[go: up one dir, main page]

CN119097328A - Dual-source CT scatter correction method and system based on deep learning model - Google Patents

Dual-source CT scatter correction method and system based on deep learning model Download PDF

Info

Publication number
CN119097328A
CN119097328A CN202411232508.4A CN202411232508A CN119097328A CN 119097328 A CN119097328 A CN 119097328A CN 202411232508 A CN202411232508 A CN 202411232508A CN 119097328 A CN119097328 A CN 119097328A
Authority
CN
China
Prior art keywords
source
dual
scatter
signal
scattering
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202411232508.4A
Other languages
Chinese (zh)
Inventor
李冬
李华勇
范毅
刘金池
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Fussen Technology Co ltd
Original Assignee
Fussen Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Fussen Technology Co ltd filed Critical Fussen Technology Co ltd
Priority to CN202411232508.4A priority Critical patent/CN119097328A/en
Publication of CN119097328A publication Critical patent/CN119097328A/en
Pending legal-status Critical Current

Links

Classifications

    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/02—Arrangements for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
    • A61B6/03—Computed tomography [CT]
    • A61B6/032—Transmission computed tomography [CT]
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/40—Arrangements for generating radiation specially adapted for radiation diagnosis
    • A61B6/4007—Arrangements for generating radiation specially adapted for radiation diagnosis characterised by using a plurality of source units
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/40—Arrangements for generating radiation specially adapted for radiation diagnosis
    • A61B6/4007—Arrangements for generating radiation specially adapted for radiation diagnosis characterised by using a plurality of source units
    • A61B6/4014—Arrangements for generating radiation specially adapted for radiation diagnosis characterised by using a plurality of source units arranged in multiple source-detector units
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/42—Arrangements for detecting radiation specially adapted for radiation diagnosis
    • A61B6/4266—Arrangements for detecting radiation specially adapted for radiation diagnosis characterised by using a plurality of detector units
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/58—Testing, adjusting or calibrating thereof
    • A61B6/582—Calibration
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods

Landscapes

  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Medical Informatics (AREA)
  • Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Surgery (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Animal Behavior & Ethology (AREA)
  • Public Health (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Radiology & Medical Imaging (AREA)
  • Pathology (AREA)
  • Optics & Photonics (AREA)
  • Theoretical Computer Science (AREA)
  • High Energy & Nuclear Physics (AREA)
  • Veterinary Medicine (AREA)
  • Computational Linguistics (AREA)
  • Software Systems (AREA)
  • Mathematical Physics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Evolutionary Computation (AREA)
  • Data Mining & Analysis (AREA)
  • Artificial Intelligence (AREA)
  • Pulmonology (AREA)
  • Image Analysis (AREA)

Abstract

本发明涉及人工智能技术，揭露了一种基于深度学习模型的双源CT散射校正方法及系统，包括：获取预设的头模的头模双源CT投影数据，获取所述头模双源CT投影数据的综合交叉散射信号，利用预设的自适应散射核叠加算法获取所述头模双源CT投影数据的前向散射信号，利用所述头模双源CT投影数据、所述综合交叉散射信号以及所述前向散射信号训练神经网络模型，得到双源CT散射分析模型，利用所述双源CT散射分析模型对预先获取的目标双源CT投影数据进行散射信号预估，得到散射信号，利用所述目标双源CT投影数据减去所述散射信号，得到目标双源CT散射矫正数据。本发明还提出一种基于深度学习模型的双源CT散射校正系统。本发明可以提高双源CT成像的图像质量。

The present invention relates to artificial intelligence technology, and discloses a dual-source CT scatter correction method and system based on a deep learning model, including: obtaining head model dual-source CT projection data of a preset head model, obtaining a comprehensive cross-scatter signal of the head model dual-source CT projection data, using a preset adaptive scatter kernel superposition algorithm to obtain a forward scatter signal of the head model dual-source CT projection data, using the head model dual-source CT projection data, the comprehensive cross-scatter signal and the forward scatter signal to train a neural network model to obtain a dual-source CT scatter analysis model, using the dual-source CT scatter analysis model to estimate the scatter signal of the pre-acquired target dual-source CT projection data to obtain a scatter signal, and using the target dual-source CT projection data to subtract the scatter signal to obtain the target dual-source CT scatter correction data. The present invention also proposes a dual-source CT scatter correction system based on a deep learning model. The present invention can improve the image quality of dual-source CT imaging.

Description

Double-source CT scattering correction method and system based on deep learning model
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a double-source CT scattering correction method and system based on a deep learning model.
Background
CT scanning technology is an important imaging method in medical imaging that can provide detailed images of the internal structures of a patient's body. The dual source CT system, an advanced form of CT technology, enables significant improvements in scanning speed and image quality through two independent X-ray sources and detectors, particularly exhibiting unique advantages in cardiac imaging and wide-range scanning.
However, the existing dual-source CT system cannot avoid the phenomenon of X-ray scattering in practical application, and the scattering signal significantly affects the image quality, increasing the complexity and difficulty of image reconstruction.
Disclosure of Invention
The invention provides a double-source CT scattering correction method and system based on a deep learning model, and mainly aims to solve the problem that X-ray scattering exists in actual imaging of the existing double-source CT system.
In order to achieve the above object, the present invention provides a dual-source CT scatter correction method based on a deep learning model, including:
Acquiring head model double-source CT projection data of a preset head model;
Acquiring comprehensive cross scattering signals of the head model double-source CT projection data;
acquiring forward scattering signals of the head model double-source CT projection data by using a preset adaptive scattering kernel superposition algorithm;
training a neural network model by using the head model double-source CT projection data, the comprehensive cross scatter signals and the forward scatter signals to obtain a double-source CT scatter analysis model;
Carrying out scattering signal estimation on pre-acquired target double-source CT projection data by using the double-source CT scattering analysis model to obtain scattering signals;
and subtracting the scattering signal from the target double-source CT projection data to obtain target double-source CT scattering correction data.
Optionally, the acquiring the integrated cross scatter signal of the head phantom dual-source CT projection data includes:
starting a first ray tube of a preset double-source CT system and a preset second detector;
Acquiring a first cross scatter signal of the first tube with the second detector;
closing the first ray tube and the second detector, and opening the second ray tube and the first detector of a preset dual-source CT system;
acquiring a second cross scatter signal of the second tube with the first detector;
And summarizing the first cross heat dissipation signal and the second cross scattering signal to obtain a comprehensive cross scattering signal.
Optionally, the acquiring the forward scattering signal of the head mode dual-source CT projection data by using a preset adaptive scattering kernel stacking algorithm includes:
Simulating scattering signals of different energy levels emitted by the high-voltage bulb tube under different currents and voltages through preset water modes with different lengths by Monte Carlo simulation to obtain a scattering signal set;
fitting the distribution of each scattering signal in the scattering signal set to obtain scattering signal distribution of the ray bundle passing through water modes with different lengths under different energy levels;
Traversing each pixel of the head model double-source CT projection data according to the scattering signal distribution to obtain the scattering distribution of each ray in the head model double-source CT projection data;
And superposing the scattering distribution of each ray in the head model double-source CT projection data to obtain forward scattering signals of the head model double-source CT projection data.
Optionally, the training a neural network model by using the head model dual-source CT projection data, the integrated cross scatter signal and the forward scatter signal to obtain a dual-source CT scatter analysis model includes:
obtaining projection image signals in the head model double-source CT projection data, and carrying out standardization processing on the projection image signals to obtain standardized projection image signals;
Combining the integrated cross scatter signal with the forward scatter signal to obtain a standard scatter signal;
Carrying out scattering signal analysis according to the standardized projection graph signals by utilizing a neural network model to obtain scattering signal estimation;
calculating a loss value of the scattered signal estimate and the standard scattered signal;
optimizing the loss value by updating weight parameters of the neural network model by using a random gradient descent algorithm;
Updating the learning rate of the neural network model by using a preset stepping learning rate algorithm to obtain an updated learning rate;
when the preset iteration times are reached, selecting a weight parameter corresponding to the minimum loss value in all the iteration times as a final weight parameter;
and inputting the updated learning rate and the final weight parameters into a neural network model to obtain a dual-source CT scattering analysis model.
Optionally, the normalizing the projection image signal to obtain a normalized projection image signal includes:
and carrying out standardization processing on the projection image signal by using the following formula to obtain a standardized projection image signal:
Wherein p is the standardized projection map signal, I is the projection map signal, and I 0 is the bright field projection map signal contained in the head model dual source CT projection data.
Optionally, the calculating the loss value of the scattered signal estimate and the standard scattered signal includes:
the loss value is calculated using the following formula:
Wherein S is the loss value, N is the number of pixels of the first detector and the second detector, I s is the standard scatter signal, I m is the scatter signal estimate, and I p is the normalized projection map signal.
Optionally, updating the learning rate of the neural network model by using a preset step learning rate algorithm to obtain an updated learning rate, including:
updating the learning rate using the following formula:
Wherein lr new is the updated learning rate, lr old is the learning rate, K is the current iteration number, and M is the preset iteration number.
Optionally, the estimating the scattering signal of the pre-acquired target dual-source CT projection data by using the dual-source CT scattering analysis model to obtain a corrected scattering signal includes:
moving a convolution kernel preset in a convolution layer of the double-source CT scattering analysis model on the target double-source CT projection data according to a preset step length, calculating dot products of the convolution kernel and data of a convolution kernel coverage area after each movement, and summing all the dot products to obtain convolution characteristics;
Continuously moving a rectangular window in a pooling layer of the dual-source CT scattering analysis model in the convolution characteristic, calculating the average value of data of a region covered by the rectangular window after each movement, and replacing the data of the region with the calculated average value of each region to obtain the pooling characteristic;
Introducing nonlinear factors into the pooling feature by utilizing an activation function preset in an activation layer of the dual-source CT scattering analysis model to obtain an activation feature;
And obtaining the output of each neuron by utilizing the full-connection layer of the dual-source CT scattering analysis model according to the activation characteristics, and summing the output of each neuron to obtain a corrected scattering signal.
Optionally, subtracting the scatter signal from the target dual-source CT projection data to obtain target dual-source CT scatter correction data includes:
And carrying out scattering correction on the target double-source CT projection data by using the following formula:
Wherein I corr is the target dual-source CT scatter correction data, I total is the target dual-source CT projection data, and I scatter,model is the scatter signal.
In order to solve the above problems, the present invention further provides a dual-source CT scatter correction system based on a deep learning model, the system comprising:
the data acquisition module is used for acquiring head model double-source CT projection data of a preset head model and acquiring comprehensive cross scattering signals of the head model double-source CT projection data;
the scattering estimation module is used for acquiring forward scattering signals of the head model double-source CT projection data by using a preset adaptive scattering kernel superposition algorithm;
the model training module is used for training a neural network model by utilizing the head model double-source CT projection data, the comprehensive cross scattering signals and the forward scattering signals to obtain a double-source CT scattering analysis model;
And the scattering correction module is used for carrying out scattering signal estimation on the pre-acquired target double-source CT projection data by utilizing the double-source CT scattering analysis model to obtain scattering signals, and subtracting the scattering signals from the target double-source CT projection data to obtain target double-source CT scattering correction data.
According to the embodiment of the invention, the comprehensive cross scatter signal of the head model double-source CT projection data is obtained by obtaining the head model double-source CT projection data of the preset head model, the forward scatter signal of the head model double-source CT projection data is obtained by utilizing the preset self-adaptive scatter kernel superposition algorithm, the efficiency of training a double-source CT scatter analysis model is improved, the neural network model is trained by utilizing the head model double-source CT projection data, the comprehensive cross scatter signal and the forward scatter signal, the double-source CT scatter analysis model is obtained, the scatter signal is estimated by utilizing the double-source CT scatter analysis model to the pre-obtained target double-source CT projection data, the scatter signal is obtained by utilizing the target double-source CT projection data to subtract the scatter signal, the target double-source CT scatter correction data is obtained, and the efficiency and the accuracy of scatter correction of the double-source CT projection data are improved. Therefore, the double-source CT scattering correction method and system based on the depth learning model can solve the problem that the existing double-source CT system actually images X-ray scattering.
Drawings
FIG. 1 is a flow chart of a dual-source CT scatter correction method based on a deep learning model according to an embodiment of the present invention;
FIG. 2 is a schematic flow chart of acquiring integrated cross scatter signals according to an embodiment of the present invention;
FIG. 3 is a flow chart of acquiring forward scattering signals according to an embodiment of the present invention;
FIG. 4 is a functional block diagram of a dual source CT scatter correction system based on a deep learning model according to an embodiment of the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The embodiment of the application provides a double-source CT scattering correction method based on a deep learning model. The execution main body of the double-source CT scattering correction method based on the deep learning model comprises, but is not limited to, at least one of a server, a terminal and the like which can be configured to execute the method provided by the embodiment of the application. In other words, the dual source CT scatter correction method based on the deep learning model may be performed by software or hardware installed in a terminal device or a server device, and the software may be a blockchain platform. The server side comprises, but is not limited to, a single server, a server cluster, a cloud server or a cloud server cluster and the like. The server may be an independent server, or may be a cloud server that provides cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (Content Delivery Network, CDN), and basic cloud computing services such as big data and artificial intelligence platforms.
Referring to fig. 1, a flow chart of a dual-source CT scatter correction method based on a deep learning model according to an embodiment of the invention is shown. In this embodiment, the dual-source CT scatter correction method based on the deep learning model includes:
S1, head model double-source CT projection data of a preset head model are obtained.
In the embodiment of the invention, the dual-source CT projection data is projection data obtained by utilizing a dual-source CT system, the dual-source CT system respectively obtains high-energy and low-energy projection attenuation information of an object under various angles by using X-rays with two energies or energy spectrums in a CT scanning process, and then obtains a distribution diagram about the atomic number and electron density of the imaged object by utilizing a dual-source CT image reconstruction algorithm according to the relation between the linear attenuation coefficient of the substance about the X-rays and the ray energy, the atomic number and density of the substance.
In detail, the dual source CT system includes a first tube, a second tube, a first detector, and a second detector.
In detail, the first detector user receives the signal of the first tube and the second detector user receives the signal of the second tube.
In the embodiment of the invention, the head model is a preset head model used for simulating the head of a patient in an actual use situation.
In the embodiment of the invention, the head model dual-source CT projection data of the preset head model is obtained by utilizing the first detector to receive a ray signal that rays emitted by the first ray tube pass through the head model, and utilizing the second detector to receive a ray signal that rays emitted by the first ray tube pass through the head model, and utilizing a dual-source CT image reconstruction algorithm to construct the head model dual-source CT projection data according to two groups of ray signals.
In the embodiment of the invention, the efficiency of subsequently establishing the double-source CT scattering analysis model is improved by acquiring the head model double-source CT projection data of the preset head model.
S2, acquiring comprehensive cross scattering signals of the head model double-source CT projection data.
In the embodiment of the invention, the comprehensive cross scattering signal refers to that in a dual-source CT system, because two ray tubes simultaneously emit X-rays, the rays may cross when passing through a human body, so as to generate the comprehensive cross scattering signal, and the comprehensive cross scattering signal increases noise of an image and reduces definition of the image.
In an embodiment of the present invention, referring to fig. 2, a flowchart of acquiring a comprehensive cross scatter signal is provided for an embodiment of the present invention, where the acquiring the comprehensive cross scatter signal of the head model dual-source CT projection data includes:
S21, starting a first ray tube of a preset dual-source CT system and a preset second detector;
s22, acquiring a first cross scattering signal of the first ray tube by using the second detector;
S23, closing the first ray tube and the second detector, and opening the second ray tube and the first detector of a preset dual-source CT system;
s24, acquiring a second cross scattering signal of the second ray tube by using the first detector;
And S25, summarizing the first cross heat dissipation signal and the second cross scattering signal to obtain a comprehensive cross scattering signal.
In detail, since the second detector is located at the side of the first tube and the first detector is located at the side of the second tube, the first tube and the second detector can be turned on separately, and the second tube and the first detector can be turned on separately, so that a real scattering signal can be obtained.
In the embodiment of the invention, the efficiency of the subsequent training of the double-source CT scattering analysis model is improved by acquiring the comprehensive cross scattering signal of the head model double-source CT projection data.
S3, acquiring forward scattering signals of the head model double-source CT projection data by using a preset self-adaptive scattering kernel superposition algorithm.
In the embodiment of the invention, the forward scattering signal refers to that when the X-ray passes through the inside of human tissue, the X-ray is deviated due to different substances, so that the forward scattering signal is generated.
In the embodiment of the invention, the adaptive scattering kernel superposition algorithm is a technology used for image processing and computer vision, and is particularly used in image segmentation, feature extraction and pattern recognition. Such algorithms enhance the local features of the image by decomposing the image into different scattering components and improve the accuracy of the image analysis.
In an embodiment of the present invention, referring to fig. 3, a schematic flow chart for acquiring a forward scattering signal is provided in an embodiment of the present invention, where the acquiring a forward scattering signal of the head mode dual-source CT projection data by using a preset adaptive scattering kernel superposition algorithm includes:
S31, simulating scattering signals of different energy levels emitted by the high-voltage bulb tube under different currents and voltages through a preset water model with different lengths by Monte Carlo simulation to obtain a scattering signal set;
S32, fitting the distribution of each scattering signal in the scattering signal set to obtain scattering signal distribution of the ray bundle passing through water modes with different lengths under different energy levels;
S33, traversing each pixel of the head model double-source CT projection data according to the scattering signal distribution to obtain the scattering distribution of each ray in the head model double-source CT projection data;
s34, superposing scattering distribution of each ray in the head model double-source CT projection data to obtain forward scattering signals of the head model double-source CT projection data.
In detail, the monte carlo simulation is a random sampling or statistical simulation based method that estimates a numerical solution or probability distribution of a complex problem by repeatedly generating random samples and calculating statistics.
In detail, simulating the scattering signals of the beams of different energy levels emitted by the high-voltage bulb tube under different currents and voltages passing through the preset water modes with different lengths through Monte Carlo simulation means that random current values and voltage values are generated through Monte Carlo simulation, the beams of different energy levels are generated according to the randomly generated current values and voltage values, and the scattering signals of the beams of different energy levels passing through the water modes with different lengths are obtained.
In the embodiment of the invention, the forward scattering signal of the head model double-source CT projection data is obtained by utilizing the preset self-adaptive scattering kernel superposition algorithm, so that the accuracy of the obtained forward scattering signal is improved, and the accuracy of a subsequently established double-source CT scattering analysis model is improved.
And S4, training a neural network model by using the head model double-source CT projection data, the comprehensive cross scattering signals and the forward scattering signals to obtain a double-source CT scattering analysis model.
In the embodiment of the invention, the head model dual-source CT projection data is a data set containing continuous frame projection map information.
In the embodiment of the present invention, the training of the neural network model by using the head-mode dual-source CT projection data, the integrated cross scatter signal, and the forward scatter signal refers to training the neural network model by using the head-mode dual-source CT projection data as training data, and using the integrated cross scatter signal and the forward scatter signal as labels.
In an embodiment of the present invention, the neural network model includes, but is not limited to, a convolutional neural network model and a recurrent neural network model.
In the embodiment of the present invention, training a neural network model by using the head model dual-source CT projection data, the integrated cross scatter signal and the forward scatter signal to obtain a dual-source CT scatter analysis model includes:
obtaining projection image signals in the head model double-source CT projection data, and carrying out standardization processing on the projection image signals to obtain standardized projection image signals;
Combining the integrated cross scatter signal with the forward scatter signal to obtain a standard scatter signal;
Carrying out scattering signal analysis according to the standardized projection graph signals by utilizing a neural network model to obtain scattering signal estimation;
calculating a loss value of the scattered signal estimate and the standard scattered signal;
optimizing the loss value by updating weight parameters of the neural network model by using a random gradient descent algorithm;
Updating the learning rate of the neural network model by using a preset stepping learning rate algorithm to obtain an updated learning rate;
when the preset iteration times are reached, selecting a weight parameter corresponding to the minimum loss value in all the iteration times as a final weight parameter;
and inputting the updated learning rate and the final weight parameters into a neural network model to obtain a dual-source CT scattering analysis model.
In detail, the normalizing the projection image signal to obtain a normalized projection image signal includes:
and carrying out standardization processing on the projection image signal by using the following formula to obtain a standardized projection image signal:
Wherein p is the standardized projection map signal, I is the projection map signal, and I 0 is the bright field projection map signal contained in the head model dual source CT projection data.
In detail, the bright-field projection image signal refers to an X-ray image without attenuation in CT imaging, that is, an image obtained assuming that all X-rays are not absorbed or scattered;
In detail, the calculating the loss value of the scattered signal estimate and the standard scattered signal comprises:
the loss value is calculated using the following formula:
Wherein S is the loss value, N is the number of pixels of the first detector and the second detector, I s is the standard scatter signal, I m is the scatter signal estimate, and I p is the normalized projection map signal.
In detail, the random gradient descent algorithm is used for updating the weight parameters of the neural network model by updating the preset weight parameters through preset weight attenuation coefficients. The preset weight decay factor may be 0.0001 and the preset weight parameter may be 0.9.
In detail, the updating the learning rate of the neural network model by using a preset step learning rate algorithm to obtain an updated learning rate includes:
updating the learning rate using the following formula:
Wherein lr new is the updated learning rate, lr old is the learning rate, K is the current iteration number, and M is the preset iteration number.
In detail, the preset number of iterations may be 300.
In an embodiment of the present invention, the weight parameters are variables of the connected neurons in the neural network model, which determine the strength of interactions between the neurons.
In the embodiment of the invention, the learning rate is an important super-parameter in the training process of the neural network model, and controls the step size of the model to be adjusted during each weight update.
In the embodiment of the invention, before training the dual-source CT scattering analysis model to reach the preset iteration times, if the loss value is smaller than the preset loss value threshold, the dual-source CT scattering analysis model is directly confirmed to be trained, and the dual-source CT scattering analysis model is prevented from being overfitted.
In the embodiment of the invention, the neural network model is trained by utilizing the head model double-source CT projection data, the comprehensive cross scattering signals and the forward scattering signals to obtain the double-source CT scattering analysis model, so that the efficiency of training the double-source CT scattering analysis model is improved.
S5, utilizing the double-source CT scattering analysis model to conduct scattering signal estimation on the pre-acquired target double-source CT projection data, and obtaining corrected scattering signals.
In the embodiment of the present invention, the pre-acquired target dual-source CT projection data may be real dual-source CT data of the patient acquired by using the dual-source CT system in an actual medical work.
In the embodiment of the present invention, the performing scattering signal estimation on pre-acquired target dual-source CT projection data by using the dual-source CT scattering analysis model to obtain corrected scattering signals includes:
moving a convolution kernel preset in a convolution layer of the double-source CT scattering analysis model on the target double-source CT projection data according to a preset step length, calculating dot products of the convolution kernel and data of a convolution kernel coverage area after each movement, and summing all the dot products to obtain convolution characteristics;
Continuously moving a rectangular window in a pooling layer of the dual-source CT scattering analysis model in the convolution characteristic, calculating the average value of data of a region covered by the rectangular window after each movement, and replacing the data of the region with the calculated average value of each region to obtain the pooling characteristic;
Introducing nonlinear factors into the pooling feature by utilizing an activation function preset in an activation layer of the dual-source CT scattering analysis model to obtain an activation feature;
And obtaining the output of each neuron by utilizing the full-connection layer of the dual-source CT scattering analysis model according to the activation characteristics, and summing the output of each neuron to obtain a corrected scattering signal.
In detail, the convolution kernel is a preset square matrix whose scale is much smaller than the target dual-source CT projection data.
In detail, the preset step size determines the amplitude of each movement of the convolution kernel, and the longer step size is, the higher the efficiency of extracting convolution characteristics is, but the information is lost, and the too short step size increases the calculation amount.
In detail, the activation function is a function added to the nodes between the neural network model layers for introducing nonlinear factors, so that the neural network model can learn and process complex problems. Common activation functions include Sigmoid functions, softmax functions, and ReLU functions.
In the embodiment of the invention, the scattering signal is estimated by utilizing the double-source CT scattering analysis model to pre-obtain the target double-source CT projection data, so that the corrected scattering signal is obtained, and the efficiency and accuracy of correcting the double-source CT signal are improved.
And S6, subtracting the corrected scattering signal from the target double-source CT projection data to obtain target double-source CT scattering correction data.
In an embodiment of the present invention, the corrected scatter signal includes a composite cross scatter signal and a forward scatter signal.
In the embodiment of the present invention, subtracting the scatter signal from the target dual-source CT projection data to obtain target dual-source CT scatter correction data includes:
And carrying out scattering correction on the target double-source CT projection data by using the following formula:
Wherein I corr is the target dual-source CT scatter correction data, I total is the target dual-source CT projection data, and I scatter,model is the scatter signal.
In the embodiment of the invention, the correction scattering signal is subtracted by utilizing the target double-source CT projection data to obtain the target double-source CT scattering correction data, so that the correction efficiency of the double-source CT projection data is improved.
FIG. 4 is a functional block diagram of a dual source CT scatter correction system based on a deep learning model according to an embodiment of the present invention.
The dual-source CT scatter correction system 100 based on the deep learning model of the present invention may be installed in an electronic device. Depending on the functionality implemented, the deep learning model-based dual source CT scatter correction system 100 may include a data acquisition module 101, a scatter prediction module 102, a model training module 103, and a scatter correction module 104. The module of the invention, which may also be referred to as a unit, refers to a series of computer program segments, which are stored in the memory of the electronic device, capable of being executed by the processor of the electronic device and of performing a fixed function.
In the present embodiment, the functions concerning the respective modules/units are as follows:
The data acquisition module 101 is configured to acquire head model dual-source CT projection data of a preset head model, and acquire a comprehensive cross scatter signal of the head model dual-source CT projection data;
The scattering estimation module 102 is configured to obtain forward scattering signals of the head-mode dual-source CT projection data by using a preset adaptive scattering kernel superposition algorithm;
the model training module 103 is configured to train a neural network model by using the head model dual-source CT projection data, the integrated cross scatter signal, and the forward scatter signal, to obtain a dual-source CT scatter analysis model;
the scatter correction module 104 is configured to perform scatter signal estimation on pre-acquired target dual-source CT projection data by using the dual-source CT scatter analysis model to obtain a scatter signal, and subtract the scatter signal from the target dual-source CT projection data to obtain target dual-source CT scatter correction data.
In detail, each module in the deep learning model-based dual-source CT scatter correction system 100 in the embodiment of the present invention adopts the same technical means as the deep learning model-based dual-source CT scatter correction method described in fig. 1 to 3, and can produce the same technical effects, which are not described herein.
In the embodiments provided in the present invention, it should be understood that the disclosed apparatus, system and method may be implemented in other manners. For example, the system embodiments described above are merely illustrative, e.g., the division of the modules is merely a logical function division, and other manners of division may be implemented in practice.
The modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional module in the embodiments of the present invention may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units can be realized in a form of hardware or a form of hardware and a form of software functional modules.
It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof.
The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.
The embodiment of the application can acquire and process the related data based on the artificial intelligence technology. Wherein artificial intelligence (ARTIFICIAL INTELLIGENCE, AI) is the theory, method, technique, and application system that uses a digital computer or a digital computer-controlled machine to simulate, extend, and expand human intelligence, sense the environment, acquire knowledge, and use knowledge to obtain optimal results.
Furthermore, it is evident that the word "comprising" does not exclude other elements or steps, and that the singular does not exclude a plurality. A plurality of units or means recited in the system claims can also be implemented by means of software or hardware by means of one unit or means. The terms first, second, etc. are used to denote a name, but not any particular order.
Finally, it should be noted that the above-mentioned embodiments are merely for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications and equivalents may be made to the technical solution of the present invention without departing from the spirit and scope of the technical solution of the present invention.

Claims (10)

1.一种基于深度学习模型的双源CT散射校正方法，其特征在于，所述方法包括：1. A dual-source CT scatter correction method based on a deep learning model, characterized in that the method comprises: 获取预设的头模的头模双源CT投影数据；Acquire dual-source CT projection data of a preset head model; 获取所述头模双源CT投影数据的综合交叉散射信号；Acquire a comprehensive cross-scatter signal of the head model dual-source CT projection data; 利用预设的自适应散射核叠加算法获取所述头模双源CT投影数据的前向散射信号；Acquire the forward scattering signal of the head model dual-source CT projection data by using a preset adaptive scattering kernel superposition algorithm; 利用所述头模双源CT投影数据、所述综合交叉散射信号以及所述前向散射信号训练神经网络模型，得到双源CT散射分析模型；Training a neural network model using the head model dual-source CT projection data, the integrated cross-scatter signal, and the forward-scatter signal to obtain a dual-source CT scatter analysis model; 利用所述双源CT散射分析模型对预先获取的目标双源CT投影数据进行散射信号预估，得到散射信号；Using the dual-source CT scattering analysis model to estimate scattering signals of pre-acquired target dual-source CT projection data, to obtain scattering signals; 利用所述目标双源CT投影数据减去所述散射信号，得到目标双源CT散射矫正数据。The scatter signal is subtracted from the target dual-source CT projection data to obtain target dual-source CT scatter correction data. 2.如权利要求1所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述获取所述头模双源CT投影数据的综合交叉散射信号，包括：2. The dual-source CT scatter correction method based on a deep learning model according to claim 1, wherein the step of obtaining a comprehensive cross-scatter signal of the head model dual-source CT projection data comprises: 开启预设双源CT系统的第一射线管以及预设的第二探测器；Turning on a first ray tube and a second detector of a preset dual-source CT system; 利用所述第二探测器获取所述第一射线管的第一交叉散射信号；Acquiring a first cross-scatter signal of the first ray tube using the second detector; 关闭所述第一射线管以及所述第二探测器，开启预设双源CT系统的第二射线管以及第一探测器；Turn off the first ray tube and the second detector, and turn on the second ray tube and the first detector of the preset dual-source CT system; 利用所述第一探测器获取所述第二射线管的第二交叉散射信号；Acquiring a second cross-scatter signal of the second ray tube using the first detector; 汇总所述第一交叉散热信号以及所述第二交叉散射信号，得到综合交叉散射信号。The first cross heat dissipation signal and the second cross scattering signal are aggregated to obtain a comprehensive cross scattering signal. 3.如权利要求2所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述利用预设的自适应散射核叠加算法获取所述头模双源CT投影数据的前向散射信号，包括：3. The dual-source CT scatter correction method based on a deep learning model according to claim 2, wherein the step of obtaining the forward scatter signal of the head model dual-source CT projection data by using a preset adaptive scatter kernel superposition algorithm comprises: 通过蒙特卡洛仿真，模拟不同电流、电压下由高压球管发射的不同能级的射线束穿过预设的不同长度水模的散射信号，得到散射信号集合；Through Monte Carlo simulation, the scattered signals of the ray beams of different energy levels emitted by the high-voltage bulb under different currents and voltages passing through the preset water models of different lengths are simulated to obtain the scattered signal set; 对所述散射信号集合中每一个散射信号的分布进行拟合，得到不同能级下射线束穿过不同长度水模的散射信号分布；Fitting the distribution of each scattering signal in the scattering signal set to obtain the scattering signal distribution of the ray beam passing through water models of different lengths at different energy levels; 根据所述散射信号分布遍历所述头模双源CT投影数据每一个像素，得到所述头模双源CT投影数据中每一条射线的散射分布；Traversing each pixel of the head model dual-source CT projection data according to the scattered signal distribution, to obtain the scattered distribution of each ray in the head model dual-source CT projection data; 将所述头模双源CT投影数据中每一条射线的散射分布叠加，得到所述头模双源CT投影数据的前向散射信号。The scattering distribution of each ray in the head model dual-source CT projection data is superimposed to obtain a forward scattering signal of the head model dual-source CT projection data. 4.如权利要求3所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述利用所述头模双源CT投影数据、所述综合交叉散射信号以及所述前向散射信号训练神经网络模型，得到双源CT散射分析模型，包括：4. The dual-source CT scatter correction method based on a deep learning model according to claim 3, characterized in that the use of the head model dual-source CT projection data, the integrated cross-scatter signal and the forward scatter signal to train a neural network model to obtain a dual-source CT scatter analysis model comprises: 获取所述头模双源CT投影数据中的投影图信号，对所述投影图信号进行标准化处理，得到标准化投影图信号；Acquire projection image signals in the head model dual-source CT projection data, and perform standardization processing on the projection image signals to obtain standardized projection image signals; 将所述综合交叉散射信号与所述前向散射信号结合，得到标准散射信号；combining the integrated cross scatter signal with the forward scatter signal to obtain a standard scatter signal; 利用神经网络模型根据所述标准化投影图信号进行散射信号分析，得到散射信号估计；Using a neural network model to perform scattering signal analysis based on the standardized projection image signal to obtain a scattering signal estimate; 计算所述散射信号估计与所述标准散射信号的损失值；Calculating a loss value between the estimated scattered signal and the standard scattered signal; 利用随机梯度下降算法通过更新所述神经网络模型的权重参数优化所述损失值；Optimizing the loss value by updating the weight parameters of the neural network model using a stochastic gradient descent algorithm; 利用预设的步进学习率算法更新所述神经网络模型的学习率，得到更新后的学习率；Using a preset step learning rate algorithm to update the learning rate of the neural network model to obtain an updated learning rate; 当达到预设的迭代次数后，选取所有迭代次数中损失值最小时对应的权重参数作为最终权重参数；When the preset number of iterations is reached, the weight parameter corresponding to the minimum loss value among all iterations is selected as the final weight parameter; 将所述更新后的学习率以及所述最终权重参数输入神经网络模型中，得到双源CT散射分析模型。The updated learning rate and the final weight parameter are input into a neural network model to obtain a dual-source CT scattering analysis model. 5.如权利要求4所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述对所述投影图信号进行标准化处理，得到标准化投影图信号，包括：5. The dual-source CT scatter correction method based on a deep learning model according to claim 4, wherein the step of normalizing the projection image signal to obtain a standardized projection image signal comprises: 利用如下公式对所述投影图信号进行标准化处理，得到标准化投影图信号：The projection image signal is standardized using the following formula to obtain a standardized projection image signal: 其中，p为所述标准化投影图信号，I为所述投影图信号，I0为所述头模双源CT投影数据中所包含的亮场投影图信号。Wherein, p is the standardized projection image signal, I is the projection image signal, and I 0 is the bright field projection image signal contained in the head model dual-source CT projection data. 6.如权利要求4所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述计算所述散射信号估计与所述标准散射信号的损失值，包括：6. The dual-source CT scatter correction method based on a deep learning model according to claim 4, wherein the calculating the loss value between the scatter signal estimate and the standard scatter signal comprises: 利用如下公式计算所述损失值：The loss value is calculated using the following formula: 其中，S为所述损失值，N为所述第一探测器以及所第二探测器像素的数量，Is为所述标准散射信号，Im为所述散射信号估计，Ip为所述标准化投影图信号。Wherein, S is the loss value, N is the number of pixels of the first detector and the second detector, Is is the standard scattering signal, Im is the scattering signal estimation, and Ip is the standardized projection image signal. 7.如权利要求4所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述利用预设的步进学习率算法更新所述神经网络模型的学习率，得到更新后的学习率，包括：7. The dual-source CT scatter correction method based on a deep learning model according to claim 4, characterized in that the step learning rate algorithm used to update the learning rate of the neural network model to obtain the updated learning rate comprises: 利用如下公式更新所述学习率：The learning rate is updated using the following formula: 其中，lrnew为所述更新后的学习率，lrold为所述学习率，K为当前的迭代次数，M为预设的迭代次数。Wherein, lr new is the updated learning rate, lr old is the learning rate, K is the current number of iterations, and M is the preset number of iterations. 8.如权利要求4所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述利用所述双源CT散射分析模型对预先获取的目标双源CT投影数据进行散射信号预估，得到矫正散射信号，包括：8. The dual-source CT scatter correction method based on a deep learning model according to claim 4, wherein the step of using the dual-source CT scatter analysis model to estimate scatter signals of pre-acquired target dual-source CT projection data to obtain corrected scatter signals comprises: 利用预设在所述双源CT散射分析模型的卷积层中的卷积核在所述目标双源CT投影数据上按照预设的步长移动，计算每一次移动后卷积核与卷积核覆盖区域的数据的点积，对所有点积进行求和，得到卷积特征；Using a convolution kernel preset in a convolution layer of the dual-source CT scattering analysis model to move on the target dual-source CT projection data according to a preset step length, calculating the dot product of the convolution kernel and the data of the area covered by the convolution kernel after each movement, summing all dot products, and obtaining a convolution feature; 利用预设在所述双源CT散射分析模型的池化层中的矩形窗口在所述卷积特征中连续移动，计算每次移动后矩形窗口所覆盖区域的数据的平均值，将每个区域计算出的平均值替换该区域的数据，得到池化特征；A rectangular window preset in the pooling layer of the dual-source CT scattering analysis model is used to continuously move in the convolution feature, an average value of data in the area covered by the rectangular window after each movement is calculated, and the average value calculated in each area replaces the data in the area to obtain a pooling feature; 利用所述双源CT散射分析模型的激活层中预设的激活函数为所述池化特征引入非线性因素，得到激活特征；Using the activation function preset in the activation layer of the dual-source CT scattering analysis model to introduce a nonlinear factor into the pooling feature, so as to obtain an activation feature; 利用所述双源CT散射分析模型的全连接层根据所述激活特征得到每个神经元的输出，对每个神经元的输出求和，得到矫正散射信号。The output of each neuron is obtained according to the activation feature using the fully connected layer of the dual-source CT scatter analysis model, and the output of each neuron is summed to obtain a corrected scatter signal. 9.如权利要求1所述的基于深度学习模型的双源CT散射校正方法，其特征在于，所述利用所述目标双源CT投影数据减去所述散射信号，得到目标双源CT散射矫正数据，包括：9. The dual-source CT scatter correction method based on a deep learning model according to claim 1, wherein the step of subtracting the scatter signal from the target dual-source CT projection data to obtain the target dual-source CT scatter correction data comprises: 利用如下公式对所述目标双源CT投影数据进行散射矫正：The target dual-source CT projection data is scatter corrected using the following formula: 其中，Icorr为所述目标双源CT散射矫正数据，Itotal为所述目标双源CT投影数据，Iscatter,model为所述散射信号。Among them, I corr is the target dual-source CT scatter correction data, I total is the target dual-source CT projection data, and I scatter,model is the scatter signal. 10.一种基于深度学习模型的双源CT散射校正系统，其特征在于，所述系统包括：10. A dual-source CT scatter correction system based on a deep learning model, characterized in that the system comprises: 数据获取模块，用于获取预设的头模的头模双源CT投影数据，获取所述头模双源CT投影数据的综合交叉散射信号；A data acquisition module, used to acquire dual-source CT projection data of a preset head model, and to acquire a comprehensive cross-scatter signal of the dual-source CT projection data of the head model; 散射预估模块，用于利用预设的自适应散射核叠加算法获取所述头模双源CT投影数据的前向散射信号；A scattering estimation module, used for acquiring a forward scattering signal of the head model dual-source CT projection data by using a preset adaptive scattering kernel superposition algorithm; 模型训练模块，用于利用所述头模双源CT投影数据、所述综合交叉散射信号以及所述前向散射信号训练神经网络模型，得到双源CT散射分析模型；A model training module, used to train a neural network model using the head model dual-source CT projection data, the integrated cross-scattering signal and the forward scattering signal to obtain a dual-source CT scattering analysis model; 散射矫正模块，用于利用所述双源CT散射分析模型对预先获取的目标双源CT投影数据进行散射信号预估，得到散射信号，利用所述目标双源CT投影数据减去所述散射信号，得到目标双源CT散射矫正数据。The scatter correction module is used to use the dual-source CT scatter analysis model to estimate the scatter signal of the pre-acquired target dual-source CT projection data to obtain the scatter signal, and subtract the scatter signal from the target dual-source CT projection data to obtain the target dual-source CT scatter correction data.
CN202411232508.4A 2024-09-04 2024-09-04 Dual-source CT scatter correction method and system based on deep learning model Pending CN119097328A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202411232508.4A CN119097328A (en) 2024-09-04 2024-09-04 Dual-source CT scatter correction method and system based on deep learning model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202411232508.4A CN119097328A (en) 2024-09-04 2024-09-04 Dual-source CT scatter correction method and system based on deep learning model

Publications (1)

Publication Number Publication Date
CN119097328A true CN119097328A (en) 2024-12-10

Family

ID=93715628

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202411232508.4A Pending CN119097328A (en) 2024-09-04 2024-09-04 Dual-source CT scatter correction method and system based on deep learning model

Country Status (1)

Country Link
CN (1) CN119097328A (en)

Similar Documents

Publication Publication Date Title
US11074506B2 (en) Estimating the amount of degradation with a regression objective in deep learning
CN108898642A (en) A kind of sparse angular CT imaging method based on convolutional neural networks
CN112348936A (en) Low-dose cone-beam CT image reconstruction method based on deep learning
CN110570492A (en) Neural network training method and device, image processing method and device, and medium
CN112598649A (en) 2D/3D spine CT non-rigid registration method based on generation of countermeasure network
CN108280859B (en) A CT sparse projection image reconstruction method and device with limited sampling angle
JP2021524631A (en) Medical image conversion
US20210330274A1 (en) Computer-implemented method, computer program, systems and x-ray facility for correction of x-ray image data with regard to noise effects
WO2024032098A1 (en) Training method and apparatus for artifact removal model, and device, medium and program product
JP2020179146A (en) Medical image processing equipment, X-ray computed tomography equipment and programs
JP2023546208A (en) Method and system for training and tuning neural network models for noise removal
Zhang et al. A locally weighted linear regression look-up table-based iterative reconstruction method for dual spectral CT
JP2017221339A (en) X-ray CT image reconstruction method and computer program
CN111080736B (en) A low-dose CT image reconstruction method based on sparse transformation
CN119097328A (en) Dual-source CT scatter correction method and system based on deep learning model
Khattak et al. High energy physics calorimeter detector simulation using generative adversarial networks with domain related constraints
CN118537489A (en) Projection domain material decomposition method and medium for dual-energy CBCT
CN119107376A (en) CT scatter correction method, device, electronic equipment and storage medium
Voss et al. BayesDose: Comprehensive proton dose prediction with model uncertainty using Bayesian LSTMs
Rehm et al. Quantum machine learning for HEP detector simulations
Lorenzon et al. Joint reconstruction and scatter estimation in cone-beam CT using diffusion posterior sampling
Li et al. Intensity Field Decomposition for Tissue-Guided Neural Tomography
CN119672146B (en) A method, system, device and medium for correcting bone sclerosis artifacts in CBCT images
CN120022015B (en) Micro-dose CT system and data acquisition method
Grama et al. Explainable AI for raising confidence in deep learning‐based tumor tracking models

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination