Medical imaging method and system

Abstract

The application provides a medical imaging method and system, and a non-transitory computer readable storage medium. The medical imaging method comprises: acquiring an original image collected by an X-ray imaging system; and performing post-processing on the original image based on a trained network to obtain a processed optimized image.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medical imaging, and more particularly to a medical imaging method and system, and a non-transitory computer readable storage medium. BACKGROUND

[0002] In an X-ray imaging system, radiation from a source is directed at a subject, typically a patient in a medical diagnostic application. A portion of the radiation passes through the patient and impinges on a detector, which is divided into a matrix of discrete elements, such as pixels. The detector elements are read out to produce output signals based on the amount or intensity of radiation that impinges on each pixel area. The signals can then be processed to produce a medical image that can be displayed for viewing, which can be displayed in a display device of the X-ray imaging system.

[0003] The image quality of the medical image based on the signals acquired by the detector is not ideal, and usually needs to be post-processed (e.g., noise reduction, detail enhancement, contrast adjustment, etc.) to improve the image quality, however, in this process, there is a mutual dependence between the image processing parameters, and the adjustment of the parameters of a certain post-processing will affect the effect of other post-processing, for example, adjusting the contrast of the medical image based on the contrast parameter, however, the adjustment of this parameter may affect the details of the medical image (e.g., blur some key details), accordingly, processing the background noise of the medical image may affect the details (or contrast or other aspects) of the medical image, therefore, it is difficult to repeatedly try and adjust the corresponding parameters in each post-processing process to obtain a more ideal optimized image. SUMMARY

[0004] The present application provides a medical imaging method and system, and a non-transitory computer readable storage medium, the imaging method constructs and trains a network based on deep learning technology, and by inputting the original image acquired by the X-ray imaging system into the network, an optimized image after post-processing can be obtained (output), which has good performance in contrast, detail enhancement, noise reduction, etc.

[0005] An exemplary embodiment of the present application provides a medical imaging method, which comprises acquiring an original image acquired by an X-ray imaging system and post-processing the original image based on a trained network to obtain an optimized image after processing.

[0006] An exemplary embodiment of the present application also provides a non-transitory computer readable storage medium for storing a computer program, which, when executed by a computer, causes the computer to execute the instructions for the medical imaging method described above.

[0007] The exemplary embodiments of the present application also provide a medical imaging system, which comprises a control module and a post-processing module. The control module is configured to acquire an original image acquired by an X-ray imaging system, and the post-processing module is configured to perform post-processing on the original image based on a trained network to obtain an optimized processed image.

[0008] Other features and aspects will become apparent from the following detailed description, drawings and claims. BRIEF DESCRIPTION OF DRAWINGS

[0009] The present application can be better understood with reference to the following examples described in conjunction with the following drawings, in which:

[0010] Figure 1 is a schematic diagram of an X-ray imaging system according to some embodiments of the present application;

[0011] Figure 2 is a schematic diagram of a medical imaging system according to some embodiments of the present application;

[0012] Figure 3 is a schematic diagram of an optimization process according to some embodiments of the present application;

[0013] Figure 4 is a schematic diagram of a network according to some embodiments of the present application;

[0014] Figure 5 is a schematic diagram of a medical imaging system according to some embodiments of the present application; Figure 2 is a schematic diagram of an image display interface of a display module in the medical imaging system shown in FIG. 8; and

[0015] Figure 6 is a flowchart of a medical imaging method according to some embodiments of the present application. DETAILED DESCRIPTION

[0016] In the following detailed description of implementations consistent with the present disclosure, various specific details are presented to provide a thorough understanding of implementations. However, it will be apparent to those skilled in the art that specific implementations can be practiced without these specific details. In other instances, well-known structures have not been described in detail in order to avoid obscuring the present disclosure. Where considered useful to clarify the present disclosure, a term commonly used in the art, and sometimes used without a specific accompanying definition, is first provided a definition herein. It will also be apparent to one skilled in the art that, as used in the specification, including the appended claims, the singular forms "a," "an," and "the" include their plural, or

[0017] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains. The terminology used in the description of the disclosure presented herein and the claims filed herewith is not intended to be interpreted in any specific manner unless as otherwise defined herein. Use of the terms "first," "second," and like terms are not intended to denote any order, quantity, or importance, but are used to distinguish one element from another, and the terms "one" and "a" and like terms used in the context of describing the disclosure are to be interpreted to cover at least one. The terms "including" and "comprising" and like terms are not intended to exclude the presence of other elements or materials. The terms "connected" and "coupled" and like terms are not intended to limit to a direct and physical connection or coupling, but can also include an indirect and / or a wireless connection or coupling.

[0018] In X-ray imaging systems, the interaction between various adjustment parameters during post-processing can cause one parameter to affect other post-processing steps. The medical imaging methods and systems in some embodiments of this invention, based on deep learning technology, can post-process the raw images obtained from the X-ray imaging system, directly acquiring optimized images after post-processing. This not only reduces the difficulty and time of parameter adjustment during post-processing but also improves the image quality of the optimized images obtained after post-processing. Furthermore, by training different learning networks according to the preferences of different users (e.g., doctors) and selecting networks based on user instructions, optimized images with different styles (i.e., different user preferences) can be obtained. It should be noted that, from the perspective of those skilled in the art or related fields, such descriptions should not be construed as limiting the invention solely to digital X-ray imaging systems. In fact, the medical imaging methods and systems described herein can be reasonably applied to other imaging fields, both medical and non-medical, such as CT systems, MR systems, PET systems, SPECT systems, or any combination thereof.

[0019] Figure 1 An X-ray imaging system 100 according to some embodiments of the present invention is shown. For example... Figure 1 As shown, the X-ray imaging system 100 includes an X-ray source 104, a detector 106, and a control subsystem 108. In some embodiments, the X-ray imaging system 100 may be a fixed X-ray imaging system located in a fixed X-ray imaging room, or it may be a mobile X-ray imaging system.

[0020] X-ray source 104 can project X-rays 114 onto a region of interest (ROI) in the object being inspected 102. Specifically, X-ray source 104 can be positioned adjacent to collimator 116, which is used to align X-rays 114 onto the ROI in the object being inspected 102. At least a portion of X-rays 114 can be attenuated by the object being inspected 102 and can be incident on detector 106.

[0021] The detector 106 includes a detector panel array that includes a pixel array of light-sensing photodiodes and switching thin-film field effect transistors (FETs) that convert photons into electrical signals. A scintillator material deposited on the pixel array of photodiodes and FETs converts incident X-ray radiation photons received on the surface of the scintillator material into lower energy photons. As mentioned above, the pixel array of photodiodes and FETs converts the photons into electrical signals. Alternatively, the detector panel array can directly convert the X-ray photons into electrical signals. These electrical signals are provided by the detector panel array interface (which provides digital signals) to the computing device 120 to convert into image data and reconstruct medical images of the detected object 102. In some embodiments, the detector 106 can be configured to convert the incident X-rays 114 into electrical signals using photoconversion, direct conversion, and / or any other suitable detection technique. In some embodiments, the detector 106 includes a wireless communication interface for wireless communication with the communication link 112 and a wired communication interface for wireless and / or wired communication with the control subsystem 108, where the wireless communication interface can utilize any suitable wireless communication protocol, such as an Ultra-Wideband (UWB) communication standard, a Bluetooth communication standard, or any IEEE 802.11 communication standard.

[0022] The detector 106 can also be configured to transmit the unprocessed or partially processed image data to a workstation or a portable detector control device via a wired or wireless connection or to transmit the processed X-ray images to a printer to generate a copy of the image. The portable detector control device can include a personal digital assistant (PDA), a palmtop computer, a laptop computer, a smart phone, a tablet computer such as an iPad™, or any suitable general or special purpose portable interface device. The portable detector control device is configured to be held by a user and to wirelessly communicate with the detector 106. Note that the detector and the portable detector control device can utilize any suitable wireless communication protocol, such as an IEEE 802.15.4 protocol, a UWB communication standard, a Bluetooth communication standard, or any IEEE 802.11 communication standard. Alternatively, the portable detector control device can be configured to be tethered or detachably tethered to the detector 106 to communicate via a wired connection.

[0023] The control subsystem 108 includes a source controller (not shown) and a detector controller (not shown). The source controller is used to command the x-ray source 104 to emit x-rays 114 for image exposure. The detector controller is used to coordinate control of various detector functions, such as, for example, various signal processing and filtering functions, specifically, initial adjustment of dynamic range, interleaving of digital image data, etc. In some embodiments, the control subsystem 108 can provide power and timing signals for controlling operation of the x-ray radiation source 104 and the detector 106. Specifically, the control subsystem 108 can provide power and timing signals to the x-ray radiation source 104 and / or the detector 106 through use of a power source 110 and one or more wired and / or wireless communication links 112, respectively, where the communication links 112 can correspond to a backplane bus, a local area network, a wide area network, and / or the Internet, etc. In some embodiments, the power source 110 includes one or more batteries, and, although Figure 1 Although the power source 110 is shown as being connected to the x-ray source 104 through a communication link, it will be appreciated by those skilled in the art that the power source 110 and the x-ray source 104 can also be directly coupled.

[0024] The control subsystem 108 can be set up and / or arranged to be used in different ways. For example, in some implementations, a single control subsystem 108 can be used; in other implementations, multiple control subsystems 108 are configured to work together (e.g., based on a distributed processing configuration) or separately, each control subsystem 108 being configured to handle specific aspects and / or functions, and / or to process data for generating models for use only with specific medical imaging systems. In some implementations, the control subsystem 108 can be local (e.g., co-located with one or more x-ray imaging systems 100, e.g., within the same facility and / or the same local network); in other implementations, the control subsystem 108 can be remote, and thus only accessible via a remote connection (e.g., via the Internet or other available remote access technology). In particular implementations, the control subsystem 108 can be configured in a cloud-like manner, and can be accessed and / or used in a manner that is substantially similar to the way other cloud-based systems are accessed and used.

[0025] In some embodiments, the system 100 also includes a computing device 120, which can be configured to use the digitized signals to reconstruct one or more desired images and / or to determine useful diagnostic information corresponding to the object 102 being imaged, where the computing device 120 can include one or more special-purpose processors, graphics processing units, digital signal processors, microcomputers, microcontrollers, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other appropriate processing devices.

[0026] In some embodiments, the system 100 further includes a storage device 122, to which the computing device 120 can store digitized signals. For example, the storage device 122 can include a hard disk drive, a floppy disk drive, a compact disk read / write (CD-R / W) drive, a digital versatile disk (DVD) drive, a flash drive, and / or a solid-state storage device. The storage device is used to store a program that can be executed by a computer, which, when the computer executes the program, can cause the plurality of components of the X-ray imaging system to implement operations corresponding to the above-described imaging sequence. When the computer executes the program, a medical imaging method can also be performed to post-process the original images to obtain optimized images after post-processing.

[0027] Although Figure 1 Although the storage device 122, the computing device 120, and the control subsystem 108 are illustrated as separate devices, in some embodiments, one or more of them can be combined into a single device to effectively use floor space and / or to meet expected imaging requirements.

[0028] In one embodiment, the system 100 further includes a display device 124, which can be used to display reconstructed images and / or diagnostic information, etc. For example, the display device 124 can display the above-described original images and / or optimized images, the graphical display interface of which will be combined with the above-described display device 124. Figure 3 Further description.

[0029] In one embodiment, the system 100 further includes an operator workstation 126, which allows a user to receive and evaluate reconstructed images, and to input control instructions (operation signals or control signals). The operator workstation 126 can include a user interface (or user input device) such as a keyboard, a mouse, a voice-activated controller, or any other suitable input device in some form of operator interface through which an operator can input operation signals / control signals to the control subsystem 108, such as one or more scanning parameters and / or request required diagnostic information and / or images to evaluate the internal structure and / or function of the subject 102.

[0030] Figure 2 A medical imaging system 200 according to some embodiments of the present application is shown, Figure 3 A schematic diagram of an optimization process according to some embodiments of the present application is shown, as Figures 2-3 As shown, the medical imaging system 200 includes a control module 210, a post-processing module 220, and a display module 230.

[0031] The control module 210 is configured to acquire original images acquired by an X-ray imaging system.

[0032] In some embodiments, the original images are acquired by the detector 106 (e.g., the detector 106 of the X-ray imaging system 100 of FIG. 1) and / or the computing device 120 (e.g., the computing device 120 of the X-ray imaging system 100 of FIG. 1). Figure 1The X-ray attenuation through the object under examination is collected and data processing and reconstruction are performed to obtain a raw image, as shown in Figure 3 The raw image 246 can be displayed in the display module 230. In some embodiments, the control module 210 is connected with or part of the control subsystem 108 (or the computing device 120) in the X-ray imaging system 100 as shown in Figure 1

[0033] The post-processing module 220 is configured to post-process the raw image based on the trained network to obtain a processed optimized image.

[0034] Specifically, as shown in Figure 3 The post-processing module 220 is configured to post-process the raw image based on the trained network to obtain a processed optimized image. Figure 1 The post-processing module 220 is configured to post-process the raw image based on the trained network to obtain a processed optimized image.

[0035] In some embodiments, the medical imaging system 200 further comprises a training module (not shown in the figure) configured to train the network 300 based on a sample raw image set (known input) and a target image set (expected output). The network is trained, tested and optimized through the preparation of training data, the selection and construction of network model, and the training of network. In some embodiments, the training data set is stored in the storage device 122 of the X-ray imaging system 100, and the training module can train and / or update the network by obtaining the training data set.

[0036] In some embodiments, the training module is further configured to implement the following functions.

[0037] First, a plurality of medical images acquired by the X-ray imaging system are obtained as a sample raw image set. In some embodiments, the sample raw image is obtained after pre-processing the medical image reconstructed by the X-ray imaging system, wherein the pre-processing includes normalization processing, etc. In other embodiments, the medical image without normalization processing can also be used as a sample image set, and the sample image set is input into the network, and then the sample image is normalized based on the normalization layer in the network. In other embodiments, the sample raw image is obtained after being reconstructed and normalized by different types of X-ray imaging systems. ​

[0038] Second: Obtain the post-processed optimized image corresponding to each image in the original sample image set, as the target image set. In some embodiments, the post-processing includes one or more of the following: noise reduction, detail enhancement, and contrast adjustment. In some embodiments, the optimized image is obtained based on conventional post-processing methods, such as by manually adjusting one or more post-processing parameters and making an empirical judgment. However, those skilled in the art will understand that the acquisition of the optimized image is not limited to the above representation and any suitable method can be used.

[0039] Third: Using the original set of sample images as input and the target set of images as output, train the network to obtain the trained network.

[0040] In other embodiments, the training module is further configured to perform the following functions.

[0041] First: Obtain multiple medical images acquired by the X-ray imaging system as the original image set of the sample.

[0042] Second: Based on the preferences of multiple users, each image in the original sample image set is post-processed to obtain multiple target image sets corresponding to each user's preferences. In some embodiments, since users (e.g., doctors) have different preferences for optimized images—some users prefer images with high contrast, while others prefer images with prominent details—different post-processing can be performed on the original sample images according to user preferences to obtain multiple target image sets. For example, the first target image set corresponds to target images with high contrast, and the second target image set corresponds to target images with prominent details.

[0043] Third: Using the original sample image set as input and each of the multiple target image sets as output, train the network to obtain one or more networks corresponding to each user's preferences. For example, a first network is trained using the original sample image set and the first target image set. This first network corresponds to users who prefer high contrast, and the optimized image obtained by post-processing using the first network also has relatively high contrast. Similarly, a second network is trained using the original sample image set and the second target image set. This second network corresponds to users who prefer prominent details, and the optimized image obtained by post-processing using the second network also has prominent details.

[0044] In some embodiments, the one or more networks are trained based on UNet or other well-known models. Figure 4 A schematic diagram of a network 300 according to some embodiments of the present invention is shown. Figure 4 As shown, network 300 includes an input layer 310, a processing layer (or hidden layer) 330, and an output layer 320.

[0045] In some embodiments, the processing layers 330 include encoding parts (331-338) and decoding parts (341-348), specifically, the processing layers 330 include a first encoding layer 331 and a first decoding layer 341 skip-connected to the first encoding layer 331, a second encoding layer 332 and a second decoding layer 342 skip-connected to the second encoding layer 332, a third encoding layer 333 and a third decoding layer 343 skip-connected to the third encoding layer 333, a fourth encoding layer 334 and a fourth decoding layer 344 skip-connected to the fourth encoding layer 334, a fifth encoding layer 335 and a fifth decoding layer 345 skip-connected to the fifth encoding layer 335, a sixth encoding layer 336 and a sixth decoding layer 346 skip-connected to the sixth encoding layer 336, a seventh encoding layer 337 and a seventh decoding layer 347 skip-connected to the seventh encoding layer 337, an eighth encoding layer 338 and an eighth decoding layer 348 skip-connected to the eighth encoding layer 338, and the first to eighth encoding layers 331-338 and the eighth to first decoding layers 348-341 are also sequentially connected. By setting the skip connection, the output of a certain encoding layer can be directly transmitted to the decoding layer skip-connected thereto, rather than the adjacent layer (i.e., the sequentially connected layer), so that feature images of different levels (e.g., different resolutions) can be output.

[0046] Although Figure 4 Although the processing layers are shown to include 8 encoding layers and 8 decoding layers, it should be understood by those skilled in the art that the processing layers can include any number of encoding layers (the number of decoding layers corresponds to the number of encoding layers), and is not limited to 8, and the number of encoding layers (or decoding layers) can be adaptively adjusted according to the number of pixels in the sample image. In addition, it should be understood by those skilled in the art that Figure 4 The structure of the network is only shown by way of example, and any other suitable configuration can be used.

[0047] In some embodiments, each of the first to eighth encoding layers 331-338 and the first to eighth decoding layers 341-348 includes at least one convolutional layer and at least one pooling layer, each convolutional layer and each pooling layer includes a plurality of neurons, and the number of neurons in each layer can be the same or can be set differently as needed. Based on the sample original image (known input) and the target image (desired output), by setting the number of processing layers in the network and the number of neurons in each processing layer, and estimating (or adjusting or calibrating) the weights and / or biases of the network, the mathematical relationship between the known input and the desired output is identified and / or the mathematical relationship between the input and the output of each layer is identified and characterized.

[0048] Specifically, when the number of neurons in one layer is n, and the corresponding values in the n neurons are X1, X2, … Xn, the mathematical relationship between the input and the output of the layer can be represented as follows: nThe number of neurons of the next layer connected with the one layer is m, and the corresponding values in the m neurons are Y1, Y2, … Ym m The two adjacent layers can be represented as:

[0049]

[0050] wherein X i represents the value corresponding to the i-th neuron of the previous layer, Y j represents the value corresponding to the j-th neuron of the next layer, W ji represents the weight, and B j represents the bias. In some embodiments, the function f is a rectified linear unit (ReLU).

[0051] Therefore, by adjusting the weight W ji and / or the bias B j , the mathematical relationship between the input and the output of each layer can be identified, so that the loss function converges, to train the network.

[0052] When the network is created or trained, as long as the original image to be post-processed is input into the network, the optimized image after post-processing can be obtained.

[0053] In one embodiment, although the configuration of the network 300 will be guided by the dimension of the prior knowledge, input, output, etc. of the estimation problem, and depends on or is specifically implemented according to the input data to achieve the best approximation of the required output data. In various alternative embodiments, certain aspects and / or features of the data, imaging geometry, reconstruction algorithm, etc. can be utilized to give explicit meaning to certain data representations in the network 300, which can help to speed up the training. Because this creates an opportunity to train (or pre-train) or define certain layers separately in the network 300.

[0054] In some embodiments, the above-mentioned sample image set (including the sample original image set and the corresponding target image set) can be updated to update or optimize the trained network based on a new sample image set, wherein the target image set can be obtained based on a traditional post-processing method (i.e., adjusting one or more post-processing parameters and obtaining based on experience), or can be obtained based on the medical imaging method according to some embodiments of the present application (i.e., through the above-mentioned trained network).

[0055] In other embodiments, the trained network described above is trained based on a training module on an external carrier (e.g., a device other than the medical imaging system). In some embodiments, the training module trains the network based on a set of sample raw images and a set of target images after post-processing of the corresponding sample raw images, based on a UNet or other known model. In some embodiments, the training system can include a first module for storing a training data set (including a set of sample raw images and a set of target images), a second module for training and / or updating based on the model, and a network for connecting the first module and the second module. In some embodiments, the first module includes a first processing unit and a first storage unit, where the first storage unit is used to store the training data set, and the first processing unit is used to receive relevant instructions (e.g., to obtain the training data set) and send the training data set according to the instructions. In addition, the second module includes a second processing unit and a second storage unit, where the second storage unit is used to store the training model, and the second processing unit is used to receive relevant instructions, train and / or update the network, etc. In other embodiments, the training data set can also be stored in the second storage unit of the second module, and the training system can not include the first module. In some embodiments, the network can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0056] Once the data (e.g., the trained network) is generated and / or configured, the data can be copied and / or loaded into the x-ray imaging system 100, which can be done in different ways. For example, the model can be loaded through a direct connection or link between the x-ray imaging system 100 and the control subsystem 108. In this regard, communication between different elements can be done using available wired and / or wireless connections and / or according to any suitable communication (and / or network) standard or protocol. Alternatively or additionally, the data can be indirectly loaded into the x-ray imaging system 100. For example, the data can be stored into a suitable machine-readable medium (e.g., a flash card, etc.), which is then used to load the data into the x-ray imaging system 100 (on-site, such as by a user or authorized personnel of the system), or the data can be downloaded into an electronic device (e.g., a laptop, etc.) that is capable of local communication, which is then used (e.g., by a user or authorized personnel of the system) on-site to upload the data into the x-ray imaging system 100 via a direct connection (e.g., a USB connector, etc.).

[0057] As discussed herein, deep learning techniques (also referred to as deep machine learning, hierarchical learning, or deep structured learning, among others) employ artificial neural networks for learning. Deep learning methods are characterized by the use of one or more network architectures to extract or model data of interest. Deep learning methods can use one or more processing layers (e.g., input layers, output layers, convolutional layers, normalization layers, sampling layers, and so on, which can have different numbers and functionalities depending on the different deep network models) to accomplish, where the configuration and number of layers allow the deep network to handle complex information extraction and modeling tasks. The specific parameters (also referred to as "weights" or "biases") of the network are typically estimated through a so-called learning process (or training process). The parameters that are learned or trained typically result in (or output) a network that corresponds to different levels of layers, and thus the extraction or modeling of different aspects of the initial data or the output of a previous layer can typically represent the hierarchy or cascade of layers. In the context of image processing or reconstruction, this can be characterized as different layers with respect to different feature levels in the data. Thus, the processing can be hierarchical, i.e., earlier or higher level layers can correspond to extracting "simple" features from the input data, followed by layers that combine these simple features into features that exhibit higher complexity. In practice, each layer (or more specifically, each "neuron" in each layer) can employ one or more linear and / or non-linear transformations (so-called activation functions) to process the input data into an output data representation. The number of "neurons" can be constant across multiple layers, or can vary from layer to layer.

[0058] As discussed herein, as part of the initial training of a deep learning process to solve a particular problem, a training data set includes known input values (e.g., sample images or pixel matrices of images undergoing coordinate transformations) and desired (target) output values of the final output of the deep learning process (e.g., images or recognition judgment results). In this way, the deep learning algorithm can process this training data set (in a supervised or guided manner or in an unsupervised or non-guided manner) until the mathematical relationship between the known input and the desired output is identified and / or the mathematical relationship between the input and output of each layer is identified and characterized. The learning process typically utilizes (part of) the input data and creates a network output for this input data, then compares the created network output with the desired output of this data set, then uses the difference between the created and desired outputs to iteratively update the parameters (weights and / or biases) of the network. The parameters of the network can typically be updated using a Stochastic gradient descent (SGD) method, however, those skilled in the art will appreciate that other methods known in the art can also be used to update the network parameters. Similarly, a separate validation data set can be employed to validate the trained network, where both the known input and the desired output are known, the network output can be obtained by providing the known input to the trained network, then the network output is compared with the (known) desired output to validate the previous training and / or prevent over-training.

[0059] Please continue to refer to Figure 2 The display module 230 is configured to display the original image and the optimized image. Figure 5 An image display interface 231 of the display module 230 in the medical imaging system 200 is shown, as Figure 5 The image display interface 231 is configured to display the optimized image 247, similarly, Figure 3 The original image 246 can also be displayed in the image display interface 231, as

[0060] The display module 230 comprises a first operation unit configured to generate a first instruction to input the original image 246 into the network 300, for example, the first operation unit is arranged in the image display interface 231, in particular, the first operation unit is arranged in the menu bar 241 of the image display interface 231, more particularly, the first operation unit is a post-processing control option 242 in the menu bar 241. In some embodiments, a user can trigger the first operation unit to obtain the first instruction, in particular, the user can select the option 242 by a mouse or an external input device, or select the option 242 by touch, etc.

[0061] The display module 230 comprises a second operation unit for generating a second instruction to select one of the one or more networks to post-process the original image 246. In some embodiments, the second operation unit is provided in the menu bar 241 of the image display interface 231, and more specifically, the second operation unit is a network selection option (not shown in the figure) in the menu bar 241, and the user can trigger the second operation unit to obtain the second instruction, for example, by selecting the option by a mouse or an external input device, or by touching the option, etc. In other embodiments, the second operation unit is provided in a user login interface, and when the user logs in his / her account, the system can automatically identify or select the network corresponding to the user, and optionally, a network selection interface can be popped up after the user logs in his / her account, so that the user can manually select the desired network (i.e. the network corresponding to the desired optimized image style).

[0062] Although the above embodiments illustrate that the second operation unit is provided in the display module, it is understood by those skilled in the art that the second operation unit can also be provided in any other control device or system, for example, a card reader provided in a control room can be used to identify the ID of the user to automatically select the network corresponding to the user.

[0063] The display module 230 comprises a third operation unit for generating a third instruction to input the original image and the optimized image into the network to optimize the network, for example, the third operation unit is provided in the menu bar 241 of the image display interface 231, and more specifically, the third operation unit is a network optimization option (not shown in the figure) in the menu bar 241. In some embodiments, the user can trigger the third operation unit to obtain the third instruction, for example, by selecting the option by a mouse or an external input device, or by touching the option, etc.

[0064] In some embodiments, the menu bar 241 in the image display interface 231 can further comprise operation units for zooming in or out, moving, etc.

[0065] Although the above embodiments illustrate that the second operation unit is provided in the display module, it is understood by those skilled in the art that the second operation unit can also be provided in any other control device or system, for example, a card reader provided in a control room can be used to identify the ID of the user to automatically select the network corresponding to the user. Figure 5 Although only the position of the menu bar 241 on the left side of the image display interface is shown, it is understood by those skilled in the art that the menu bar 241 can be displayed in any manner and / or position and / or style, for example, a floating window can be provided, and all menu options can be expanded by selection, or the menu bar can be provided in a row below the display interface, etc. Furthermore, although the above embodiments illustrate that the menu bar 241 is provided in the image display interface 231, it is understood by those skilled in the art that the menu bar 241 can also be provided in any other interface, for example, the menu bar 241 can be provided in the image display interface 231 and the user login interface, or the menu bar 241 can be provided in the image display interface 231 and the image selection interface, etc. Figure 5The image display interface 231 in FIG. 2 only shows the optimized image 247 and the menu bar 241, however, those skilled in the art should understand that the image display interface 231 can further display other types of content, for example, the left side of the optimized image display part can also include the address of the image, the switching control of multiple images (original image and optimized image), and other content. In some embodiments, the original image and the optimized image can also be displayed in the same graphical user interface.

[0066] In some embodiments, the display module 230 is the display device 124 in the X-ray imaging system (as shown in FIG. 2), and the display module 230 includes a control display arranged in the scanning room (or device room) and / or a display arranged in the control room. Figure 1

[0067] Figure 6 A flowchart of a medical imaging method 400 according to some embodiments of the present application is shown. As shown in FIG. 4, the medical imaging method 400 in some embodiments of the present application includes steps 410 and 420. Figure 6

[0068] In step 410, an original image acquired by an X-ray imaging system is obtained.

[0069] In some embodiments, the medical image can be obtained by detecting the attenuation of X-rays passing through the detected object by a detector and performing data processing and reconstruction, and the original image can be displayed in the display device or display module in the X-ray imaging system.

[0070] In step 420, the original image is post-processed based on the trained network to obtain an optimized image after processing. Specifically, by inputting the original image into the trained network, the optimized image after post-processing can be obtained (or output), and the optimized image can also be displayed in the display device in the X-ray imaging system.

[0071] In some embodiments, the network is completed by preparing training data, selecting and constructing a network model, training, testing, and optimizing the network. In some embodiments, the one or more networks are trained based on UNet or other known models, wherein the network is trained based on a sample original image set and a target image set. Specifically, the training of the network includes the following steps one to three.

[0072] ​​Step one: Obtain a plurality of medical images acquired by an X-ray imaging system as a sample original image set. In some embodiments, the sample original images are obtained after pre-processing of the medical images reconstructed by the X-ray imaging system, wherein the pre-processing includes normalization processing, etc. In some embodiments, the sample original images are obtained after normalization processing of medical images reconstructed by X-ray imaging systems of different models.

[0073] Step two: Obtain an optimized image corresponding to each image in the sample original image set after post-processing, as a target image set. In some embodiments, the post-processing includes one or more of noise reduction, detail enhancement, and contrast adjustment. In some embodiments, the optimized image is obtained based on a conventional post-processing method, for example, by manually adjusting one or more post-processing parameters and obtaining a result based on experience. However, those skilled in the art can understand that the above-mentioned obtaining of the optimized image is not limited to the above-mentioned manner, and any suitable manner can be used.

[0074] Step three: Train the network with the sample original image set as input and the target image set as output, to obtain the trained network.

[0075] In other embodiments, the training of the network includes the following steps one to three.

[0076] Step one: Obtain a plurality of medical images acquired by an X-ray imaging system as a sample original image set.

[0077] Step two: Based on the preferences of a plurality of users, respectively post-process each image in the sample original image set to obtain a plurality of target image sets corresponding to the preferences of each user. In some embodiments, since the preferences of users (e.g., doctors) for optimized images are different, i.e., some users prefer images with high contrast, and some users prefer images with prominent details, different post-processing can be performed on the sample original images according to the preferences of the users to obtain a plurality of target image sets, for example, a first target image set corresponding to target images with high contrast, and a second target image set corresponding to target images with prominent details.

[0078] Step three: training the network with the sample original image set as input and each of the plurality of target image sets as output to obtain one or more networks corresponding to the preference of each user. For example, a first network is trained by the sample original image set and the first target image set, the first network corresponding to a user who prefers high contrast, and the optimized image obtained by post-processing through the first network is also high in contrast. Similarly, a second network is trained by the sample original image set and the second target image set, the second network corresponding to a user who prefers details, and the optimized image obtained by post-processing through the second network is also detailed.

[0079] In some embodiments, before step 420 (post-processing the original image based on the trained network to obtain a processed optimized image), it further includes inputting the original image into the network based on a first instruction of the user, specifically, the first instruction is triggered based on a first operation unit in the image display interface.

[0080] In some embodiments, the trained network includes one or more networks, and the medical imaging method 400 further includes selecting one of the one or more networks based on a second instruction of the user to post-process the original image. In some embodiments, the second instruction of the user is triggered based on the identity recognition of the user. In other embodiments, the second instruction of the user is triggered based on a second operation unit in the image display interface.

[0081] In some embodiments, the medical imaging method 400 further includes inputting the original image and the optimized image into the network based on a third instruction of the user to optimize the network, specifically, the third instruction is triggered based on a third operation unit in the image display interface.

[0082] The medical imaging method based on deep learning proposed in the present application can more accurately and timely post-process medical original images, no longer relying on manual adjustment by on-site engineers and experience judgment, not only reducing the difficulty and time of parameter adjustment in the post-processing process, but also improving the image quality of the optimized image obtained after post-processing. In addition, different style sample image sets can be set according to the preference (or style) of the user to train different style learning networks, and the corresponding learning network can be selected by the user to obtain the optimized image style corresponding to the user's preference.

[0083] The present application can also provide a non-transitory computer readable storage medium for storing a set of instructions and / or a computer program, which, when executed by a computer, cause the computer to perform the above-mentioned medical imaging method. The computer executing the set of instructions and / or the computer program can be a computer of the X-ray imaging system, or can be another device / module of the X-ray imaging system. In an embodiment, the set of instructions and / or the computer program can be programmed in a processor / controller of the computer.

[0084] In particular, the set of instructions and / or the computer program, when executed by the computer, cause the computer to:

[0085] obtain a raw image acquired by the X-ray imaging system; and

[0086] post-process the raw image based on the trained network to obtain a processed optimized image.

[0087] The instructions as described above can be executed as one instruction, or any of the instructions can be split into multiple instructions to be executed. Furthermore, the order of the instructions as described above is not limited.

[0088] In some embodiments, the method further comprises inputting the raw image into the network based on a first instruction of a user before post-processing the raw image.

[0089] In some embodiments, the trained network comprises one or more networks, and the method further comprises selecting one of the one or more networks based on a second instruction of a user to post-process the raw image.

[0090] In some embodiments, the method further comprises inputting the raw image and the optimized image into the network based on a third instruction of a user to optimize the network.

[0091] As used herein, the term "computer" can include any processor-based or microprocessor-based system including systems using microcontrollers, reduced instruction set computers (RISC), application specific integrated circuits (ASICs), logic circuits, and any other circuit or processor capable of executing the functions described herein. The above examples are exemplary only, and are thus not intended to limit in any way the definition and / or meaning of the term "computer".

[0092] The instruction set may include various commands that instruct a computer or processor, acting as a processor, to perform specific operations, such as methods and processes according to various embodiments. The instruction set may take the form of a software program, which may be part of one or more tangible, non-transitory computer-readable media. The software may take various forms, such as system software or application software. Furthermore, the software may take the form of a collection of independent programs or modules, a program module within a larger program, or part of a program module. The software may also include modular programming in the form of object-oriented programming. Input data processing by the processor may be in response to operator commands, previous processing results, or requests made by another processor.

[0093] Some exemplary embodiments have been described above; however, it should be understood that various modifications can be made. For example, suitable results may be achieved if the described techniques are performed in a different order and / or if components in the described system, architecture, device, or circuit are combined in a different manner and / or replaced or supplemented by other components or their equivalents. Accordingly, other embodiments also fall within the scope of the claims.

Claims

1. A medical imaging method comprising: obtaining an original image acquired by an X-ray imaging system; and based on an identified identity of a user, selecting one of one or more networks to post-process the original image to obtain a processed optimized image; wherein the one or more networks are trained based on a set of sample original images and a set of target images, the training comprising: obtaining a plurality of original images acquired by an X-ray imaging system to obtain the set of sample original images; post-processing the original images respectively based on one or more user preferences to obtain a plurality of sets of target images corresponding to each user preference; and training a neural network with the set of sample original images as input and each of the plurality of sets of target images as output to obtain the one or more networks corresponding to each user preference.

2. The method of claim 1, wherein, further comprising inputting the original image into the network based on a first instruction of the user before post-processing the original image.

3. The method of claim 2, wherein, the first instruction is triggered based on a first operation unit in an image display interface.

4. The method of claim 1, wherein, the trained network comprises one or more networks, and the method further comprises: selecting one of the one or more networks based on a second instruction of the user to post-process the original image.

5. The method of claim 4, wherein, the second instruction of the user is triggered based on an identity recognition of the user.

6. The method of claim 1, wherein, further comprising: further inputting the original image and the optimized image into the network to optimize the network based on a third instruction of the user.

7. The method of claim 6, wherein, the third instruction is triggered based on a third operation unit in the image display interface. 8.A non-transitory computer-readable storage medium for storing a computer program which, when executed by a computer, causes the computer to perform the medical imaging method of any one of claims 1-7. 9.A medical imaging system comprising: a control module for obtaining an original image acquired by an X-ray imaging system; a post-processing module for post-processing the original image based on an identified identity of a user to obtain a processed optimized image, by selecting one of one or more networks; and a display module for displaying the original image and the optimized image; a training module for being trained based on a set of sample original images and a set of target images, wherein the training module is further configured to: obtain a plurality of original images acquired by an X-ray imaging system to obtain the set of sample original images; post-process the original images respectively based on one or more user preferences to obtain a plurality of sets of target images corresponding to each user preference; and train a neural network with the set of sample original images as input and each of the plurality of sets of target images as output to obtain the one or more networks corresponding to each user preference. the display module comprises a first operation unit for generating a first instruction to input the original image into the network.

10. The system of claim 9, wherein, the display module comprises a third operation unit for generating a third instruction to input the original image and the optimized image into the network to optimize the network.

11. The system of claim 9, wherein, ​

Images