Multi-source CT scattering correction method and device based on deep learning

By using deep learning methods to correct scattering signals in multi-source CT systems, the problem of scattering signals affecting image quality in multi-source CT imaging is solved, achieving efficient scattering correction and image enhancement.

CN118967857APending Publication Date: 2024-11-15TSINGHUA UNIVERSITY
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411064509.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-15

AI Technical Summary

Technical Problem

In multi-source CT imaging systems, the simultaneous emission of beams from multiple sources results in the detector receiving both forward-scattered and cross-scattered signals, which severely affects the quality of the reconstructed image. Furthermore, existing correction methods may lead to significant numerical errors and are prone to producing high-frequency artifacts.

Method used

A deep learning-based multi-source CT scattering correction method is adopted. By acquiring multi-source CT scan data, the low-frequency multi-source cross-scattering field is learned and the cross-scattering signal is corrected. Combined with a preset forward scattering network, the estimated forward scattering signal is expressed to complete the correction of multi-source CT scattering.

Benefits of technology

It effectively reduces network parameters, suppresses the generation of high-frequency artifacts, ensures the low-frequency characteristics of scattering estimation results, and improves the accuracy and clarity of reconstructed images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118967857A_ABST
    Figure CN118967857A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of radiation imaging, in particular to a multi-source CT scattering correction method and device based on deep learning, and the method comprises the steps: obtaining multi-source CT scanning data when at least two ray sources in a multi-source CT system emit beams at the same time; learning from the total signal of the multi-source CT scanning data to obtain a corresponding low-frequency multi-source cross scattering field so as to correct a cross scattering signal; and estimating a forward scattering signal based on a preset forward scattering network expression so as to correct the forward scattering signal and complete the correction of the multi-source CT scattering. Therefore, the problems that the quality of a reconstructed image is seriously influenced because scattering signals received by a detector comprise forward scattering signals and cross scattering signals due to simultaneous beam output of multiple sources in the related technology, and a related correction method is large in numerical error, high-frequency artifacts are easily generated and the like are solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of radiation imaging technology, and in particular to a multi-source CT scatter correction method and device based on deep learning. Background Art

[0002] X-ray CT imaging systems are widely used in medical treatment, security inspection, industrial non-destructive testing and other fields. By collecting projection data in all directions through a single ray source and detectors surrounding the object, the internal structure of the object can be reconstructed, but the imaging speed is slow and very limited.

[0003] In the related technology, multiple radiation sources can be used to simultaneously emit beams for scanning and imaging. Whether it is a dual-source CT or a triple-source CT with a small number of radiation sources, or a static CT with a large number of radiation sources (the radiation sources and detectors are arranged in a circle, and projections at different angles are obtained by rapid switching of the radiation sources), a beam-emitting mode with multiple sources emitting beams simultaneously can be used to give full play to the imaging speed potential of multi-source CT.

[0004] However, in the related art, due to the simultaneous beam emission of multiple sources, the scattering signals received by the detector include forward scattering signals and cross scattering signals, which seriously affects the quality of the reconstructed image. In addition, the use of kernel methods to correct multi-source CT scattering may lead to large numerical errors, and the Monte Carlo method has a large amount of computation and cannot accurately show the low-frequency characteristics of the scattering signal, resulting in the estimated scattering signal containing large noise. At the same time, it is difficult to estimate forward scattering and cross scattering, which is prone to high-frequency artifacts, and needs to be improved urgently. Summary of the invention

[0005] The present application provides a multi-source CT scatter correction method and device based on deep learning to solve the problems in the related art that due to the simultaneous beam emission of multiple sources, the scatter signals received by the detector include forward scatter signals and cross scatter signals, which seriously affects the quality of the reconstructed image, and the related correction method may lead to large numerical errors and easily generate high-frequency artifacts.

[0006] The first aspect of the present application provides a multi-source CT scatter correction method based on deep learning, comprising the following steps: acquiring multi-source CT scanning data when at least two radiation sources emit beams simultaneously in a multi-source CT system; learning from the total signal of the multi-source CT scanning data to obtain a corresponding low-frequency multi-source cross-scattering field to correct the cross-scattering signal; estimating the forward scattering signal based on a preset forward scattering network expression to correct the forward scattering signal, thereby completing the correction of multi-source CT scattering.

[0007] Optionally, in one embodiment of the present application, the learning of a corresponding low-frequency multi-source cross-scattering field from the total signal of the multi-source CT scan data to correct the cross-scattering signal includes: learning a corresponding large-angle scattering signal from the total signal; calculating the low-frequency multi-source cross-scattering field based on the large-angle scattering signal, and subtracting the low-frequency multi-source cross-scattering field from the total signal to correct the cross-scattering signal.

[0008] Optionally, in one embodiment of the present application, the learning to obtain the corresponding large-angle scattering signal includes: converting the single-frequency channel signal in the imaging field of view formed by each ray source in the multi-source CT scanning data into a multi-frequency channel signal; performing projection signal encoding on the multi-frequency channel signal, and learning the mapping relationship from the signal encoding of the total signal to the large-angle scattering signal, so as to generate the corresponding large-angle scattering signal from the feature space.

[0009] Optionally, in one embodiment of the present application, the preset forward scattering network is expressed as:

[0010]

[0011] in, For additional detection information, For additional detection information features, is the encoding and decoding module, F is the frequency attention mechanism module, S i,p is the predicted forward scattering signal, T i is the total signal of multi-source CT scanning, C i To predict the cross-scattering signal of the multi-source scattering field in region i.

[0012] The second aspect of the present application provides a multi-source CT scatter correction device based on deep learning, including: an acquisition module, used to acquire multi-source CT scanning data when at least two radiation sources emit beams simultaneously in a multi-source CT system; a learning module, used to learn from the total signal of the multi-source CT scanning data to obtain a corresponding low-frequency multi-source cross-scattering field to correct the cross-scattering signal; a correction module, used to estimate the forward scattering signal based on a preset forward scattering network expression to correct the forward scattering signal and complete the correction of multi-source CT scattering.

[0013] Optionally, in one embodiment of the present application, the learning module includes: a learning unit, used to learn and obtain a corresponding large-angle scattering signal from the total signal; and a distribution unit, used to calculate the low-frequency multi-source cross-scattering field based on the large-angle scattering signal, and subtract the low-frequency multi-source cross-scattering field from the total signal to correct the cross-scattering signal.

[0014] Optionally, in one embodiment of the present application, the learning unit includes: a conversion subunit, used to convert a single-frequency channel signal in an imaging field of view formed by each ray source in the multi-source CT scanning data into a multi-frequency channel signal; an encoding subunit, used to perform projection signal encoding on the multi-frequency channel signal, and learn a mapping relationship from the signal encoding of the total signal to the large-angle scattering signal, so as to generate the corresponding large-angle scattering signal from the feature space.

[0015] Optionally, in one embodiment of the present application, the preset forward scattering network is expressed as:

[0016]

[0017] in, For additional detection information, For additional detection information features, is the encoding and decoding module, F is the frequency attention mechanism module, S i,p is the predicted forward scattering signal, T i is the total signal of multi-source CT scanning, C i To predict the cross-scattering signal of the multi-source scattering field in region i.

[0018] A third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-source CT scatter correction method based on deep learning as described in the above embodiment.

[0019] A fourth aspect of the present application provides a computer-readable storage medium, which stores a computer program. When the program is executed by a processor, it implements the above-mentioned deep learning-based multi-source CT scatter correction method.

[0020] The fifth aspect of the present application provides a computer program product, which, when executed, is used to implement the above-mentioned deep learning-based multi-source CT scatter correction method.

[0021] The embodiment of the present application can obtain multi-source CT scanning data when at least two ray sources emit beams simultaneously in a multi-source CT system, and then can learn from the total signal of the multi-source CT scanning data to obtain the corresponding low-frequency multi-source cross-scattering field, complete the correction of the cross-scattering signal, and estimate the forward scattering signal based on the preset forward scattering network expression, and can correct the forward scattering signal, thereby completing the correction of multi-source CT scattering, which can ensure the low-frequency characteristics of the scattering estimation result, effectively reduce the network parameters, suppress the generation of high-frequency artifacts, and ensure the quality of the reconstructed image, improve the accuracy and clarity of the imaging. Therefore, the problem in the related art that due to the simultaneous beam emission of multiple sources, the scattering signals received by the detector include forward scattering signals and cross-scattering signals, which seriously affects the quality of the reconstructed image, and the related correction method may cause large numerical errors and easily generate high-frequency artifacts.

[0022] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0024] Figure 1 A flowchart of a multi-source CT scatter correction method based on deep learning provided according to an embodiment of the present application;

[0025] Figure 2 Schematic diagram of two multi-source simultaneous beam-emitting imaging scanning modes according to an embodiment of the present application;

[0026] Figure 3 A schematic diagram of the principle of a multi-source CT scatter correction method based on deep learning according to an embodiment of the present application;

[0027] Figure 4 A schematic diagram of a deep cross scattering estimation network according to an embodiment of the present application;

[0028] Figure 5 A schematic diagram of a deep forward scattering estimation network according to an embodiment of the present application;

[0029] Figure 6 A schematic diagram of a frequency attention mechanism module according to an embodiment of the present application;

[0030] Figure 7 A schematic diagram of a cross-scattering signal estimation result according to an embodiment of the present application;

[0031] Figure 8 A schematic diagram of a forward scattering signal estimation result according to an embodiment of the present application;

[0032] Fig. 9 A schematic diagram of an overall prediction result of multi-source CT scatter correction based on deep learning according to an embodiment of the present application;

[0033] Fig.10 A block diagram of a multi-source CT scatter correction device based on deep learning provided according to an embodiment of the present application;

[0034] Fig.11 It is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0035] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0036] The following describes the multi-source CT scatter correction method and device based on deep learning of the embodiment of the present application with reference to the accompanying drawings. In view of the related technology mentioned in the background technology center above, due to the simultaneous beam emission of multiple sources, the scattering signals received by the detector include forward scattering signals and cross scattering signals, which seriously affects the quality of the reconstructed image, and the related correction method may cause large numerical errors and easily generate high-frequency artifacts. The present application provides a multi-source CT scatter correction method based on deep learning. In this method, when at least two ray sources emit beams simultaneously in a multi-source CT system, multi-source CT scanning data can be obtained, and then, the corresponding low-frequency multi-source cross scattering field can be learned from the total signal of the multi-source CT scanning data to complete the correction of the cross scattering signal, and the forward scattering signal can be estimated based on the preset forward scattering network expression, and the forward scattering signal can be corrected, thereby completing the correction of the multi-source CT scattering, which can ensure the low-frequency characteristics of the scattering estimation result, effectively reduce the network parameters, suppress the generation of high-frequency artifacts, and ensure the quality of the reconstructed image, and improve the accuracy and clarity of the imaging. This solves the problem in the related technology that, due to the simultaneous beam emission from multiple sources, the scattering signals received by the detector include forward scattering signals and cross scattering signals, which seriously affects the quality of the reconstructed image, and the related correction method may lead to large numerical errors and easily produce high-frequency artifacts.

[0037] Specifically, Figure 1 A flowchart of a multi-source CT scatter correction method based on deep learning provided in an embodiment of the present application.

[0038] like Figure 1 As shown, the multi-source CT scatter correction method based on deep learning includes the following steps:

[0039] In step S101, when at least two ray sources in a multi-source CT system emit beams simultaneously, multi-source CT scanning data is acquired.

[0040] It can be understood that the multi-source CT scanning data includes but is not limited to forward attenuation signals, forward scattering signals, and cross-scattering signals from other ray sources.

[0041] Specifically, in the embodiment of the present application, when at least two ray sources emit beams simultaneously, for the total signal T in the imaging field formed by each ray source, i , including the forward attenuation signal P of the ray source corresponding to the imaging field of view i and forward scatter S i , and cross-scattering C from other ray sources i , which can be shown as follows:

[0042] T i =P i +S i +C i

[0043]

[0044] Among them, C i j is the cross-scatter signal from ray source j in imaging field of view i.

[0045] It is worth noting that the combination Figure 2 As shown, the embodiment of the present application uses a cross-scattering depth estimation method and a forward-scattering depth estimation method, both of which are multi-layer CNN networks, to achieve end-to-end mapping, and directly learn the corresponding scattering signal from the corresponding total signal.

[0046] In step S102, a corresponding low-frequency multi-source cross-scattering field is learned from the total signal of the multi-source CT scanning data to correct the cross-scattering signal.

[0047] Specifically, the embodiment of the present application can learn the corresponding low-frequency multi-source cross-scattering field from the total signal of the multi-source CT scanning data through cross-scattering depth estimation to correct the cross-scattering signal, wherein, combined with Figure 3 As shown in the figure, the cross-scattering depth estimation method is mainly divided into a frequency attention mechanism module F and a projection signal encoding module Encoding space multi-source interaction module Additional detection information encoding module and decoding and reconstruction modules

[0048] The embodiment of the present application can learn from the total signal of the multi-source CT scan data through a deep learning model to obtain the corresponding low-frequency multi-source cross-scattering field to correct the cross-scattering signal, effectively remove or reduce scattering artifacts from the original scan data, thereby improving the accuracy and clarity of image reconstruction.

[0049] Optionally, in one embodiment of the present application, a corresponding low-frequency multi-source cross-scattering field is learned from the total signal of the multi-source CT scanning data to correct the cross-scattering signal, including: learning to obtain a corresponding large-angle scattering signal from the total signal; calculating the low-frequency multi-source cross-scattering field based on the large-angle scattering signal, and subtracting the low-frequency multi-source cross-scattering field from the total signal to correct the cross-scattering signal.

[0050] Specifically, the embodiment of the present application can directly obtain the total signal T from the corresponding i Directly learn the corresponding large-angle scattering signal C i , which can be shown as follows:

[0051]

[0052]

[0053] in, Reconstruction module for decoding.

[0054] Furthermore, for large-angle scattering signals {C 1 ,C 2 ,C 3 ...}, C i j In fact, it is the large-angle scattering signal C from the ray source j. j The total cross-scattering signal in region i is C i , can be as follows:

[0055]

[0056] Among them, j1, j2…j m represents all other ray sources that emit beams simultaneously with ray source i, It represents a part of the large-angle scattering signal from the ray source j1 in region i.

[0057] The embodiment of the present application obtains a large-angle scattering signal from the total signal through deep learning, and can more accurately identify and separate scattering signals caused by different ray sources, thereby improving the accuracy of scattering correction and reducing image processing time.

[0058] Optionally, in one embodiment of the present application, learning to obtain corresponding large-angle scattering signals includes: converting a single-frequency channel signal in an imaging field of view formed by each ray source in multi-source CT scanning data into a multi-frequency channel signal; performing projection signal encoding on the multi-frequency channel signal, and learning a mapping relationship from the signal encoding of the total signal to the large-angle scattering signal, so as to generate a corresponding large-angle scattering signal from the feature space.

[0059] It can be understood that a multi-frequency channel signal refers to a signal that is decomposed into different frequency domains, and each channel represents a specific frequency segment of the signal.

[0060] Specifically, the embodiment of the present application can introduce a frequency attention mechanism module F, through which the single-frequency channel signal in the imaging field formed by each ray source in the multi-source CT scan data can be converted into a multi-frequency channel signal, which can then enter the projection signal encoding module. In feature space Generate representation in and enter the decoding and reconstruction module, which can be obtained from the feature space The corresponding large-angle scattering signal is generated in

[0061] It is worth noting that the data of different channels are different frequencies of the signal. The embodiment of the present application can effectively reduce the frequency of the initial signal without losing high-frequency information by introducing the channel attention mechanism method. When there is no other corresponding auxiliary information, the embodiment of the present application can directly obtain the total forward signal T i estimate However, there are other sources of radiation emitting at the same time. and additional detection information We can use both to calculate z in the feature space i Get a better estimate.

[0062] The embodiments of the present application convert the signal into multi-frequency channels and encode it, which can accurately estimate the cross-scattering signal from a richer feature space, improve the effect of scattering correction, reduce image artifacts, enhance the performance of the image reconstruction algorithm, and better adapt to scene changes, significantly improve image quality and processing efficiency. In addition, a frequency attention mechanism module is proposed for the first time, which can effectively reduce the frequency of the initial signal without losing high-frequency information, ensure the low-frequency nature of the scattering estimation result, suppress high-frequency artifacts, effectively reduce network parameters, and facilitate actual work deployment.

[0063] In step S103, the forward scattering signal is estimated based on a preset forward scattering network expression to correct the forward scattering signal, thereby completing the correction of multi-source CT scattering.

[0064] It can be understood that the preset forward scattering network expression refers to a pre-set model for estimating the low-frequency characteristics of the forward scattering signal from the total signal. It can be specifically set by technical personnel in this field based on factors such as the physical characteristics of the scattering signal and the imaging mechanism of the multi-source CT system, and no specific limitation is made here.

[0065] Specifically, combined Figure 4 As shown, the forward scattering depth estimation method of the embodiment of the present application is mainly divided into a frequency attention mechanism module F, a signal encoding and decoding module Additional detection information encoding module After the cross-scattering estimation is completed, the input of the network is from T i Became T i -C i =P i +S i At this time, the network input is consistent with the single ray source scattering correction, so the embodiment of the present application can directly learn from P i +S i To S i Mapping, optionally, in one embodiment of the present application, the preset forward scattering network is expressed as:

[0066]

[0067] in, For additional detection information, For additional detection information features, is the encoding and decoding module, F is the frequency attention mechanism module, S i,p is the predicted forward scattering signal, T i is the total signal of multi-source CT scanning, C i To predict the cross-scattering signal of the multi-source scattering field in region i.

[0068] It is worth noting that the encoding and decoding module of the embodiment of the present application In P i +S i Before signal encoding, the frequency attention mechanism module F is used, and additional detection information features are introduced in the highest dimensional feature layer of the encoding and decoding module. It can better guide the network to learn the scattered signals and achieve better estimation.

[0069] The embodiment of the present application is based on a preset forward scattering network expression, uses a cross-scattering signal to estimate a low-frequency multi-source scattering field, and uses the estimated scattering signal in a subsequent image reconstruction process to correct image artifacts caused by scattering, thereby improving the quality of the reconstructed image, completing the correction of multi-source CT scattering, and introducing additional detection information to assist in the estimation, so that the depth estimation method can obtain better results in the verification data.

[0070] Combination Figures 2 to 9 As shown, the multi-source CT scatter correction method based on deep learning of an embodiment of the present application is described in detail by taking an embodiment as an example.

[0071] Specifically, the embodiments of the present application can take the scatter correction of dual-ring static CT as an example. In the dual-ring static CT, in order to maximize the time resolution, multiple radiation sources (greater than 3) are often used to emit beams simultaneously, so as to further shorten the scanning time. Then, since both the radiation source and the detector are ring-shaped, it is very difficult to manufacture the corresponding static collimator. It is not hopeful to directly remove the scattered signal physically, and the scattered signal of the system needs to be corrected.

[0072] like Figure 2 and Figure 4 As shown in the figure, in a dual-ring static CT, take five ray sources emitting beams at the same time as an example. Five imaging fields of view are formed on the annular detector. The signals in the five imaging fields of view are T1, T2, T3, T4 and T5, which come from ray sources 1, 2, 3, 4 and 5 respectively. The angle range of the annular detector corresponding to each imaging field of view is 50 degrees, and the angle between the imaging fields of view is 22 degrees. Among them, for the first imaging field of view T1:

[0073] T1=P1+S1+C1

[0074] C1=C1 2 +C1 3 +C1 4 +C1 5 =S 2,1 c +S 3,1 c +S 4,1 c +S 5,1 c

[0075] Among them, S 2,1 c and C1 2 The area in the middle of the five imaging fields is the additional detection information, which only receives the scattered signal and is recorded as O. T , then the cross-scattering estimation method can be expressed as:

[0076]

[0077] C 2,p 1 =FOV2(C 1 p )

[0078]

[0079] Note that the subscript of T here is expressed modulo 5, that is, T6 represents T1.

[0080] Combination Figure 3 As shown, in the embodiment of the present application, the predictions between different imaging fields of view in terms of cross-scatter prediction are decoupled, but the same geometric relationship makes the parameters shared, and thus, by predicting the large-angle scattering signals corresponding to multiple ray sources, consistent estimation of different imaging areas can be achieved, and finally the cross-scattering estimation of the entire multi-source scattering field can be completed.

[0081] It is worth noting that when implementing cross scatter estimation in the embodiment of the present application, the projection signal encoding module Encoding space multi-source interaction module Additional detection information encoding module and decoding and reconstruction modules It is consistent with the common network coding structure. Among them, the frequency attention mechanism module is as follows Figure 5 As shown, multi-frequency channel operation can be achieved through wavelet transform or Fourier transform. Figure 6 The cross-scattering prediction results of this embodiment are shown.

[0082] Combination Figure 7 As shown, when the estimated value C of C1 is obtained by the cross-scattering estimation method 1 p The next goal is to estimate S1. At this time, the input of the network is T1-C 1 p , the target predicted by the network is S1. The forward scattering network used is expressed as:

[0083]

[0084] In the actual implementation process, the network includes the introduced frequency attention mechanism module and the common Unet network structure. The two parts are combined to form a multi-source scattering field depth estimation method. Obviously, the multi-source scattering field depth estimation method proposed in this application is an end-to-end network consisting of two networks. The general strategy adopted by the network during training is to train separately and then train jointly. Finally, the entire depth estimation architecture is formed together. After completing the training, you only need to input the obtained data to quickly infer and obtain the estimation results of the multi-scattering field to achieve scattering correction.

[0085] Combination Figure 6 , Figure 7 and Figure 8 As shown, the embodiment of the present application can obtain the difference between the prediction results, true value and true value of the two sub-networks and the entire method. Among them, the network parameter amount of the deep cross scattering estimation network is 2.8 million, and the average absolute error and average absolute percentage error of the prediction results are 0.43 and 1.97%. The parameter amount of the deep forward scattering estimation network is 1 million, and the average absolute error and average absolute percentage error of the prediction results are 0.58 and 2.45%. Compared with the 100M parameter amount of Unet in the related art, this application has also achieved very excellent results with a smaller parameter amount.

[0086] In addition, combined Fig. 9 As shown, the prediction accuracy of the network with and without additional detection information is also trained and tested in the embodiment of the present application, and the average absolute error of the entire network test data is reduced from 1.16 to 0.8, and the average absolute percentage error is reduced from 1.32% to 0.9%. It can be seen that with the assistance of additional detection information, the prediction of scattering can achieve a more accurate estimate.

[0087] According to the multi-source CT scatter correction method based on deep learning proposed in the embodiment of the present application, when at least two ray sources in the multi-source CT system emit beams at the same time, multi-source CT scanning data can be obtained, and then the corresponding low-frequency multi-source cross-scattering field can be learned from the total signal of the multi-source CT scanning data to complete the correction of the cross-scattering signal, and the forward scattering signal can be estimated based on the preset forward scattering network expression, and the forward scattering signal can be corrected, thereby completing the correction of multi-source CT scattering, which can ensure the low-frequency characteristics of the scattering estimation result, effectively reduce the network parameters, suppress the generation of high-frequency artifacts, and ensure the quality of the reconstructed image, improve the accuracy and clarity of the imaging. Therefore, the problem in the related art that due to the simultaneous beam emission of multiple sources, the scattering signals received by the detector include forward scattering signals and cross scattering signals, which seriously affects the quality of the reconstructed image, and the related correction method may cause large numerical errors and easily generate high-frequency artifacts.

[0088] Next, a multi-source CT scatter correction device based on deep learning proposed in accordance with an embodiment of the present application is described with reference to the accompanying drawings.

[0089] Fig.10 It is a block diagram of a multi-source CT scatter correction device based on deep learning according to an embodiment of the present application.

[0090] like Fig.10 As shown, the multi-source CT scatter correction device 10 based on deep learning includes: an acquisition module 100, a learning module 200 and a correction module 300.

[0091] Specifically, the acquisition module 100 is used to acquire multi-source CT scanning data when at least two ray sources emit beams simultaneously in a multi-source CT system;

[0092] A learning module 200 is used to learn and obtain a corresponding low-frequency multi-source cross-scattering field from a total signal of the multi-source CT scanning data to correct the cross-scattering signal;

[0093] The correction module 300 is used to estimate the forward scattering signal based on a preset forward scattering network expression to correct the forward scattering signal and complete the correction of multi-source CT scattering.

[0094] Optionally, in one embodiment of the present application, the learning module 200 includes: a learning unit and an allocation unit.

[0095] Wherein, the learning unit is used to learn and obtain the corresponding large-angle scattering signal from the total signal;

[0096] The allocation unit is used for calculating the low-frequency multi-source cross-scattering field based on the large-angle scattering signal, and subtracting the low-frequency multi-source cross-scattering field from the total signal to correct the cross-scattering signal.

[0097] Optionally, in one embodiment of the present application, the learning unit includes: a conversion subunit and an encoding subunit.

[0098] The conversion subunit is used to convert the single frequency channel signal in the imaging field formed by each ray source in the multi-source CT scanning data into a multi-frequency channel signal;

[0099] The encoding subunit is used for encoding the projection signals of the multi-frequency channel signals and learning the mapping relationship from the signal encoding of the total signal to the large-angle scattering signal to generate the corresponding large-angle scattering signal from the feature space.

[0100] Optionally, in one embodiment of the present application, the preset forward scattering network is expressed as:

[0101]

[0102] in, For additional detection information, For additional detection information features, is the encoding and decoding module, F is the frequency attention mechanism module, S i,p is the predicted forward scattering signal, T i is the total signal of multi-source CT scanning, C i To predict the cross-scattering signal of the multi-source scattering field in region i.

[0103] It should be noted that the aforementioned explanation of the embodiment of the multi-source CT scatter correction method based on deep learning is also applicable to the multi-source CT scatter correction device based on deep learning in this embodiment, and will not be repeated here.

[0104] According to the multi-source CT scatter correction device based on deep learning proposed in the embodiment of the present application, when at least two ray sources in the multi-source CT system emit beams at the same time, multi-source CT scanning data can be obtained, and then the corresponding low-frequency multi-source cross-scattering field can be learned from the total signal of the multi-source CT scanning data to complete the correction of the cross-scattering signal, and the forward scattering signal can be estimated based on the preset forward scattering network expression, and the forward scattering signal can be corrected, thereby completing the correction of multi-source CT scattering, which can ensure the low-frequency characteristics of the scattering estimation result, effectively reduce the network parameters, suppress the generation of high-frequency artifacts, and ensure the quality of the reconstructed image, improve the accuracy and clarity of the imaging. Therefore, the problem in the related art that due to the simultaneous beam emission of multiple sources, the scattering signals received by the detector include forward scattering signals and cross scattering signals, which seriously affects the quality of the reconstructed image, and the related correction method may cause large numerical errors and easily generate high-frequency artifacts.

[0105] Fig.11 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0106] A memory 1101 , a processor 1102 , and a computer program stored in the memory 1101 and executable on the processor 1102 .

[0107] When the processor 1102 executes the program, the multi-source CT scatter correction method based on deep learning provided in the above embodiment is implemented.

[0108] Furthermore, the electronic device further comprises:

[0109] The communication interface 1103 is used for communication between the memory 1101 and the processor 1102 .

[0110] The memory 1101 is used to store computer programs that can be executed on the processor 1102 .

[0111] The memory 1101 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0112] If the memory 1101, the processor 1102 and the communication interface 1103 are implemented independently, the communication interface 1103, the memory 1101 and the processor 1102 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Fig.11 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0113] Optionally, in a specific implementation, if the memory 1101, the processor 1102 and the communication interface 1103 are integrated on a chip, the memory 1101, the processor 1102 and the communication interface 1103 can communicate with each other through an internal interface.

[0114] The processor 1102 may be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0115] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned multi-source CT scatter correction method based on deep learning.

[0116] An embodiment of the present application also provides a computer program product, which can run computer instructions. When the computer instructions are executed by a processor, the multi-source CT scatter correction method based on deep learning as described above is implemented.

[0117] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0118] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, "N" means at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0119] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or N executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0120] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in combination with these instruction execution systems, devices or apparatuses. For the purpose of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in combination with these instruction execution systems, devices or apparatuses. More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or N wirings (electronic devices), a portable computer disk box (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), a fiber optic device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically by optically scanning the paper or other medium and then editing, interpreting or processing in other suitable ways as necessary and then storing it in a computer memory.

[0121] It should be understood that the various parts of the present application can be implemented by hardware, software, firmware or a combination thereof. In the above embodiment, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented by hardware, as in another embodiment, it can be implemented by any one or a combination of multiple of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0122] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0123] In addition, each functional unit in each embodiment of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0124] The storage medium mentioned above may be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limiting the present application. A person of ordinary skill in the art may change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A multi-source CT scatter correction method based on deep learning, characterized in that: The following steps are involved: When at least two radiation sources in a multi-source CT system emit beams simultaneously, multi-source CT scanning data is acquired; Learning a corresponding low-frequency multi-source cross-scattering field from the total signal of the multi-source CT scanning data to correct the cross-scattering signal; The forward scattering signal is estimated based on a preset forward scattering network expression to correct the forward scattering signal, thereby completing the correction of multi-source CT scattering.

2. The method according to claim 1, characterized in that The learning from the total signal of the multi-source CT scanning data to obtain a corresponding low-frequency multi-source cross-scattering field to correct the cross-scattering signal includes: From the total signal, learning to obtain a corresponding large-angle scattering signal; The low-frequency multi-source cross-scattering field is calculated based on the large-angle scattering signal, and the low-frequency multi-source cross-scattering field is subtracted from the total signal to correct the cross-scattering signal.

3. The method according to claim 2, characterized in that The learning to obtain the corresponding large-angle scattering signal includes: Converting a single frequency channel signal in an imaging field formed by each ray source in the multi-source CT scanning data into a multi-frequency channel signal; Projection signal encoding is performed on the multi-frequency channel signal, and a mapping relationship from the signal encoding of the total signal to the large-angle scattering signal is learned to generate the corresponding large-angle scattering signal from a feature space.

4. The method according to claim 1, characterized in that: The preset forward scattering network is expressed as: in, For additional detection information, For additional detection information features, is the encoding and decoding module, F is the frequency attention mechanism module, S i,p is the predicted forward scattering signal, T i is the total signal of multi-source CT scanning, C i To predict the cross-scattering signal of the multi-source scattering field in region i.

5. A multi-source CT scatter correction device based on deep learning, characterized in that: include: An acquisition module, used for acquiring multi-source CT scanning data when at least two ray sources in a multi-source CT system emit beams simultaneously; A learning module, used for learning from the total signal of the multi-source CT scanning data to obtain a corresponding low-frequency multi-source cross-scattering field to correct the cross-scattering signal; The correction module is used to estimate the forward scattering signal based on a preset forward scattering network expression to correct the forward scattering signal and complete the correction of multi-source CT scattering.

6. The device according to claim 5, characterized in that The learning modules include: A learning unit, used for learning to obtain a corresponding large-angle scattering signal from the total signal; The allocation unit is used to calculate the low-frequency multi-source cross-scattering field based on the large-angle scattering signal, and subtract the low-frequency multi-source cross-scattering field from the total signal to correct the cross-scattering signal.

7. The device according to claim 6, characterized in that The learning units include: A conversion subunit, used for converting a single frequency channel signal in an imaging field formed by each ray source in the multi-source CT scanning data into a multi-frequency channel signal; The encoding subunit is used to perform projection signal encoding on the multi-frequency channel signal, and learn the mapping relationship from the signal encoding of the total signal to the large-angle scattering signal, so as to generate the corresponding large-angle scattering signal from the feature space.

8. The device according to claim 5, characterized in that The preset forward scattering network is expressed as: in, For additional detection information, For additional detection information features, is the encoding and decoding module, F is the frequency attention mechanism module, S i,p is the predicted forward scattering signal, T i is the total signal of multi-source CT scanning, C i To predict the cross-scattering signal of the multi-source scattering field in region i.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the multi-source CT scatter correction method based on deep learning as described in any one of claims 1 to 4.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the multi-source CT scatter correction method based on deep learning as described in any one of claims 1 to 4.

11. A computer program product, comprising a computer program, characterized in that The computer program is executed to implement the multi-source CT scatter correction method based on deep learning as described in any one of claims 1 to 4.

Citation Information

Cited By

  • Multi-energy CT scattering correction method and device based on Boltzmann transport equation

    CN120031995A

  • Multi-energy CT scattering correction method and device based on Boltzmann transport equation

    CN120031995B