Medical image generation method and device, computer equipment and storage medium

Abstract

The invention relates to a medical image generation method and device, computer equipment and a storage medium. A target signal evolution curve graph of a scanned object in magnetic resonance fingerprint imaging is obtained; and obtaining a quantitative image of the scanned object according to the target signal evolution curve graph and a preset target quantitative prediction model. In the embodiment of the invention, since the sample weighted image has the quantitative value consistent with that of the gold standard quantitative image, the sample weighted image is utilized to train the quantitative prediction model to obtain the target quantitative prediction model, and the quantitative image is obtained through the target quantitative prediction model, so that the influence of factors such as a magnetization transfer effect in an MRF scanning process can be avoided; and a quantitative image obtained based on an MRF technology is different from a gold standard quantitative image. Moreover, the sample weighted image is used as a gold standard for quantitative prediction model training, a large number of gold standard quantitative graphs do not need to be obtained, and the determination efficiency of the target quantitative prediction model is improved.

Description

Technical Field

[0001] The present application relates to the field of medical technology, and in particular to a method, apparatus, computer device and storage medium for generating medical images. Background Art

[0002] Magnetic resonance fingerprint imaging (MRF) is a rapid quantitative imaging technology that can obtain a variety of quantitative images such as longitudinal relaxation time T1 / transverse relaxation time T2 / proton density image (PD).

[0003] The MRF acquisition sequence uses a non-constant acquisition module to acquire K-space data from patients, and obtains a signal evolution curve based on the K-space data. Quantitative images are further obtained by dictionary matching / AI model prediction. However, due to the influence of factors such as the magnetization transfer effect during the scanning process, there are certain differences between the quantitative images obtained based on MRF technology and the gold standard quantitative images, which brings obstacles to application scenarios that require the simultaneous combination of two quantitative images. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment and storage medium for generating medical images that can reduce the difference between quantitative images and gold standard quantitative images in order to address the above technical problems.

[0005] In a first aspect, the present application provides a method for generating a medical image, comprising:

[0006] Obtaining a target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging;

[0007] According to the target signal evolution curve diagram and the preset target quantitative prediction model, a quantitative image of the scanned object is obtained; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image.

[0008] In one embodiment, the process of obtaining the target quantitative prediction model includes:

[0009] Acquire a sample signal evolution curve graph acquired based on a magnetic resonance fingerprint imaging sequence, and acquire a sample weighted image acquired based on a magnetic resonance imaging sequence;

[0010] Inputting the sample signal evolution curve graph into an initial quantitative prediction model to obtain a plurality of predicted quantitative images;

[0011] The initial quantitative prediction model is trained according to the plurality of predicted quantitative images and the sample weighted images to obtain the target quantitative prediction model.

[0012] In one embodiment, the training of the initial quantitative prediction model according to the plurality of predicted quantitative images and the sample weighted images to obtain the target quantitative prediction model includes:

[0013] Performing simulation processing on the plurality of predicted quantitative images to obtain a plurality of predicted weighted images;

[0014] The initial quantitative prediction model is trained according to the multiple predicted weighted images and the sample weighted images corresponding to the multiple predicted weighted images to obtain the target quantitative prediction model.

[0015] In one embodiment, the training of the initial quantitative prediction model according to the plurality of predicted weighted images and the sample weighted images corresponding to the plurality of predicted weighted images to obtain the target quantitative prediction model includes:

[0016] For each type of predicted weighted image, obtaining a loss value between the predicted weighted image and a corresponding sample weighted image;

[0017] The initial quantitative prediction model is trained according to the loss values ​​corresponding to a plurality of different types of prediction weighted images to obtain the target quantitative prediction model.

[0018] In one embodiment, the step of acquiring a sample signal evolution curve graph acquired based on a magnetic resonance fingerprint imaging sequence includes:

[0019] Acquire sample K-space data acquired based on a magnetic resonance fingerprint imaging sequence;

[0020] The sample K-space data is reconstructed to obtain the sample signal evolution curve diagram.

[0021] In one embodiment, reconstructing the sample K-space data to obtain the sample signal evolution curve diagram includes:

[0022] Compressing the sample K-space data to obtain compressed sample K-space data;

[0023] The compressed sample K-space data is reconstructed to obtain the sample signal evolution curve diagram.

[0024] In a second aspect, the present application also provides a device for generating a medical image, comprising:

[0025] A first acquisition module is used to acquire a target signal evolution curve diagram of a scanned object in magnetic resonance fingerprint imaging;

[0026] The first determination module is used to obtain a quantitative image of the scanned object according to the target signal evolution curve diagram and a preset target quantitative prediction model; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with a gold standard quantitative image.

[0027] In a third aspect, the present application further provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the following steps are implemented:

[0028] Obtaining a target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging;

[0029] According to the target signal evolution curve diagram and the preset target quantitative prediction model, a quantitative image of the scanned object is obtained; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image.

[0030] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the following steps are implemented:

[0031] Obtaining a target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging;

[0032] According to the target signal evolution curve diagram and the preset target quantitative prediction model, a quantitative image of the scanned object is obtained; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image.

[0033] In a fifth aspect, the present application further provides a computer program product, including a computer program, which implements the following steps when executed by a processor:

[0034] Obtaining a target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging;

[0035] According to the target signal evolution curve diagram and the preset target quantitative prediction model, a quantitative image of the scanned object is obtained; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image.

[0036] The above-mentioned method, device, computer equipment and storage medium for generating medical images obtain the target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging; obtain the quantitative image of the scanned object according to the target signal evolution curve diagram and the preset target quantitative prediction model; the target quantitative prediction model is obtained based on the sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image. In the embodiment of the present application, since the sample weighted image has a quantitative value consistent with the gold standard quantitative image, the quantitative prediction model is trained using the sample weighted image to obtain the target quantitative prediction model, so as to obtain the quantitative image through the target quantitative prediction model, which can avoid the problem that the quantitative image obtained based on the MRF technology is different from the gold standard quantitative image due to factors such as the magnetization transfer effect during the MRF scanning process. Moreover, by using the sample weighted image as the gold standard for training the quantitative prediction model, there is no need to obtain a large number of gold standard quantitative images, which improves the determination efficiency of the target quantitative prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments of the present application or related technical descriptions will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0038] FIG1 is a diagram of an application environment of a method for generating a medical image in one embodiment;

[0039] FIG2 is a schematic diagram of a flow chart of a method for generating a medical image in one embodiment;

[0040] FIG3 is a schematic diagram of a flow chart of a method for determining a target quantitative prediction model in one embodiment;

[0041] FIG4 is a schematic flow chart of a method for determining a target quantitative prediction model in another embodiment;

[0042] FIG5 is a schematic flow chart of a method for determining a target quantitative prediction model in another embodiment;

[0043] FIG6 is a schematic diagram of a flow chart of a method for acquiring a sample signal evolution curve diagram in one embodiment;

[0044] FIG7 is a schematic flow chart of a method for generating a medical image in another embodiment;

[0045] FIG8 is a structural block diagram of a device for generating medical images in one embodiment. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0047] MRF is a fast quantitative imaging technology that can obtain multiple quantitative images such as T1 / T2 / PD. The MRF acquisition sequence uses non-constant acquisition modules (such as variable repetition time (TR), echo time (TE), flip angle (FA), etc.) to acquire transient signals, obtain K-space data, and obtain signal evolution curves based on K-space data. The signal evolution curves make the evolution curves of tissues with different characteristics along the time dimension distinguishable.

[0048] After collecting the signal evolution curve of each pixel, a quantitative image is usually obtained by dictionary matching, that is, the analog signal evolution curves corresponding to different tissue characteristics are calculated in advance in the form of simulation to generate a dictionary, and the analog signal evolution curve that is most similar to each signal evolution curve is obtained through dictionary matching, and the quantitative parameters corresponding to the matched analog signal evolution curve are assigned to the pixels corresponding to the corresponding signal evolution curve to obtain a quantitative image. Another way to obtain a quantitative image is to input the signal evolution curve diagram (the signal evolution curve of each pixel) into the AI ​​model to obtain a quantitative image. That is, the AI ​​model is pre-inputted into the model through the signal evolution curve diagram sample, and the corresponding reference quantitative image is used as supervision for training to obtain the AI ​​model. The reference quantitative image used for supervision is usually obtained by dictionary matching, or it can be a gold standard quantitative image set in the simulation.

[0049] However, the above two mapping methods from signal evolution curves to quantitative images, whether dictionary or AI model, cannot overcome a limitation of MRF itself: in the scanning process of actual subjects, due to the influence of unpredictable and difficult to quantify practical factors such as MT effect, the quantitative images obtained based on MRF and the gold standard quantitative images obtained based on the traditional gold standard quantitative sequence are usually different, which brings obstacles to the application scenarios that need to combine two quantitative images at the same time. For example, the generated weighted images of quantitative medical images obtained based on MRF are usually different from the weighted images collected in actual clinical practice.

[0050] The method for generating a medical image provided in an embodiment of the present application can be applied in an application environment as shown in FIG1. ​​The application environment includes a computer device, which can be a server, and its internal structure diagram can be shown in FIG1. ​​The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store relevant data for the generation of medical images. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for generating a medical image is implemented. The server may be an independent physical server, a server cluster or a distributed system composed of multiple physical servers, or a cloud server that provides cloud computing services.

[0051] Those skilled in the art will understand that the structure shown in FIG. 1 is merely a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components.

[0052] In an exemplary embodiment, as shown in FIG2 , a method for generating a medical image is provided, which is described by taking the method applied to the computer device in FIG1 as an example, and includes the following S201 to S202. Wherein:

[0053] S201, obtaining a target signal evolution curve diagram of a scanned object in magnetic resonance fingerprint imaging.

[0054] In the embodiment of the present application, the FA and TR of the radio frequency pulses in the magnetic resonance fingerprint imaging sequence can be changed, and the scanned object can be scanned after each change to obtain the corresponding K-space data, and the collected K-space data can be reconstructed to obtain the corresponding medical scan image. Based on the medical scan images at multiple scanning moments, the target signal evolution curve corresponding to each pixel point is obtained, that is, the target signal evolution curve diagram, which contains rich information about tissue characteristics, such as T1 longitudinal relaxation time, T2 transverse relaxation time and PD proton density.

[0055] In a possible implementation, in order to shorten the scanning time, the scanning object can also be scanned using a magnetic resonance fingerprint imaging sequence to obtain undersampled K-space data, and the uncollected data can be restored using interpolation and other methods, and the restored K-space data rows can be reconstructed to obtain a target signal evolution curve graph.

[0056] S202, obtaining a quantitative image of the scanned object according to a target signal evolution curve diagram and a preset target quantitative prediction model; the target quantitative prediction model is obtained based on training of a sample weighted image, and the sample weighted image has a quantitative value consistent with a gold standard quantitative image.

[0057] Among them, the quantitative value refers to the quantitative value of physiological parameters, such as the quantitative values ​​of physiological parameters such as T1\T2\PD. Physiological parameters reflect the characteristics of human tissues themselves and are independent of the scanning sequence. Therefore, the quantitative value can theoretically intuitively reflect the physiological state of tissues. Sample weighted images refer to images that emphasize the influence of specific physiological parameters on image signal intensity. Sample weighted images reflect a certain aspect of the gold standard quantitative image, and the underlying quantitative values ​​are consistent in different sample weighted images and gold standard quantitative images. By adjusting the imaging parameters, the influence of a certain physiological parameter on the image signal intensity can be enhanced, highlighting the role of the physiological parameter in image contrast, thereby helping doctors or researchers to better diagnose and analyze. For example, T1-weighted images mainly reflect the difference in T1 relaxation time of tissues, while T2-weighted images highlight the difference in T2 relaxation time of tissues.

[0058] Different types of sample weighted images can be obtained in clinical diagnosis by adjusting the two parameters of repetition time TR and echo time TE during scanning. For example, short TR and short TE are used to obtain T1-weighted images, which are suitable for displaying anatomical structures; long TR and long TE are used to obtain T2-weighted images, which are suitable for displaying lesion areas; long TR and short TE will produce proton density-weighted images, which are mainly used to display normal tissues.

[0059] In an embodiment of the present application, the trained target quantitative prediction model includes multiple output nodes, the target signal evolution curve graph is input into the target quantitative prediction model, and the target quantitative prediction model is used to extract the features of the target signal evolution curve graph to obtain multiple different types of quantitative images of the scanned object.

[0060] In the above-mentioned method for generating medical images, a target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging is obtained; a quantitative image of the scanned object is obtained according to the target signal evolution curve diagram and a preset target quantitative prediction model; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image. In the embodiment of the present application, since the sample weighted image has a quantitative value consistent with the gold standard quantitative image, the quantitative prediction model is trained using the sample weighted image to obtain the target quantitative prediction model, so as to obtain the quantitative image through the target quantitative prediction model, which can avoid the problem that the quantitative image obtained based on the MRF technology is different from the gold standard quantitative image due to factors such as the magnetization transfer effect during the MRF scanning process. Moreover, by using the sample weighted image as the gold standard for training the quantitative prediction model, there is no need to obtain a large number of gold standard quantitative images, which improves the determination efficiency of the target quantitative prediction model.

[0061] FIG3 is a flow chart of a method for determining a target quantitative prediction model in an embodiment, as shown in FIG3 , comprising the following steps:

[0062] S301, obtaining a sample signal evolution curve diagram obtained based on a magnetic resonance fingerprint imaging sequence, and obtaining a sample weighted image obtained based on a magnetic resonance imaging sequence.

[0063] S302, inputting the sample signal evolution curve graph into the initial quantitative prediction model to obtain a plurality of predicted quantitative images.

[0064] S303, training an initial quantitative prediction model according to a plurality of predicted quantitative images and a sample weighted image to obtain a target quantitative prediction model.

[0065] Optionally, the predicted quantitative image may be a T1 predicted image, a T2 predicted image, a PD predicted image, or the like.

[0066] In the embodiment of the present application, as shown in FIG. 4 , the sample signal evolution curve diagram is input into the initial quantitative prediction model to obtain a plurality of predicted quantitative images.

[0067] In one possible implementation, a signal evolution model can be used to perform weighted processing on multiple predicted quantitative images to obtain multiple predicted weighted images. Based on the differences between the multiple predicted quantitative images and the sample weighted images, an initial quantitative prediction model is trained to obtain a target quantitative prediction model.

[0068] In another possible implementation, the sample weighted image can be converted into a sample quantitative image, and the initial quantitative prediction model can be trained according to the difference between the sample quantitative image and the predicted quantitative image to obtain a target quantitative prediction model. For example, a T1 sample weighted image can be quantitatively analyzed to generate a T1 sample quantitative image that can directly reflect the T1 value of the tissue.

[0069] In an embodiment of the present application, a sample signal evolution curve diagram obtained based on a magnetic resonance fingerprint imaging sequence is obtained, and a sample weighted image obtained based on a magnetic resonance imaging sequence is obtained, the sample signal evolution curve diagram is input into an initial quantitative prediction model to obtain a plurality of predicted quantitative images, the initial quantitative prediction model is trained based on the plurality of predicted quantitative images and the sample weighted images to obtain a target quantitative prediction model, thereby improving the efficiency of determining the target quantitative prediction model.

[0070] FIG5 is a flow chart of a method for determining a target quantitative prediction model in another embodiment. As shown in FIG5 , the embodiment of the present application relates to a possible implementation method of how to train an initial quantitative prediction model according to multiple predicted quantitative images and sample weighted images to obtain a target quantitative prediction model, including the following steps:

[0071] S501, performing simulation processing on a plurality of predicted quantitative images to obtain a plurality of predicted weighted images.

[0072] In an embodiment of the present application, physiological parameters such as T1 / T2 / PD corresponding to each predicted quantitative image and sequence acquisition parameters such as FA / TI / TR / TE corresponding to the magnetic resonance scanning sequence can be simulated through physical models such as Bloch simulation and EPG simulation to obtain multiple predicted weighted images.

[0073] S502 , training an initial quantitative prediction model according to a plurality of prediction weighted images and sample weighted images corresponding to the plurality of prediction weighted images to obtain a target quantitative prediction model.

[0074] In the embodiment of the present application, the multiple prediction weighted images may be prediction weighted images of the same type or prediction weighted images of different types. When the prediction weighted images include prediction weighted images of the same type, the initial quantitative prediction model is trained according to the loss value between the prediction weighted image and the corresponding sample weighted image to obtain the target quantitative prediction model.

[0075] When the predicted weighted image includes different types of predicted weighted images, the loss value between the predicted weighted image and the sample weighted image under each type is obtained, and the initial quantitative prediction model is trained in combination with multiple loss values ​​to obtain the target quantitative prediction model.

[0076] Furthermore, an initial quantitative prediction model is trained based on multiple predicted weighted images and sample weighted images corresponding to the multiple predicted weighted images to obtain a target quantitative prediction model, including: for each type of predicted weighted image, obtaining a loss value between the predicted weighted image and the corresponding sample weighted image; and the initial quantitative prediction model is trained based on the loss values ​​corresponding to multiple different types of predicted weighted images to obtain a target quantitative prediction model.

[0077] In an embodiment of the present application, the loss value between the predicted weighted image and the corresponding sample weighted image of the same type can be obtained, and the initial quantitative prediction model can be trained according to each loss value in turn. When the loss value under each type is less than the corresponding loss threshold, an intermediate quantitative prediction model is obtained. The sample signal evolution curve diagram is input into the intermediate quantitative prediction model to obtain a new predicted weighted image, and a new loss value between the new predicted weighted image and the corresponding sample weighted image of the same type is obtained. The total loss value is determined according to each new loss value, and the intermediate quantitative prediction model is trained based on the total loss value. When the total loss value is less than the preset loss value, the target quantitative prediction model is obtained.

[0078] In an embodiment of the present application, a plurality of predicted quantitative images are simulated to obtain a plurality of predicted weighted images, and an initial quantitative prediction model is trained based on the plurality of predicted weighted images and sample weighted images corresponding to the plurality of predicted weighted images to obtain a target quantitative prediction model, thereby laying a foundation for subsequently obtaining a quantitative image based on the target quantitative prediction model.

[0079] FIG6 is a flow chart of a method for acquiring a sample signal evolution curve diagram in an embodiment. As shown in FIG6 , the embodiment of the present application relates to a possible implementation method of how to acquire a sample signal evolution curve diagram acquired based on a magnetic resonance fingerprint imaging sequence, including the following steps:

[0080] S601, acquiring sample K-space data acquired based on a magnetic resonance fingerprint imaging sequence.

[0081] S602, reconstructing the sample K-space data to obtain a sample signal evolution curve diagram.

[0082] In an embodiment of the present application, in one possible implementation, the obtained sample K-space data can be directly reconstructed to obtain a sample medical scan image, thereby obtaining a sample signal evolution curve diagram based on the sample medical scan image; in another possible implementation, the sample K-space data can also be compressed to remove redundant information in the sample K-space data, and the compressed sample K-space data can be reconstructed to obtain a sample signal evolution curve diagram. That is, in an embodiment of the present application, when reconstructing the sample K-space data, the sample K-space data can be compressed first and then reconstructed, or the sample K-space data can be reconstructed without compression to obtain a sample signal evolution curve diagram.

[0083] Furthermore, the sample K-space data is reconstructed to obtain a sample signal evolution curve diagram, including: compressing the sample K-space data to obtain compressed sample K-space data; and reconstructing the compressed sample K-space data to obtain a sample signal evolution curve diagram.

[0084] In an embodiment of the present application, the sample K-space data is compressed to obtain compressed sample K-space data. For example, the TR of the radio frequency pulse is changed, and all the K-space data under 1000 TRs are reconstructed, and the obtained sample signal evolution curve diagram is 256*256*1000; all the sample K-space data under 1000 TRs are compressed to obtain compressed sample K-space data, and the compressed sample K-space data is reconstructed, and the obtained sample signal evolution curve diagram can be 256*256*10.

[0085] The compressed sample K-space data can represent the characteristic information of the sample K-space data and will not affect the characteristics.

[0086] In an embodiment of the present application, sample K-space data obtained based on magnetic resonance fingerprint imaging sequence acquisition is obtained, and the sample K-space data is compressed to obtain compressed sample K-space data; the compressed sample K-space data is reconstructed to obtain a sample signal evolution curve diagram, which can eliminate redundant data in the sample K-space data and improve the training speed of the target quantitative prediction model.

[0087] FIG. 7 is a flow chart of a method for generating a medical image in another embodiment. As shown in FIG. 7 , the method includes the following steps:

[0088] S701, acquiring sample K-space data acquired based on a magnetic resonance fingerprint imaging sequence, and acquiring a sample weighted image acquired based on a magnetic resonance imaging sequence;

[0089] S702, compressing the sample K-space data to obtain compressed sample K-space data;

[0090] S703, reconstructing the compressed sample K-space data to obtain a sample signal evolution curve diagram;

[0091] S704, inputting the sample signal evolution curve graph into the initial quantitative prediction model to obtain a plurality of predicted quantitative images;

[0092] S705, performing simulation processing on the multiple predicted quantitative images to obtain multiple predicted weighted images;

[0093] S706, for each type of predicted weighted image, obtaining a loss value between the predicted weighted image and the corresponding sample weighted image;

[0094] S707, training the initial quantitative prediction model according to the loss values ​​corresponding to the multiple different types of prediction weighted images to obtain a target quantitative prediction model;

[0095] S708, obtaining a target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging;

[0096] S709, obtaining a quantitative image of the scanned object according to the target signal evolution curve diagram and a preset target quantitative prediction model.

[0097] In the embodiment of the present application, since the sample weighted image has a quantitative value consistent with the gold standard quantitative image, the sample weighted image is used to train the quantitative prediction model to obtain the target quantitative prediction model, so as to obtain the quantitative image through the target quantitative prediction model, which can avoid the problem that the quantitative image obtained based on the MRF technology is different from the gold standard quantitative image due to factors such as the magnetization transfer effect during the MRF scanning process. Moreover, by using the sample weighted image as the gold standard for training the quantitative prediction model, there is no need to obtain a large number of gold standard quantitative images, which improves the determination efficiency of the target quantitative prediction model.

[0098] It should be understood that, although the steps in the flowcharts involved in the above embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above embodiments may include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0099] Based on the same inventive concept, the embodiment of the present application also provides a medical image generation device for implementing the medical image generation method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the one or more medical image generation device embodiments provided below can refer to the limitations of the medical image generation method above, and will not be repeated here.

[0100] In an exemplary embodiment, as shown in FIG8 , a medical image generation device is provided, comprising: a first acquisition module 11 and a first determination module 12, wherein:

[0101] A first acquisition module 11 is used to acquire a target signal evolution curve diagram of a scanned object in magnetic resonance fingerprint imaging;

[0102] The first determination module 12 is used to obtain a quantitative image of the scanned object according to the target signal evolution curve diagram and a preset target quantitative prediction model; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image.

[0103] In one embodiment, the medical image generating device further comprises:

[0104] A second acquisition module is used to acquire a sample signal evolution curve diagram acquired based on a magnetic resonance fingerprint imaging sequence, and to acquire a sample weighted image acquired based on a magnetic resonance imaging sequence;

[0105] A second determination module is used to input the sample signal evolution curve diagram into the initial quantitative prediction model to obtain a plurality of predicted quantitative images;

[0106] The training module is used to train the initial quantitative prediction model according to multiple predicted quantitative images and sample weighted images to obtain a target quantitative prediction model.

[0107] In one embodiment, the training module is specifically used to simulate multiple predicted quantitative images to obtain multiple predicted weighted images; train the initial quantitative prediction model according to the multiple predicted weighted images and the sample weighted images corresponding to the multiple predicted weighted images to obtain the target quantitative prediction model.

[0108] In one embodiment, the training module is specifically used to obtain the loss value between the predicted weighted image and the corresponding sample weighted image for each type of predicted weighted image; the initial quantitative prediction model is trained according to the loss values ​​corresponding to multiple different types of predicted weighted images to obtain the target quantitative prediction model.

[0109] In one embodiment, the second acquisition module is specifically used to acquire sample K-space data acquired based on the magnetic resonance fingerprint imaging sequence; reconstruct the sample K-space data to obtain a sample signal evolution curve diagram.

[0110] In one embodiment, the second acquisition module is specifically used to compress the sample K-space data to obtain compressed sample K-space data; and reconstruct the compressed sample K-space data to obtain a sample signal evolution curve diagram.

[0111] Each module in the above-mentioned medical image generation device can be implemented in whole or in part by software, hardware and a combination thereof. Each of the above-mentioned modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0112] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps of any of the above method embodiments when executing the computer program.

[0113] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above method embodiments are implemented.

[0114] In one embodiment, a computer program product is provided, including a computer program, which implements the steps of any of the above method embodiments when executed by a processor.

[0115] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0116] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment method can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., but are not limited to this.

[0117] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0118] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for generating a medical image, characterized in that: The method comprises: Obtaining a target signal evolution curve diagram of the scanned object in magnetic resonance fingerprint imaging; According to the target signal evolution curve diagram and the preset target quantitative prediction model, a quantitative image of the scanned object is obtained; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with the gold standard quantitative image.

2. The method according to claim 1, characterized in that The process of obtaining the target quantitative prediction model includes: Acquire a sample signal evolution curve graph acquired based on a magnetic resonance fingerprint imaging sequence, and acquire a sample weighted image acquired based on a magnetic resonance imaging sequence; Inputting the sample signal evolution curve graph into an initial quantitative prediction model to obtain a plurality of predicted quantitative images; The initial quantitative prediction model is trained according to the plurality of predicted quantitative images and the sample weighted images to obtain the target quantitative prediction model.

3. The method according to claim 2, characterized in that The step of training the initial quantitative prediction model according to the plurality of predicted quantitative images and the sample weighted images to obtain the target quantitative prediction model comprises: Performing simulation processing on the plurality of predicted quantitative images to obtain a plurality of predicted weighted images; The initial quantitative prediction model is trained according to the multiple predicted weighted images and the sample weighted images corresponding to the multiple predicted weighted images to obtain the target quantitative prediction model.

4. The method according to claim 3, characterized in that The step of training the initial quantitative prediction model according to the plurality of predicted weighted images and the sample weighted images corresponding to the plurality of predicted weighted images to obtain the target quantitative prediction model comprises: For each type of predicted weighted image, obtaining a loss value between the predicted weighted image and a corresponding sample weighted image; The initial quantitative prediction model is trained according to the loss values ​​corresponding to a plurality of different types of prediction weighted images to obtain the target quantitative prediction model.

5. The method according to claim 2, characterized in that: The step of acquiring a sample signal evolution curve graph acquired based on a magnetic resonance fingerprint imaging sequence includes: Acquire sample K-space data acquired based on a magnetic resonance fingerprint imaging sequence; The sample K-space data is reconstructed to obtain the sample signal evolution curve diagram.

6. The method according to claim 5, characterized in that The reconstructing the sample K-space data to obtain the sample signal evolution curve diagram includes: Compressing the sample K-space data to obtain compressed sample K-space data; The compressed sample K-space data is reconstructed to obtain the sample signal evolution curve diagram.

7. A device for generating a medical image, characterized in that: The device comprises: A first acquisition module is used to acquire a target signal evolution curve diagram of a scanned object in magnetic resonance fingerprint imaging; The first determination module is used to obtain a quantitative image of the scanned object according to the target signal evolution curve diagram and a preset target quantitative prediction model; the target quantitative prediction model is obtained based on sample weighted image training, and the sample weighted image has a quantitative value consistent with a gold standard quantitative image.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images