WO2023169565A1 - Image registration method and system - Google Patents

Abstract

Disclosed in embodiments of the present description are an image registration method and system. The method comprises: obtaining respiration data and initial dynamic PET data of a subject under test in a preset time period; determining a plurality of frames of dynamic MR-like images in the preset time period on the basis of the respiration data; obtaining, on the basis of the initial dynamic PET data, a plurality of frames of dynamic PET-like images corresponding to the plurality of frames of dynamic MR-like images, and determining registered dynamic PET-like images on the basis of the plurality of frames of dynamic PET-like images; and performing registration on the basis of the registered dynamic PET-like images and a plurality of frames of initial dynamic PET images to determine registered dynamic PET images, wherein the plurality of frames of initial dynamic PET images are obtained by reconstructing on the basis of the initial dynamic PET data.

Description

An image registration method and system

cross reference

This specification claims the priority of Chinese application number 202210237538.9 submitted on March 10, 2022, the entire content of which is incorporated herein by reference.

Technical field

This description relates to the field of imaging diagnosis, and in particular to an image registration method and system.

Background technique

Kinetic parameter analysis based on dynamic positron emission computed tomography (PET) can reveal the biochemical process of tracers and provide a basis for clinically revealing pathological mechanisms.

In order to analyze the biochemical process of the tracer, it is necessary to first inject the tracer into the detection imaging part of the object to be tested, then collect initial PET images over a continuous period of time, and then analyze the kinetic parameters of the initial PET images. Since it is difficult for the object to be tested to keep its body posture completely fixed for a long time, the initial PET images collected at different time points will be displaced. Therefore, before analyzing the dynamic parameters, it is generally necessary to perform registration processing on the initial PET images to improve the power. The accuracy of quantification of learning parameters.

Therefore, it is necessary to provide an image registration method and system to improve the accuracy of image registration.

Contents of the invention

One aspect of the embodiments of this specification provides an image registration method. The method includes: obtaining the respiratory data and initial dynamic PET data of the subject to be measured within a preset time period; based on the respiratory data, determining the preset time Multi-frame dynamic-like MR images within the segment; obtain multi-frame dynamic-like PET images corresponding to the multi-frame dynamic-like MR images based on the initial dynamic PET data, and determine the registration based on the dynamic-like PET images of each frame A quasi-dynamic PET image; registration is performed based on the registered quasi-dynamic PET image and a multi-frame initial dynamic PET image to determine the registered dynamic PET image; wherein the multi-frame initial dynamic PET image is based on the Initial dynamic PET data are reconstructed and obtained.

Another aspect of the embodiments of this specification provides an image registration system. The system includes: an acquisition module for acquiring respiratory data and initial dynamic PET data of the subject to be measured within a preset time period; a first image acquisition module, for determining a multi-frame dynamic MR image within the preset time period based on the respiratory data; a second image acquisition module for obtaining a corresponding multi-frame dynamic MR image based on the initial dynamic PET data A multi-frame dynamic-like PET image, and determine a registered dynamic-like PET image based on the dynamic-like PET image of each frame; a third image acquisition module, configured to determine the registered dynamic-like PET image and the multi-frame based on the registered dynamic-like PET image The initial dynamic PET images are registered to determine the registered dynamic PET images; wherein the multi-frame initial dynamic PET images are reconstructed and obtained based on the initial dynamic PET data.

Another aspect of the embodiments of this specification provides an image registration device, including at least one storage medium and at least one processor. The at least one storage medium is used to store computer instructions; the at least one processor is used to execute the computer instructions. Instructions to implement the above image registration method.

Another aspect of the embodiments of this specification provides a computer-readable storage medium. The storage medium stores computer instructions. After the computer reads the computer instructions in the storage medium, the computer executes the above image registration method.

In some embodiments of this specification, by acquiring the respiratory data and initial dynamic PET data of the object to be measured within a preset time period, determining corresponding multi-frame dynamic MR images within the preset time period based on the respiratory data, and obtaining each frame of dynamic MR images. The quasi-dynamic PET image corresponding to the MR image is determined, and the registered quasi-dynamic PET image is determined through each frame of the quasi-dynamic PET image. Based on the registered quasi-dynamic PET image, multiple frames of initial dynamic PET images are registered to obtain the registration. Accurate dynamic PET images; the above method can directly obtain quasi-dynamic MR images through respiratory data, and then obtain quasi-dynamic PET images with tissue/organ structural characteristics of the imaging site based on the quasi-dynamic MR images and initial dynamic PET data, and based on the configuration The registered dynamic PET image is used to register multiple frames of initial dynamic PET images, so that the registered dynamic PET image can reflect structural information and obvious image features, thereby improving the accuracy of dynamic PET image registration results.

Description of the drawings

This specification is further explained by way of example embodiments, which are described in detail by means of the accompanying drawings. These embodiments are not limiting. In these embodiments, the same numbers represent the same structures, where:

Figure 1 is a schematic diagram of an exemplary application scenario of an image registration system according to some embodiments of this specification;

Figure 2 is an exemplary flow chart of an image registration method according to some embodiments of this specification;

Figure 3 is an exemplary flow chart of a model training method according to some embodiments of this specification;

Figure 4 is an exemplary flow chart for determining model training according to other embodiments of this specification;

Figure 5 is an exemplary flowchart of determining the result of a multi-frame dynamic PET image according to other embodiments of this specification;

Figure 6 is an exemplary flowchart of a method for determining registered dynamic PET images according to some embodiments of the present specification;

Figure 7 is an exemplary module diagram of an image registration system according to some embodiments of this specification;

Figure 8 is an exemplary schematic diagram of multiple frames of different types of images in the image registration process according to some embodiments of this specification;

FIG. 9 is an exemplary schematic diagram illustrating the correspondence between time points, sample MR data, and respiratory data according to some embodiments of this specification.

Detailed ways

In order to explain the technical solutions of the embodiments of this specification more clearly, the accompanying drawings needed to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some examples or embodiments of this specification. For those of ordinary skill in the art, without exerting any creative efforts, this specification can also be applied to other applications based on these drawings. Other similar scenarios. Unless obvious from the locale or otherwise stated, the same reference numbers in the figures represent the same structure or operation.

It should be understood that the terms "system", "apparatus", "unit" and/or "module" as used herein are a means of distinguishing between different components, elements, parts, portions or assemblies at different levels. However, said words may be replaced by other expressions if they serve the same purpose.

As shown in this specification and claims, words such as "a", "an", "an" and/or "the" do not specifically refer to the singular and may include the plural unless the context clearly indicates an exception. Generally speaking, the terms "comprising" and "comprising" only imply the inclusion of clearly identified steps and elements, and these steps and elements do not constitute an exclusive list. The method or apparatus may also include other steps or elements.

Flowcharts are used in this specification to illustrate operations performed by systems according to embodiments of this specification. It should be understood that preceding or following operations are not necessarily performed in exact order. Instead, the steps can be processed in reverse order or simultaneously. At the same time, you can add other operations to these processes, or remove a step or steps from these processes.

PET imaging is widely used in tumor staging and grading, preoperative evaluation, and prognosis evaluation. Compared with traditional single-frame static images, dynamic parameter analysis based on dynamic multi-frame PET imaging can further reveal the biochemical process of tracers and provide scientific basis for clinically revealing pathological mechanisms. In recent years, it has received widespread attention from the clinical and scientific research communities.

Limited by the time characteristics of tracer accumulation in the patient's body, PET dynamic imaging requires continuous data collection for a certain period of time, and the collection time for many common tracers is as long as one hour. For patients undergoing dynamic PET imaging, it is difficult to keep the body completely fixed for a long time, causing displacement between data collected at different time points. Therefore, before analyzing the kinetic parameters of PET dynamic data, it is generally necessary to perform registration processing on the data to improve the accuracy of quantification of the kinetic parameters.

Currently, related technologies use direct registration of multi-time point PET data, or registration based on single-shot MR (Magnetic Resonance)/CT (Computed Tomography) data for subsequent use. data analysis. However, in dynamic PET imaging, with the accumulation of tracers, the contrast of the PET image will change significantly, affecting the registration effect. At the same time, the PET image lacks structural information and obvious image feature points, resulting in related problems. There are ways to achieve unsatisfactory results.

In response to the above problems, this specification proposes some improvement methods to improve at least part of them. The image registration method provided by some embodiments of this specification can be applied to the image registration system as exemplarily shown in Figure 1, and can be applied to medical image registration or medical image calibration scenarios. For example, in medical scenarios, dynamic parameter analysis of scanned medical images can provide a basis for clinically revealing pathological mechanisms. Kinetic parameter analysis can reveal the biochemical process of the tracer. In this process, the tracer can be administered to the subject to be tested, and then the subject to be tested with the tracer can be scanned to obtain medical images, and then the medical image can be obtained at this time. Medical images are analyzed for dynamic parameters. Optionally, the subject to be tested is subject to the characteristics of the accumulation time of the tracer in the body. Therefore, in some embodiments, it is necessary to continuously collect data for a certain period of time, generate dynamic medical images after reconstruction, and then perform dynamic parameter analysis on the dynamic medical images to reveal the biochemical process of the tracer. Further, medical staff can use the indications to The biochemical process of the tracer determines the pathological results of the subject to be tested. The above dynamic medical images may be dynamic positron emission computed tomography (Positron Emission Computed Tomography, PET) images, electronic computed tomography (Computed Tomography, CT) images, MR images, etc. For example, in this specification, the dynamic medical image is a dynamic PET image as an example.

Figure 1 is a schematic diagram of an exemplary application scenario of an image registration system according to some embodiments of this specification.

As shown in FIG. 1 , the image registration system 100 may include an imaging device 110 , a processing device 120 , a network 130 , a storage device 140 and a terminal 150 .

Imaging device 110 may be used to image a target object to produce an image. The imaging device 110 may be a medical imaging device (for example, CT (Computed Tomography), PET (Positron Emission Computed Tomography), MRI (Magnetic Resonance Imaging), SPECT (Single-Photon Emission Computed Tomography, PET-CT imaging equipment, PET-MR imaging equipment, etc.). In some embodiments, imaging device 110 may be a PET-MR device. PET-MR equipment is a mixed-modal imaging equipment that integrates a PET (Positron Emission Tomography, positron emission tomography equipment) scanner and an MRI (Magnetic Resonance Imaging, MR) scanner. It also has PET imaging With MR imaging function, it has high sensitivity and accuracy.

In some embodiments, the PET-MR device can be used to scan the object to be tested and collect MR signals and photon signals related to the object to be tested. In some embodiments, PET-MR can scan a region of interest (eg, a tumor site) of a scanning object (eg, a patient, etc.) and collect corresponding data, such as PET data, MR data. PET data and MR data can be used for imaging respectively, for example, acquiring PET images, MR images, PET-MR fusion images, etc. In some embodiments, the PET-MR device may include a respiratory gating device, and the respiratory gating device may be used to assist in collecting accurate respiratory data during imaging by the PET-MR device. In some embodiments, the respiratory gating device can also be set up independently of the PET/MR equipment, which is not limited in this specification. In some embodiments, respiratory data may be used to acquire PET images or MR images corresponding to respiratory time points.

The processing device 120 may process data and/or information obtained from the imaging device 110, the storage device 140, and/or the terminal 150. For example, the processing device 120 can process the PET data and MR data acquired by the imaging device 110 and generate corresponding images. For example, PET data and MR data can be used to obtain PET images, MR images, PET-MR fusion images, etc. In some embodiments, processing device 120 may acquire MR images based on respiratory data. In some embodiments, PET data, MR data, respiratory data or generated images can be sent to the terminal 150 and displayed on a display device in the terminal 150 for the user to view and analyze. In some embodiments, processing device 120 may be a single server or a group of servers. Server groups can be centralized or distributed. In some embodiments, processing device 120 may be local or remote. In some embodiments, processing device 120 may be directly connected to imaging device 110, storage device 140, and/or terminal 150 to access information and/or data stored thereon. In some embodiments, processing device 120 may be integrated into imaging device 110 . In some embodiments, the processing device 120 may be implemented on a cloud platform. For example only, the cloud platform may include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-cloud, etc. or any combination thereof.

Network 130 may include any suitable network that may facilitate the exchange of information and/or data for image registration system 100 . In some embodiments, one or more components of image registration system 100 (eg, imaging device 110, processing device 120, storage device 140, or terminal 150) may be connected to other components of image registration system 100 through network 130 and/or or communication. For example, processing device 120 may obtain ultrasound data from imaging device 110 over network 130. For another example, the processing device 120 can obtain the user's instructions from the terminal 150 through the network 130, and the instructions can be used to instruct the imaging device 110 to scan and image the object to be measured, etc. In some embodiments, network 130 may include one or more network access points. For example, network 130 may include wired and/or wireless network access points, such as base stations and/or Internet access points, through which one or more components of image registration system 100 may be connected to network 130 to exchange data and/or information. .

Storage device 140 may store data and/or instructions. In some embodiments, storage device 140 may store data obtained from terminal 150 and/or processing device 120. In some embodiments, storage device 140 may store data and/or instructions that processing device 120 may perform or be used to perform the example methods described in this specification. In some embodiments, the storage device 140 may be implemented on a cloud platform.

In some embodiments, storage device 140 may be connected to network 130 to communicate with one or more components of image registration system 100 (eg, processing device 120, terminal 150, etc.). One or more components of image registration system 100 may access data or instructions stored in storage device 140 via network 130 . In some embodiments, storage device 140 may be directly connected to or in communication with one or more components of image registration system 100 (eg, processing device 120, terminal 150, etc.). in some In embodiments, storage device 140 may be part of processing device 120.

The terminal 150 may include a mobile device 150-1, a tablet 150-2, a laptop 150-3, etc., or any combination thereof. In some embodiments, terminal 150 can operate imaging device 110 remotely. In some embodiments, terminal 150 may operate imaging device 110 via a wireless connection. In some embodiments, terminal 150 may receive information and/or instructions input by a user and send the received information and/or instructions to imaging device 110 or processing device 120 via network 130 . In some embodiments, terminal 150 may receive data and/or information from processing device 120 . In some embodiments, terminal 150 may be part of processing device 120. In some embodiments, terminal 150 may be omitted.

Figure 2 is an exemplary flowchart of an image registration method according to some embodiments of this specification. In some embodiments, process 200 may be performed by a processing device (eg, processing device 120). For example, the process 200 may be stored in a storage device (such as a self-contained storage unit of a processing device or an external storage device) in the form of a program or instructions, and when executed, the process 200 may be implemented. Process 200 may include the following operations.

Step 202: Obtain the respiratory data and initial dynamic PET data of the subject to be measured within a preset time period. In some embodiments, step 202 may be performed by an acquisition module.

Objects to be tested may include biological objects and/or non-biological objects. For example, the object to be tested may include a specific part of the patient's body, such as the neck, chest, abdomen, etc., or a combination thereof. For another example, the subject to be tested may be a medical experimental animal (for example, a mouse) or the like. For another example, the object to be tested may also be a phantom built to simulate human body characteristics, etc.

The respiratory data may be parameters used to reflect the respiratory physiological condition of the subject to be measured. In some embodiments, the respiratory data may include respiratory signals, which are physiological electrical signals generated by periodic deformations in the respiratory tract, chest and abdomen accompanied by periodic changes in exhalation and inhalation.

Initial dynamic PET data refers to signal data collected through PET imaging equipment. Dynamic can refer to the presence of signals at multiple points in time in the collected data. For example, the object to be tested is detected by a PET imaging device, and data is collected for a period of time (for example, within twenty minutes). The data obtained by averaging the data within this period of time is static PET data. Dynamic PET data refers to multiple time points within this time period. For example, a set of data is released in the first minute, a set of data is released in the second minute, and finally at the 20th minute, a total of twenty sets of data are released. There are twenty time points in total, and these data can be called dynamic PET data.

The preset time period may refer to a certain time period specified in advance, or a time period that meets certain preset conditions. For example, the preset time period may be a specified time period, such as the time period from 12:00 to 12:20, or it may be the time period after the subject to be tested is administered a tracer. For example, 5 minutes, 20 minutes, 30 minutes, one hour, etc. after the tracer is administered. In some embodiments, the preset time period after the tracer is administered may be referred to as the first preset time period.

In some embodiments, before scanning any one or more imaging parts of the subject to be tested through the imaging device, the medical staff may first intravenously administer a certain amount of tracer to one or more imaging parts of the subject to be tested, and then With the accumulation of tracer in the tissue, within a specific period of time, there will be a certain biochemical reaction with the tissue/organ in the imaging area where the tracer is administered, and during this biochemical reaction, the tissue/organ in the lesion area will interact with the tissue/organ. Tissues/organs in non-lesion areas will have obvious differences, which can be reflected in dynamic PET images (see the description below for dynamic PET images).

Optionally, during the biochemical reaction of the tissue/organ, the imaging device can scan the imaging site where the tracer is applied to obtain a dynamic PET image, where as the tracer accumulates in the tissue/organ at the imaging site, The contrast of the dynamic PET image will change significantly. However, the longer the time, the greater the contrast of the dynamic PET image. Within a period of time, the contrast of the dynamic PET image will reach a peak. Therefore, the above-mentioned specific time period includes The time point when the contrast of the dynamic PET image reaches its peak. Optionally, the starting time point of the specific time period may be the time point when the tracer administration ends, or it may be a time point after the time point when the tracer administration ends. Optionally, the above-mentioned tracers can be different nuclide drugs, which are harmless to the subject to be tested. In some embodiments, the specific time period here may be the preset time period.

In practical applications, the imaging equipment can scan one or more imaging parts of the subject to be tested where the tracer is administered, obtain the initial dynamic PET data scanned within a preset time period, and collect the initial dynamic PET data at the same time The respiratory data of the subject to be measured can also be collected simultaneously. When collecting respiratory data, the respiratory gating device can be used to ensure the accuracy of the collected respiratory data. The imaging device can send both the initial dynamic PET data and respiratory data collected by scanning within a preset time period to the processing device for processing. Alternatively, the processing device can also obtain the initial dynamic PET data collected within the historical time period from the cloud or local data, that is, the initial dynamic PET data within the first preset time period. In some embodiments, the method of obtaining the initial dynamic PET data within the first preset time period may not be limited.

Step 204: Based on the respiratory data, determine multi-frame dynamic MR images within the preset time period. In some embodiments, step 204 may be performed by the first image acquisition module.

Before introducing quasi-dynamic MR images, dynamic MR images will be introduced first. Dynamic MR is a scan in medical clinical method), the data collected at the same time are also MR data at multiple time points, and the corresponding generated images correspond to multiple time points.

The difference between quasi-dynamic MR images and dynamic MR images is that no MR tracer is injected into quasi-dynamic MR images during acquisition. Dynamic MR images still collect data at multiple time points during magnetic resonance scanning. The data at each time point can be called sequence 1, sequence 2, and sequence 3. Each sequence can correspond to one or more images. images. Quasi-dynamic MR images have the characteristics of multi-time acquisition of traditional dynamic MR images, but do not have the characteristics of traditional dynamic MR tracers. Therefore, quasi-dynamic MR images can also be defined as a set of original MR images collected.

In some embodiments, the processing device can process the respiratory data through a preset prediction model to determine multi-frame dynamic MR images within a preset time period. Specifically, the processing device can input the respiratory data into a prediction model for processing. After black-box processing of the respiratory data within the prediction model, the prediction model can output multi-frame dynamic MR images corresponding to the respiratory data. The preset prediction model can obtain the correspondence between respiratory data and MR images after pre-training. Therefore, after inputting the respiratory data into the prediction model for processing, the corresponding dynamic MR image can be obtained. Respiration data can be represented by a respiration signal similar to a waveform state. Each MR sequence can correspond to a segment of the respiration signal. After the segment of the respiration signal is input into the prediction model, the prediction model can output an MR image corresponding to the segment of the respiration signal. The MR image corresponds to the MR sequence corresponding to the respiratory signal. For example, the MR image and the MR sequence corresponding to the respiratory signal should be relatively similar in terms of image information (for example, the tissue in the image, the size and position of the structure, etc.) .

In some embodiments, the preset prediction model may be a pre-trained machine learning model, and its type may be a deep neural network model, a convolutional neural network model, a recurrent neural network model, an adversarial neural network model, or other combination models, etc. , this manual does not limit this. Regarding the training of the prediction model, please refer to the description in Figure 3 or Figure 4.

Although the collected MR or PET images can be directly registered, their acquisition frame rate will be relatively low. In this case, the acquisition frame rate of the MR image will be predicted by synchronizing the collected respiratory data, and the subsequent Image registration will also be more accurate. The acquisition frame rate is used to reflect the scanning efficiency. For example, for a timeline, if it is imaged through an imaging device, it may take several minutes of data to generate a picture. However, if breathing data is used to predict the image, the breathing signal of the subject under test will continue to exist. , there is a corresponding respiratory signal every second, which is equivalent to predicting an MR image in a few seconds. The number of images and the efficiency of image output will be greatly improved.

In some embodiments, the processing device can also perform arithmetic operations, data conversion, analysis, data comparison and/or reconstruction on the respiratory data within the preset time period to obtain corresponding multi-frame dynamics within the preset time period. MR images. Alternatively, the processing device can also perform preprocessing such as arithmetic operations, data conversion, analysis, data comparison and/or reconstruction on the respiratory data within a preset time period, and then use a specific algorithm (for example, a deep convolutional network algorithm, etc. ) perform specific processing on the preprocessing results to obtain corresponding multi-frame dynamic MR images within a preset time period. Optionally, the above arithmetic operations may be addition operations, subtraction operations, division operations, multiplication operations, exponential operations and/or logarithm operations, etc.

It should be noted that the lengths of the time periods (or sub-time periods, time points) corresponding to each frame type of dynamic MR images may be equal or unequal. Optionally, the duration of the sub-time period of the dynamic MR image of each frame may be greater than or equal to the duration of the sub-time period of the initial dynamic PET image of the corresponding frame (the acquisition method can be found in the description of Figure 5 below). Optionally, multiple frames of quasi-dynamic MR images can be acquired within a preset time period; each frame of quasi-dynamic MR image can be a dynamic MR image generated within a certain sub-time period within the preset time period. Among them, the sub-time period corresponding to each frame of dynamic MR image is different from the sub-time period corresponding to each frame of dynamic PET image.

Step 206: Acquire multi-frame dynamic-like PET images corresponding to the multi-frame dynamic-like MR images based on the initial dynamic PET data, and determine the registered dynamic-like PET images based on the dynamic-like PET images of each frame. In some embodiments, step 206 may be performed by the second image acquisition module.

A quasi-dynamic PET image refers to a PET image reconstructed based on the initial dynamic PET data.

In some embodiments, the processing device may first obtain corresponding initial dynamic PET data based on each frame of dynamic-like MR images (or respiratory data corresponding to the dynamic-like MR images), for example, determine the corresponding dynamic PET data based on the dynamic-like MR images or respiratory data. The time point is then mapped through the corresponding relationship between the time point and the initial dynamic PET data to determine the corresponding initial dynamic PET data. Afterwards, the processing device can use an image reconstruction algorithm to reconstruct the multi-frame dynamic PET image corresponding to the multi-frame dynamic MR image based on the initial dynamic PET data. For example, the processing device may use a direct back-projection method, an iterative method or a Fourier transform reconstruction method to reconstruct the image.

In some embodiments, the initial dynamic PET data within a preset time period has corresponding multi-frame dynamic PET images. For example, the initial dynamic PET data within a time period can reconstruct multiple corresponding dynamic PET images. In some embodiments, the initial dynamic PET data may also be referred to as dynamic PET data or initial PET data.

In some embodiments, the processing device can perform mapping processing, conversion processing, and/or analysis processing, etc., on the dynamic MR images of each frame and the dynamic PET images of the corresponding frames, to obtain the dynamic MR images of each frame corresponding to the dynamic PET images of each frame. PET image. Optionally, the mapping process can be understood as mapping between the quasi-dynamic MR image of the same size and the pixel values of the pixels at the corresponding positions in the corresponding dynamic PET image, and then superimposing the pixel values of the two pixels at the mapped positions. Or the process of operations such as subtraction. Optionally, the conversion process can be understood as a process of performing arithmetic operations on one or more preset values and pixel values of pixels at different positions in the quasi-dynamic MR image. Analysis processing can be understood as the process of analyzing the pixel resolution of each pixel in the quasi-dynamic MR image.

In some embodiments, the processing device can use any frame of the dynamic-like MR image as a reference image, perform image registration on each frame of the dynamic-like PET image, and obtain a registered dynamic-like PET image. Among them, the above-mentioned quasi-dynamic PET image can reflect the boundary, shape and other information of the tissue/organ in the imaging part of the object to be tested. That is, the quasi-dynamic PET image carries the structural information of the tissue/organ, and the quasi-dynamic PET image is different from other Compared with medical images, it can reflect obvious feature points in the image.

For example, the processing device can determine the deformation field used to register the dynamic PET images of each frame based on the multi-frame dynamic MR images, and then register the dynamic PET images of each frame based on the deformation field to obtain the registration. Accurate motion-like PET images.

Specifically, the processing device can perform arithmetic operations, conversion, analysis, and/or comparison processing on each frame type of dynamic MR image to obtain the deformation field corresponding to each frame type of dynamic MR image. In some embodiments, the image may be represented in the form of a matrix, and the size of the matrix may be the same as the size of the image. If the size of the image is 3×3, the size of the matrix is also 3×3. The data in the first row and first column of the matrix can be the pixel value of the pixel in the first row and first column of the image and the first The position (1, 1) of the pixel in the first row and column of the matrix. The data in the first row and second column of the matrix can be the pixel value of the pixel in the first row and second column of the image and the pixel value of the first row and second column in the image. The pixel position (1, 2) of the pixel point also has a corresponding relationship between the pixel values of other pixel points in the image and the data at the corresponding position in the matrix.

It should be noted that the deformation field can be represented by a deformation field matrix, and the size of the deformation field matrix can be equal to the pixel matrix size of the quasi-dynamic MR image. Optionally, the values at different positions in the deformation field matrix can represent the deformation values of corresponding pixels in the quasi-dynamic MR image.

In some embodiments, the processing device may determine the registered dynamic-like PET image based on each frame of the dynamic-like PET image in the manner described in the embodiments below.

The processing device may determine the deformation field between the dynamic-like MR image and the reference image in each frame; perform registration processing on the dynamic-like PET image in each frame based on the deformation field, and determine the registered dynamic-like PET image. image. The reference image can be any image selected from a multi-frame dynamic MR image. For example, the reference image can be a certain moment (for example, a certain moment before the preset time period, the initial moment of the preset time period, the middle moment of the preset time period, etc.) time or end time, etc.) corresponding to the MR image.

In some embodiments, during a period of time before applying a certain amount of tracer to one or more imaging sites of the subject to be tested (the preset period may be referred to as the second preset period, optionally , the length of the second preset time period can be greater than, less than, or equal to the length of the first preset time period), the imaging device can scan the imaging part of the object to be measured, obtain the sample MR data, and send the sample MR data to the processor equipment. The processing device can reconstruct the sample MR data within the time period to obtain multi-frame sample MR images. Each frame of the sample MR image has a corresponding sub-time period, and the time period in which the sub-time periods corresponding to each frame of the sample MR image are combined is equal to the time period of the sample MR data collected by the medical scanning equipment.

Wherein, the processing device can select any one frame of MR image from the multi-frame sample MR image as the reference image. For example, the processing device may select a frame of sample MR image corresponding to the sub-time period closest to the starting time point of the first preset time period as the reference image. Among them, the corresponding sample MR data can carry the relaxation properties of the tissue/organ, that is, the sample MR data can include T1WI sequences and/or T2WI sequences, etc. The T1WI sequence represents the T1 sequence of nuclear magnetic resonance, and the T2WI sequence represents the T2 sequence of nuclear magnetic resonance. In some embodiments, the sample MR images acquired within the second preset time period can be used to train and update the initial prediction model to improve the prediction effect of the images.

In some embodiments, the processing device may use an image registration algorithm to perform image registration on dynamic MR images of each frame type through the selected reference image to obtain the deformation field corresponding to the dynamic MR image of each frame type. The deformation field matrix corresponding to the deformation field may be equal to the difference between the pixel value of each pixel point in the reference image and the dynamic MR image of each frame.

In this embodiment, the deformation field corresponding to the dynamic MR image of each frame type can be obtained, and then through the deformation field corresponding to the dynamic MR image of each frame type, the dynamic PET images of each frame type in the corresponding time period can be more conveniently registered. This enables the registration of quasi-dynamic PET images to take into account the structural density of quasi-dynamic MR images, thereby improving the accuracy of image registration results.

Step 208: Perform registration based on the registered dynamic-like PET image and multiple frames of initial dynamic PET images to determine the registered dynamic PET image. In some embodiments, step 208 may be performed by a third image acquisition module.

The initial dynamic PET image refers to the image obtained by reconstruction based on the initial dynamic PET data. Multiple frames of initial dynamic PET images can be reconstructed and obtained based on multiple initial dynamic PET data within a preset time period. It should be noted that although the multi-frame initial dynamic PET image is similar to the quasi-dynamic PET image, both are reconstructed based on the initial dynamic PET data, but the difference lies in the emphasis on The time period corresponding to the initial dynamic PET data for constructing the quasi-dynamic PET image is the same as the time period corresponding to the quasi-dynamic MR image, and the initial dynamic PET image can be based on a certain time period within the preset time period or the entire time. The initial dynamic PET data reconstruction within the segment is obtained. The reconstructed initial dynamic PET image can correspond to multiple time periods. For example, one frame of the initial dynamic PET image can correspond to one time period.

In some embodiments, the processing device may determine that the registered dynamic-like PET image determines a corresponding initial dynamic PET image. For example, the registered quasi-dynamic PET images and the initial dynamic PET images can be grouped according to the time points corresponding to them, and the registered quasi-dynamic PET images and initial dynamic PET images with the same or similar time points can be divided into a group. And register the quasi-dynamic PET image and the initial dynamic PET image in the group to obtain the registered dynamic PET image.

When performing registration, the processing device can register multiple frames of initial dynamic PET images through the registered quasi-dynamic PET images carrying tissue/organ structure information to improve the accuracy of the dynamic PET image registration results. Among them, the processing device can use the registered dynamic PET image as a reference image, use a registration algorithm to register the initial dynamic PET images of each frame, and obtain the registered dynamic PET image. Optionally, the processing device can also perform arithmetic operations on the registered quasi-dynamic PET images and the initial dynamic PET images of each frame, so as to register the initial dynamic PET images of each frame and obtain the registered dynamic PET images. .

In some embodiments, the object to be measured has almost the same posture within the preset time period. In actual processing, the sizes of dynamic PET images, quasi-dynamic PET images and quasi-dynamic MR images can be the same.

In some embodiments, the processing device may use an image registration algorithm to register the corresponding dynamic-like PET images through the deformation fields corresponding to the dynamic-like MR images of each frame, to obtain the registered dynamic-like PET images.

In some embodiments, the above-mentioned image registration algorithm may be a matching algorithm based on image grayscale or a matching algorithm based on image features, or may also be other image matching algorithms, which is not limited in this embodiment. Among them, the above-mentioned matching algorithms based on image grayscale can be average absolute difference algorithm, absolute error sum algorithm, error sum of squares algorithm, average error sum of squares algorithm, normalized product correlation algorithm, sequential similarity algorithm, etc.; the above-mentioned based on The matching algorithm of image features can be feature extraction, feature matching, model parameter estimation, image transformation, grayscale interpolation algorithm, etc.

Figure 8 shows the multi-frame dynamic quasi-MR image corresponding to a certain imaging part of an object to be measured during the image registration process, the registered quasi-dynamic MR image, the quasi-dynamic PET image, the registered quasi-dynamic PET image, Schematic diagram of dynamic PET images and registered dynamic PET images.

In some embodiments of this specification, the respiratory signal (respiratory data) of the subject to be tested can be collected while collecting dynamic PET data, and a pre-trained prediction model can be used to predict the class at the corresponding time using the respiratory signal as input. Dynamic MR images. Since MR and the corresponding PET are scanned at the same time, the two sets of data are completely synchronized in space and time. Therefore, registration can be directly based on multi-frame dynamic MR images, the corresponding deformation field is recorded, and then the deformation field is directly It is applied to PET images at corresponding time points to assist in the registration of dynamic PET images and improve the accuracy of registration.

Figure 3 is an exemplary flow chart of a model training method according to some embodiments of this specification. In some embodiments, process 300 may be performed by a processing device (eg, processing device 120). For example, the process 300 may be stored in a storage device (such as a self-contained storage unit of a processing device or an external storage device) in the form of a program or instructions, and when executed, the process 300 may be implemented. Process 300 may include the following operations.

Step 302: Obtain multi-frame sample MR images and sample respiratory data of the subject to be tested.

Training samples refer to the data used for model training. In some embodiments, the training samples may include multiple frames of sample MR images and sample respiratory data. Among them, sample respiratory data can be used as input data for model training, and sample MR images can be used as the gold standard for model training. There is a corresponding relationship between the multi-frame sample MR image and the sample respiration data. Each frame of the sample MR image corresponds to one (or a piece of) sample respiration data.

In some embodiments, the sample MR images may be dynamic-like MR images.

In some embodiments, the processing device may obtain historically collected and stored multi-frame sample MR images and sample respiratory data from a storage device, a database, an imaging device reading, or the like.

In some embodiments, the processing device may also perform a magnetic resonance scan to acquire multi-frame sample MR images of the object to be tested before the preset time period (this period may be referred to as the second preset time period). and sample breath data. By performing data collection online in real time, training samples that are closer to the current state of the object to be tested can be obtained, which can improve the prediction effect of the trained model to a certain extent.

For example, within the second preset time period before applying the tracer to the imaging part of the object to be tested, the imaging device may scan the imaging part of the object to be tested to obtain sample MR data. The processing device may segment the second preset time period to obtain multiple sub-time periods, and then reconstruct the sample MR data in each time period to obtain multi-frame sample MR images corresponding to the multiple sub-time periods. The duration of the sub-time periods corresponding to the sample MR images of each frame may be equal or unequal. Correspondence of sample MR images of each frame The corresponding time period when the sub-time periods are combined together may be equal to the second preset time period.

Step 304: Train an initial prediction model based on the multi-frame sample MR images and sample respiratory data, and determine a preset prediction model.

In some embodiments, the processing device can input the sample respiratory data into the initial prediction model and obtain the prediction results of the initial prediction model; the prediction results can be predicted dynamic-like MR images. The sample MR image corresponding to the sample respiratory data can be used as the gold standard, and a loss function is constructed based on the sample MR image and the predicted dynamic-like MR image to obtain the value of the loss function. The processing device can adjust the parameters of the initial prediction model based on the value of the loss function. For example, the processing device can adjust the parameters of the initial prediction model with minimizing the loss function value as the optimization goal. When the loss function value converges (minimizes) When, determine the preset prediction model.

The initial prediction model may be a model that has been trained at a certain stage. For example, the initial prediction model may be trained using MR images and respiratory data obtained by scanning normal subjects (eg, normal human individuals).

In some embodiments, by using the sample MR images and sample respiratory data of the subject to be tested to train the initial prediction model before using the prediction model to process the respiratory data of the subject to be tested, the trained model can be made more accurate for the subject to be tested. Strong pertinence, thus improving the prediction effect.

At the same time, the above image registration method can use the respiratory data collected synchronously with the dynamic PET data to obtain quasi-dynamic MR images with structural information, thus shortening the data collection time. At the same time, quasi-dynamic MR images can be obtained directly from the respiratory data. image, it is not necessary to first collect dynamic MR data, then reconstruct the dynamic MR data into dynamic MR images, and then perform indirect processing on the dynamic MR images to obtain a quasi-dynamic MR image, thereby reducing the data processing process of the entire image registration. , shortening the image registration time.

Figure 4 is an exemplary flowchart of determining model training according to other embodiments of this specification. In some embodiments, process 400 may be performed by a processing device (eg, processing device 120). For example, the process 400 can be stored in a storage device (such as a self-contained storage unit of a processing device or an external storage device) in the form of a program or instructions. When the program or instructions are executed, the process 400 can be implemented. Process 400 may include the following operations.

Step 402: Process the sample respiratory data based on the initial prediction model to obtain a prediction dynamic MR image.

Predictive dynamic MR images refer to images output by the initial prediction model.

In some embodiments, the processing device may input sample respiratory data into an initial prediction model for processing, and the prediction model outputs a prediction-like dynamic MR image.

Step 404: Determine whether preset conditions are met based on the predicted dynamic MR image and the sample MR image corresponding to the sample respiratory data.

The preset condition may be a condition that needs to be satisfied to predict the similarity between the dynamic-like MR image and the sample MR image corresponding to the sample respiratory data. For example, the similarity is less than a preset threshold, for example, the similarity is less than 80%, 90%, etc. The preset threshold can be determined according to actual needs.

In some embodiments, the processing device can determine the similarity between the predicted dynamic MR image and the corresponding sample MR image through various similarity calculation methods, and determine whether the similarity is satisfied based on the similarity and the preset threshold. Preset conditions.

When the similarity is greater than the preset threshold, that is, when the preset conditions are not met, it means that the prediction model has achieved good performance for the current object to be tested. In order to save the time required for the workflow, the initial prediction may not be made. The model is trained and updated.

When the similarity is less than the preset threshold, that is, when the preset conditions are met, step 406 can be performed to train and update the initial prediction model, improve the prediction performance of the model, lay a strong foundation for subsequent image registration, and improve image registration. accurate effect.

Step 406: When the preset condition is met, train and update the initial prediction model.

For instructions on training and updating the initial prediction model, please refer to the relevant description in Figure 3 and will not be repeated here.

In some embodiments, by judging the performance of the initial prediction model before using the prediction model to process the respiratory data of the subject to be tested, time can be saved and the performance of the initial prediction model can be improved if the performance of the initial prediction model meets the usage requirements. When the performance cannot meet the needs of use, the initial prediction model is trained and updated to improve the performance of the model, ensure the prediction effect of the model, and ensure the accuracy of subsequent image registration.

FIG. 5 is an exemplary flowchart of determining a result of a multi-frame dynamic PET image according to other embodiments of this specification. In some embodiments, process 500 may be performed by a processing device (eg, processing device 120). For example, the process 500 may be stored in a storage device (such as a built-in storage unit of a processing device or an external storage device) in the form of a program or instructions, and when executed, the process 500 may be implemented. Process 500 may include the following operations.

Step 502: Determine the time point corresponding to each frame type dynamic MR image within the preset time period.

The corresponding time point within the preset time period may be the respiratory data corresponding to each frame of dynamic MR image within the preset time period. Collection time. For example, collect data at the 1st minute, 5th minute, and 10th minute within the preset time period. In some embodiments, each frame-type dynamic MR image may correspond to multiple time points within a preset time period, and these multiple time points may constitute a time period.

In some embodiments, the processing device may determine the corresponding time point within the preset time period based on the acquisition time of the respiratory data corresponding to each frame of dynamic MR image.

Figure 9 shows the corresponding relationship diagram of the first preset time period, the second preset time period, the time point of administering the tracer, and the corresponding initial dynamic PET data, sample MR data and respiratory data on the same time axis. . The training sequence in Figure 9 can be the sample MR data of multiple sub-time periods collected when training the prediction model; sequence 1, sequence 2 and sequence 3 can be the sample MR data of the corresponding three sub-time periods after the tracer is administered. However, the sample MR data corresponding to the three sub-time periods after the tracer is administered does not need to be used in the image registration process.

Step 504: Determine PET data corresponding to the time points corresponding to the dynamic MR images of each frame type within the preset time period from the initial dynamic PET data.

In some embodiments, the processing device can find the same time point from the time points corresponding to the initial dynamic PET data according to the corresponding time points within the preset time period, and then convert the initial dynamic PET data corresponding to these time points into The data was determined to be the PET data.

Step 506: Reconstruct based on the PET data to determine the multi-frame dynamic PET image.

In some embodiments, the processing device can perform reconstruction based on the PET data based on various common image reconstruction algorithms to determine multi-frame dynamic PET images, which is not limited in this specification.

For a more detailed description of the reconstructed PET image, please refer to the relevant description in Figure 6 below.

6 is an exemplary flowchart of a method of determining registered dynamic PET images according to some embodiments of the present specification. In some embodiments, process 600 may be performed by a processing device (eg, processing device 120). For example, the process 600 can be stored in a storage device (such as a built-in storage unit of a processing device or an external storage device) in the form of a program or instructions, and when executed, the process 600 can be implemented. Process 600 may include the following operations.

Step 602: Based on the dynamic PET images of each frame type, obtain the corresponding initial dynamic PET data within the preset time period.

Each frame-type dynamic PET image within the preset time period corresponds to part of the initial dynamic PET data. Therefore, the processing device can be based on the sub-time period corresponding to each frame-type dynamic PET image and the sub-time period corresponding to each frame's initial dynamic PET data. Perform time mapping on the dynamic PET images of each frame type and the initial dynamic PET data of each frame to obtain the mapped initial dynamic PET data of the dynamic PET images of each frame type.

Step 604: Reconstruct based on the corresponding initial dynamic PET data within the preset time period to determine the multiple frames of initial dynamic PET images.

The processing device can reconstruct the initial dynamic PET data in multiple sub-time periods within the preset time period to obtain multiple frames of initial dynamic PET images corresponding to the preset time period. Each sub-time period corresponds to one frame of initial dynamic PET image. There is a one-to-one correspondence between the multi-frame initial dynamic PET image and the multi-frame dynamic PET image, and the correspondence can be determined by the time point of the PET data corresponding to the image. For example, if the quasi-dynamic PET image and the initial dynamic PET image have the same time point, it can be considered that there is a corresponding relationship between the two.

It should be noted that each frame of the initial dynamic PET image may correspond to the initial dynamic PET data in a sub-time period within the preset time period. Optionally, the preset time period can include multiple sub-time periods, each sub-time period has a corresponding frame of initial dynamic PET image; the duration corresponding to each sub-time period can be equal or unequal; multiple sub-time periods The combined duration may be less than or equal to the duration corresponding to the first preset time period.

In some embodiments, the processing device may statically reconstruct the initial dynamic PET data scanned by the imaging device to obtain a static PET image. In some embodiments, the processing device can also reconstruct the initial dynamic PET data scanned by the imaging device into a dynamic PET image, and a dynamic reconstruction algorithm can be used when reconstructing the dynamic PET image. Among them, the contrast of dynamic PET images is greater than the contrast of static PET images.

For example, the processing device can use the direct back-projection method, the iterative method or the Fourier transform reconstruction method to reconstruct the initial dynamic PET data of each frame type dynamic PET image, and obtain the initial dynamic PET image corresponding to each frame type dynamic PET image ( Or called mapping initial dynamic PET image).

For example, if there are two frames of quasi-dynamic PET images and four frames of initial dynamic PET images within the preset time period, the two frames of quasi-dynamic PET images are quasi-dynamic PET image 1 and quasi-dynamic PET image 2 respectively, and the four frames of initial dynamic PET images are respectively are the initial dynamic PET image 1, the initial dynamic PET image 2, the initial dynamic PET image 3, and the initial dynamic PET image 4. Among them, the sub-time period corresponding to the dynamic PET image 1 is [0, 2], and the sub-time period corresponding to the dynamic PET image 2 is The sub-time period of is [2, 4], the sub-time period corresponding to the initial dynamic PET image 1 is [0, 1], the sub-time period corresponding to the initial dynamic PET image 2 is [1, 2], the initial dynamic PET image The sub-time period corresponding to image 3 is [2, 3], and the sub-time period corresponding to the initial dynamic PET image 4 is [3, 4] (the data in the interval all represent time points), then the processing device can convert the initial dynamic PET image The sub-time period [0, 1] corresponding to 1 and the sub-time period [1, 2] corresponding to the initial dynamic PET image 2 are mapped to the sub-time period [0, 2] corresponding to the dynamic PET image 1, and the initial dynamic The sub-time period [2, 3] corresponding to PET image 3 and the sub-time period [3, 4] corresponding to the initial dynamic PET image 4 are mapped to the sub-time period [2, 4] corresponding to the dynamic PET image 2, respectively. The mapping of dynamic PET image 1 is the mapping of initial dynamic PET image 1 and the mapping of initial dynamic PET image 2, and the mapping of dynamic PET image 2 is the mapping of initial dynamic PET image 3 and the mapping of initial dynamic PET image 4.

Or, if the sub-time period corresponding to the quasi-dynamic PET image 1 in the preset time period is [0, 4], the sub-time period corresponding to the quasi-dynamic PET image 2 is [5, 9], and the sub-time period corresponding to the initial dynamic PET image 1 The sub-time period is [0, 1.5], the sub-time period corresponding to the initial dynamic PET image 2 is [2.5, 4], the sub-time period corresponding to the initial dynamic PET image 3 is [5, 6.5], and the corresponding sub-time period to the initial dynamic PET image 4 The sub-time period is [7.5, 9] (the data in the interval all represent time points), then the computer device can compare the sub-time period [0, 1.5] corresponding to the initial dynamic PET image 1 and the sub-time period corresponding to the initial dynamic PET image 2 The time period [2.5, 4] is mapped to the sub-time period [0, 4] corresponding to the dynamic PET image 1, and the sub-time period [5, 6.5] corresponding to the initial dynamic PET image 3 corresponds to the initial dynamic PET image 4 The sub-time period [7.5, 9] of is mapped to the sub-time period [5, 9] corresponding to the class dynamic PET image 2, respectively, the mapping initial dynamic PET image 1 of the class dynamic PET image 1 and the mapping initial dynamic PET image 2, class The mapped initial dynamic PET image 3 of the dynamic PET image 2 and the mapped initial dynamic PET image 4 .

In some embodiments, the duration of the sub-time periods corresponding to the initial dynamic PET images of each frame may be equal; the duration of the sub-time periods corresponding to the dynamic PET images of each frame is greater than the duration of the sub-time periods corresponding to the corresponding initial dynamic PET images of each frame. duration.

Step 606: Based on the registered dynamic-like PET image, register the multiple frames of initial dynamic PET images to determine the registered dynamic PET image.

In some embodiments, the processing device may group the registered dynamic-like PET images and the initial dynamic PET images based on the correspondence between the registered dynamic-like PET images and the multi-frame initial dynamic PET images. The images corresponding to the same acquisition time point are divided into one group, and then the registered quasi-dynamic PET images and the initial dynamic PET images in each group are registered to obtain the corresponding registered dynamic PET images.

In some embodiments, each frame of dynamic PET image has a corresponding registered dynamic PET image. Among them, the processing device can use the registered dynamic-like PET images of each frame in each group as reference images, and use an image registration algorithm to match the mapped initial dynamic PET images corresponding to the registered dynamic-like PET images of each frame. Accurately, the registered dynamic PET image is obtained.

The above image registration method can register the corresponding initial dynamic PET image (mapping the initial dynamic PET image) based on the registered dynamic PET image, so that the registered dynamic PET image can reflect structural information and obvious images. characteristics, thereby improving the accuracy of dynamic PET image registration results; at the same time, the above method can also allow medical staff to accurately obtain the pathological mechanism of the object to be tested from the registered dynamic PET images, further improving the accuracy of diagnosis and treatment methods. , and can take timely and effective treatment methods to treat the subjects to be tested.

In order to facilitate a better understanding of the technical solutions disclosed in this specification, taking the execution subject as the processing device as an example, the complete process of the image registration method provided in some embodiments of this specification may include:

(1) Obtain the respiratory data and initial dynamic PET data within the first preset time period after the subject is administered the tracer.

(2) Obtain multi-frame sample MR images and sample respiratory data within the second preset time period before the tracer is administered to the subject.

(3) Train the initial prediction model through multi-frame sample MR images and sample respiratory data to obtain the prediction model.

(4) Input the respiratory data into the prediction model to obtain corresponding multi-frame dynamic MR images within the first preset time period.

(5) Perform time mapping on the dynamic MR images of each frame type and the initial PET data within the first preset time period to obtain the mapped PET data of the dynamic MR images of each frame type.

(6) Reconstruct the mapped PET data to obtain the quasi-dynamic PET image corresponding to each frame of the quasi-dynamic MR image.

(7) Determine the reference image based on multi-frame sample MR images.

(8) Perform image registration on the dynamic MR images of each frame type through reference images to obtain the deformation field corresponding to the dynamic MR images of each frame type.

(9) According to the deformation fields corresponding to each frame of dynamic-like MR images, register the corresponding dynamic-like PET images to obtain registered dynamic-like PET images.

(10) Based on the dynamic PET images of each frame type, perform time mapping on the initial dynamic PET data of each frame within the first preset time period to obtain the mapped initial dynamic PET data of the dynamic PET images of each frame type.

(11) Reconstruct the mapped initial dynamic PET data of each frame type dynamic PET image to obtain the corresponding mapped initial dynamic PET image.

(12) By registering the registered quasi-dynamic PET image and the corresponding mapped initial dynamic PET image, we get Registered dynamic PET image.

For details on the execution processes of (1) to (12) above, please refer to the descriptions of the above embodiments. The implementation principles and technical effects are similar and will not be described again here.

It should be noted that the above relevant descriptions are only for examples and illustrations, and do not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to each process under the guidance of this specification. However, such modifications and changes remain within the scope of this specification. For example, changes to the process steps in this manual, such as adding preprocessing steps and storage steps, etc.

Figure 7 is an exemplary block diagram of an image registration system according to some embodiments of this specification. As shown in FIG. 7 , system 700 may include an acquisition module 710 , a first image acquisition module 720 , a second image acquisition module 730 , and a third image acquisition module 740 .

The acquisition module 710 can be used to acquire the respiratory data and initial dynamic PET data of the subject to be measured within a preset time period. In some embodiments, the acquisition module 710 is further configured to: perform a magnetic resonance scan to acquire multi-frame sample MR images and sample respiratory data of the object to be tested before the preset time period.

The first image acquisition module 720 may be configured to determine multi-frame dynamic MR images within the preset time period based on the respiratory data. In some embodiments, the first image acquisition module 720 is further configured to process the respiratory data through a preset prediction model to determine multi-frame dynamic MR images within the preset time period.

The second image acquisition module 730 may be configured to acquire multi-frame dynamic-like PET images corresponding to the multi-frame dynamic-like MR images based on the initial dynamic PET data, and determine the registered dynamic-like PET image based on each frame. Class dynamic PET image. In some embodiments, the second image acquisition module 730 is further configured to: determine the time point corresponding to each frame type dynamic MR image within the preset time period; determine from the initial dynamic PET data the corresponding time point of the dynamic MR image. PET data corresponding to the corresponding time point of each frame-type dynamic MR image within the preset time period; reconstruction is performed based on the PET data to determine the multi-frame type dynamic PET image. In some embodiments, the second image acquisition module 730 is further configured to: determine the deformation field between the dynamic-like MR image and the reference image in each frame; and compare the dynamic-like PET image in each frame based on the deformation field. Perform registration processing to determine the registered dynamic-like PET image.

The third image acquisition module 740 may be used to perform registration based on the registered dynamic-like PET image and the multi-frame initial dynamic PET image, and determine the registered dynamic PET image; wherein the multi-frame initial dynamic PET image Reconstruction is obtained based on the initial dynamic PET data. In some embodiments, the third image acquisition module 740 is further configured to: acquire corresponding initial dynamic PET data within the preset time period based on each frame type dynamic PET image; based on the corresponding initial dynamic PET data within the preset time period; The initial dynamic PET data of the multiple frames are reconstructed to determine the multiple frames of initial dynamic PET images; based on the registered quasi-dynamic PET images, the multiple frames of initial dynamic PET images are registered to determine the registered Dynamic PET images.

In some embodiments, the system 700 may further include a training module configured to: obtain multi-frame sample MR images and sample respiratory data of the object to be tested; based on the multi-frame sample MR images and sample Breathing data is used to train the initial prediction model and determine the preset prediction model. In some embodiments, the training module is further configured to: process the sample respiratory data based on the initial prediction model to obtain a predicted dynamic MR image; based on the predicted dynamic MR image and the sample respiratory data The corresponding sample MR image is used to determine whether the preset conditions are met; when the preset conditions are met, the initial prediction model is trained and updated. For detailed descriptions of each module of the above system, please refer to the corresponding process section of this manual, for example, the relevant descriptions of Figures 2 to 6, which will not be described again here.

It should be understood that the system and its modules shown in Figure 7 can be implemented in various ways. For example, in some embodiments, the system and its modules may be implemented in hardware, software, or a combination of software and hardware. Among them, the hardware part can be implemented using dedicated logic; the software part can be stored in the memory and executed by an appropriate instruction execution system, such as a microprocessor or specially designed hardware. Those skilled in the art will understand that the above-mentioned methods and systems can be implemented using computer-executable instructions and/or included in processor control code, for example on a carrier medium such as a disk, CD or DVD-ROM, such as a read-only memory (firmware). Such code is provided on a programmable memory or a data carrier such as an optical or electronic signal carrier. The system and its modules in this specification may not only be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc. , can also be implemented by, for example, software executed by various types of processors, or can also be implemented by a combination of the above-mentioned hardware circuits and software (for example, firmware).

It should be noted that the above description of the image registration system and its modules is only for convenience of description and does not limit this specification to the scope of the embodiments. It can be understood that for those skilled in the art, after understanding the principle of the system, it is possible to arbitrarily combine various modules or form a subsystem to connect with other modules without departing from this principle. For example, in some embodiments, the acquisition module 710, the first image acquisition module 720, the second image acquisition module 730 and the third image acquisition module 740 can be different modules in one system, or one module can implement the above two. one or more modules Function. For example, each module can share a storage module, or each module can have its own storage module. Such deformations are within the scope of this manual.

It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the possible beneficial effects may be any one or a combination of the above, or any other possible beneficial effects.

The basic concepts have been described above. It is obvious to those skilled in the art that the above detailed disclosure is only an example and does not constitute a limitation of this specification. Although not explicitly stated herein, various modifications, improvements, and corrections may be made to this specification by those skilled in the art. Such modifications, improvements, and corrections are suggested in this specification, and therefore such modifications, improvements, and corrections remain within the spirit and scope of the exemplary embodiments of this specification.

At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment," "an embodiment," and/or "some embodiments" means a certain feature, structure, or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that “one embodiment” or “an embodiment” or “an alternative embodiment” mentioned twice or more at different places in this specification does not necessarily refer to the same embodiment. . In addition, certain features, structures or characteristics in one or more embodiments of this specification may be appropriately combined.

Furthermore, those skilled in the art will appreciate that aspects of the specification may be illustrated and described in several patentable categories or circumstances, including any new and useful process, machine, product, or combination of matter, or combination thereof. any new and useful improvements. Accordingly, various aspects of this specification may be entirely executed by hardware, may be entirely executed by software (including firmware, resident software, microcode, etc.), or may be executed by a combination of hardware and software. The above hardware or software may be referred to as "data block", "module", "engine", "unit", "component" or "system". Additionally, aspects of this specification may be represented by a computer product including computer-readable program code located on one or more computer-readable media.

Computer storage media may contain a propagated data signal embodying the computer program code, such as at baseband or as part of a carrier wave. The propagated signal may have multiple manifestations, including electromagnetic form, optical form, etc., or a suitable combination. Computer storage media may be any computer-readable media other than computer-readable storage media that enables communication, propagation, or transfer of a program for use in connection with an instruction execution system, apparatus, or device. Program code located on a computer storage medium may be transmitted via any suitable medium, including radio, electrical cable, fiber optic cable, RF, or similar media, or a combination of any of the foregoing.

The computer program coding required to operate 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#, VB.NET, Python etc., conventional procedural programming languages such as C language, Visual Basic, Fortran 2003, Perl, COBOL 2002, PHP, ABAP, dynamic programming languages such as Python, Ruby and Groovy, or other programming languages. The program code may run entirely on the user's computer, as a stand-alone software package, or partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In the latter case, the remote computer can be connected to the user computer via any form of network, such as a local area network (LAN) or a wide area network (WAN), or to an external computer (e.g. via the Internet), or in a cloud computing environment, or as a service Use software as a service (SaaS).

In addition, unless explicitly stated in the claims, the order of the processing elements and sequences, the use of numbers and letters, or the use of other names in this specification are not intended to limit the order of the processes and methods in this specification. Although the foregoing disclosure discusses by various examples some embodiments of the invention that are presently considered useful, it is to be understood that such details are for purposes of illustration only and that the appended claims are not limited to the disclosed embodiments. To the contrary, rights The claims are intended to cover all modifications and equivalent combinations consistent with the spirit and scope of the embodiments of this specification. For example, although the system components described above can be implemented through hardware devices, they can also be implemented through software-only solutions, such as installing the described system on an existing server or mobile device.

Similarly, it should be noted that, in order to simplify the expression disclosed in this specification and thereby help understand one or more embodiments of the invention, in the previous description of the embodiments of this specification, multiple features are sometimes combined into one embodiment. accompanying drawings or descriptions thereof. However, this method of disclosure does not imply that the subject matter of the description requires more features than are mentioned in the claims. In fact, embodiments may have less than all features of a single disclosed embodiment.

In some embodiments, numbers are used to describe the quantities of components and properties. It should be understood that such numbers used to describe the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Grooming. Unless otherwise stated, "about," "approximately," or "substantially" means that the stated number is allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximations that may vary depending on the desired features of the individual embodiment. In some embodiments, numerical parameters should account for the specified number of significant digits and use general digit preservation methods. Although the numerical ranges and parameters used to identify the breadth of ranges in some embodiments of this specification are approximations, in specific embodiments, such numerical values are set as accurately as is feasible.

Each patent, patent application, patent application publication and other material, such as articles, books, instructions, publications, documents, etc. cited in this specification is hereby incorporated by reference into this specification in its entirety. It is inconsistent with the contents of this manual or the product Conflicting application history documents are excluded, as are documents (currently or later appended to this specification) that limit the broadest scope of the claims in this specification. It should be noted that if there is any inconsistency or conflict between the descriptions, definitions, and/or the use of terms in the accompanying materials of this manual and the content described in this manual, the descriptions, definitions, and/or the use of terms in this manual shall prevail. .

Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Accordingly, by way of example and not limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to those expressly introduced and described in this specification.

Claims (20)

1. An image registration method, the method includes:

   Obtain the respiratory data and initial dynamic PET data of the subject under test within a preset time period;

   Based on the respiratory data, determine a multi-frame dynamic MR image within the preset time period;

   Acquire multi-frame dynamic-like PET images corresponding to the multi-frame dynamic-like MR images based on the initial dynamic PET data, and determine the registered dynamic-like PET images based on the dynamic-like PET images of each frame;

   Registration is performed based on the registered dynamic-like PET image and the multi-frame initial dynamic PET image to determine the registered dynamic PET image; wherein the multi-frame initial dynamic PET image is reconstructed based on the initial dynamic PET data get.
2. The method of claim 1, wherein determining, based on the respiratory data, a multi-frame dynamic MR image within the preset time period includes:

   The respiratory data is processed through a preset prediction model to determine multi-frame dynamic MR images within the preset time period.
3. The method of claim 2, further comprising:

   Obtain multi-frame sample MR images and sample respiratory data of the subject to be tested;

   Based on the multi-frame sample MR images and sample respiratory data, an initial prediction model is trained to determine the preset prediction model.
4. The method according to claim 3, before training and updating the initial prediction model, the method further includes:

   Process the sample respiratory data based on the initial prediction model to obtain predicted dynamic MR images;

   Based on the predicted dynamic MR image and the sample MR image corresponding to the sample respiratory data, determine whether the preset conditions are met;

   When the preset conditions are met, the initial prediction model is trained and updated.
5. The method according to claim 3, obtaining multiple sample MR images and sample respiratory data of the subject to be tested includes:

   Before the preset time period, a magnetic resonance scan is performed to acquire multiple frames of sample MR images and sample respiratory data of the subject to be tested.
6. The method according to claim 1, said obtaining a multi-frame dynamic PET image corresponding to the multi-frame dynamic MR image based on the initial dynamic PET data, including:

   Determine the time point corresponding to each frame of dynamic MR image within the preset time period;

   Determine PET data corresponding to the time points corresponding to the dynamic MR images of each frame type within the preset time period from the initial dynamic PET data;

   Reconstruction is performed based on the PET data to determine the multi-frame dynamic PET image.
7. The method according to claim 1, wherein determining the registered dynamic-like PET image based on each frame of the dynamic-like PET image includes:

   Determine the deformation field between the dynamic-like MR image and the reference image in each frame;

   The dynamic-like PET images of each frame are registered based on the deformation field, and the registered dynamic-like PET images are determined.
8. The method according to claim 1, wherein the registration is performed based on the registered dynamic-like PET image and the multi-frame initial dynamic PET image, and the registered dynamic PET image is determined, including:

   Based on each frame type dynamic PET image, obtain the corresponding initial dynamic PET data within the preset time period;

   Perform reconstruction based on the corresponding initial dynamic PET data within the preset time period to determine the multi-frame initial dynamic PET images;

   Based on the registered dynamic-like PET image, the multiple frames of initial dynamic PET images are registered to determine the registered dynamic PET image.
9. According to the method of claim 1, the preset time period is a time period after the tracer is administered to the subject to be measured.
10. An image registration system, the system includes:

    The acquisition module is used to acquire the respiratory data and initial dynamic PET data of the subject to be measured within a preset time period;

    A first image acquisition module, configured to determine multi-frame dynamic MR images within the preset time period based on the respiratory data;

    The second image acquisition module is configured to acquire multi-frame dynamic quasi-PET images corresponding to the multi-frame dynamic quasi-MR images based on the initial dynamic PET data, and determine the registered quasi-dynamic PET image based on each frame of the quasi-dynamic PET image. Dynamic PET images;

    The third image acquisition module is used to perform registration based on the registered dynamic PET image and the multi-frame initial dynamic PET image, and determine the registered dynamic PET image; wherein the multi-frame initial dynamic PET image is based on The initial dynamic PET data is reconstructed and obtained.
11. According to the system of claim 10, the first image acquisition module is further used for:

    The respiratory data is processed through a preset prediction model to determine multi-frame dynamic MR images within the preset time period.
12. The system according to claim 11, said system further comprising a training module, said training module being used for:

    Obtain multi-frame sample MR images and sample respiratory data of the subject to be tested;

    Based on the multi-frame sample MR images and sample respiratory data, an initial prediction model is trained to determine the preset prediction model.
13. According to the system of claim 12, the training module is further used for:

    Process the sample respiratory data based on the initial prediction model to obtain predicted dynamic MR images;

    Based on the predicted dynamic MR image and the sample MR image corresponding to the sample respiratory data, determine whether the preset conditions are met;

    When the preset conditions are met, the initial prediction model is trained and updated.
14. According to the system of claim 12, the acquisition module is further used for:

    Before the preset time period, a magnetic resonance scan is performed to acquire multiple frames of sample MR images and sample respiratory data of the subject to be tested.
15. According to the system of claim 10, the second image acquisition module is further used for:

    Determine the time point corresponding to each frame of dynamic MR image within the preset time period;

    Determine PET data corresponding to the time points corresponding to the dynamic MR images of each frame type within the preset time period from the initial dynamic PET data;

    Reconstruction is performed based on the PET data to determine the multi-frame dynamic PET image.
16. According to the system of claim 10, the second image acquisition module is further used for:

    Determine the deformation field between the dynamic-like MR image and the reference image in each frame;

    The dynamic-like PET images of each frame are registered based on the deformation field, and the registered dynamic-like PET images are determined.
17. According to the system of claim 10, the third image acquisition module is further used for:

    Based on each frame type dynamic PET image, obtain the corresponding initial dynamic PET data within the preset time period;

    Perform reconstruction based on the corresponding initial dynamic PET data within the preset time period to determine the multi-frame initial dynamic PET images;

    Based on the registered dynamic-like PET image, the multiple frames of initial dynamic PET images are registered to determine the registered dynamic PET image.
18. According to the system of claim 10, the preset time period is a time period after the tracer is administered to the object to be measured.
19. An image registration device, including at least one storage medium and at least one processor, the at least one storage medium is used to store computer instructions; the at least one processor is used to execute the computer instructions to implement claims 1-9 The image registration method described in any one of the above.
20. A computer-readable storage medium that stores computer instructions. After the computer reads the computer instructions in the storage medium, the computer executes the image registration method according to any one of claims 1-9.