

[0145] 3. The method according to Supplementary Note 2, further comprising:

[0146] The square root of the sum of the square of the real part and the square of the imaginary part corresponding to each echo signal is taken to obtain the corresponding input image.

[0147] 4. The method according to any one of Notes 1 to 3, characterized in that the first number is equal to 8.

[0148] 5. The method according to any one of Notes 1 to 4 is characterized in that the training model is a convolutional neural network with an encoder-decoder structure, the encoder module and the decoder module of the convolutional neural network each include at least one convolution layer, and each convolution layer adopts a 1×1 convolution kernel.

[0149] 6. The method according to Note 5 is characterized in that the number of channels of the first convolutional layer of the encoder module is equal to the first number.

[0150] 7. The method according to Note 5 or 6 is characterized in that the convolutional neural network uses a leakage correction unit activation function.

[0151] 8. The method according to any one of Notes 1 to 7 is characterized in that the medical function parameter map includes a quantitative parameter map and a weighted parameter map.

[0152] 9. The method according to any one of Notes 1 to 8, further comprising:

[0153] Acquire training data, and use the training data to train parameters of the training model.

[0154] 10. A magnetic resonance imaging system, wherein the system comprises:

[0155] a gradient coil assembly configured to generate a gradient magnetic field;

[0156] a radio frequency coil assembly configured to transmit radio frequency excitation pulses;

[0157] a controller in communication with the gradient coil assembly and the radio frequency coil assembly, the controller being configured to:

[0158] instructing a gradient coil assembly to generate magnetic field gradients for use during an MRI scan;

[0159] instructing the radio frequency coil assembly to generate and transmit a radio frequency excitation pulse sequence, and to receive MR signals from the scanned object;

[0160] An original image signal is determined based on the MR signal, and the original image signal is preprocessed to obtain a first number of input images; an analytical relationship between pixels at the same position in the first number of input images is determined using a training model, and a second number of medical function parameter maps corresponding to the original image signal are determined based on the analytical relationship.

[0161] 11. A medical image acquisition device, characterized in that the device comprises: a memory and a processor, the memory storing a computer program, and the processor being configured to execute the computer program to implement the medical image acquisition method as described in any one of Notes 1-9.

[0162] 12. A storage medium storing a computer-readable program, wherein the computer-readable program enables a computer to execute the medical image acquisition method as described in any one of Notes 1 to 9 in a medical image acquisition device.

Claims

1. A medical image acquisition device, characterized in that: The device comprises: a preprocessing unit, configured to preprocess the original image signal to obtain a first number of input images; a determining unit configured to determine an analytical relationship between pixels at the same position in the first number of input images using a training model, and determine a second number of medical function parameter maps corresponding to the original image signals based on the analytical relationship; The pre-processing unit obtains a first number of echo signals, and performs square root on the sum of the square of the real part and the square of the imaginary part corresponding to each of the echo signals to obtain the corresponding input image.

2. The device according to claim 1, characterized in that The preprocessing unit obtains the first number of echo signals corresponding to different excitation pulses, and determines the first number of input images according to the first number of echo signals.

3. The device according to claim 1, characterized in that The first number is equal to 8.

4. The device according to claim 1, characterized in that The training model is a convolutional neural network with an encoder-decoder structure, wherein the encoder module and the decoder module of the convolutional neural network respectively include at least one convolutional layer, and each convolutional layer adopts a 1×1 convolution kernel.

5. The device according to claim 4, characterized in that The number of channels of the first convolutional layer of the encoder module is equal to the first number.

6. The device according to claim 4, characterized in that The convolutional neural network uses a leakage-corrected unit activation function.

7. The device according to claim 1, characterized in that The medical function parameter map includes a quantitative parameter map and a weighted parameter map.

8. A medical image acquisition method, characterized in that: The method comprises: Preprocessing the original image signal to obtain a first number of input images, including: acquiring a first number of echo signals and taking the square root of the sum of the square of the real part and the square of the imaginary part corresponding to each of the echo signals to obtain the corresponding input image; The training model is used to determine an analytical relationship between pixels at the same position in the first number of input images, and a second number of medical function parameter maps corresponding to the original image signals are determined based on the analytical relationship.

9. A storage medium storing a computer-readable program, wherein the computer-readable program causes a computer to execute the medical image acquisition method according to claim 8 in a medical image acquisition device.

Images