CN112052885A

CN112052885A - Image processing method, device and equipment and PET-CT system

Info

Publication number: CN112052885A
Application number: CN202010845685.5A
Authority: CN
Inventors: 王亮; 王浩; 李明
Original assignee: Neusoft Medical Systems Co Ltd
Current assignee: Shenyang Zhihe Medical Technology Co ltd
Priority date: 2020-08-20
Filing date: 2020-08-20
Publication date: 2020-12-08
Anticipated expiration: 2040-08-20
Also published as: CN112052885B

Abstract

The present disclosure relates to an image processing method, apparatus, device and PET-CT system, so as to provide a new method for obtaining a plasma input function, thereby performing parameter reconstruction on a PET image according to the plasma input function more efficiently. The image processing method comprises the following steps: acquiring a Positron Emission Tomography (PET) image sequence; clustering according to the time variation characteristics of the radioactivity activity corresponding to the plasma points in the PET image sequence within a preset time length to obtain plasma point classes for representing the plasma in the scanned object; and performing function fitting according to the time change characteristics of the radioactivity activity corresponding to each voxel point in the plasma voxel point classes within a preset time length to obtain a plasma input function, wherein the plasma input function is used for performing parameter reconstruction on the PET image sequence to obtain a parameter image for representing the in-vivo metabolic rate of the scanning object.

Description

Image processing method, device and equipment and PET-CT system

Technical Field

The present disclosure relates to the field of positron emission computed tomography, and in particular, to an image processing method, apparatus, device, and PET-CT system.

Background

The PET-CT system is an imaging device combining two imaging devices of PET (Positron Emission Tomography) and CT (Computed Tomography). Each pixel point of the PET image generated by the PET-CT system represents a Standard Uptake Value (SUV) which represents a ratio of the radioactivity of the local tissue tracer to the radioactivity of the whole body. In order to visually check the human body metabolic rate, parameter reconstruction is usually performed on the PET image to obtain a parameter image for representing the human body metabolic rate index. The parameter reconstruction takes the radioactivity in the plasma as an input function, and a parameter image is obtained by utilizing the data of multiple PET scans. Therefore, the acquisition of the plasma input function is an important step in the parametric reconstruction.

In the related art, the plasma input function can be obtained by periodically sampling blood of a scanning object and acquiring the content of radioactive substances in the blood through a counter, or by searching empirical values in the literature. The former method needs to continuously take blood from a scanned object in the scanning process, and a gamma counter is generally needed to obtain the content of radioactive substances in the blood, so that the cost is high, and the method is not suitable for wide application. The latter approach does not take into account individual differences of the scanned objects, and the same plasma input function may be obtained for different scanned objects, thereby affecting the accuracy of the parameter reconstruction result.

Disclosure of Invention

The invention aims to provide an image processing method, an image processing device, an image processing equipment and a PET-CT system, so as to provide a new method for acquiring a plasma input function, and further more efficiently perform parameter reconstruction on a PET image according to the plasma input function.

In order to achieve the above object, in a first aspect, the present disclosure provides an image processing method, the method comprising:

acquiring a Positron Emission Tomography (PET) image sequence, wherein the PET image sequence comprises a plurality of three-dimensional PET images obtained by scanning the same part of a scanned object at different scanning moments within a preset time length, and each voxel point in each three-dimensional PET image is provided with radioactivity;

clustering according to the time variation characteristics of the radioactivity activity corresponding to the plasma points in the PET image sequence within the preset time length to obtain plasma point classes for representing the plasma in the scanning object;

and performing function fitting according to the time change characteristics of the radioactivity activity corresponding to each voxel point in the plasma voxel point classes in the preset time length to obtain a plasma input function, wherein the plasma input function is used for performing parameter reconstruction on the PET image sequence to obtain a parameter image for representing the in-vivo metabolic rate of the scanned object.

Optionally, the method further comprises:

acquiring a three-dimensional Computed Tomography (CT) image aiming at the scanning object, wherein the scanning part corresponding to the CT image is the same as the scanning part corresponding to the PET image sequence;

performing semantic segmentation processing on the CT image, and determining a first target region including the scanning part in the CT image;

the clustering according to the time variation characteristics of the radioactivity activity corresponding to the voxel points in the PET image sequence within the preset time duration comprises the following steps:

for each three-dimensional PET image in the PET image sequence, mapping the first target region in the CT image to the three-dimensional PET image to obtain a second target region including the scanning part in the three-dimensional PET image;

and clustering according to the time change characteristics of the radioactivity corresponding to the voxel points in the second target region within the preset time length.

Optionally, the method further comprises:

performing semantic segmentation processing on the CT image, and determining a non-plasma region corresponding to the non-plasma of the scanning part in the CT image;

performing function fitting according to the time variation characteristics of the radioactivity of each voxel point in the plasma voxel point classes within the preset time length to obtain a plasma input function, wherein the function fitting comprises the following steps:

for each three-dimensional PET image in the PET image sequence, mapping the non-plasma region into the three-dimensional PET image to obtain a target non-plasma region in the three-dimensional PET image;

performing correction processing on voxel points included in the plasma voxel point class to remove voxel points belonging to the target non-plasma region in the plasma voxel point class;

and performing function fitting according to the time change characteristics of the radioactivity of each voxel point in the plasma voxel point classes within the preset time length after correction processing to obtain a plasma input function.

Optionally, the PET image sequence includes T three-dimensional PET images obtained at T different scanning times, each three-dimensional PET image includes N voxel points, where T and N are positive integers, and an image feature of each voxel point is a spatial feature used for characterizing a position of the voxel point in the scanning portion, and the clustering is performed according to a time variation feature of the radioactivity corresponding to the voxel point in the PET image sequence within the preset time duration, including:

performing feature conversion on voxel points included in the PET image sequence to obtain N sample points, wherein each sample point has T time features, and the T time features are used for representing the radioactivity of the sample points at T scanning moments;

and clustering according to the T time characteristics of each sample point in the N sample points.

Optionally, when the clustering results in at least three clustering results including the plasma voxel point class, the method further comprises:

determining the plasma body point class used for characterizing the plasma in the body of the scanning object and a non-plasma body point class used for characterizing the tissue in the body of the scanning object according to the time variation characteristics of each voxel point in the at least three clustering results;

for each voxel point in other voxel point classes except the plasma voxel point class and the non-plasma voxel point class in the at least three clustering results, if the voxel point in the preset domain of the voxel point belongs to the plasma voxel point, determining that the voxel point belongs to the plasma voxel point class, and if the voxel point in the preset domain of the voxel point belongs to the non-plasma voxel point, determining that the voxel point belongs to the non-plasma voxel point class.

Optionally, the method further comprises:

preserving said plasma pixel population;

obtaining the preserved plasma pigment point class;

and performing function fitting according to the acquired time change characteristics of the radioactivity of each voxel point in the plasma voxel point class in the preset time length to obtain the plasma input function.

In a second aspect, the present disclosure also provides an image processing apparatus, the apparatus comprising:

the system comprises an acquisition module, a processing module and a display module, wherein the acquisition module is used for acquiring a Positron Emission Tomography (PET) image sequence, the PET image sequence comprises a plurality of three-dimensional PET images obtained by scanning the same part of a scanned object at different scanning moments within a preset time length, and each voxel point in each three-dimensional PET image corresponds to radioactivity;

the clustering module is used for clustering according to the time variation characteristics of the radioactivity activity corresponding to the plasma points in the PET image sequence within the preset time length to obtain plasma point classes for representing the plasma in the scanning object;

and the fitting module is used for performing function fitting according to the time change characteristics of the radioactivity activity corresponding to each voxel point in the plasma voxel point classes in the preset time length to obtain a plasma input function, and the plasma input function is used for performing parameter reconstruction on the PET image sequence to obtain a parameter image for representing the in-vivo metabolic rate of the scanned object.

In a third aspect, the present disclosure also provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the method of any one of the first aspect.

In a fourth aspect, the present disclosure also provides an electronic device, including:

a memory having a computer program stored thereon;

a processor for executing the computer program in the memory to implement the steps of the method of any one of the first aspect.

In a fifth aspect, the present disclosure also provides a positron emission tomography PET-CT system, comprising: the radiation source, the first detector, the second detector and the electronic device of the fourth aspect;

the ray source is used for emitting rays;

the first detector is used for detecting attenuated ray signals after passing through a scanned object, converting the attenuated ray signals into electric signals and sending the electric signals to the electronic equipment to obtain a CT image;

the second detector is used for detecting high-energy photons emitted from the body of the scanned object, converting the high-energy photons into pulse signals and sending the pulse signals to the electronic equipment so as to obtain a PET image sequence.

Through the technical scheme, the function fitting can be carried out according to the time variation characteristics of the radioactivity activity corresponding to the voxel points in the PET image sequence within the preset time length, the plasma input function is obtained, the whole process can be automatically executed after the PET image sequence is obtained, artificial participation is not needed, and therefore the efficiency of obtaining the plasma input function is improved. In addition, in the method, the PET image sequence corresponding to the scanning object is processed to further obtain the plasma input function, the individual difference among the scanning objects is considered, the plasma input function which is more in line with the actual situation of the scanning object can be obtained, and therefore the result error caused by the fact that the individual difference of the scanning objects is not considered in the subsequent parameter reconstruction process is reduced.

Additional features and advantages of the disclosure will be set forth in the detailed description which follows.

Drawings

The accompanying drawings, which are included to provide a further understanding of the disclosure and are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and together with the description serve to explain the disclosure without limiting the disclosure. In the drawings:

FIG. 1 is a schematic diagram of a dynamic compartmental model for parametric reconstruction in the related art;

FIG. 2 is a schematic diagram of a PET-CT system for implementing the image processing method in embodiments of the present disclosure;

FIG. 3 is a flow chart illustrating a method of image processing according to an exemplary embodiment of the present disclosure;

FIG. 4 is a graph of the change in radioactivity of plasma and tissue over time in humans;

FIG. 5 is a schematic diagram of a neural network structure for performing semantic segmentation processing on a CT image in an image processing method according to an exemplary embodiment of the present disclosure;

FIG. 6 is a flowchart illustrating an image processing method according to another exemplary embodiment of the present disclosure;

FIG. 7 is a block diagram illustrating an image processing apparatus according to an exemplary embodiment of the present disclosure;

fig. 8 is a block diagram illustrating an electronic device according to an exemplary embodiment of the present disclosure.

Detailed Description

The following detailed description of specific embodiments of the present disclosure is provided in connection with the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present disclosure, are given by way of illustration and explanation only, not limitation.

The PET-CT system is an imaging device combining two imaging devices of PET (Positron Emission Tomography) and CT (Computed Tomography). The PET receives a pair of gamma photons with opposite directions, wherein the gamma photons are generated by annihilation of a positron generated in the decay process of a radionuclide injected into a human body and a free electron in the human body, and the distribution condition of a tracer in the human body can be deduced through a series of processing, so that the changes of physiology, pathology, biochemistry, metabolism and the like of human tissues are reflected on the molecular level. CT utilizes the characteristic that various tissues of a human body have different absorption capacities on X rays, obtains an image matrix through the attenuation of the X rays in the human body and the reconstruction calculation, and has higher density resolution ratio on the tissues. In the PET-CT system, PET is used to reflect tissue physiological uptake and CT is used to reflect organ structure.

However, each pixel point of the PET image generated by the PET-CT system represents a Standard Uptake Value (SUV), which represents the ratio of the radioactivity of the local tissue tracer to the radioactivity of the whole body. In order to visually check the human body metabolic rate, parameter reconstruction is usually performed on the PET image to obtain a parameter image for representing the human body metabolic rate index.

The parameter reconstruction is based on a dynamic compartment model as a theoretical basis, the compartment model divides the body into one or more independent units (tissues with the same drug transport rate are divided into the same compartment), and the characteristics of absorption, distribution, elimination and the like of the drug in the body can be simulated. The models are often classified into a one-room model, a two-room model and a three-room model according to the complexity of the models. FIG. 1 shows a three-compartment model, where k is₁Denotes the rate of FDG (fluorodeoxyglucose) entering the tissue from the blood, k₂Representing the rate of FDG return from the tissue into the blood, k₃Denotes the rate at which FDG in tissue is phosphorylated to FDG-6-P catalyzed by phosphohexokinase, k₄Represents the rate at which FDG-6-P is converted back to FDG by glucose-6-phosphatase, k₁、k₂、k₃、k₄Referred to as individual rate constants or micro-parameters, whose dimension is the inverse of the time dimension.

In the process of parameter reconstruction, the radioactivity in plasma is generally used as an input function, and k is used₃Or k₃In combination with other rate constants as a parametric image. It can be seen that the acquisition of the plasma input function is a parameterAn important step in reconstruction.

The inventor researches and discovers that in the related art, the plasma input function can be obtained by taking blood of a scanning object regularly and acquiring the content of radioactive substances in the blood through a counter, or by searching empirical values in the literature. The former method needs to continuously take blood from a scanned object in the scanning process, and a gamma counter is generally needed to obtain the content of radioactive substances in the blood, so that the cost is high, and the method is not suitable for wide application. The latter approach does not take into account individual differences of the scanned objects, and the same plasma input function may be obtained for different scanned objects, thereby affecting the accuracy of the parameter reconstruction result.

In addition, the inventor also researches and discovers that the related art can obtain the plasma input function by measuring the activity change curve of the blood of the aorta or the heart in the pre-reconstructed dynamic image, but the method needs to artificially pre-mark the aorta or the heart area, and only can mark a part of the aorta or the heart area, thereby leading to higher signal-to-noise ratio of the calculated plasma input function.

In view of this, embodiments of the present disclosure provide an image processing method, an image processing apparatus, an image processing device, and a PET-CT system, so as to provide a new way for acquiring a plasma input function, reduce human involvement in the process of acquiring the plasma input function, and achieve automatic acquisition of the plasma input function.

First, an implementation scenario of the image processing method provided by the embodiment of the present disclosure is exemplarily described. FIG. 2 is a schematic diagram of a PET-CT system for implementing the image processing methods of the present disclosure. Referring to fig. 2, the PET-CT system may include: a radiation source 11, a first detector 12, a second detector 13, electronics 14 and other relevant components shown in fig. 2 but not labeled. Wherein the radiation source 11 can be used for emitting radiation, and the detector 12 can be used for detecting attenuated radiation signals after passing through the scanned object, and converting the attenuated radiation signals into electric signals to be sent to the electronic device 14 to obtain a CT image. The second detector 13 is used to detect high-energy photons emitted from the body of the scanned object and convert them into pulse signals which are sent to the electronic device 14 to obtain a PET image sequence. The electronic device 14 may be configured to execute the image processing method provided by the embodiment of the present disclosure to obtain a plasma input function, so as to implement parametric reconstruction of a PET image sequence.

Fig. 3 is a flowchart illustrating an image processing method according to an exemplary embodiment of the present disclosure. Referring to fig. 3, the image processing method may include:

step 301, acquiring a Positron Emission Tomography (PET) image sequence. The PET image sequence comprises a plurality of three-dimensional PET images obtained by scanning the same part of a scanning object at different scanning moments within a preset time length, and each voxel point in each three-dimensional PET image is provided with radioactivity.

And step 302, clustering according to the time variation characteristics of the radioactivity activity corresponding to the plasma points in the PET image sequence within a preset time length to obtain plasma point classes for representing the plasma in the scanned object.

And 303, performing function fitting according to the time change characteristics of the radioactivity activity corresponding to each voxel point in the plasma voxel point classes within a preset time length to obtain a plasma input function, wherein the plasma input function is used for performing parameter reconstruction on the PET image sequence to obtain a parameter image for representing the in-vivo metabolic rate of the scanned object.

By the mode, function fitting can be carried out according to the time variation characteristics of the radioactivity activity corresponding to the voxel points in the PET image sequence within the preset time length, the plasma input function is obtained, the whole process can be automatically executed after the PET image sequence is obtained, artificial participation is not needed, and therefore the efficiency of obtaining the plasma input function is improved. In addition, in the method, the PET image sequence corresponding to the scanning object is processed to further obtain the plasma input function, the individual difference among the scanning objects is considered, the plasma input function which is more in line with the actual situation of the scanning object can be obtained, and therefore the result error caused by the fact that the individual difference of the scanning objects is not considered in the subsequent parameter reconstruction process is reduced.

In order to make those skilled in the art understand the multi-sequence scanning method provided by the embodiments of the present disclosure, the following describes the above steps in detail.

It should be understood at first that the specific acquisition process of the PET image sequence is not limited by the embodiments of the present disclosure, and in practical applications, the scan parameters for acquiring the PET image sequence may be set according to different scan requirements. The PET image sequence may include a plurality of three-dimensional PET images obtained by scanning the same portion of the scan object at different scanning times within a preset time period. That is, each three-dimensional PET image shows the same scanning region, but the distribution of the tracer in the scanning region is different in each three-dimensional PET image due to the different scanning time, that is, the radioactivity corresponding to each three-dimensional PET image is different.

It should also be understood that each voxel point in each three-dimensional PET image corresponds to a radioactivity. The voxel points in the three-dimensional PET image are equivalent to pixel points in the two-dimensional image and are the minimum representation units of the three-dimensional PET image. The radioactivity at the voxel point locations is used to characterize the radioactivity at the voxel point locations. Plasma voxel points for characterizing plasma in the body of the scanned subject and tissue voxel points (i.e., non-plasma voxel points) for characterizing tissue in the body of the scanned subject may be included in the three-dimensional PET image.

The inventor researches and finds that the change characteristic of the radioactivity activity in the plasma with time is obviously different from the change characteristic of the radioactivity activity in the tissue with time. For example, the change in radioactivity over time in plasma can be characterized as curve A in FIG. 4, while the change in radioactivity over time in tissue can be characterized as curve B in FIG. 4. Referring to FIG. 4, the radioactivity in plasma is shown at T₀And T₁The time is higher, the rest time is reduced, and the radioactivity in the tissue is T₀And T₁The time is lower, and the rest times are higher, which are obviously different. Therefore, after the PET image sequence is acquired, the embodiments of the present disclosure can perform clustering according to the time variation characteristics of the radioactivity corresponding to the plasma points in the PET image sequence within the preset time duration, so as to obtain plasma point classes for characterizing the plasma in the scanned object. Then, each voxel point in the plasma voxel point classes is used as a basisAnd performing function fitting on the corresponding time change characteristics of the radioactivity activity to obtain a plasma input function.

In a possible manner, the PET image sequence may include T three-dimensional PET images obtained at T different scanning times, each three-dimensional PET image may include N voxel points, where T and N are positive integers, and an image feature of each voxel point is a spatial feature used for characterizing a position of the voxel point in a scanning portion, and accordingly, the clustering may be performed according to a time variation feature of a radioactivity activity corresponding to the voxel point in the PET image sequence within a preset time duration, where: firstly, performing feature conversion on voxel points included in a PET image sequence to obtain N sample points, wherein each sample point has T time features, the T time features are used for representing the radioactivity of the sample points at T scanning moments, and then clustering is performed according to the T time features of each sample point in the N sample points.

In the embodiment of the present disclosure, the clustering is performed to obtain voxel points in the three-dimensional PET image for characterizing plasma in the scanned object, and as can be seen from the above, the change characteristic of radioactivity in plasma with time is significantly different from the change characteristic of radioactivity in tissue with time. Therefore, in order to improve the clustering efficiency, before clustering, feature transformation may be performed on the PET image sequence to obtain the radioactivity activity temporal feature of each voxel point in the PET image sequence.

Illustratively, the initial PET image sequence includes T three-dimensional PET images obtained at T different scan times, each three-dimensional PET image includes N voxel points, and the image feature of each voxel point is a spatial feature used to characterize the position of the voxel point in the scan region. During feature conversion, voxel points representing the same position of a scanning part in a PET image sequence can be traversed in sequence, and the radioactivity of the voxel points at T different scanning moments is obtained. Thus, after feature conversion, N sample points can be obtained, each sample point corresponding to the radioactivity at T different scan instants, i.e. each sample point corresponding to T temporal features.

Then, clustering is performed according to the T time characteristics corresponding to each sample point in the N sample points, so as to cluster sample points whose change characteristics of the radioactivity activity over time conform to the curve a in fig. 4 into one class, and cluster sample points whose change characteristics of the radioactivity activity over time conform to the curve B in fig. 4 into another class. For example, the clustering may adopt any clustering algorithm in the related art, and the embodiment of the present disclosure does not limit this. For example, a hierarchical clustering algorithm may be adopted, each sample point is initially and separately classified into one class, and two closest sample points are iteratively combined to form a new class until a convergence condition is satisfied (for example, the iteration number reaches a preset number). Or, a K-means clustering algorithm may be adopted, K clustering centers are pre-specified for iterative clustering, and clustering is stopped until a convergence condition is satisfied, and so on.

After clustering in the above manner, a plasma dot class for characterizing plasma in the scanned subject can be obtained. And because the conversion from the spatial characteristic to the time characteristic is carried out before the clustering, the clustering efficiency is improved, so that the subsequent execution process can be more quickly carried out, the whole image processing efficiency is improved, and the plasma input function is more efficiently obtained.

In practical applications, the clustering results may be more than two, that is, besides obtaining a plasma point class and a non-plasma point class for characterizing tissue, other clustering results may be obtained, in which plasma points may also be present. Therefore, it is necessary to further determine whether the voxel points included in the other clustering results belong to the plasma voxel points.

In a possible manner, when the clustering result includes at least three clustering results of plasma pixel point classes, a plasma pixel point class for characterizing plasma in the body of the scanning object and a non-plasma pixel point class for characterizing non-plasma in the body of the scanning object may be determined according to the time variation characteristics of each pixel point in the at least three clustering results. Then, for each voxel point in the other voxel points except the plasma voxel point class and the non-plasma voxel point class in the at least three clustering results, if the voxel point in the preset domain of the voxel point belongs to the plasma voxel point, determining that the voxel point belongs to the plasma voxel point class, and if the voxel point in the preset domain of the voxel point belongs to the non-plasma voxel point, determining that the voxel point belongs to the non-plasma voxel point class.

Illustratively, a neighborhood of a point is defined as the set of points inside or on the boundary of a circle centered on the point. The preset neighborhood of the voxel point can be set according to actual conditions, and the embodiment of the disclosure does not limit the preset neighborhood. For example, the preset domain may be set to 4 neighborhoods around the voxel point, 8 domains, or the like. In addition, the voxel points in the preset region of voxel points may be all voxel points in the preset neighborhood of voxel points, or may also be some voxel points in the preset neighborhood of voxel points, which is not limited in this embodiment of the disclosure.

For example, the preset neighborhood is an 8-neighborhood, and the clustering results in three clustering results, that is, all sample points are clustered into three classes. Wherein, T time characteristics corresponding to the sample points in the first class of results can be obviously seen₀Eigenvalues and T₁The characteristic values are all higher, and accord with the characteristics of the plasma body element points. And in T time characteristics corresponding to the sample points in the first class of results, T₀Eigenvalues and T₁The eigenvalues are all lower, conforming to the characteristics of the non-plasma voxel points. For the rest of classes, it can be determined whether 6 voxel points in the voxel point 8 field belong to plasma voxel points or not for each voxel point in the class, that is, whether the change characteristic of the radioactivity activity over time corresponding to the partial voxel points in the preset neighborhood of the voxel point conforms to the time change characteristic of the radioactivity activity of the plasma voxel points or not. If 6 voxel points in the 8 fields of the voxel points belong to plasma voxel points, the voxel points are determined to belong to the plasma voxel point class, otherwise, the voxel points are determined to represent tissues in the body of the scanning object and belong to non-plasma voxel points.

By the method, when at least three clustering results are obtained by clustering, whether the voxel point belongs to the plasma voxel point or not can be determined according to whether the voxel point in the preset field of the voxel point belongs to the plasma voxel point, so that the at least three clustering results can be divided into two classification results of the plasma voxel point class and the non-plasma voxel point class, the voxel point of plasma in a scanned object, which is actually represented, is avoided being omitted in the function fitting process, the accuracy of the result of subsequent function fitting is improved, and a more accurate plasma input function is obtained.

In practical cases, each three-dimensional PET image included in the PET image sequence may include a human body region and a non-human body region, and the non-human body region is determined not to include a plasma voxel point. If the human body region and the voxel point corresponding to the human body region are used as the initial voxel point of the clustering together in the clustering process, more invalid clustering processes may be caused, and the clustering efficiency is affected. Therefore, in order to improve the calculation efficiency of clustering, the voxel points corresponding to the human body region in the three-dimensional PET image may be clustered.

In a possible embodiment, a three-dimensional CT image can be acquired for the scanning object, which CT image corresponds to the same scanning region as the PET image sequence. Then, semantic segmentation processing is carried out on the CT image, and a first target region including the scanning part in the CT image is determined. Correspondingly, clustering is performed according to the time variation characteristics of the radioactivity corresponding to the voxel points in the PET image sequence within the preset time duration, which may be: the method comprises the steps of firstly mapping a first target region in a CT image to a three-dimensional PET image aiming at each three-dimensional PET image in a PET image sequence to obtain a second target region including a scanning part in the three-dimensional PET image, and then clustering according to the time change characteristics of the radioactivity corresponding to a voxel point in the second target region within a preset time length.

It should be understood that the CT image can accurately reflect the structure of the human organ, and therefore, performing semantic segmentation processing on the CT image can obtain a more accurate first target region for characterizing the scanned portion of the scanned object. In the PET-CT system, the CT image and the PET image sequence scan the same scanning region as the scanning target, that is, the scanning region displayed by the CT image is the same as the scanning region displayed by the PET image sequence. Therefore, after obtaining the first target region for characterizing the scanning portion of the scanning object in the CT image, the first target region may be mapped into each three-dimensional PET image in the PET image sequence, and the second target region for characterizing the scanning portion of the scanning object in each three-dimensional PET image is obtained. Under the condition, subsequent clustering can be carried out on the voxel points in the second target region, and all voxel points in the three-dimensional PET image are not required to be subjected to, so that the clustering efficiency can be effectively improved, the overall efficiency of image processing is improved, and the plasma input function can be obtained more quickly.

For example, the CT image may be processed by any semantic segmentation processing algorithm in the related art, so as to obtain a first target region including a scanning portion in the CT image, which is not limited in this disclosure. For example, a U-shaped neural network (U-Net) as shown in fig. 5 may be used to perform semantic segmentation processing on the CT image. Referring to fig. 5, the U-shaped neural network uses 3-dimensional CT data as input, performs downsampling through a plurality of convolution and pooling operations to obtain image features, restores the image size through deconvolution operations and retains image features as much as possible, and simultaneously combines a feature map in the downsampling process directly with a deconvolution feature map, so that two aspects of features of the image and the contour can be retained simultaneously in the upsampling process, and the segmentation accuracy is improved. It should be understood that the structure shown in fig. 5 is only a schematic structure, and details such as the number of layers in the network may be modified. Of course, other semantic segmentation algorithms may be used, such as Full Convolutional Network (FCN), Residual Network (ResNet), and so on.

After the first target region is obtained by semantic segmentation processing, the mapping of the first target region into the three-dimensional PET image may be: firstly, carrying out voxel point coordinate alignment on the CT image and the three-dimensional PET image. After the voxel point coordinates are aligned, the CT image and the three-dimensional PET image have one-to-one correspondence with each other, so that in the mapping process, a voxel point which is consistent with the voxel point coordinates in the first target region can be searched in the three-dimensional PET image. And the searched region formed by all the voxel points is a second target region including the scanning part in the three-dimensional PET image.

It should be understood that, in order to further improve the accuracy of the clustering result, in a possible manner, an initial target region may be determined in the three-dimensional PET image according to the first target region of the CT image, and then for each edge voxel point of the initial target region, other voxel points in the neighborhood of the edge voxel point may be determined. And then determining a region formed by all the voxel points included in the initial target region and all other voxel points in the neighborhood of the edge voxel points as a second target region including the scanning part in the three-dimensional PET image. Under the condition, the range of the second target area is larger than that of the first target area, so that the clustering operation amount can be reduced, the number of clustered voxel points is not too small, and a more accurate clustering result is obtained.

After the clustering is carried out in the mode, plasma voxel point classes used for representing the plasma in the scanned object can be obtained, and then function fitting can be carried out according to the time change characteristics of the radioactivity activity of each voxel point in the plasma voxel point classes within the preset time length, so that a plasma input function is obtained. It should be understood that the disclosed embodiments do not limit the function used for fitting.

In the plasma voxel point class, the radioactivity associated with each voxel point at each scan time is known. If there are M (M is a positive integer) points in the plasma voxel point class and Y passes through₀、Y₁、……、Y_MAre respectively shown at T₀、T₁、……、T_MThe radioactivity at the time of the scan corresponds to the following data points: (T)₀，Y₀)、(T₁，Y₁)、……、(T_M，Y_M) A function can then be fitted to these data points to obtain the plasma input function. For example, the fitting may be performed by the following function:

where t denotes the scanning time, τ, A₁、A₂、A₃、λ₁、λ₂、λ₃And λ₄The coefficients to be fitted are characterized.

It should be understood that the fitting process may be to substitute the obtained data points into the function to be fitted, determine the coefficients of the function, and obtain the plasma input function, which is similar to the function fitting process in the related art and will not be described herein again.

In a possible embodiment, in order to improve the accuracy of the function fitting to obtain the plasma input function, a three-dimensional CT image of the scan object may be acquired, the CT image corresponding to the same scan region as the PET image sequence. Then, semantic segmentation processing is performed on the CT image, and a non-plasma region corresponding to the non-plasma of the scanning part in the CT image is determined. Correspondingly, performing function fitting according to the time variation characteristics of the radioactivity of each voxel point in the plasma voxel point class within a preset time length to obtain a plasma input function, which can be: the method comprises the steps of mapping a non-plasma region to a three-dimensional PET image in the PET image sequence aiming at each three-dimensional PET image to obtain a target non-plasma region in the three-dimensional PET image, then performing correction processing on voxel points included in plasma voxel point classes to remove the voxel points belonging to the target non-plasma region in the plasma voxel point classes, and finally performing function fitting according to the time change characteristics of the radioactivity of each voxel point in the plasma voxel point classes after correction processing within a preset time length to obtain a plasma input function.

It should be understood that the CT image can accurately reflect the structure of the human organ, and therefore, performing semantic segmentation processing on the CT image can obtain more accurate plasma regions for characterizing plasma in the scanned region and non-plasma regions for characterizing non-plasma in the scanned region. In the PET-CT system, the CT image and the PET image sequence scan the same scanning region as the scanning target, that is, the scanning region displayed by the CT image is the same as the scanning region displayed by the PET image sequence. Therefore, after obtaining the non-plasma region in the CT image that is used to characterize the non-plasma in the scan region, the non-plasma region can be mapped into the three-dimensional PET image to obtain the target non-plasma region in the three-dimensional PET image. And then, correcting the voxel points included in the plasma voxel point classes obtained by clustering according to the target non-plasma region to remove the voxel points belonging to the target non-plasma region in the plasma voxel point classes, so that the result accuracy of the subsequent function fitting process is improved, and the plasma input function more conforming to the actual condition of the scanned object is obtained.

For example, the CT image may be processed by any semantic segmentation processing algorithm in the related art, so as to obtain a non-plasma region corresponding to non-plasma of the scanned part in the CT image, which is not limited in this disclosure. For example, the CT image may be processed by semantic segmentation using a U-Net (U-Net) as shown in fig. 5, or may be processed by a semantic segmentation algorithm such as a full convolution network or a residual error network to obtain a non-plasma region in the CT image.

After obtaining the non-plasma region in the CT image, the non-plasma region is mapped into the three-dimensional PET image, which may be: firstly, carrying out voxel point coordinate alignment on the CT image and the three-dimensional PET image. After the voxel point coordinates are aligned, the CT image and the three-dimensional PET image have one-to-one correspondence, so that in the mapping process, voxel points which are consistent with the voxel point coordinates in the non-plasma region can be searched in the three-dimensional PET image. The region composed of all the voxel points found is the target non-plasma region in the three-dimensional PET image.

After obtaining the target non-plasma region in the three-dimensional PET image, correction processing may be performed on voxel points included in the plasma voxel point class obtained by clustering to remove voxel points belonging to the target non-plasma region in the plasma voxel point class. For example, the coordinates of each voxel point included in the clustered plasma voxel point class may be sequentially compared with the coordinates of the voxel points in the target non-plasma region. If the coordinate of a certain voxel point in the plasma voxel point class is consistent with the coordinate of another voxel point in the target non-plasma region, the voxel point is considered to belong to the target non-plasma region. In this case, in order to improve the accuracy of the finally obtained plasma input function, the voxel points may be removed from the plasma voxel point groups, and then function fitting may be performed on the basis of the remaining voxel points in the plasma voxel point groups to obtain the plasma input function.

By the method, after the plasma voxel point classes are obtained through the time characteristics of the voxel points, the spatial characteristics of the voxel points can be obtained through the CT image to perform correction processing on the plasma voxel point classes so as to remove the voxel points which actually belong to non-plasma areas in the plasma voxel point classes and obtain the plasma voxel point classes which are more in line with the actual situation, and therefore the accuracy of obtaining the plasma input function by performing function fitting according to the plasma voxel point classes is improved.

In a possible manner, the plasma voxel point classes may also be stored, and accordingly, function fitting is performed according to a time variation characteristic of the radioactivity of each voxel point in the plasma voxel point classes within a preset time duration to obtain a plasma input function, which may be: the method comprises the steps of firstly obtaining stored plasma voxel point classes, and then carrying out function fitting according to the time change characteristics of the radioactivity activity of each voxel point in the obtained plasma voxel point classes within a preset time length to obtain a plasma input function.

For example, after the plasma voxel point class used for characterizing the plasma in the body of the scanned object is obtained by clustering, the obtained plasma voxel point class is saved. Alternatively, the plasma voxel point group may be subjected to the correction processing described above, and the corrected plasma voxel point group may be stored. The embodiment of the present disclosure is not limited to this, and may be selected according to actual situations. It should be understood that, if the plasma voxel point class obtained by clustering is stored, after the plasma voxel point class is obtained, in order to improve the accuracy of the result, the obtained plasma voxel point class may be corrected first, and then function fitting may be performed according to the corrected plasma voxel point class.

By the method, the plasma body pigment point classes are stored, then the stored plasma body pigment point classes can be obtained at any time according to actual requirements to carry out function fitting, a plasma input function is obtained, and the accuracy of the obtained plasma input function is verified at any time.

The image processing method provided by the present disclosure is explained below by another exemplary embodiment. Referring to fig. 6, the image processing method may include:

step 601, acquiring a PET image sequence and a three-dimensional CT image aiming at the same scanning object. The PET image sequence comprises a plurality of three-dimensional PET images obtained by scanning the same part of a scanning object at different scanning moments within a preset time length, each voxel point in each three-dimensional PET image is provided with radioactivity, and the scanning part corresponding to the CT image is the same as the scanning part corresponding to the PET image sequence.

Step 602, performing semantic segmentation processing on the CT image, and determining a first target region including the scanning part in the CT image and a non-plasma region corresponding to non-plasma of the scanning part.

Step 603, for each three-dimensional PET image in the PET image sequence, mapping the first target region in the CT image to the three-dimensional PET image to obtain a second target region including the scanning site in the three-dimensional PET image.

And step 604, clustering according to the time variation characteristics of the radioactivity corresponding to the voxel points in the second target region within a preset time length to obtain a plurality of clustering results.

In step 605, a plasma dot class for characterizing plasma in the scanned subject is determined among the plurality of clustering results.

And 606, mapping the non-plasma region to the three-dimensional PET image according to each three-dimensional PET image in the PET image sequence to obtain a target non-plasma region in the three-dimensional PET image.

Step 607, the voxel points included in the plasma voxel point class are corrected to remove the voxel points belonging to the target non-plasma region in the plasma voxel point class.

At step 608, the plasma pixel class is saved.

And step 609, performing function fitting according to the time change characteristics of the radioactivity of each voxel point in the plasma voxel point classes after correction processing within a preset time length to obtain a plasma input function.

The detailed description of the above steps is given above for illustrative purposes, and will not be repeated here. It will also be appreciated that for simplicity of explanation, the above-described method embodiments are all presented as a series of acts or combination of acts, but those skilled in the art will recognize that the present disclosure is not limited by the order of acts or combination of acts described above. Further, those skilled in the art will also appreciate that the embodiments described above are preferred embodiments and that the steps involved are not necessarily required for the present disclosure.

By the mode, function fitting can be carried out according to the time variation characteristics of the radioactivity activity corresponding to the voxel points in the PET image sequence within the preset time length, the plasma input function is obtained, the whole process can be automatically executed after the PET image sequence is obtained, artificial participation is not needed, and therefore the efficiency of obtaining the plasma input function is improved. In addition, in the method, the PET image sequence corresponding to the scanning object is processed to further obtain the plasma input function, the individual difference among the scanning objects is considered, the plasma input function which is more in line with the actual situation of the scanning object can be obtained, and therefore the result error caused by the fact that the individual difference of the scanning objects is not considered in the subsequent parameter reconstruction process is reduced. In addition, the above method makes full use of the time characteristic of the PET dynamic image and refers to the space characteristic of the CT image, so that the plasma input function which is more in line with the actual situation can be obtained. In addition, the plasma voxel point classes obtained from the PET image can be stored, so that the accuracy of the obtained plasma input function can be verified at any time.

Based on the same inventive concept, the disclosed embodiments also provide an image processing apparatus, which may be a part or all of a PET-CT system through software, hardware or a combination of both. Referring to fig. 7, the image processing apparatus 700 includes:

an obtaining module 701, configured to obtain a Positron Emission Tomography (PET) image sequence, where the PET image sequence includes multiple three-dimensional PET images obtained by scanning a same portion of a scanned object at different scanning times within a preset time duration, and each voxel point in each three-dimensional PET image has a radioactivity;

a clustering module 702, configured to perform clustering according to a time variation characteristic of radioactivity corresponding to a voxel point in the PET image sequence within the preset time duration to obtain a plasma voxel point class used for representing plasma in the scanned object;

a fitting module 703, configured to perform function fitting according to a time variation characteristic of radioactivity corresponding to each voxel point in the plasma voxel point classes within the preset time duration to obtain a plasma input function, where the plasma input function is used to perform parameter reconstruction on the PET image sequence to obtain a parameter image used to represent the in-vivo metabolic rate of the scanned object.

Optionally, the apparatus 700 further comprises:

a first acquisition module, configured to acquire a three-dimensional Computed Tomography (CT) image of the scan object, where a scan region corresponding to the CT image is the same as a scan region corresponding to the PET image sequence;

the first segmentation module is used for performing semantic segmentation processing on the CT image and determining a first target region comprising the scanning part in the CT image;

the clustering module 702 is configured to:

Optionally, the apparatus 700 further comprises:

a second acquisition module, configured to acquire a three-dimensional Computed Tomography (CT) image of the scan object, where a scan region corresponding to the CT image is the same as a scan region corresponding to the PET image sequence;

the second segmentation module is used for performing semantic segmentation processing on the CT image and determining a non-plasma region corresponding to the non-plasma of the scanning part in the CT image;

the fitting module 703 is configured to:

Optionally, the PET image sequence includes T three-dimensional PET images obtained at T different scanning time instants, each three-dimensional PET image includes N voxel points, where T and N are positive integers, and an image feature of each voxel point is a spatial feature for characterizing a position of the voxel point in the scanning portion, the clustering module 702 is configured to:

Optionally, when the clustering results in at least three clustering results including the plasma voxel point class, the apparatus 700 further comprises:

a first determining module, configured to determine, according to the time variation characteristic of each voxel point in the at least three clustering results, the class of plasma voxel points used for characterizing plasma in the body of the scanned subject and the class of non-plasma voxel points used for characterizing tissues in the body of the scanned subject;

a second determining module, configured to, for each voxel point in other voxel point classes except the plasma voxel point class and the non-plasma voxel point class in the at least three clustering results, determine that the voxel point belongs to the plasma voxel point class when a voxel point in the preset domain of the voxel point belongs to the plasma voxel point, and determine that the voxel point belongs to the non-plasma voxel point class when a voxel point in the preset domain of the voxel point belongs to the non-plasma voxel point.

Optionally, the apparatus 700 further comprises:

a preservation module for preserving the plasma voxel point class;

the fitting module 703 is configured to:

obtaining the preserved plasma pigment point class;

With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.

Based on the same inventive concept, an embodiment of the present disclosure further provides an electronic device, including:

a memory having a computer program stored thereon;

a processor for executing the computer program in the memory to implement the steps of any of the image processing methods described above.

In a possible approach, a block diagram of the electronic device may be as shown in fig. 8. Referring to fig. 8, the electronic device 800 may include: a processor 801, a memory 802. The electronic device 800 may also include one or more of a multimedia component 803, an input/output (I/O) interface 804, and a communications component 805.

The processor 801 is configured to control the overall operation of the electronic device 800, so as to complete all or part of the steps in the image processing method. The memory 802 is used to store various types of data to support operation at the electronic device 800, such as instructions for any application or method operating on the electronic device 800 and application-related data, such as contact data, transmitted and received messages, pictures, audio, video, and so forth. The Memory 802 may be implemented by any type of volatile or non-volatile Memory device or combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic Memory, flash Memory, magnetic disk or optical disk. The multimedia components 803 may include screen and audio components. Wherein the screen may be, for example, a touch screen and the audio component is used for outputting and/or inputting audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may further be stored in the memory 802 or transmitted through the communication component 805. The audio assembly also includes at least one speaker for outputting audio signals. The I/O interface 804 provides an interface between the processor 801 and other interface modules, such as a keyboard, mouse, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 805 is used for wired or wireless communication between the electronic device 800 and other devices. Wireless Communication, such as Wi-Fi, bluetooth, Near Field Communication (NFC), 2G, 3G, 4G, NB-IOT, eMTC, or other 5G, etc., or a combination of one or more of them, which is not limited herein. The corresponding communication component 805 may therefore include: Wi-Fi module, Bluetooth module, NFC module, etc.

In an exemplary embodiment, the electronic Device 800 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components for performing the image Processing methods described above.

In another exemplary embodiment, there is also provided a computer readable storage medium comprising program instructions which, when executed by a processor, implement the steps of the image processing method described above. For example, the computer readable storage medium may be the memory 802 described above that includes program instructions executable by the processor 801 of the electronic device 800 to perform the image processing method described above.

In another exemplary embodiment, a computer program product is also provided, which comprises a computer program executable by a programmable apparatus, the computer program having code portions for performing the image processing method described above when executed by the programmable apparatus.

Based on the same inventive concept, the embodiment of the present disclosure further provides a positron emission tomography PET-CT system, which includes: the device comprises a ray source, a first detector, a second detector and the electronic equipment. Wherein the radiation source is used for emitting radiation. The first detector is used for detecting attenuated ray signals after passing through a scanned object, converting the attenuated ray signals into electric signals and sending the electric signals to the electronic equipment so as to obtain a CT image. The second detector is used for detecting high-energy photons emitted from the body of the scanned object, converting the high-energy photons into pulse signals and sending the pulse signals to the electronic equipment so as to obtain a PET image sequence.

In a possible manner, the PET-CT system may be as shown in FIG. 2, which has already been described above and will not be described in detail here. It should be understood that the positions of the first and second detectors in fig. 2 are for illustration only, and in practical applications, the positions of the first and second detectors may be different from the positions illustrated in fig. 2.

The preferred embodiments of the present disclosure are described in detail with reference to the accompanying drawings, however, the present disclosure is not limited to the specific details of the above embodiments, and various simple modifications may be made to the technical solution of the present disclosure within the technical idea of the present disclosure, and these simple modifications all belong to the protection scope of the present disclosure.

It should be noted that, in the foregoing embodiments, various features described in the above embodiments may be combined in any suitable manner, and in order to avoid unnecessary repetition, various combinations that are possible in the present disclosure are not described again.

In addition, any combination of various embodiments of the present disclosure may be made, and the same should be considered as the disclosure of the present disclosure, as long as it does not depart from the spirit of the present disclosure.

Claims

1. An image processing method, characterized in that the method comprises:

2. The method of claim 1, further comprising:

3. The method of claim 1, further comprising:

4. The method according to any one of claims 1 to 3, wherein the PET image sequence includes T three-dimensional PET images obtained at T different scanning moments, each of the three-dimensional PET images includes N voxel points, wherein T and N are positive integers, and an image feature of each voxel point is a spatial feature for characterizing a position of the voxel point in the scanning portion, and the clustering according to the time-varying feature of the radioactivity corresponding to the voxel points in the PET image sequence within the preset time duration includes:

5. The method according to any one of claims 1-3, wherein when clustering results in at least three clustering results including the plasma voxel point class, the method further comprises:

6. The method according to any one of claims 1-3, further comprising:

preserving said plasma pixel population;

obtaining the preserved plasma pigment point class;

7. An image processing apparatus, characterized in that the apparatus comprises:

8. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 6.

9. An electronic device, comprising:

a memory having a computer program stored thereon;

a processor for executing the computer program in the memory to carry out the steps of the method of any one of claims 1 to 6.

10. A positron emission computed tomography PET-CT system, the PET-CT system comprising: a source of radiation, a first detector, a second detector, and the electronic device of claim 9;

the ray source is used for emitting rays;