A medical image processing method and system

Abstract

The present specification generally relates to the field of image processing, and more particularly, to a method and system for medical image processing, the method comprising: obtaining a plurality of scanning data of a scanning object; processing the plurality of scanning data based on a first model to obtain a reference frame; processing the reference frame and the plurality of scanning data to obtain correction information of the plurality of scanning data relative to the reference frame.

Description

TECHNICAL FIELD

[0001] The present specification generally relates to the field of image processing, and particularly relates to a method and system for medical image processing. BACKGROUND

[0002] Commonly used technical means of medical image imaging include, but are not limited to, magnetic resonance imaging (MRI), computed tomography (CT), and positron emission tomography (PET), etc. When multiple medical images are acquired by an imaging device, due to the change in the position of the scanning object, the scanning object may be offset in the medical images.

[0003] Therefore, it is desirable to provide a medical image processing method that can perform motion calibration between different frames of multiple medical images, and has high calibration efficiency and accuracy. SUMMARY

[0004] According to a first aspect of the present specification, a medical image processing method is provided, which includes: acquiring multiple frames of scanning data of a scanning object; processing the multiple frames of scanning data based on a first model to obtain a reference frame; processing the reference frame and the multiple frames of scanning data based on a second model to obtain correction information of the multiple frames of scanning data relative to the reference frame.

[0005] According to another aspect of the present specification, a medical image processing system is provided, which includes: a scanning data acquisition module configured to acquire multiple frames of scanning data of a scanning object; a reference frame determination module configured to process the multiple frames of scanning data based on a first model to obtain a reference frame; and a correction information acquisition module configured to process the reference frame and the multiple frames of scanning data based on a second model to obtain correction information of the multiple frames of scanning data relative to the reference frame.

[0006] According to another aspect of the present specification, a medical image processing device is provided, which includes at least one storage medium and at least one processor; the at least one storage medium is configured to store computer instructions; and the at least one processor is configured to execute the computer instructions to implement the medical image processing method described above.

[0007] Some of the additional features of the present specification can be explained in the following description. Some of the additional features of the present specification will become apparent to those skilled in the art from the following description and the accompanying drawings, or can be learned by practice of the production or operation of the embodiments. The features of the present specification can be realized and achieved by practicing or using various aspects of the embodiments described below. BRIEF DESCRIPTION OF DRAWINGS

[0008] The present specification will be further described by way of exemplary embodiments. These exemplary embodiments will be described in detail with reference to the accompanying drawings. These embodiments are non-limiting exemplary embodiments in which the same numbers are used to represent similar structures in each of the drawings in which:

[0009] Figure 1 is an exemplary application scenario diagram of a medical image processing system according to some embodiments of the present specification;

[0010] Figure 2 is an exemplary flowchart of a medical image processing method according to some embodiments of the present specification;

[0011] Figure 3 is an exemplary flowchart of obtaining target scan data according to some embodiments of the present specification;

[0012] Figure 4 is a schematic diagram of training and functions of a first model according to some embodiments of the present specification;

[0013] Figure 5 is a schematic diagram of training and functions of a second model according to some embodiments of the present specification;

[0014] Figure 6 is an exemplary schematic diagram of a medical image processing method according to some embodiments of the present specification;

[0015] Figure 7 is an exemplary module diagram of a medical processing system according to some embodiments of the present specification. DETAILED DESCRIPTION

[0016] In order to more clearly illustrate the technical solutions of the embodiments of the present specification, the drawings required to be used in the embodiment description will be briefly introduced below. However, it should be understood by those skilled in the art that the present specification can be implemented without these details. In other cases, in order to avoid unnecessary obscurity of some aspects of the present specification, well-known methods, procedures, systems, components and / or circuits have been described in more detail. It is obvious for those skilled in the art that various changes can be made to the disclosed embodiments, and the general principles defined in the present specification can be applied to other embodiments and application scenarios without departing from the principles and scope of the present specification. Therefore, the present specification is not limited to the shown embodiments, but conforms to the broadest range consistent with the scope of the patent application.

[0017] The terminology used in the present specification is for the purpose of describing particular example embodiments only and is not intended to limit the scope of the present specification. As used in the present specification, the singular forms "a," "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0018] It should be understood that the terms "system," "unit" and / or "module" used herein are a method for distinguishing different levels of different components, elements, parts, or assemblies. However, if these terms achieve the same purpose, they can be replaced by another term.

[0019] Generally, the terms "module," "unit," or "block" used herein refer to logic embodied in hardware or firmware, or a collection of software instructions. The modules, units, or blocks described herein can be implemented as software and / or hardware and can be stored in any type of non-transitory computer-readable medium or another storage device. In some embodiments, the software modules / units / blocks can be compiled and linked into executable programs. It should be understood that the software modules can be invoked from other modules / units / blocks or from themselves, and / or can be invoked in response to detected events or interrupts. Software modules configured for execution on a computing device (e.g., processing device in server 130 in Figure 1

[0020] These and other features, characteristics, and advantages of the present specification can become more apparent from the following detailed description considered in conjunction with the accompanying drawings, in which:

[0021] ​Systems and components for medical imaging and / or medical treatment are provided herein. In some embodiments, a medical system can include an imaging system. The imaging system can include a multi-modality imaging system. The multi-modality imaging system can include, for example, a computed tomography-magnetic resonance imaging (MRI-CT) system, a positron emission tomography-magnetic resonance imaging (PET-MRI) system, a single photon emission computed tomography-magnetic resonance imaging (SPECT-MRI) system, and / or a digital subtraction angiography-magnetic resonance imaging (DSA-MRI) system, etc. The processing device can include an accelerator of particle species, including, for example, photons, electrons, protons, and / or heavy ions, etc. The imaging device can include an MRI scanner, a CT scanner (e.g., a cone beam computed tomography (CBCT) scanner), a digital radiography (DR) scanner, and / or an electronic portal imaging device (EPID), etc.

[0022] In some medical image processing scenarios, a plurality of medical images of a scanning object are acquired by a medical image acquisition device, but during the imaging process, the patient can move due to a long scanning time, or the scanning view angle changes due to other reasons, which can cause the organs or lesions at different time points to be offset, and ultimately cause the imaging results to be abnormal. Therefore, it is necessary to calibrate the motion of the medical images obtained by scanning to reduce or eliminate the influence of the lesion offset.

[0023] In some embodiments, the motion calibration of the medical images can be implemented by using non-rigid registration. However, due to the long time consumption of non-rigid registration, it is not suitable for scenarios that require registration of a large number of images, and it is also not suitable for situations that require high registration speed.

[0024] In some embodiments, some medical images can have low resolution, and therefore, when registering such low-resolution images, the difficulty of registration increases and the efficiency is low due to the insufficiently obvious anatomical structure.

[0025] Therefore, in some embodiments, a machine learning model is used to implement the calibration process of the medical images to improve the calibration speed and accuracy. In some embodiments, other clearer images can also be used to assist in the calibration of the medical images to break through the resolution limit of the original medical images and further improve the calibration accuracy.

[0026] Figure 1 is an example application scenario diagram of a medical image processing system according to some embodiments of the present specification. As shown in Figure 1 The application scenario 100 of the medical image processing system can include a first medical image acquisition device 110, a second medical image acquisition device 120, a server 130, and / or a network 140, etc.

[0027] The first medical image acquisition device 110 can refer to an imaging device that acquires medical images. For example, a positron emission tomography (PET) scanner or a single photon emission computed tomography (SPECT-CT) scanning device. In some embodiments, the first medical image acquisition device can acquire scan data. For more information about scan data, see Figure 2 and related descriptions thereof, which are not repeated here.

[0028] The second medical image acquisition device 120 can refer to another imaging device that acquires medical images. For example, the imaging device can include an MRI scanner and / or a CT scanner, etc. In some embodiments, the second medical image acquisition device can acquire a contrast image. For more information about the contrast image, see Figure 2 and related descriptions thereof, which are not repeated here.

[0029] In some embodiments, the first medical image acquisition device 110 and the second medical image acquisition device 120 can be different. For example, the first medical image acquisition device 110 can be a PET scanner, and the second medical image acquisition device 120 can be a CT / MR scanner. In some embodiments, the first medical image acquisition device 110 and the second medical image acquisition device 120 can be the same, such as both being PET-CT scanning devices or SPECT-CT scanning devices.

[0030] In some embodiments, the first medical image acquisition device 110 and the second medical image acquisition device 120 scan the same scan object to acquire different types of medical images of the scan object at corresponding time points. For example, a patient is scanned to obtain a PET image and a corresponding CT image of the patient. For more information about the scan object, see Figure 2 and related descriptions thereof, which are not repeated here.

[0031] The server 130 refers to a system with computing capability. The server 130 can include a processing device to correct medical images. In some embodiments, the server can acquire scan data obtained by the first medical image acquisition device 110 and correct the scan data. For example, the server 130 acquires a contrast image obtained by the second medical image acquisition device 120 and corrects scan data at different time points based on the contrast image. For more information about the correction, see Figure 2and the related description thereof will not be repeated here. In some embodiments, one or more machine learning models can be included in the server 130, through which the correction of the scan data is implemented. For example, a first model and a second model can be included in the server 130, and the correction of the scan data is implemented based on the first model and the second model. For specific content of the first model and the second model, see Figure 2 , Figure 4 and Figure 5 and the related description thereof will not be repeated here.

[0032] The network 140 can connect the components of the system and / or connect the system with external resource parts. The network 140 enables communication between the components and with other parts outside the system. For example, the first medical image acquisition device 110 and the second medical image acquisition device 120 transmit the medical images obtained thereby to the server 130 for processing through the network 140. For another example, the server 130 transmits the corrected medical images to a doctor's computer or for printing through the network 140.

[0033] Figure 2 is an exemplary flowchart of a medical image processing method according to some embodiments of the present specification. As shown in Figure 2 , the flow 200 can include one or more of the following steps. In some embodiments, one or more steps in the flow 200 can be performed by the server 130 in Figure 1 .

[0034] Step 210, obtaining a plurality of frames of scan data of a scan object. In some embodiments, step 210 can be performed by the scan data obtaining module 710.

[0035] The scan object can be one or more organs, parts and / or the whole body. For example, the scan object can include organs of the human body (e.g., the lungs, heart and / or liver of a patient, etc.) and / or parts of the human body (e.g., the head, chest and / or abdomen of a patient, etc.), etc.

[0036] The scan data can be medical images related to the scan object. For example, the scan data can include a plurality of scan pictures, such as positron emission tomography (PET) images. In some embodiments, the scan data can be obtained by Figure 1The first medical image acquisition device 110 in the medical image acquisition system 100 acquires scan data. For example, dynamic PET images or static PET images are acquired by a PET scanner. In some embodiments, the multi-frame scan data can be derived based on multi-frame PET images. For example, the PET images are divided according to their corresponding time information to obtain the multi-frame PET images. In some embodiments, the multi-frame scan data can be acquired by multi-bed scan or multi-scan. For example, the scan data can include medical images obtained by a patient performing PET scan for the first time and medical images obtained by the patient performing PET scan for the second time.

[0037] At step 220, the multi-frame scan data is processed based on the first model to obtain a reference frame. In some embodiments, step 220 can be performed by the reference frame determination module 720.

[0038] The first model can be a machine learning model used to obtain the reference frame. For more information about the first model, please refer to Figure 4 and related descriptions thereof.

[0039] The reference frame can refer to a medical image used as a reference for correction or registration. For example, the reference frame can be selected from the scan data. For another example, the reference frame can be selected from the processed scan data.

[0040] In some embodiments, a plurality of similarities between each of the scan data and a specific image can be determined by the first model, and the reference frame can be determined based on the plurality of similarities. For another example, the first model specifies the scan data with the highest signal-to-noise ratio in the scan data as the reference frame.

[0041] In some embodiments, the first model can determine the reference frame based on a contrast image. Wherein, determining the reference frame based on the contrast image includes:

[0042] Acquiring a contrast image of the scan object.

[0043] The contrast image can be another medical image of the scan object. In some embodiments, the contrast image can be acquired by Figure 1 the second medical image acquisition device 120 in the medical image acquisition system 100.

[0044] In some embodiments, the contrast image and the scan data can be of different types. For example, the scan data is a PET image, and the contrast image is a CT image. In some embodiments, the resolution of the contrast image can be higher than that of the scan data. For example, the anatomical structure or organ outline of the scan object in the contrast image is more obvious than that in the scan data.

[0045] In some embodiments, the contrast image can be obtained before the scan data is acquired. In some embodiments, the scan region of the contrast image contains the scan region of the scan data. For example, the scan region of the contrast image is the same as the scan region of the scan data. In some embodiments, the contrast image can be acquired in various feasible ways, including but not limited to, acquiring a CT image of the scan object by a CT scanner.

[0046] The contrast image and the multi-frame scan data are input into the first model to obtain a reference frame.

[0047] In some embodiments, the first model can determine the reference frame based on the similarity between the contrast image and the multi-frame scan data. For example, the first model can identify the similarity of the positions (e.g., the positions of the liver top and / or the lung bottom, etc.) and / or shapes, contours (e.g., the shapes, contours of the liver and / or the heart, etc.) of the organs in the contrast image and the scan data by using mutual information, and take the scan data with the highest similarity as the reference frame.

[0048] Step 230, processing the reference frame and the multi-frame scan data to obtain correction information of the multi-frame scan data relative to the reference frame. In some embodiments, step 230 can be performed by the correction information acquisition module 730.

[0049] In some embodiments, the correction information can refer to information for correcting or registering the scan data to the reference frame. For example, the correction information can include registration information, movement information of pixel points, change information of the scan object (e.g., shape change of the liver, etc.), etc.

[0050] In some embodiments, the reference frame and the multi-frame scan data can be processed by manual or computer-aided methods to obtain the correction information of the multi-frame scan data relative to the reference frame.

[0051] In some embodiments, the reference frame and the multi-frame scan data can be processed based on the second model to obtain the correction information of the multi-frame scan data relative to the reference frame.

[0052] The second model can be a machine learning model for obtaining the correction information. For more information about the second model, see Figure 5 and related descriptions thereof.

[0053] It should be noted that in some embodiments, the first model and the second model can be one machine learning model with both functions.

[0054] In some embodiments, the second model can identify the scan data and the contrast image by image recognition technology to determine the correction information.

[0055] In some embodiments, the second model can determine the correction information based on the deformation field. Wherein, determining the correction information based on the deformation field can include:

[0056] The multi-frame scanning data and the reference frame are input into the second model to obtain a deformation field of each of the multi-frame scanning data relative to the reference frame.

[0057] The deformation field can include displacement vectors of the pixel points in the scanning data to the reference frame. In some embodiments, the deformation field can be determined based on the difference between each pixel point in the scanning data and the reference frame. For example, the second model can obtain the displacement and direction of the pixel points in the scanning data to the reference frame, and take the displacement and direction corresponding to all the pixel points as the deformation field.

[0058] The correction information is determined based on the deformation field. For example, the deformation field can be directly determined as the correction information. For another example, the deformation field is processed to obtain the correction information. For example, only the information corresponding to the pixel points with non-zero element values in the deformation field can be extracted as the correction information.

[0059] At step 240, the multi-frame scanning data is registered based on the correction information to obtain multi-frame target scanning data. In some embodiments, step 240 can be performed by the registration processing module 740.

[0060] The target scanning data can refer to the scanning data after correction or registration. In some embodiments, the target scanning data can be the scanning data registered to the reference frame based on the correction information. For example, the positions of the feature points in the target scanning data are registered one by one with the positions in the reference frame based on the information of the deformation field. In some embodiments, the target scanning data can be a single modality image (such as a PET image) or a multi-modality image (such as a PET-CT image or a PET-MR image).

[0061] In some embodiments, for more information about obtaining the target scanning data, see Figure 3 and the related description. It should be noted that in some embodiments, the registration processing can not be performed, and in this case, the flow 200 can not include step 240.

[0062] In some embodiments, the multi-frame target scanning data after registration processing can also be image reconstructed to obtain a static PET image or a dynamic PET image.

[0063] Some embodiments of the present specification obtain the reference frame and the correction information through a machine learning model, so that the selected reference frame is more efficient, and the accuracy of the correction is improved.

[0064] Figure 3 is an exemplary flowchart for obtaining target scanning data according to some embodiments of the present specification. As shown inFigure 3 The flow 300 can include one or more of the following steps. In some embodiments, one or more steps in the flow 300 can be performed by the registration processing module 740.

[0065] At step 310, the multi-frame scanning data is processed based on the correction information to obtain multi-frame registered scanning data.

[0066] In some embodiments, the correction information can include a deformation field, and the scanning data can be deformed based on the deformation field to obtain the registered scanning data. The correction information can also be in other forms, and the registered scanning data can be obtained in various feasible ways based on different correction information.

[0067] At step 320, multi-frame target scanning data is obtained based on the multi-frame registered scanning data and the multi-frame matching images.

[0068] The matching images can refer to medical images used for matching. In some embodiments, the matching images can be the same number as the registered scanning data. In some embodiments, the corrected or registered scanning data can be matched by the corresponding matching images to obtain the corresponding target scanning data. For example, the PET frame images at multiple time points can be registered with at least one CT image based on the corrected or registered multi-frame to obtain PET-CT images.

[0069] Some embodiments of the present specification correct or register the scanning data by clearer contrast images obtained by other medical image devices, break through the limitation of the original resolution of the scanning data, improve the accuracy of the correction or registration, improve the resolution of the target scanning data, and automate the correction or registration process.

[0070] Figure 4 is an exemplary schematic diagram of the training and functions of the first model according to some embodiments of the present specification.

[0071] As Figure 4 indicated, the contrast image 410 and the corresponding multiple scanning data 420 can be input into the first model 430, and the first model can output the reference frame 440. In other words, the first model 430 can select the reference frame 440 from the multiple scanning data 420. For more information about the contrast image, the scanning data, and the reference frame 440, see Figure 2 and Figure 3 and the related descriptions thereof, which will not be repeated here.

[0072] In some embodiments, the first model 430 can be obtained from one or more components of the medical image processing system or external sources through the network 140. For example, the first model can be pre-trained by a processing device and stored in the server 130. In some embodiments, the first model can be generated by a machine learning algorithm. The machine learning algorithm can include an artificial neural network algorithm, a deep learning algorithm, etc., which are not limited in the present specification.

[0073] For example, the mutual information of the whole scan object (e.g., shape, contour, etc.) and / or different parts of the scan object (e.g., the scan object is an abdomen, and the parts can include the heart, liver, and / or stomach, etc.) can be calculated by different layers in the neural network, respectively, and the scan data with the highest similarity to the contrast image can be determined by calculation.

[0074] In some embodiments, the first model can be trained by a supervised learning algorithm (e.g., a logistic regression algorithm, etc.). In some embodiments, the initial first model 450 can be supervised trained based on the first training samples 460 to obtain the first model 430; wherein the first training samples 460 include a plurality of first training sample sets 470, each of which can include a sample contrast image 470-1 and a plurality of first sample scan data 470-2, and the sample contrast image 470-1 and the first sample scan data 470-2 are used as training samples, and the image with the highest similarity to the contrast image 410 in the first sample scan data 470-2 is used as a label. It should be noted that in some embodiments, the first model can also be trained by a semi-supervised learning algorithm.

[0075] In some embodiments, the sample contrast image corresponds to the contrast image, and the first sample scan data corresponds to the scan data. In some embodiments, the sample contrast image and the contrast image can be images of the same type, for example, CT images. In some embodiments, the sample contrast image and the contrast image can also be images of the same scan object. For example, an abdominal image. In some embodiments, the first sample scan data can be images of the same type as the scan data, for example, PET frame images. In some embodiments, the first sample scan data and the scan data can be images of the same scan object. For example, an abdominal image. In some embodiments, the label can be labeled by various feasible ways, including but not limited to manual, automatic, semi-automatic labeling, etc. In some embodiments, the sample contrast image and the first sample scan data can be obtained based on historical visit information, wherein the label can be obtained by manual labeling by a doctor.

[0076] In some embodiments, the first training sample set can be input to an initial first model, the initial first model can output a predicted reference frame 440. The processing device can determine a value of a first loss function based on the label and the predicted value. For example, the first loss function can take the difference between the predicted reference frame 440 and the label for corresponding pixels. In some embodiments, the processing device can iteratively train the initial first model to minimize the first loss function using various algorithms. For example, adjusting the parameters of the first model by gradient descent. In some embodiments, the initial first model that the loss function converges or the number of iterations reaches a preset value can be designated as a trained first model.

[0077] Figure 5 is an exemplary schematic diagram of the training and function of the second model according to some embodiments of the present specification.

[0078] As Figure 5 indicated, the reference frame 510 and the corresponding scan data 520 can be input into the second model 530, the second model can output a deformation field 540 corresponding to the scan data 520, and in some embodiments, the processing device can also determine the correction information 550 based on the deformation field 540. For more information about the reference frame, the scan data, and the correction information 550, see Figure 2 and related descriptions, which will not be repeated here.

[0079] In some embodiments, the second model 530 can be obtained from one or more components of the medical image processing system or external sources through the network 140. For example, the second model can be pre-trained by the processing device and stored in the server 130. In some embodiments, the second model can be generated by a machine learning algorithm.

[0080] In some embodiments, the second model can be trained and generated using an unsupervised algorithm such as the K-Means algorithm. In some embodiments, the initial second model 560 can be unsupervised trained based on the second training sample 570 to obtain the second model 530; wherein the second training sample 570 includes a plurality of second training sample sets 580, each second training sample set includes a sample reference frame 580-1 and a plurality of second sample scan data 580-2.

[0081] In some embodiments, before training the initial second model, one or more parameters can be assigned to it, each parameter having one or more initial values. When training the initial second model, the values of the parameters in the model can be updated. In some embodiments, the initial second model can include multiple rounds of iterations, and the processing device can determine the iterated initial second model as the second model.

[0082] In some embodiments, the reference frame corresponds to a sample reference frame; and the scan data corresponds to second sample scan data. In some embodiments, the reference frame can be obtained by manually specifying a frame image in the second sample scan data, and the second sample scan data can be obtained from historical medical information of a hospital.

[0083] In some embodiments, the sample reference frame and the second sample scan data can be input into an initial second model, the model calculates the position and / or contour information of the sample reference frame and the second sample scan data, then adjusts the parameters of the initial second model based on the position and / or contour information, adjusts the second sample scan data using the adjusted initial second model, and takes the adjusted second sample scan data as the second sample scan data, repeats the above steps until the position and / or contour information between the sample reference frame and the second sample scan data are similar, and determines the last obtained initial second model as the second model.

[0084] In some embodiments, the second model is trained by using an unsupervised learning method, and the advantage of processing images by the method makes it unnecessary to obtain label data, i.e., the deformation field 540 between two images, thereby reducing the cost of obtaining label data.

[0085] Figure 6 is an exemplary schematic diagram of a medical image processing method according to some embodiments of the present specification.

[0086] As shown in Figure 6 , in some embodiments, the scan data in the flow 600 can be PET frame images, and the contrast image can be a CT / MR image. The medical image processing system 100 can obtain a plurality of CT / MR images 610 and PET frame images 630 by using an image acquisition device, and determine the contrast image from the CT / MR images. In some embodiments, the first model can extract one PET frame image as a reference frame 620 from the plurality of PET frame images based on the similarity between the PET frame images and the contrast image. In some embodiments, the second model can obtain a deformation field 640 corresponding to a PET frame image based on the PET frame image and the reference frame, respectively. In some embodiments, the PET frame image can be corrected or registered based on the deformation field 640 to obtain a corrected or registered PET frame image 660. In order to improve the resolution of the target scan data, in some embodiments, the contrast image can also be corrected or registered based on the deformation field 640 to obtain a matching image 650, and the corrected or registered PET frame image can be registered with the CT / MR image based on the matching image 650 to obtain the target scan data 670.

[0087] Figure 7 is an exemplary module diagram of a medical processing system according to some embodiments of the present specification. As shown in Figure 7 , the medical processing system 700 can include one or more of the following modules.

[0088] The scan data obtaining module 710 is configured to obtain a plurality of frames of scan data of a scan object. For more information about obtaining a plurality of frames of scan data of a scan object, see Figure 2 and related descriptions thereof, which will not be repeated here.

[0089] The reference frame determining module 720 is configured to process the plurality of frames of scan data based on a first model to obtain a reference frame. For more information about processing the plurality of frames of scan data based on a first model to obtain a reference frame, see Figure 2 and related descriptions thereof, which will not be repeated here.

[0090] The correction information obtaining module 730 is configured to process the reference frame and the plurality of frames of scan data based on a second model to obtain correction information of the plurality of frames of scan data relative to the reference frame. For more information about processing the reference frame and the plurality of frames of scan data based on a second model to obtain correction information of the plurality of frames of scan data relative to the reference frame, see Figure 2 and related descriptions thereof, which will not be repeated here.

[0091] The registration processing module 740 is configured to perform registration processing on the plurality of frames of scan data based on the correction information to obtain a plurality of frames of target scan data. For more information about performing registration processing on the plurality of frames of scan data based on the correction information to obtain a plurality of frames of target scan data, see Figure 2 and Figure 3 and related descriptions thereof, which will not be repeated here.

[0092] It should be noted that the above is provided for illustrative purposes only and is not intended to limit the scope of the present specification. Various changes and modifications can be made to the present specification by one of ordinary skill in the art based on the description of the present specification. However, these changes and modifications will not depart from the scope of the present specification.

[0093] Some embodiments of the present specification improve the efficiency and accuracy of registration by correcting or registering medical images through a machine learning model. Some embodiments of the present specification also improve the resolution of the registered medical images by registering medical images with obvious features using medical images with non-obvious features.

[0094] The above has described the basic concepts, and it is obvious to one of ordinary skill in the art after reading this application that the above invention disclosure is only as an example and does not constitute a limitation on the present specification. Although it is not explicitly stated here, one of ordinary skill in the art can make various modifications, improvements and corrections to the present specification. Such modifications, improvements and corrections are suggested in the present specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of the present specification.

[0095] Also, various aspects of the present description can be described in terms of sequences of actions to be performed by, for example, elements of a computing device. It will be understood that such descriptions are used by way of illustration only, and that the described sequences of actions can be performed in other sequences or in other items without departing from the scope of the present description. Furthermore, some of the items described herein can be implemented in hardware, software, firmware, middleware or any combination thereof, and can be implemented in any of a variety of system configurations, including personal computers, laptop computers, tablet computers, netbooks, e-readers, consumer electronics, gaming devices, cellular telephones, smartphones, PDAs, multiprocessor systems, microprocessor-based systems, set-top boxes, wearable devices, programmable consumer electronics, minicomputers, mainframe computers, distributed computing environments that include any of the above systems or devices, processing circuitry or other components, or the like.

[0096] Furthermore, those of ordinary skill in the art will appreciate that the various aspects of the present description can be implemented in any of a variety of ways, including as a computer program product that comprises computer program code configured to be implemented by one or more computers. Such program code can be stored, for example, on a computer-readable storage medium, which can be any available media that can be accessed by one or more computers.

[0097] A computer readable signal medium can include a propagated data signal with computer program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated signal can take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium can be any computer readable medium that can be

[0098] The computer program code required for the operation of each part of this manual can be written in any one or more programming languages, including object-oriented programming languages ​​such as Java, Scala, Smalltalk, Eiffel, JADE, Emerald, C++, C#, .NET, Python, etc.; conventional procedural programming languages ​​such as C, Visual Basic, Fortran2003, Perl, COBOL 2002, PHP, ABAP; dynamic programming languages ​​such as Python, Ruby, and Groovy; or other programming languages. This program code can run entirely on the user's computer, or as a standalone software package on the user's computer, or partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the latter case, the remote computer can be connected to the user's computer through any network, such as a local area network (LAN) or wide area network (WAN), or connected to an external computer (e.g., via the Internet), or in a cloud computing environment, or used as a service such as Software as a Service (SaaS).

[0099] Furthermore, unless expressly stated in the claims, the order of processing elements and sequences, the use of numbers and letters, or other names described in this specification are not intended to limit the order of the processes and methods described herein. Although various examples have been discussed in the foregoing disclosure of some embodiments of the invention that are currently considered useful, it should be understood that such details are for illustrative purposes only, and the appended claims are not limited to the disclosed embodiments; rather, the claims are intended to cover all modifications and equivalent combinations that conform to the spirit and scope of the embodiments described herein. For example, while the system components described above can be implemented using hardware devices, they can also be implemented solely using software solutions, such as installing the described system on existing servers or mobile devices.

[0100] Similarly, it should be noted that, in order to simplify the description disclosed herein and thus aid in the understanding of one or more embodiments of the invention, multiple features may sometimes be grouped into a single embodiment, drawing, or description thereof in the foregoing description of the embodiments. However, the method described herein should not be construed as reflecting an intention that the claimed object to be scanned requires more features than expressly recited in each claim. In fact, the embodiments contain fewer features than all the features of the individual embodiments disclosed above.

Claims

1. A medical image processing method, comprising: Acquire multi-frame scan data of the scanned object; The reference frame is selected from the multi-frame scan data based on the first model, including: Obtain a comparison image of the scanned object; wherein the comparison image is of a different data type than the multi-frame scan data, and the resolution of the comparison image is higher than the resolution of the multi-frame scan data; The comparison image and the multi-frame scan data are input into the first model to obtain the reference frame; wherein, the reference frame is the scan data with the highest similarity to the comparison image among the multi-frame scan data; The reference frame and the multi-frame scan data are processed to obtain correction information of the multi-frame scan data relative to the reference frame.

2. The method of claim 1, wherein the training method of the first model includes: The first model is obtained by supervised training of the initial first model based on the first training samples; wherein the first training samples include multiple first training sample sets, each first training sample set includes sample comparison images and multiple frames of first sample scanning data, the sample comparison images and the first sample scanning data are used as training samples, and the image with the highest similarity to the comparison images in the first sample scanning data is used as the label.

3. The method of claim 1, further comprising: Based on the correction information, the multi-frame scan data is registered to obtain multi-frame target scan data; wherein, the correction information includes registration information.

4. The method of claim 3, wherein, The registration process based on the correction information to obtain multi-frame target scanning data includes: Based on the correction information, the multi-frame scan data is registered to obtain multi-frame registered scan data. The scanning data is obtained by combining the multi-frame registered scanning data and the multi-frame matched images to obtain the multi-frame target scanning data.

5. The method of claim 1, wherein processing the reference frame and the multi-frame scan data to obtain correction information of the multi-frame scan data relative to the reference frame includes: The multi-frame scan data and the reference frame are input into the second model to obtain the deformation field of each frame scan data relative to the reference frame. The correction information is determined based on the deformation field.

6. The method of claim 5, wherein the training method of the second model includes: The initial second model is trained unsupervised based on the second training samples to obtain the second model; wherein, the second training samples include multiple second training sample sets, and each second training sample set includes a sample reference frame and multiple frames of second sample scan data.

7. The method of claim 1, wherein, The scan data includes data acquired by PET or SPECT scanning equipment.

8. A medical image processing system, comprising: The scan data acquisition module is used to acquire multi-frame scan data of the scanned object; A reference frame determination module is used to select a reference frame from the multi-frame scan data based on a first model, including: Obtain a comparison image of the scanned object; wherein the comparison image is of a different data type than the multi-frame scan data, and the resolution of the comparison image is higher than the resolution of the multi-frame scan data; The comparison image and the multi-frame scan data are input into the first model to obtain the reference frame; wherein, the reference frame is the scan data with the highest similarity to the comparison image among the multi-frame scan data; The correction information acquisition module is used to process the reference frame and the multi-frame scan data to obtain correction information of the multi-frame scan data relative to the reference frame.

9. A medical image processing apparatus, the apparatus comprising at least one frame storage medium and at least one frame processor; The at least one frame of storage medium is used to store computer instructions; The at least one frame processor is used to execute the computer instructions to implement the medical image processing method according to any one of claims 1 to 7.

Images