Image optimization system and method

Abstract

Systems and methods are provided for image optimization. The system may acquire an initial image of a target object (601). The system may also acquire a correlation reference image (603) generated based on a reference image (602) associated with the target object. The second image quality of the reference image (602) may be higher than the first image quality of the initial image (601), while the third image quality of the relevant reference image (603) may be lower than the second image quality. The system may also determine an optimized image (606) of the initial image (601) by inputting the initial image (601) and the correlation reference image (603) into an optimization model (605). The optimization model (605) may refer to a machine learning model configured for high resolution and noise reduction reconstruction using existing priori information in the reference image (602). A fourth image quality of the optimized image (606) may be higher than the first image quality.

Description

Technical Field

[0001] This specification mainly relates to image processing, and in particular to a system and method for image optimization. Background Art

[0002] Magnetic resonance imaging (MRI) is an important imaging technique widely used in disease diagnosis and / or treatment of various medical conditions. Magnetic resonance (MR) images obtained using MRI equipment may not have image quality that meets clinical requirements. This may be due to the resolution of the MRI equipment and / or the acquisition mode of the MR images. For example, MR images may have low resolution and / or a low signal-to-noise ratio (SNR). Therefore, it is desirable to provide systems and methods for image optimization to improve the image quality of MR images. Summary of the Invention

[0003] In one aspect of the present specification, an image optimization method is provided. The method may include obtaining an initial image of a target object. The initial image may have a first image quality. The method may also include obtaining a correlation reference image generated based on a reference image associated with the target object. The reference image may have a second image quality higher than the first image quality, and the correlation reference image may have a third image quality lower than the second image quality. The method may further include determining an optimized image of the initial image by inputting the initial image and the correlation reference image into an optimization model. The optimization model may refer to a machine learning model configured to perform high-resolution and noise reduction reconstruction using prior information present in a reference image. The optimized image may have a fourth image quality higher than the first image quality.

[0004] In another aspect of the present specification, a system for image optimization is provided. The system may include a storage device including a set of instructions and at least one processor in communication with the storage device. When executing the set of instructions, the at least one processor may be configured to instruct the system to perform the following operations. The system may obtain an initial image of a target object. The system may also obtain a correlation reference image generated based on a reference image associated with the target object. The system may further determine an optimized image of the initial image by inputting the initial image and the correlation reference image into the optimization model as described above.

[0005] In another aspect of the present disclosure, a system for image optimization is provided. The system may include an acquisition module configured to acquire an initial image of a target object and obtain a correlation reference image generated based on a reference image associated with the target object. The initial image may have a first image quality. The system may also include a determination module configured to determine an optimized image of the initial image by inputting the initial image and the correlation reference image into an optimization model as described above.

[0006] In another aspect of the present specification, a non-transitory computer-readable medium is provided, which may include executable instructions that, when executed by at least one processor, instruct the at least one processor to perform the method for image optimization as described above.

[0007] In another aspect of the present specification, a method for generating an optimization model is provided. The method can be implemented by a computing device. The method can include obtaining a plurality of training samples, each training sample including a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image. The sample image can have a first image quality, the sample reference image can have a second image quality higher than the first image quality, and the sample gold standard image can have a third image quality higher than the first image quality. The method can also include obtaining an initial machine learning model. The method can also include generating an optimization model through training, using a plurality of training samples, and generating an initial machine learning model according to a training process. The training process can include, for each of the plurality of training samples, determining a sample correlation reference image based on the sample reference image, the sample correction reference image having a fourth image quality lower than the second image quality; and generating an optimization model using each of the plurality of training samples and the corresponding sample correlation reference image.

[0008] In another aspect of the present specification, a system for generating an optimization model is provided. The system may include a storage device comprising a set of instructions and at least one processor in communication. When executing the set of instructions, the at least one processor may be configured to instruct the system to perform the following operations. The system may obtain a plurality of training samples. The system may also obtain an initial machine learning model. The system may further generate the optimization model through training, using a plurality of training samples to generate the initial machine learning model according to the training process described above. The training process may include determining a sample correlation reference image based on a sample reference image for each of the plurality of training samples. The sample correction reference image may have a fourth image quality lower than the second image quality. The training process may also include generating the optimization model using each of the plurality of training samples and the corresponding sample correlation reference image.

[0009] In another aspect, a system for generating an optimization model is provided. The system may include an acquisition module and a training module. The acquisition module may be configured to acquire multiple training samples. The acquisition module may also be configured to obtain an initial machine learning model. The training module may be configured to generate an initial machine learning model through training using the multiple training samples according to the training process described above, thereby generating an optimization model.

[0010] In another aspect of the present disclosure, a non-transitory computer-readable medium is provided, which may include executable instructions that, when executed by at least one processor, instruct the at least one processor to perform the method for generating an optimization model as described above.

[0011] Additional features will be set forth in part in the description which follows and in part will become apparent to those skilled in the art upon examination of the following and accompanying drawings, or may be learned by manufacture or operation of the examples. The features of this specification may be realized and obtained by practice or use of the various aspects of the methods, tools, and combinations described in the detailed examples discussed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] This specification will be further described in the form of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting, and in these embodiments, like numbers represent like structures, wherein:

[0013] Figure 1 is a schematic diagram of a medical imaging system according to some embodiments of the present specification.

[0014] Figure 2 is a schematic diagram of hardware and / or software components of a computing device according to some embodiments of the present specification.

[0015] Figure 3 is a schematic diagram of hardware and / or software components of a mobile device according to some embodiments of the present specification.

[0016] Figure 4A and Figure 4B is a block diagram of a processing device according to some embodiments of the present specification.

[0017] Figure 5 is a flowchart of an exemplary process for image optimization according to some embodiments of the present specification.

[0018] Figure 6 is a schematic diagram of the application of the optimization model according to some embodiments of this specification.

[0019] Figure 7is a flowchart of an exemplary process of generating an optimization model according to some embodiments of the present specification.

[0020] Figure 8 is a flowchart of a training process according to some embodiments of the present specification. DETAILED DESCRIPTION

[0021] In order to more clearly illustrate the technical solutions in this specification, many specific details will be described in the form of examples in the detailed content below. Obviously, for those skilled in the art, this specification can be implemented without these details. In other cases, in order to avoid unnecessary confusion between the various aspects of this specification, well-known methods, processes, systems, components and / or circuits have been described at a relatively high level and without details. It is obvious to those skilled in the art that various changes can be made to the disclosed embodiments, and the general principles defined in this specification can be applied to other embodiments and application scenarios without departing from the spirit and scope of this specification. Therefore, this specification is not limited to the embodiments shown, but is in the broadest scope consistent with the scope of the patent application.

[0022] The terms used herein are used only for the purpose of describing specific example embodiments and are not intended to be restrictive. As used herein, the singular forms "one", "an" and "said" may also be intended to include the plural forms, unless the context clearly indicates otherwise. It should be further understood that the terms "comprise", "include" and / or "include", "include", "include" and / or "include", when used in this specification, specify the presence of the features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups. The term "image" in this specification is used to collectively refer to image data (e.g., scan data) and / or various forms of images, including two-dimensional (2D) images, three-dimensional (3D) images, four-dimensional (4D) images, etc. In this specification, the terms "pixel" and "voxel" are used interchangeably to refer to elements of an image.

[0023] It should be understood that the terms "system", "engine", "unit", "module" and / or "block" as used herein are a way of distinguishing different components, elements, parts, sections or assemblies at different levels in ascending order. However, these terms may be replaced by another way of expression if they achieve the same purpose. In general, the term "module", "unit" or "block" as used herein refers to logic embodied in hardware or firmware, or to a collection of software instructions. The modules, units or blocks described herein may be implemented as software and / or hardware and may be stored in any type of non-transitory computer-readable medium or another storage device. In some embodiments, software modules / units / blocks may be compiled and linked into an executable program. It should be understood that a software module may be called from other modules / units / blocks or from itself, and / or may be called in response to a detected event or interrupt. Configuration for use on a computing device (e.g. Figure 2 The software modules / units / blocks executed on the processor 210 shown in the figure may be provided on a computer-readable medium, such as a compact disc, a digital video disc, a flash drive, a disk, or any other tangible medium, or as a digital download (and may be initially stored in a compressed or installable format, requiring installation, decompression, or decryption before execution). Such software code may be stored in part or in whole on a storage device of the executing computing device for execution by the computing device. The software instructions may be embedded in firmware, such as an EPROM. It will be further understood that hardware modules / units / blocks may be included in connected logical components, such as gates and flip-flops, and / or may include programmable units, such as programmable gate arrays or processors. The modules / units / blocks or computing device functions described herein may be implemented as software modules / units / blocks, but may be represented in hardware or firmware. Generally speaking, the modules / units / blocks described herein refer to logical modules / units / blocks that, although physically organized or stored differently, may be combined with other modules / units / blocks or divided into sub-modules / sub-units / sub-blocks. This description may apply to a system, an engine, or a portion thereof.

[0024] These and other features, as well as the features of the present specification, as well as the methods of operation and function of the related structural elements and the combination and manufacturing economy of the parts, may become more apparent after considering the following description with reference to the accompanying drawings, all of which form a part of this specification. However, it is to be expressly understood that the drawings are for illustration and description purposes only and are not intended to limit the scope of this specification. It should be understood that the drawings are not drawn to scale.

[0025] Provided herein are systems and methods for non-invasive biomedical imaging, for example, for disease diagnosis or research purposes. In some embodiments, the system may include a single-modality imaging system and / or a multi-modality imaging system. A single-modality imaging system may include, for example, a PET system and a computed tomography (CT) system. A multi-modality imaging system may include, for example, a positron emission tomography-computed tomography (PET-CT) system, etc. It should be noted that the imaging systems described below are provided for illustrative purposes only and are not intended to limit the scope of this specification.

[0026] As used herein, the term "imaging method" or "modality" broadly refers to an imaging method or technology that collects, generates, processes, and / or analyzes imaging information of an object. The object may include biological objects and / or non-biological objects. A biological object may be a human, animal, plant, or part thereof (e.g., a cell, tissue, organ, etc.). In some embodiments, the object may be an artificial composition of organic and / or inorganic matter with or without life. The terms "object" and "subject" are used interchangeably.

[0027] In this specification, the term "image" may refer to a two-dimensional (2D) image, a three-dimensional (3D) image, or a four-dimensional (4D) image. As used herein, for the sake of brevity, a representation of a subject (e.g., a patient or a portion thereof) in an image may be referred to as a subject. For example, a representation of an organ or tissue (e.g., a patient's heart, liver, lungs, etc.) in an image may be referred to simply as an organ or tissue. For the sake of brevity, an image including a representation of a subject may be referred to as an image of the subject or an image including the subject. As used herein, operations on the representation of a subject in an image may be referred to as operations on the subject. For example, segmentation of a portion of an image including a representation of an organ or tissue (e.g., a patient's heart, liver, lungs, etc.) from an image may be referred to as segmentation of the organ or tissue.

[0028] In some embodiments, an MRI device can be used to image a subject to obtain MR images of the subject under various conditions. The MRI device may include a volume transfer coil (VTC) and / or one or more surface coils. For example, the surface coil of an MRI device can be used for imaging during radiation therapy planning. When placed on a subject, the surface coil may compress the subject, causing changes in the morphology and / or position of muscles and / or organs. Such changes may be reflected in the MR images obtained, which in turn may affect the accuracy of radiation therapy planned based on the MR images. As another example, in cases where a subject has suffered severe trauma (e.g., a fracture) or is extremely debilitated, using a surface coil may increase pain. Therefore, the subject may need to be imaged without a surface coil, using the MRI device's VTC instead. MR images of the subject can be generated based on MR signals received by the MRI device's VTC. Using a VTC for MR imaging scans eliminates the need for an operator (e.g., a physician or technician) to set up the surface coil, thereby improving the efficiency of imaging and / or radiation therapy planning (e.g., in multimodality scans such as MRI / PET scans). However, because the VTC covers a larger area (e.g., tissue) than a surface coil, and the VTC is farther from the area to be imaged (e.g., tissue) than a surface coil, the SNR of the signal obtained using the VTC may be lower than the SNR of the signal obtained using the surface coil at the same spatial resolution. Furthermore, an MR image of a subject generated based on the signal obtained using the VTC may have relatively lower image quality than an MR image of the subject generated based on the signal obtained using the surface coil. Alternatively, a subject may be imaged using an MRI device with low spatial resolution (e.g., a low-field MRI device), for example, where the subject is affected by a medical condition. Therefore, an image of the subject acquired by an MRI device with low spatial resolution may have lower image quality and / or a lower SNR than an image of the subject acquired by an MRI device with high spatial resolution (e.g., a high-field MRI device). Therefore, there is a need to improve the resolution and / or SNR of MR images acquired using a VTC and / or a low-field MRI device.

[0029] One aspect of the present invention relates to a system and method for image optimization. The system can obtain an initial image of a target object (e.g., an initial MR image). The initial image can have a first image quality. The system can obtain a reference image associated with the target object (e.g., a reference MR image). The reference image can have a second image quality higher than the first image quality. The system can generate a correlation reference image based on the reference image. The correlation reference image can have a third image quality lower than the second image quality (e.g., consistent with the first image quality). The system can determine an optimized image of the initial image by inputting the initial image, the reference image, and the correlation reference image into an optimization model. The optimization model may include a deep feature extraction component and a correlation search component. The optimized image can have a fourth image quality higher than the first image quality (e.g., consistent with the second image quality).

[0030] According to some embodiments of the present invention, images with low image quality (e.g., MRI images) can be accurately processed to obtain images with improved image quality (e.g., images with improved image quality have higher resolution and / or lower noise than images with low image quality). For example, an image with high image quality can be used as a reference image, and deep features of the reference image can be extracted as prior information. Structural features of the low-quality image can be extracted using a structural feature extraction component (e.g., a structure extraction network with multiple channels and modules), which can also remove noise that may eliminate structural features. Using a correlation search component (e.g., a correlation search network), a mapping relationship between features of the high-quality image and features of the low-quality image can be determined. Based on the mapping relationship, the high-quality image features can be transferred to the low-quality image to restore the fine structure corresponding to the subject in the low-quality image, thereby achieving high resolution and / or low noise in the image with improved image quality. In this case, VTC imaging using an MRI device can be used for image-guided radiotherapy, avoiding interference with radiotherapy positioning by surface coils and their stents, fully utilizing the bore space of the MRI device, allowing for a variety of desired radiotherapy positioning, and simulating the patient's physical state during radiotherapy. In addition, using VTC imaging instead of surface coils can simplify setup operations and improve the stability of MRI equipment imaging quality. In addition, MRI equipment with low spatial resolution can be used directly to scan the target object, and improved images can be obtained by using high-quality reference images and optimized models, thereby reducing costs and improving inspection efficiency, combined with scanning using high-spatial-resolution MRI equipment.

[0031] Another aspect of the present specification relates to a system and method for generating an optimization model. The system may obtain multiple training samples, each training sample including a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image. The sample image may have a first image quality. The sample reference image may have a second image quality higher than the first image quality. The sample gold standard image may have a third image quality higher than the first image quality. For example, the second image quality and the third image quality may be consistent with each other (e.g., equal to or the difference between them is less than a threshold). The system may obtain an initial machine learning model. The system may generate the optimization model through training, using multiple training samples, and generating the initial machine learning model according to a training process. The training process may include, for each of the multiple training samples, determining a sample correlation reference image based on the sample reference image. The sample correction reference image may have a fourth image quality lower than the second image quality. The training process may also include generating the optimization model using each of the multiple training samples and the corresponding sample correlation reference image.

[0032] According to some embodiments of the present invention, the optimization model may be trained using sample images of the sample object, sample reference images associated with the sample object, and sample gold standard images corresponding to the sample images of each training sample. For a certain training sample, the corresponding reference image and the corresponding gold standard image may be images of the sample object. The optimization model may be applied to a case where a reference image of the target object is already available, and an optimized image of the target object is generated based on the optimization model, the reference image, and a low-quality image of the target object. Alternatively or additionally, the trained optimization model may be applied to a case where no existing reference image of the target object exists, and an improved image of the target object is to be generated based on the optimization model, the low-quality image of the target object, and a reference image of another object.

[0033] Figure 1 1 is a schematic diagram of an imaging system according to some embodiments of the present invention. The imaging system 100 may include a single-modality system (e.g., an MRI system) or a multi-modality system (e.g., an MRI-guided radiation therapy device). For illustrative purposes, Figure 1 The imaging system 100 shown in FIG may include an MRI system. The MRI system may include an imaging device such as an MRI scanner (also referred to as an MRI device) 110, a network 120, a terminal device 130, a processing device 140, and a storage device 150. The components of the imaging system 100 may be operatively connected in one or more of various ways. For example, Figure 1As shown, the imaging device 110 can be connected to the processing device 140 via the network 120. As another example, the imaging device 110 can be directly operatively connected to the processing device 140. As a further example, the storage device 150 can be directly or operatively connected to the processing device 140 via the network 120. As yet another example, terminal devices (e.g., 131, 132, 133, etc.) can be directly or operatively connected to the processing device 140 via the network 120. It should be noted that the imaging system 100 can include any other imaging device other than an MRI device and is not limited herein.

[0034] In this manual, Figure 1 The x-axis, y-axis, and z-axis shown may form an orthogonal coordinate system. Figure 1 The x-axis and z-axis shown may be horizontal, and the y-axis may be vertical. As shown, the positive x-direction along the x-axis may be from the right side to the left side of the imaging device 110 as seen from the front of the imaging device 110; Figure 1 The positive y-direction of the illustrated y-axis may be from the bottom to the top of the imaging device 110 ; Figure 1 The positive z-direction of the illustrated z-axis may refer to a direction that moves the object out of the detection area (or aperture) of the imaging device 110 .

[0035] The imaging device 110 may be configured to scan at least a portion of the subject and acquire image data (or scan data) related to the subject. The imaging device (e.g., MRI device) 110 may include a magnet (not shown), a coil (not shown), a gantry 112, a patient support 114, etc. The magnet of the imaging device may include a main magnet (e.g., a resistive magnet, a superconducting magnet, or a permanent magnet). Figure 1 As shown, the main magnet can form a hole (e.g., including the detection area) with an axis parallel to the z-direction and surround an object that moves or is positioned in the detection area along the z-direction. The main magnet can also control the uniformity of the generated main magnetic field. The coil may include a gradient coil, a radio frequency (RF) coil, etc. The gradient coil can be located inside the main magnet (e.g., in the hole formed by the main magnet). The gradient coil may be surrounded by the main magnet in the z-direction and be closer to the object than the main magnet. The gradient coil can be configured to generate a gradient magnetic field. The gradient magnetic field can be superimposed on the main magnetic field generated by the main magnet and distort the main magnetic field so that the magnetic orientation of the protons of the object can change with their position within the gradient magnetic field, thereby encoding spatial information into MR signals generated in the imaged object area. The RF coil can be located in the hole formed by the main magnet and serve as a transmitter, a receiver, or both.

[0036] When used as a transmitter, the RF coil can generate an RF signal that provides a magnetic field for generating MR signals associated with the imaged subject region. When used as a receiver, the RF coil may be responsible for detecting MR signals (e.g., echoes). For example, the RF coil used to detect MR signals may include volume transmit coils (VTCs), surface coils, or the like, or any combination thereof. Surface coils may be closer to the imaged region than VTCs. Surface coils may include specialized coils for different regions of the subject (e.g., head coils, rotating coils, body surface coils, neck coils, limb coils, etc.), which can improve signal transmission efficiency and / or image quality of images determined based on the acquired MR signals. However, surface coils may need to be positioned adjacent to the subject's surface and secured with auxiliary devices (e.g., bandages, supports, etc.), which limits their practical use. The gantry 112 may be configured to support the magnet (e.g., the main magnet), the coils (e.g., the gradient coils and / or the RF coils), etc. The gantry 112 may encompass the subject in the z-direction while it is moved into or positioned within the detection region. The patient support 114 may be configured to support the subject. Accordingly, the position of the subject within the detection area can be adjusted by adjusting the patient support 114. For example only, the patient support 114 can be positioned along Figure 1 Move the object into the detection area in the z direction.

[0037] The network 120 may include any suitable network that can facilitate the exchange of information and / or data for the imaging system 100. In some embodiments, one or more components of the imaging system 100 (e.g., the imaging device 110, the terminal device 130, the processing device 140, or the storage device 150) can communicate information and / or data with one or more other components of the imaging system 100 via the network 120. In some embodiments, the network 120 can be any type of wired or wireless network, or a combination thereof.

[0038] The terminal device 130 may include a mobile device 131, a tablet computer 132, a laptop computer 133, or the like, or any combination thereof. In some embodiments, the mobile device 131 may include a smart home device, a wearable device, a smart mobile device, a virtual reality device, an augmented reality device, or the like, or any combination thereof. In some embodiments, the terminal device 130 may remotely operate the imaging device 110 and / or the processing device 140. In some embodiments, the terminal device 130 may operate the imaging device 110 and / or the processing device 140 via a wireless connection. In some embodiments, the terminal device 130 may receive information and / or instructions input by a user and send the received information and / or instructions to the imaging device 110 or the processing device 140 via the network 120. In some embodiments, the terminal device 130 may receive data and / or information from the processing device 140. In some embodiments, the terminal device 130 may be part of the processing device 140. In some embodiments, the terminal device 130 may be omitted.

[0039] The processing device 140 can process data and / or information obtained from the imaging device 110, the terminal device 130 and / or the storage device 150. For example, the processing device 140 can obtain an initial image of the target object. The initial image can have a first image quality. The processing device 140 can obtain a reference image associated with the target object. The reference image can have a second image quality higher than the first image quality. The processing device 140 can generate a correlation reference image based on the reference image. The processing device 140 can determine an optimized image of the initial image by inputting the initial image, the reference image and the correlation reference image into an optimization model. The optimization model may include a deep feature extraction component and a correlation search component. The optimized image can have a fourth image quality higher than the first image quality. As another example, the processing device 140 can generate an optimization model through training, using multiple training samples, and generating an initial machine learning model according to a training process.

[0040] In some embodiments, generation (e.g., training) and / or updating of the optimization model can be performed on a processing device, while application of the optimization model can be performed on a different processing device. In some embodiments, generation and / or updating of the optimization model can be performed on a processing device of a system different from the imaging system 100, or on a server different from the server including the processing device 140 on which application of the optimization model is executed. For example, generation and / or updating of the optimization model can be performed on a first system of a vendor that provides and / or maintains the optimization model and / or has access to training samples used to generate the optimization model, while image optimization based on the provided optimization model can be performed on a second system of a client of the vendor. In some embodiments, generation and / or updating of the optimization model can be performed on a first processing device of the imaging system 100, while application of the optimization model can be performed on a second processing device of the imaging system 100. In some embodiments, generation and / or updating of the optimization model can be performed online in response to an image optimization request. In some embodiments, generation and / or updating of the optimization model can be performed offline.

[0041] In some embodiments, the optimization model can be generated (e.g., trained) and / or updated (or maintained) by, for example, the manufacturer or supplier of the imaging device 110. For example, the manufacturer or supplier can load the optimization model into the imaging system 100 or a portion thereof (e.g., the processing device 140) before or during installation of the imaging device 110 and / or the processing device 140, and maintain or update the optimization model from time to time (regularly or irregularly). Maintenance or updating can be achieved by installing a program stored on a storage device (e.g., a compact disc, a USB drive, etc.) or retrieving the program from an external source (e.g., a server maintained by the manufacturer or supplier) via the network 120. The program can include a new model (e.g., a new optimization model) or a portion thereof that replaces or supplements a corresponding portion of the optimization model.

[0042] In some embodiments, the processing device 140 may be a single server or a server group. The server group may be centralized or distributed. In some embodiments, the processing device 140 may be local or remote. In some embodiments, the processing device 140 may be implemented on a cloud platform. For example, the cloud platform may include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an inter-cloud, a multi-cloud, or any combination thereof. In some embodiments, the processing device 140 may be implemented by having a Figure 2 One or more of the components shown may be implemented on a computing device 200 .

[0043] The storage device 150 can store data and / or instructions. In some embodiments, the storage device 150 can store data obtained from the imaging device 110, the terminal device 130 and / or the processing device 140. In some embodiments, the storage device 150 can store data and / or instructions that the processing device 140 can execute or use to execute the exemplary methods described in the present disclosure. In some embodiments, the storage device 150 may include a large-capacity storage device, a removable storage device, a volatile read-write memory, a read-only memory (ROM), etc., or any combination thereof. In some embodiments, the storage device 150 can be implemented on a cloud platform. For example, the cloud platform can include a private cloud, a public cloud, a hybrid cloud, a community cloud, a distributed cloud, an inter-cloud, a multi-cloud, etc., or any combination thereof.

[0044] In some embodiments, the storage device 150 can be connected to the network 120 to communicate with one or more components of the imaging system 100 (e.g., the imaging device 110, the processing device 140, the terminal device 130, etc.). One or more components of the imaging system 100 can access data or instructions stored in the storage device 150 via the network 120. In some embodiments, the storage device 150 can be part of the processing device 140, or it can be independent and directly or indirectly connected to the processing device 140.

[0045] It should be noted that the above description of imaging system 100 is provided for illustrative purposes only and is not intended to limit the scope of the present disclosure. Numerous variations and modifications may be made by one skilled in the art based on the teachings of this specification. For example, imaging system 100 may include one or more additional components, and / or one or more components of imaging system 100 described above may be omitted. Additionally or alternatively, two or more components of imaging system 100 may be integrated into a single component. Components of imaging system 100 may be implemented in two or more subassemblies.

[0046] Figure 2 is a schematic diagram illustrating hardware and / or software components of an exemplary computing device 200 that may be implemented according to some embodiments of the present specification. The computing device 200 may be used to implement any component of the imaging system described herein. For example, the processing device 140 and / or the terminal device 130 may be implemented on the computing device 200 through its hardware, software program, firmware, or a combination thereof. Although only one such computing device is shown, for convenience, the computer functions associated with the imaging system 100 described herein may be implemented in a distributed manner across many similar platforms to distribute the processing load. Figure 2 As shown, computing device 200 may include processor 210 , storage device 220 , input / output (I / O) 230 , and communication port 240 .

[0047] Processor 210 can execute computer instructions (program code) and perform the functions of processing device 140 in accordance with the techniques described herein. Computer instructions may include, for example, routines, programs, objects, components, signals, data structures, processes, modules, and functions that perform the specific functions described herein. For example, processor 210 can perform attenuation correction on PET images to generate optimized images of a target object. As another example, processor 210 can generate an optimization model based on machine learning techniques. In some embodiments, processor 210 can execute instructions obtained from terminal device 130. In some embodiments, processor 210 may include one or more hardware processors, such as a microcontroller, a microprocessor, a reduced instruction set computer (RISC), an application-specific integrated circuit (ASIC), an application-specific instruction set processor (ASIP), a central processing unit (CPU), a graphics processing unit (GPU), a physical processing unit (PPU), a microcontroller unit, a digital signal processor (DSP), a field-programmable gate array (FPGA), an advanced RISC machine (ARM), a programmable logic device (PLD), any circuit or processor capable of performing one or more functions, or the like, or any combination thereof.

[0048] For example purposes only, only one processor is described in computing device 200. However, it should be noted that computing device 200 in this specification may also include multiple processors. Therefore, operations and / or method steps described in this specification as being performed by one processor may also be performed jointly or individually by multiple processors. For example, if in this specification, a processor of computing device 200 is described as performing operation A and operation B simultaneously, it should be understood that operation A and operation B may also be performed jointly or individually by two or more different processors in computing device 200 (e.g., a first processor performs operation A and a second processor performs operation B, or the first processor and the second processor perform operations A and B jointly).

[0049] The storage device 220 can store data / information obtained from the imaging device 110, the terminal device 130, the storage device 150, or any other component of the imaging system 100. In some embodiments, the storage device 220 can include a mass storage device, a removable storage device, a volatile read-write memory, a read-only memory (ROM), or the like, or any combination thereof. In some embodiments, the storage device 220 can store one or more programs and / or instructions for executing the exemplary methods described herein. For example, the storage device 220 can store a program for the processing device 140 to perform attenuation correction on PET images.

[0050] The (I / O) 230 may input or output signals, data, and / or information. In some embodiments, the (I / O) 230 may enable user interaction with the processing device 140. In some embodiments, the (I / O) 230 may include input devices and output devices. Exemplary input devices may include a keyboard, a mouse, a touch screen, a microphone, or the like, or a combination thereof. Exemplary output devices may include a display device, a speaker, a printer, a projector, or the like, or a combination thereof. Exemplary display devices may include a liquid crystal display (LCD), a light emitting diode (LED)-based display, a flat panel display, a curved screen, a television device, a cathode ray tube (CRT), or the like, or a combination thereof.

[0051] The communication port 240 can be connected to a network (e.g., network 120) to facilitate data communication. The communication port 240 can establish a connection between the processing device 140 and the imaging device 110, the terminal device 130, or the storage device 150. The connection can be a wired connection, a wireless connection, or a combination of the two that enable data transmission and reception. The wired connection may include an electrical cable, an optical cable, a telephone line, etc., or any combination thereof. The wireless connection may include a Bluetooth network, a Wi-Fi network, a WiMax network, a WLAN, a ZigBee network, a mobile network (e.g., 3G, 4G, 5G, etc.), or any combination thereof. In some embodiments, the communication port 240 can be a standardized communication port, such as RS232, RS485, etc. In some embodiments, the communication port 240 can be a specially designed communication port. For example, the communication port 240 can be designed according to the Digital Imaging and Communications (DICOM) protocol in medicine.

[0052] Figure 3 is a diagram illustrating hardware and / or software components of an exemplary mobile device 300 according to some embodiments of the present specification. In some embodiments, one or more components of the imaging system 100 (eg, the terminal device 130 and / or the processing device 140) can be implemented on the mobile device 300.

[0053] like Figure 3As shown, mobile device 300 may include a communication platform 310, a display 320, a graphics processing unit (GPU) 330, a central processing unit (CPU) 340, (I / O) 350, memory 360, and storage 390. In some embodiments, any other suitable components, including but not limited to a system bus or controller (not shown), may also be included in mobile device 300. In some embodiments, a mobile operating system 370 (e.g., iOS, Android, Windows Phone, etc.) and one or more applications 380 may be loaded from storage 390 into memory 360 for execution by CPU 340. Application 380 may include a browser or any other suitable mobile application for receiving and presenting information related to image processing or other information from processing device 140. User interaction with the information stream may be implemented through (I / O) 350 and provided to processing device 140 and / or other components of imaging system 100 via network 120.

[0054] In order to implement the various modules, units and functions thereof described in this specification, a computer hardware platform can be used as the hardware platform for one or more elements described herein. The hardware components, operating systems and programming languages of these computers are conventional in nature, and it is assumed that those skilled in the art are familiar enough with this to adjust these technologies to generate the images described herein. A computer with user interface elements can be used to implement a personal computer (PC) or another type of workstation or terminal device, although the computer can also act as a server if properly programmed. It is believed that those skilled in the art are familiar with the structure, programming and general operation of such computer equipment, and therefore, the accompanying drawings should be self-explanatory.

[0055] Figure 4A and Figure 4B is a block diagram illustrating exemplary processing devices 140a and 140b according to some embodiments of the present specification. In some embodiments, processing devices 140a and 140b may be based on Figure 1 In some embodiments, the processing devices 140a and 140b may be respectively in a processing unit (e.g., Figure 2 The processor 210 shown in Figure 3 340). By way of example only, processing device 140a may be implemented on the (CPU) 340 of the terminal device, and processing device 140b may be implemented on the computing device 200. Alternatively, processing devices 140a and 140b may be implemented on the same computing device 200 or the same (CPU) 340. For example, processing devices 140a and 140b may be implemented on the same computing device 200.

[0056] like Figure 4AAs shown, the processing device 140a may include an acquisition module 401 , a generation module 403 and an optimization module 405 .

[0057] The acquisition module 401 may be configured to acquire data / information related to image optimization from one or more components of the imaging system 100. For example, the acquisition module 401 may acquire an initial image of a target object to be optimized, a reference image associated with the target object, and / or an optimized image from a storage device (e.g., storage device 150, storage device 220, etc.). As another example, the acquisition module 401 may acquire image / scan data of the target object from a storage device. The acquisition module 401 may reconstruct the initial image of the target object based on the scan data according to an image reconstruction algorithm. Further description of acquisition operations may be found elsewhere in this specification (e.g., operations 501 and 503 and their related descriptions).

[0058] The generation module 403 may be configured to generate a correlation reference image based on the reference image. For example, the generation module 403 may generate the correlation reference image by performing one or more processing operations on the reference image, more descriptions of which may be found elsewhere in this specification (e.g., operation 505 and its related description).

[0059] The optimization module 405 can be configured to determine an optimized image of the initial image by inputting the initial image, the reference image, and the correlation reference image into an optimization model. The optimization model can be a machine learning model such as a deep learning network, or a model configured for high-resolution and noise reduction reconstruction. More description of the determination of the optimized image can be found elsewhere in this specification (e.g., operation 507 and its related description).

[0060] like Figure 4B As shown, the processing device 140 b may include an acquisition module 407 and a training module 409 .

[0061] The acquisition module 407 may be configured to acquire data / information for model training. For example, the acquisition module 407 may acquire multiple training samples, each of which includes a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image. For another example, the acquisition module 407 may acquire an initial machine learning model for training an optimization model. Further description of obtaining multiple training samples and / or an initial machine learning model can be found elsewhere in this specification (e.g., operations 701 and 703 and their related descriptions).

[0062] The training module 409 may be configured to train an initial machine learning model using a plurality of training samples according to a training process, thereby generating an optimized image. The training process may include, for each of the plurality of training samples, determining a sample correlation reference image based on the sample reference image, and generating an optimization model using each of the plurality of training samples and the corresponding sample correlation reference image. Further description of training the optimization model may be found elsewhere in this specification (e.g., operations 705 and 801-805 and their associated descriptions).

[0063] It should be noted that the above description is provided for illustrative purposes only and is not intended to limit the scope of this specification. Obviously, for those with ordinary skills in the art, various changes and modifications can be made under the guidance of this specification. However, these variations and modifications do not depart from the scope of this specification. Each of the above modules can be a hardware circuit designed to perform certain actions, for example, according to a set of instructions stored in one or more storage media, and / or any combination of hardware circuits and one or more storage media. In some embodiments, the processing device 140a and / or the processing device 140b can share the two or more modules, and any one of the modules can be divided into two or more units. For example, the processing devices 140a and 140b can share the same acquisition module, that is, the acquisition module 401 and the acquisition module 407 are the same module. In some embodiments, the processing device 140a and / or the processing device 140b can include one or more additional modules, such as a storage module (not shown) for storing data. In some embodiments, the processing device 140a and the processing device 140b can be integrated into the same processing device.

[0064] Figure 5 is a flow chart illustrating an exemplary process for image optimization according to some embodiments of the present specification. In some embodiments, process 500 may be implemented as a set of instructions (e.g., an application) stored in a storage device (e.g., storage device 150, storage device 220, and / or storage device 390). The processing device 140a (e.g., processor 210, CPU 340, and / or Figure 4A The one or more modules shown in the figure may execute the set of instructions, and when executing the instructions, the processing device 140a may be configured to perform the process 500. The operations of the process shown below are intended to be illustrative. In some embodiments, the process 500 may be completed by one or more additional operations not described and / or without one or more operations discussed. In addition, in Figure 5 The order in which the operations of process 500 are illustrated and described below is not intended to be limiting.

[0065] In 501, the processing device 140a (e.g., the acquisition module 401) can obtain an initial image of a target object. The target object can be biological or non-biological. For example, the target object can include a patient, an artificial object, etc., as described in other parts of this specification.

[0066] In some embodiments, the initial image of the target object may be a medical image (e.g., an MRI image, a CT image, etc.) that does not meet clinical needs and requires optimization. For example, the initial image of the target object may have a low image resolution and / or a low signal-to-noise ratio. In some embodiments, the initial image of the target object may have a first image quality. For example, the first image quality may be less than or equal to the clinically required image quality. For example, the initial image of the target object may be generated based on MR signals received by a VTC of an imaging device (e.g., an imaging device 110 such as an MRI device). As another example, the initial image of the target object may be obtained by an imaging device with a low spatial resolution.

[0067] In some embodiments, the initial image of the target object can be pre-generated and stored in a storage device (e.g., storage device 150, storage device 220, etc.), and the processing device 140a can retrieve the initial image of the target object from the storage device. Alternatively, the initial image of the target object can be generated by the processing device 140a. For example, an imaging device (e.g., imaging device 110) can be instructed to scan the target object to obtain scan data of the target object. The processing device 140a can reconstruct the initial image of the target object based on the scan data according to an image reconstruction algorithm. Exemplary image reconstruction algorithms may include two-dimensional Fourier transform technology, back projection technology (e.g., convolution back projection technology, filtered back projection technology), iterative reconstruction technology, etc., or any combination thereof.

[0068] In 503 , the processing device 140 a (eg, the acquisition module 401 ) may obtain a reference image associated with the target object.

[0069] In some embodiments, the reference image associated with the target object may be a medical image with high image quality. For example, the reference image may have a second image quality that is higher than the first image quality. In some embodiments, the reference image associated with the target object may have clearer and / or more detailed structural information than the initial image. In some embodiments, the reference image associated with the target object may include a previous image of the target object or another object (i.e., an object other than the target object). The object other than the target object may be similar to the target object. That is, the object may have similar profile information (e.g., gender, age, weight, height, etc.) and / or medical condition as the target object. For example, the object may have the same gender, the same age (or substantially the same age), the same weight (or substantially the same weight), and / or the same height (or substantially the same height) as the target object. As used herein, "substantially," when used to define a characteristic (e.g., "equivalent to"), means that the deviation from the characteristic is below a threshold, e.g., 30%, 25%, 20%, 15%, 10%, 5%, etc. For example, "substantially the same age / weight / height" means the age / weight / height is plus or minus (±) the corresponding threshold. For another example, subjects other than the target subject may have the same medical disease (eg, breast cancer, lung cancer, etc.) that requires treatment as the target subject.

[0070] In some embodiments, the reference image may be acquired by the same or different imaging device (e.g., imaging device 110) as the imaging device used to acquire the initial image. For example, the reference image may be acquired by the same imaging device as the initial image. The initial image may be generated using the imaging device's VTC, while the reference image may be generated from MR signals acquired using the imaging device's surface coil. As another example, the reference image and the initial image may be acquired by different imaging devices. The initial image may be acquired by an imaging device with low spatial resolution, while the reference image may be acquired by an imaging device with high spatial resolution. In some embodiments, the reference image may be acquired using the same imaging sequence as the initial image. For example, the initial image may be acquired using a T1 sequence, and the reference image may be acquired using a T1 sequence. As another example, the initial image may be acquired using a T2 sequence, and the reference image may be acquired using a T2 sequence. In some embodiments, the reference image may correspond to the same portion / region as the initial image. For example, when the initial image is an MR image of the target patient's chest, the reference image may be a previous MR image of the target patient's chest or an MR image of the chest of a patient similar to the target patient. For another example, when the initial image is an MR image of the target patient's head, the reference image may be a previous MR image of the target patient's head or an MR image of the head of a patient similar to the target patient.

[0071] In some embodiments, a reference image of the target object may have been previously stored in a storage device (e.g., storage device 150, storage device 220, etc.), and the processing device 140a may retrieve the reference image from the storage device. For example, the storage device may store a reference image pool containing a plurality of reference images. The processing device 140a may determine whether there is a previous image of the target object with high image quality in the reference image pool. In order to determine whether there is a previous image of the target object with high image quality in the reference image pool, the processing device 140a may obtain an image of the target object with high image quality as a reference image associated with the target object. In order to determine whether there is a previous image of the target object with high image quality in the reference image pool, the processing device 140a may obtain an object image similar to the target object from the reference image pool, and determine that the reference image associated with the target object is an object image similar to the target object.

[0072] In 505 , the processing device 140 a (eg, the generation module 403 ) may generate a correlation reference image based on the reference image.

[0073] In some embodiments, the correlation reference image may have a third image quality that is lower than the second image quality. For example, the third image quality may be consistent with the first image quality (e.g., the difference between the third image quality and the first image quality is less than a threshold). As another example, the correlation reference image may have a lower image resolution and / or higher noise than the reference image. In some embodiments, the correlation reference image may have the same fine features (e.g., depth features, shallow features such as texture features, etc.) as the reference image in a high-dimensional space.

[0074] In some embodiments, processing device 140a may generate a correlation reference image by performing one or more processing operations on a reference image to ensure that a third image quality of the correlation reference image is consistent with the first image quality of the initial image, thereby enabling further processing of the initial image using the correlation reference image. The one or more operations may not destroy fine features of the reference image in the high-dimensional space. The one or more processing operations may include at least one of a downsampling operation, an upsampling operation, a noise addition operation, or a filtering operation. By way of example only, processing device 140a may generate a downsampled image by performing a downsampling operation on the reference image. The downsampled image may have an image resolution lower than that of the reference image. Processing device 140a may generate an upsampled image by performing an upsampling operation on the downsampled image. The upsampled image may have the same image resolution as the reference image. Processing device 140a may generate the correlation reference image by performing a noise addition operation on the upsampled image. For example, processing device 140a may add one or more noises to the upsampled image. The one or more noises may include white noise, Gaussian noise, Lil noise, noise with a non-central chi-square distribution, or any combination thereof. In some embodiments, parameters such as the degree of downsampling and / or upsampling and the type and / or amount of one or more noises may be default settings or adjusted based on different circumstances (e.g., based on user experience or automatically). For example, processing device 140a may determine the degree of downsampling and the degree of upsampling based on user experience. In another example, processing device 140a may determine the degree of downsampling and the degree of upsampling by analyzing the initial image. In another example, processing device 140a may determine the type of one or more noises based on the type of optimized model. Different types of optimized models may correspond to different types of noise. As a further example, processing device 140a may determine the type of one or more noises by analyzing the noise in the initial image. As yet another example, processing device 140a may determine the type of one or more noises based on the type of the initial image. For example, if the initial image is an MR image, processing device 140a may determine the type of one or more noises based on the classical noise of MR images. Alternatively, parameters such as the degree of downsampling and / or upsampling and the type and / or amount of one or more noises may be determined during the training of the optimized model used in step 507. In some embodiments, the processing device 140a may generate a correlation reference image by performing one or more processing operations on the reference image without damaging the detailed features (e.g., depth features) of the reference image in a high-dimensional space. In other words, the correlation reference image may retain the detailed features of the reference image.

[0075] In 507, the processing device 140a (eg, the acquisition module 401, the optimization module 405, etc.) may determine an optimized image of the initial image by inputting the initial image, the reference image, and the correlation reference image into the optimization model.

[0076] In some embodiments, an optimized image of an initial image is an image generated by performing an optimization operation on the initial image based on a reference image using an optimization model. The optimized image of the initial image may have a fourth image quality that is higher than the first image quality. For example, the optimized image may have a higher image resolution and / or lower noise than the initial image.

[0077] As used herein, an optimization model refers to a machine learning model, such as a deep learning network, that is configured to use prior information present in a reference image for high-resolution and noise-reduced reconstruction. In some embodiments, the optimization model can be any type of deep learning network or model. For example, the optimization model can include a trained neural network model, such as a trained convolutional neural network (CNN) model, a trained generative adversarial network (GAN) model, or any other suitable type of model. In some embodiments, the processing device 140a (e.g., the acquisition module 401) can obtain the optimization model from one or more components of the imaging system 100 (e.g., the storage device 150, the terminal 130) or an external source via a network (e.g., the network 120). For example, the optimization model can be pre-trained by a computing device (e.g., the processing device 140b) of the imaging system 100 and stored in a storage device (e.g., the storage device 150, the storage device 220, and / or the storage device 390) of the imaging system 100. The processing device 140a can access the storage device and retrieve the optimization model. In some embodiments, the optimization model can be generated based on a machine learning algorithm. The machine learning algorithm may include, but is not limited to, an artificial neural network algorithm, a deep learning algorithm, a decision tree algorithm, an association rule algorithm, an inductive logic programming algorithm, a support vector machine algorithm, a clustering algorithm, a Bayesian network algorithm, a reinforcement learning algorithm, a representation learning algorithm, a similarity and metric learning algorithm, a sparse dictionary learning algorithm, a genetic algorithm, a rule-based machine learning algorithm, or the like, or any combination thereof. The machine learning algorithm used to generate the optimization model may be a supervised learning algorithm, a semi-supervised learning algorithm, an unsupervised learning algorithm, or the like. In some embodiments, the optimization model may be generated by a computing device (e.g., processing device 140b) by executing a process (e.g., process 700) for generating the optimization model disclosed herein. More descriptions of the generation of the optimization model can be found elsewhere in this specification. See, for example, Figure 7 -9 and its related description.

[0078] In some embodiments, the optimization model can be a trained deep learning model comprising a plurality of networks (also referred to as components) that are operably connected. For example, the plurality of components can be connected via a plurality of operations. The plurality of components may include a deep feature extraction network (also referred to as a deep feature extraction component), a structural feature extraction network (also referred to as a structural feature extraction component), an association search network (also referred to as a correlation search component), an image generation network (also referred to as an image generation component), and the like, or any combination thereof. The deep feature extraction component can be configured to extract multiple layers of features from an image (e.g., an initial image, a reference image, and / or a correlation reference image). The structural feature extraction component can be configured to extract structural features from an image (e.g., an initial image). The correlation search component can be configured to construct a correspondence between features of the initial image (e.g., multiple layers of features and / or structural features) and features of the reference image (e.g., multiple layers of features and / or structural features). More description of the connections and / or functions of the plurality of components of the optimization model can be found elsewhere in this specification (e.g., Figure 6 and its related descriptions).

[0079] In some embodiments, during the application of the optimization model, the input of the optimization model may include an initial image, a reference image, and a correlation reference image, and the output of the optimization model may include an optimized image. Multiple components of the optimization model may perform optimization operations on the initial image based on the reference image and the correlation reference image. For example, the processing device 140a may directly input the initial image, the reference image, and the correlation reference image into the optimization model, and the optimization model may output an optimized image of the initial image. Alternatively, the processing device 140a may pre-process the initial image, the reference image, and / or the correlation reference image (for example, perform denoising operations, normalization operations, etc. on them), and / or post-process the output of the optimization model (for example, perform denormalization operations on them) to generate an optimized image of the initial image. More descriptions about generating optimized images of target objects by applying optimization models can be found elsewhere in this specification (for example, Figure 6 and its description).

[0080] In some embodiments, processing device 140a can transmit the optimized image of the initial image to a terminal (e.g., terminal device 130) for display. Optionally, a user of the terminal (e.g., a doctor or operator) can input a response regarding the optimized image of the initial image through, for example, an interface of the terminal. For example, the user can evaluate whether the optimized image of the initial image meets preset conditions (e.g., clinical needs, such as clinically required image quality). Based on the evaluation results, the user can send a request to processing device 140a, for example, to repeat or redo the optimization.

[0081] It should be noted that the above description of process 500 is provided for illustrative purposes only and is not intended to limit the scope of the present disclosure. Numerous variations and modifications may be made by those skilled in the art based on the teachings of this specification. However, such variations and modifications do not depart from the scope of this specification. In some embodiments, one or more operations of process 500 may be omitted, and / or one or more additional operations may be added to method 500. For example, a storage operation may be added elsewhere in process 500. During the storage operation, processing device 140a may store information and / or data related to the image optimization process in a storage device (e.g., storage device 150) disclosed elsewhere in this specification. For another example, operation 505 may be omitted, and an optimized image of the initial image may be generated without a correlation reference image by applying the optimization model. In some embodiments, during the application of the optimization model, the input to the optimization model may include the initial image without a reference image and the correlation reference image, and the output of the optimization model may include an optimized image of the initial image. For example, acquisition module 401 may acquire the initial image and the correlation reference image and input them into the optimization model to output the optimized image. Alternatively, the input to the optimization model may include only the initial image, and the output of the optimization model may include an optimized image of the initial image.

[0082] Figure 6 is a schematic diagram illustrating the application of an exemplary optimization model according to some embodiments of the present specification. Optimization model 605 provides an example of the optimization model described in operation 507 and is not intended to limit the scope of the present specification. A description of the application of optimization model 605 can be found elsewhere in this specification. For example, see Figure 5 In some embodiments, the application of the optimization model 605 may be performed by Figure 5 The processing device 140a shown is implemented.

[0083] like Figure 6As shown, processing device 140a may obtain an initial image 601 of a target object. Initial image 601 may be similar to the initial image described in operation 501. For example, initial image 601 may be a medical image (e.g., an MR image) whose first image quality does not meet clinical requirements and requires optimization. Processing device 140a may obtain a reference image 602 associated with the target object. Reference image 602 may be similar to the reference image described in operation 503. For example, reference image 602 may be a previous image of the target object whose second image quality is higher than the first image quality. Processing device 140a may generate a correlation reference image 603 based on reference image 602. Correlation reference image 603 may be similar to the correlation reference image described in operation 505. For example, processing device 140a may generate correlation reference image 603 according to operation 505. Processing device 140a may generate optimized image 606 of initial image 601 by inputting initial image 601, reference image 602, and correlation reference image 603 into optimization model 605. The optimization model 605 may be similar to the optimization model comprising multiple components (ie, networks) described in operation 507. Figure 6 For the purpose of illustration, the optimization model 605 may include a deep feature extraction network 610, a correlation search network 620, a structural feature extraction network 630, an image generation network 640, etc., or any combination thereof.

[0084] The deep feature extraction network 610 can be configured to extract multi-layer features from an image (e.g., the initial image 601, the reference image 602, or the correlation reference image 603). The multi-layer features may include deep features and / or shallow features. As used herein, the deep features of an image may include abstract information of the image (e.g., semantic information). As used herein, the shallow features of an image may include basic information of the image (e.g., texture information, contour information, such as color, shape, counter, etc.). In some embodiments, the deep feature extraction network 610 may be a multi-layer residual connection network that can extract deep features and shallow features from an image, respectively, to generate a feature map corresponding to the image. The feature map corresponding to the image may indicate the extracted deep features of the image and the extracted shallow features of the image. That is, the input of the deep feature extraction network 610 may include one or more images, and the output of the deep feature extraction network 610 may include one or more feature maps corresponding to the one or more images, respectively. For example, the processing device 140a may extract a first multi-layer feature from the initial image 601 to generate a first feature map 611. The processing device 140a may extract second multi-layer features from the reference image 602 to generate a second feature map 612. The processing device 140a may extract third multi-layer features from the correlation reference image 603 to generate a third feature map 613. The first multi-layer features, the second multi-layer features, and / or the third multi-layer features may include deep features and / or shallow features of their corresponding images.

[0085] The correlation search network 620 can be configured to establish a correspondence between features of the initial image (multi-layer features and / or structural features) and features of the reference image. For example, since the correlation reference image 603 has the same image quality as the initial image 601, the correlation search network 620 can determine a correlation map 621 (not shown) between the initial image 601 and the correlation reference image 603 based on the third feature map 613 corresponding to the correlation reference image 603 and the first feature map 611 corresponding to the initial image 601. Figure 6 ). That is, the input of the association search network 620 may include the first feature map 611 and the third feature map 613, and the output of the association search network 620 may include the correlation map 621. The correlation map 621 may indicate the correspondence between the initial image 601 and the correlation reference image 603, which may in turn be used to indicate the correspondence between the initial image 601 and the reference image 602.

[0086] The structural feature extraction network 630 can be configured to extract structural features from an image (eg, an initial image). In some embodiments, the structural feature extraction network 630 can extract structural features from the initial image 601 to generate a fourth feature map (not shown in FIG. Figure 6), for example, by using a plurality of operation units (e.g., residual connection units, dense connection units, etc.) to remove noise that may occlude structural features. That is, the input of the structural feature extraction network 630 may include the initial image 601, and the output of the structural feature extraction network 630 may include a fourth feature map that may indicate the structural features extracted from the initial image 601.

[0087] The image generation network 640 can be configured to generate an optimized image 606 of the initial image 601 by performing image reconstruction. In some embodiments, the image generation network 640 can be a deep learning network that can generate the optimized image 606 by performing image reconstruction based on the second feature map 612, the correlation map 621, and the fourth feature map. For example, the input of the image generation network 640 may include the second feature map 612, the correlation map 621, and the fourth feature map, and the output of the image generation network 640 may include the optimized image 606. In some embodiments, the second feature map may be processed before being input to the image generation network 640 to improve the accuracy and / or image quality of the optimized image 606. For example, the deep feature extraction network 610 can be operably connected to the image generation network 640 through one or more operations (e.g., an attention algorithm). The attention algorithm can be configured to fuse features (e.g., deep features of the second feature map 612 and the multi-scale correlation map) to generate a correlation feature map 622 (not in Figure 6As used herein, multi-scale features refer to features whose dimensions are arranged in different orders. For example, multi-scale features may include features with dimensions on the order of millimeters and features with dimensions on the order of microns. By using an attenuation algorithm, one or more attention weights may be generated for features of interest (i.e., depth features). The relevant feature map 622 may include one or more attention weights. The larger the attention weight, the more attention is given to the features corresponding to the attention weight during the image reconstruction process. In other words, during the image reconstruction process, features corresponding to high attention weights may receive more attention than features corresponding to low attention weights. For example, an attenuation algorithm may be used to process (e.g., compare or search) the second feature map 612 and the correlation map 621 in a first scale (e.g., features with respect to the first order of dimension), a second scale (e.g., features with respect to the second order of dimension), etc. to determine the relevant feature map 622. The dimension of the first order may be greater than the dimension of the second order. In addition, the image generation network 640 may perform image reconstruction by fusing the features of the relevant feature map 622, the correlation map 621, and the fourth feature map to generate an optimized image 606. In this case, the input of the image generation network 640 may include the correlation map 621, the fourth feature map, and the related feature map 622, and the output of the image generation network 640 may include the optimized image 606. During the image reconstruction process, the related feature map 622 may guide the image reconstruction to focus on features in the high-dimensional space according to one or more corresponding attention weights, and accordingly, accurate local details and key structures may be transferred from the reference image 602 to the initial image 601 to generate the optimized image 606.

[0088] It should be noted that the above description of the optimization model 605 is provided for illustrative purposes only and is not intended to limit the scope of this specification. For those with ordinary skills in the art, various changes and modifications can be made according to the teachings of this specification. However, such variations and modifications do not depart from the scope of this specification. In some embodiments, one or more networks or components can be added to the optimization model 605 and / or one or more networks or components of the optimization model 605 can be omitted. In some embodiments, networks or components can be added to generate a correlation reference image 603 based on the reference image 602. In this case, the input of the optimization model 605 may include only the initial image 601 and the reference image 602, without including the correlation reference image 603.

[0089] Figure 77 is a flow chart illustrating an exemplary process 700 for generating an optimization model according to some embodiments of the present specification. In some embodiments, the process 700 may be implemented as a set of instructions (e.g., an application program) stored in a storage device (e.g., storage device 150, storage device 220, and / or storage device 390). The processing device 140b (e.g., processor 210, CPU 340, and / or Figure 4B The one or more modules shown in the figure may execute the set of instructions, and when executing the instructions, the processing device 140b may be configured to perform the process 700. The operations of the process shown below are intended to be illustrative. In some embodiments, the process 700 may be completed by one or more additional operations not described and / or without one or more operations discussed. In addition, in Figure 7 The order of operations of process 700 shown in FIG and described below is not intended to be limiting. In some embodiments, the process 700 may be obtained according to Figure 5 The optimization model described in operation 503 or Figure 6 Optimization model 605 is described. In some embodiments, process 700 may be performed by another device or system other than imaging system 100, such as a device or system of a manufacturer's supplier. For illustrative purposes, implementation of process 700 by processing device 140b is described as an example.

[0090] In 701, the processing device 140b (eg, the acquisition module 407) may obtain a plurality of training samples. The training samples may include a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image.

[0091] As used herein, a sample object refers to an object whose image is used to train an optimization model. Figure 5One or more features of the subject of the description. By way of example only, if an optimization model is used to optimize an image of a patient's head (or a portion thereof), the sample object may be the head (or a portion thereof) of another patient. As used herein, a sample image of a sample object refers to a sample image used to train the optimization model, and a sample gold standard image corresponding to the sample image refers to a sample image used as the ground truth for the sample image. In some embodiments, for one of a plurality of training samples, the sample image may have a first image quality, the sample reference image may have a second image quality higher than the first image quality, and the sample gold standard image may have a third image quality higher than the first image quality. For example, for the training sample, the sample image may be generated based on MR signals received by a VTC of an imaging device or acquired by the imaging device at a low spatial resolution; the sample reference image and / or the sample gold standard image of the training sample may be generated based on MR signals received by a surface coil of the imaging device or acquired by an imaging device with a high spatial resolution. In some embodiments, the second image quality may be consistent with the third image quality (e.g., the difference between the second image quality and the third image quality is below a threshold). In some embodiments, for a training sample, the same imaging sequence (eg, T1 sequence, T2 sequence, etc.) may be used to obtain a sample image, a sample reference image, and a sample gold standard image of the training sample.

[0092] In some embodiments, for a training sample, the sample reference image associated with the sample object and / or the sample gold standard image corresponding to the sample image may include a previous image of the sample object or an image of another sample object other than the sample object that is similar to the description of the reference image associated with the target object. The other sample object may be similar to the sample object, for example, having similar contour information as the sample object. For example, each of the sample reference image and the sample gold standard image may include a previous image of the sample object. As another example, the sample reference image may include a sample image of another sample object other than the sample object, and the sample gold standard image may include a previous image of the sample object. In some embodiments, the sample reference image and the sample gold standard image of the same training sample may be acquired using the same imaging device under the same imaging conditions. For example, for a training sample, the sample reference image and the sample gold standard image may be the same image. As another example, for a training sample, the sample reference image may be acquired by performing a first scan of the sample object using the first scanning parameters using the imaging device, and the sample gold standard image may be acquired by performing a second scan of the same sample object using the same first scanning parameters using the same imaging device. By acquiring sample reference images and sample gold standard images of the same training sample based on different scans under the same imaging conditions, multiple training samples can be made random in imaging quality, thereby further improving the accuracy of the optimization model.

[0093] In some embodiments, the training samples may be pre-generated and stored in a storage device (e.g., storage device 150, storage device 220, storage device 390, or an external database). The processing device 140b may retrieve the training samples directly from the storage device. In some embodiments, at least a portion of the training samples may be generated by the processing device 140b. By way of example only, for the training samples, the processing device 140b may obtain a sample image and a sample reference image from a storage device; the processing device 140b may use an image optimization technique other than an optimization model to determine a sample gold standard image based on the sample image. In another embodiment, for the training samples, the processing device 140b may obtain a sample reference image and a sample gold standard image from a storage device; the processing device 140b may use image processing techniques including, for example, downsampling, upsampling, noise addition, filtering, etc. to determine one or more sample images based on the sample reference image or the sample gold standard image.

[0094] In 703 , the processing device 140 b (eg, the acquisition module 407 ) may obtain an initial machine learning model.

[0095] The initial machine learning model may be any type of machine learning model, similar to the optimization model described in operation 507. For example, the initial machine learning model may include multiple components or networks, such as a deep feature extraction component, a structural feature extraction component, a correlation search component, an image generation component, etc., or any combination thereof, similar to Figure 6 Multiple components of the optimization model 605. In some embodiments, the initial machine learning model may include one or more additional components, such as skip connections, residual blocks, dense blocks, etc., or any combination thereof. These additional components can be configured to combine different features extracted by different components or networks of the initial machine learning model, thereby accelerating convergence during model training and improving the accuracy of the resulting optimized model.

[0096] In some embodiments, the initial machine learning model may include one or more model parameters. The processing device 140b may initialize the parameter values of the model parameters before training, and the model parameter values of the initial machine learning model may be updated during the training process of the initial machine learning model. For illustrative purposes, the initial machine learning model may be a deep learning model, such as a neural network. Exemplary model parameters of the initial machine learning model may include a loss function, the number (or count) of convolutional layers, the number (or count) of kernels, kernel size, stride, padding of each convolutional layer, etc., or any combination thereof.

[0097] In 705, the processing device 140b (e.g., the training module 409) can train the initial machine learning model using multiple training samples according to a training process to generate an optimized model.

[0098] In some embodiments, the training process may include, for each of the plurality of training samples, determining a sample correlation reference image based on a sample reference image. The sample correction reference image may have a fourth image quality that is lower than the second image quality. For example, for the training sample, the processing device 140b may generate the sample correlation reference image by performing one or more operations on the sample reference image, the operations being similar to the generation of the correlation reference image based on the reference image described in operation 505. The training process may also include generating an optimization model using each of the plurality of training samples and the corresponding sample correlation reference image.

[0099] In some embodiments, the components or networks of the initial machine learning model (e.g., the depth feature extraction component, the structural feature extraction component, the correlation search component, the image generation component, etc.) can be trained in parallel during the training process. That is, multiple components or networks of the initial machine learning model can be operably connected and updated as a whole. As an example only, the processing device 140b can train the initial machine learning model by iteratively updating the model parameters of the initial machine learning model based on multiple training samples and corresponding sample correlation reference images.

[0100] In some embodiments, the training process may include one or more iterations. For one of the multiple iterations, the processing device 140b may, for one of the multiple training samples, generate a sample prediction optimized sample image by inputting the training sample (e.g., a sample image of the training sample and a sample reference image) into the updated machine learning model determined in the previous iteration. For example, the processing device 140b may generate a sample prediction optimized image by inputting the sample image, the sample reference image of the training sample, and the sample correlation reference image into the initial machine learning model (e.g., in the first iteration) or the updated machine learning model, where the sample correlation reference image may be determined based on the sample reference image. The processing device 140b may determine a sample evaluation result of the updated machine learning model based on the sample predicted optimized image and the sample gold standard image of the training sample. The processing device 140b may update the parameter values of the updated machine learning model based on the sample evaluation result. More description of one or more iterations of the training process may be found elsewhere in this specification (e.g., Figure 8 and its related descriptions).

[0101] It should be noted that the above description of process 700 is provided for illustrative purposes only and is not intended to limit the scope of this specification. A person having ordinary skill in the art may make various changes and modifications based on the teachings of this specification. However, such changes and modifications do not depart from the scope of this specification. In some embodiments, one or more operations may be added to or omitted from the process 700. For example, a storage operation for storing the optimization model may be added after operation 705 for further use (e.g., in conjunction with Figure 5 As another example, after generating the optimization model, processing device 140b may further test the optimization model using a set of test samples. Additionally or alternatively, processing device 140b may periodically or irregularly update the optimization model based on one or more newly generated training images (e.g., new samples generated in medical diagnosis).

[0102] Figure 8800 is a flow chart illustrating an exemplary training process according to some embodiments of the present specification. In some embodiments, process 800 may be implemented as a set of instructions (e.g., an application) stored in a storage device (e.g., storage device 150, storage device 220, and / or storage device 390). The processing device 140b (e.g., processor 210, (CPU) 340, and / or Figure 4B The one or more modules shown in FIG. 8 may execute the set of instructions, and when executing the instructions, the processing device 140b may be configured to perform the process 800. In some embodiments, one or more operations of the process 800 may be performed to implement the Figure 7 At least a portion of operation 705 described in connection with the present invention may be performed. For example, process 800 may be performed to implement an iteration (e.g., a current iteration) of the training process described in operation 705, during which components of the initial machine learning model are trained in parallel. The current iteration may be performed based on at least a portion of the training samples. In some embodiments, the same set or different sets of training samples may be used in different iterations of the training process.

[0103] In 801, for one of a plurality of training samples, the processing device 140b (e.g., the training module 409) may generate a predicted optimized image by inputting the training sample (e.g., a sample image and a sample reference image of the training sample) into an updated machine learning model determined in a previous iteration.

[0104] When applying the updated machine learning model to a training sample, the updated machine learning model's deep feature extraction component can be configured to extract a first sample multi-layer feature from the training sample's sample image to generate a first sample feature map; extract a second sample multi-layer feature from the training sample's sample reference image to generate a second sample feature map; and extract a third sample multi-layer feature from a sample correlation reference image generated based on the sample reference image to generate a third sample feature map. At least one of the first sample multi-layer feature, the second sample multi-layer feature, or the third sample multi-layer feature can include sample deep features and / or sample shallow features. The updated machine learning model's structural feature extraction component can be configured to extract sample structural features from the training sample's sample image to generate a fourth sample feature map. The updated machine learning model's correlation search component can be configured to determine a sample correlation map between the training sample's sample image and the corresponding sample correlation reference image based on the first sample feature map and the third sample feature map. The updated machine learning model's image generation component can be configured to generate a sample predicted optimized image of the training sample's sample image based on the second sample feature map, the sample correlation map, and the fourth sample feature map. In some embodiments, the deep feature extraction component can be operably connected to the image generation component via an attention algorithm. The attenuation algorithm can be configured to fuse features of the second sample feature map and the sample correlation map at multiple scales. For example, the attenuation algorithm can be used to generate a sample correlation feature map based on the second sample feature map and the sample correlation map. In this case, the input of the image generation component can include the fourth sample feature map, the sample correlation map, and the sample correlation feature map, and the output of the image generation component can include a sample prediction optimized image of the sample image of the training sample.

[0105] In 803, the processing device 140b (e.g., the training module 409) can determine the evaluation result of the updated machine learning model based on the sample predicted optimization image and the sample gold standard image of the training sample.

[0106] The sample evaluation result may indicate the accuracy and / or efficiency of the updated machine learning model. In some embodiments, the sample evaluation result may be associated with the difference between the sample prediction optimization image and the sample gold standard image of the training sample. For example, the value of the loss function may be determined to measure the difference between the sample prediction optimization image and the sample gold standard image of the training sample. The processing device 140b may determine the sample evaluation result based on the value of the loss function. As another example, the value of the overall loss function may be determined to measure the overall difference between each sample prediction optimization image and the sample gold standard image in multiple training samples (i.e., the training sample in operation 801 and other training samples of the multiple training samples). The processing device 140b may determine the sample evaluation result based on the value of the overall loss function. In some embodiments, the sample evaluation result may be associated with the time required for the updated machine learning model to generate the sample prediction optimization image of the training sample. For example, the shorter the time required, the higher the efficiency the updated machine learning model may have.

[0107] In some embodiments, the example evaluation results may include a determination as to whether a termination condition is met in the current iteration. In some embodiments, the termination condition may be related to the value of the loss function and / or the overall loss function in the updated machine learning model. For example, if the value of the loss function and / or the overall loss function is minimum or less than a threshold value (e.g., a constant), the termination condition may be met. As another example, the termination condition may be met if the value of the loss function and / or the overall loss function converges. In some embodiments, if, for example, the change in the value of the loss function and / or the overall loss function in two or more consecutive iterations is equal to or less than a threshold value (e.g., a constant), a certain iteration count may be performed, or the like, then convergence may be considered to have occurred. In some embodiments, the termination condition may be met if the time required for the updated machine learning model to generate a sample prediction optimization image for the training sample is less than a threshold value.

[0108] In 805, the processing device 140b (e.g., the training module 409) may update the parameter values of the updated machine learning model based on the sample evaluation results.

[0109] In some embodiments, processing device 140b may determine whether a termination condition is satisfied based on the sample evaluation results. In response to a determination that the termination condition is satisfied, processing device 140b may determine the updated machine learning model as the optimized model. In response to a determination that the termination condition is not satisfied, processing device 140b may update parameter values of the updated machine learning model based on the sample evaluation results for use in the next iteration.

[0110] It should be noted that the above description of process 800 is provided for illustrative purposes only and is not intended to limit the scope of this specification. For those with ordinary skills in the art, various changes and modifications can be made according to the teachings of this specification. However, these variations and modifications do not depart from the scope of this specification. The operations of the method shown above are intended to be illustrative. In some embodiments, the process 800 can be completed by one or more additional operations not described and / or without one or more operations discussed.

Claims

1. An image optimization method, implemented by a computing device, comprising: obtaining an initial image of a target object, wherein the initial image has a first image quality; obtaining a correlation reference image generated based on a reference image associated with the target object, wherein the reference image has a second image quality higher than the first image quality, and the correlation reference image has a third image quality lower than the second image quality; An optimized image of the initial image is determined by inputting the initial image and the correlation reference image into an optimization model, wherein the optimization model refers to a machine learning model configured for high-resolution and noise reduction reconstruction based on prior information present in the reference image, and the optimized image has a fourth image quality higher than the first image quality.

2. The method of claim 1, wherein the optimization model includes a deep feature extraction component configured to: Extracting first multi-layer features from the initial image to generate a first feature map; extracting second multi-layer features from the reference image to generate a second feature map; and A third multi-layer feature is extracted from the correlation reference image to generate a third feature map.

3. The method of claim 2, wherein at least one of the first multi-layer feature, the second multi-layer feature, or the third multi-layer feature comprises a deep feature and / or a shallow feature.

4. The method of claim 2 or 3, wherein the optimization model includes a correlation search component configured to determine a correlation map between the initial image and the correlation reference image based on the first feature map and the third feature map. 5 . The method according to claim 2 , wherein the optimization model further comprises a structural feature extraction component configured to extract structural features from the initial image to generate the fourth feature map.

6. The method of claim 4 or 5, wherein the optimization model further comprises an image generation component configured to generate the optimized image based on the second feature map, the correlation map and the fourth feature map.

7. The method of claim 6, wherein the deep feature extraction component is operably connected to the image generation component via an attention algorithm.

8. The method of claim 7, wherein the attention algorithm is configured to fuse features of the second feature map and the correlation map at multiple scales.

9. The method of claim 7 or 8, further comprising: Based on the second feature map and the correlation map, a correlation feature map is generated using the attention algorithm, wherein The inputs of the image generation component include the fourth feature map, the correlation map and the correlation feature map, and, The output of the image generation component includes the optimized image.

10. The method of claims 1 to 9, wherein the initial image is obtained using the same imaging sequence as the reference image.

11. The method of claims 1 to 10, wherein the initial image is generated based on magnetic resonance (MR) signals received by a volume transmit coil (VTC); and the reference image is generated based on MR signals received by a surface coil.

12. The method of claim 1 , wherein generating the correlation reference image based on the reference image comprises the following operations: The correlation reference image is generated by performing one or more processing operations on the reference image, wherein the one or more processing operations include at least one processing operation selected from a downsampling operation, an upsampling operation, a noise addition operation, or a filtering operation.

13. The method of claims 1 to 12, wherein the reference image associated with the target object comprises a previous image of the target object or an image of an object other than the target object.

14. The method of claims 1 to 13, wherein the optimization model is trained using a plurality of training samples.

15. The method of claim 14, wherein each of at least one training sample in the plurality of training samples comprises a sample image of a sample object and a sample reference image of the sample object.

16. The method of claim 14 or 15, wherein each of at least one training sample of the plurality of training samples comprises a sample image of a first sample object and a sample reference image of a second sample object, the object of the second sample being different from the first sample object.

17. A method for generating an optimization model, the method being implemented by a computing device, the method comprising: obtaining a plurality of training samples, each training sample comprising a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image, the sample image having a first image quality, the sample reference image having a second image quality higher than the first image quality, and the sample gold standard image having a third image quality higher than the first image quality; Obtain an initial machine learning model; and Generating an optimized model by training the initial machine learning model using the multiple training samples according to a training process includes: for each of the plurality of training samples, determining a sample-related reference image based on the sample reference image, the sample-related corrected image having a fourth image quality lower than the second image quality; and The optimization model is generated using each of the plurality of training samples and the corresponding sample-correlation reference image.

18. The method of claim 17, wherein the initial machine learning model comprises a deep feature extraction component, a structural feature extraction component, a correlation search component, and an image generation component that are trained in parallel.

19. The method of claim 18, wherein for each of the plurality of training samples and the corresponding sample correlation reference image, The deep feature extraction component is configured as follows: Extracting first sample multi-layer features from the sample image of the training sample to generate a first sample feature map; Extracting second sample multi-layer features from the sample reference image of the training sample to generate a second sample feature map; and Extracting a third sample multi-layer feature from the corresponding sample correlation reference image to generate a third sample feature map; The structural feature extraction component is configured to extract sample structural features from the sample image of the training sample to generate a fourth sample feature map; The correlation search component is configured to determine a sample correlation map between a sample image of the training sample and a corresponding sample correlation reference image based on the first sample feature map and the third sample feature map; and The image generation component is configured to generate a sample prediction optimized image of the sample image of the training sample based on the second sample feature map, the sample correlation map and the fourth sample feature map.

20. The method of claim 19, wherein at least one of the first sample multi-layer feature, the second sample multi-layer feature, or the third sample multi-layer feature comprises a sample deep layer feature and / or a sample shallow layer feature.

21. The method of any one of claims 18 to 20, wherein the deep feature extraction component is operably connected to the image generation component via an attention algorithm.

22. The method of claim 21, wherein the attenuation algorithm is configured to fuse features of the second sample feature map and the sample correlation map at multiple scales.

23. The method of claim 21 or 22, further comprising: Based on the second sample feature map and the sample correlation map, a sample correlation feature map is generated using the attenuation algorithm, wherein The input of the image generation component includes the fourth sample feature map, the sample correlation map and the sample correlation feature map, and The output of the image generation component includes a sample predicted optimized image of the sample image of the training sample.

24. The method of claims 17 to 23, wherein the training process comprises a plurality of iterations, each of the plurality of iterations comprising: For one of the plurality of training samples, generating a sample prediction optimization image by inputting the training sample into the updated machine learning model determined in the previous iteration; Determining a sample evaluation result of the updated machine learning model based on the optimized image predicted by the sample and the sample gold standard image of the training sample; and Update parameter values of the updated machine learning model based on the sample evaluation results.

25. The method of claim 24, wherein the sample evaluation result is determined based on at least one of the following conditions: The difference between the sample prediction optimization image of the training sample and the sample gold standard image of the training sample, or The time required for the updated machine learning model to generate a sample prediction optimized image of the training sample.

26. The method of claims 17 to 25, wherein the sample reference image associated with the sample object or the sample gold standard image corresponding to each of the sample images in the plurality of training samples includes a previous image of the sample object or a sample object image of another sample object other than the sample object.

27. The method of claims 17 to 25, wherein the sample image, the sample reference image, and the sample gold standard image corresponding to the sample image of the training sample are obtained using the same imaging sequence.

28. An image optimization system, comprising: a storage device comprising a set of instructions; There is at least one processor in communication with the storage device, wherein, when executing the instruction set, the at least one processor is configured to direct the system to perform the following: Acquire an initial image of the target object, wherein the initial image has a first image quality; obtaining a correlation reference image generated based on a reference image associated with the target object, the reference image having a second image quality higher than the first image quality, and the correlation reference image having a third image quality lower than the second image quality; An optimized image of the initial image is determined by inputting the initial image and the correlation reference image into an optimization model, wherein the optimization model refers to a machine learning model configured for high-resolution and noise reduction reconstruction based on prior information present in the reference image, and the optimized image has a fourth image quality higher than the first image quality.

29. An image optimization system, comprising: An acquisition module, wherein the acquisition module is aligned as follows: Acquire an initial image of the target object, wherein the initial image has a first image quality; obtaining a correlation reference image generated based on a reference image associated with the target object, the reference image having a second image quality higher than the first image quality, and the correlation reference image having a third image quality lower than the second image quality; and A determination module is configured to determine an optimized image of the initial image by inputting the initial image and the correlation reference image into an optimization model, wherein the optimization model refers to a machine learning model configured for high-resolution and noise reduction reconstruction using prior information present in the reference image, and the optimized image has a fourth image quality higher than the first image quality.

30. A non-transitory computer-readable medium comprising executable instructions that, when executed by at least one processor, direct the at least one processor to perform a method for image optimization, the method comprising: obtaining an initial image of a target object, the initial image having a first image quality; obtaining a correlation reference image generated based on a reference image associated with the target object, wherein the reference image has a second image quality higher than the first image quality, and the correlation reference image has a third image quality lower than the second image quality; An optimized image of the initial image is determined by inputting the initial image and the correlation reference image into an optimization model, wherein the optimization model refers to a machine learning model configured for high-resolution and noise reduction reconstruction based on prior information present in the reference image, and the optimized image has a fourth image quality higher than the first image quality.

31. A system for generating an optimization model, the system comprising: a storage device comprising a set of instructions; There is at least one processor in communication with the storage device, wherein, when executing the instruction set, the at least one processor is configured to direct the system to perform the following: obtaining a plurality of training samples, each of the training samples comprising a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image, the sample image having a first image quality, the sample reference image having a second image quality higher than the first image quality, and the sample gold standard image having a third image quality higher than the first image quality; Obtaining an initial machine learning model; and Generating an optimized model by training the initial machine learning model using a plurality of training samples according to a training process includes: For each of the plurality of training samples, determining a sample correlation reference image based on the sample reference image, the sample correction reference image having a fourth image quality lower than the second image quality; and The optimization model is generated using each of the plurality of training samples and the corresponding sample-correlation reference image.

32. A system for generating an optimization model, the system comprising: Configured as an acquisition module obtaining a plurality of training samples, each of the training samples comprising a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image, the sample image having a first image quality, the sample reference image having a second image quality higher than the first image quality, and the sample gold standard image having a third image quality higher than the first image quality; and Obtain an initial machine learning model; and The training module is configured to generate the optimization model through training, using the multiple training samples, and the initial machine learning model generated according to the training process includes: for each of the plurality of training samples, determining a sample correlation reference image based on the sample reference image, the sample correction reference image having a fourth image quality lower than the second image quality; and The optimization model is generated using each of the plurality of training samples and the corresponding sample-correlation reference image.

33. A non-transitory computer-readable medium comprising executable instructions that, when executed by at least one processor, direct the at least one processor to perform a method for generating an optimization model, the method comprising: obtaining a plurality of training samples, each of the training samples comprising a sample image of a sample object, a sample reference image associated with the sample object, and a sample gold standard image corresponding to the sample image, the sample image having a first image quality, the sample reference image having a second image quality higher than the first image quality, and the sample gold standard image having a third image quality higher than the first image quality; Obtain an initial machine learning model; and The step of generating an optimized model by training the initial machine learning model using the multiple training samples according to a training process includes: for each of the plurality of training samples, determining a sample correlation reference image based on the sample reference image, the sample correction reference image having a fourth image quality lower than the second image quality; and The optimization model is generated using each of the plurality of training samples and the corresponding sample-correlation reference image.

Images