WO2024036760A1 - 能谱图像处理方法、装置、计算机设备及存储介质 - Google Patents

能谱图像处理方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2024036760A1
WO2024036760A1 PCT/CN2022/129561 CN2022129561W WO2024036760A1 WO 2024036760 A1 WO2024036760 A1 WO 2024036760A1 CN 2022129561 W CN2022129561 W CN 2022129561W WO 2024036760 A1 WO2024036760 A1 WO 2024036760A1
Authority
WO
WIPO (PCT)
Prior art keywords
energy spectrum
data
image sequence
spectrum data
material image
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.)
Ceased
Application number
PCT/CN2022/129561
Other languages
English (en)
French (fr)
Inventor
李印生
梁栋
刘新
郑海荣
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.)
Shenzhen Institute of Advanced Technology of CAS
Original Assignee
Shenzhen Institute of Advanced Technology of CAS
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 Shenzhen Institute of Advanced Technology of CAS filed Critical Shenzhen Institute of Advanced Technology of CAS
Publication of WO2024036760A1 publication Critical patent/WO2024036760A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T12/00—Tomographic reconstruction from projections
    • G06T12/30—Image post-processing, e.g. metal artefact correction
    • 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
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T11/00—Two-dimensional [2D] image generation
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T12/00—Tomographic reconstruction from projections
    • G06T12/10—Image preprocessing, e.g. calibration, positioning of sources or scatter correction

Definitions

  • Embodiments of the present application relate to the field of medical image processing, and in particular, to an energy spectrum image processing method, device, computer equipment and storage medium.
  • Spectral CT which collects two or more X-ray spectrum data sets, can be used to solve this long-standing technical limitation in traditional CT imaging.
  • Energy spectrum CT involves many research directions, among which data acquisition technology and quantitative material decomposition technology are the most important research directions.
  • model-driven material decomposition methods are mainly divided into indirect material decomposition methods and direct material decomposition methods.
  • Indirect material decomposition method The first subcategory of this type of method first performs material decomposition in projected space, followed by analysis or iterative image reconstruction of material baseline integration data. Since substance decomposition is an unstable process, this type of method is sensitive to noise, especially under low-dose conditions, and it is difficult to obtain accurate substance decomposition.
  • the second subcategory of this class of methods begins with iterative reconstruction of spectral data, followed by image space material decomposition. This type of method is relatively simple, but cannot eliminate wire harness hardening artifacts and metal artifacts.
  • Direct material decomposition method is relatively simple, but cannot eliminate wire harness hardening artifacts and metal artifacts.
  • the method of reconstructing material-based images directly from energy spectrum data models the physical process of data collection and introduces prior knowledge about the base material images to improve the quantitative accuracy of material decomposition.
  • the nonlinear multi-energy X-ray measurement model makes the direct matter decomposition problem highly ill-posed, especially under low-dose data acquisition conditions.
  • data-driven methods have begun to be applied to energy spectrum CT material decomposition, including image space decomposition network, projection space decomposition network, and projection-image end-to-end decomposition network.
  • Embodiments of the present application provide an energy spectrum image processing method, device, computer equipment and storage medium, which solve the problem of low quantitative accuracy of material decomposition in existing energy spectrum image processing methods.
  • embodiments of the present application provide an energy spectrum image processing method, including:
  • the initial material image sequence is used as the expected material image sequence
  • the material data is adjusted according to the difference, and the material image sequence corresponding to the adjusted material data is used as the expected material image sequence.
  • embodiments of the present application also provide an energy spectrum image processing method, including:
  • the material image module is used to determine the initial material image sequence corresponding to the measured energy spectrum data
  • the energy spectrum data estimation module inputs the material data corresponding to the initial material image sequence into the trained energy spectrum data model to obtain estimated energy spectrum data;
  • a first output module configured to use the initial material image sequence as the expected material image sequence if the difference between the estimated energy spectrum data and the measured energy spectrum data is within a set error range
  • the second output module is used to adjust the material data according to the difference if the difference between the estimated energy spectrum data and the measured energy spectrum data is not within the set error range, and store the adjusted material data corresponding to The material image sequence serves as the desired material image sequence.
  • embodiments of the present application further provide a computer device, where the computer device includes:
  • processors one or more processors
  • a storage device for storing one or more programs
  • the one or more processors When the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the energy spectrum image processing method as described in any embodiment.
  • a fourth aspect is a storage medium containing computer-executable instructions, which when executed by a computer processor are used to perform the energy spectrum image processing method according to any embodiment.
  • the technical solution of the energy spectrum image processing method determines the initial material image sequence corresponding to the measured energy spectrum data; inputs the material data corresponding to the initial material image sequence into the trained energy spectrum data model to obtain the estimated energy spectrum data ; If the difference between the estimated energy spectrum data and the measured energy spectrum data is within the set error range, the initial material image sequence is used as the expected material image sequence; if the difference between the estimated energy spectrum data and the measured energy spectrum data is not within the set error range within, the material data is adjusted according to the difference, and the material image sequence corresponding to the adjusted material data is used as the expected material image sequence.
  • This material data adjustment method can ensure the speed and effectiveness of material data adjustment; estimated energy spectrum data and measurement
  • the calculation amount of determining the difference between energy spectrum data and adjusting the material data is relatively small, so it can meet the requirements of clinical image processing speed.
  • Figure 1 is a flow chart of an energy spectrum image processing method provided by an embodiment of the present application.
  • Figure 2A is a schematic diagram of dual-source dual-detector acquisition of energy spectrum data provided by an embodiment of the present application
  • Figure 2B is a schematic diagram of obtaining energy spectrum data using the fast tube voltage switching method provided by the embodiment of the present application.
  • Figure 2C is a schematic diagram of obtaining energy spectrum data using the wire harness filtering method provided by the embodiment of the present application.
  • Figure 2D is a schematic diagram of the photon counting detector provided by the embodiment of the present application for acquiring energy spectrum data
  • Figure 2E is a schematic diagram of the double-layer detector provided by the embodiment of the present application for acquiring energy spectrum data
  • FIG. 3 is a schematic flow chart of the energy spectrum image processing method provided by the embodiment of the present application.
  • Figure 4A is a schematic diagram of a phantom provided by an embodiment of the present application.
  • Figure 4B is a schematic diagram of experimental results provided by the embodiment of the present application.
  • Figure 5A is a flow chart of the energy spectrum data model training method provided by the embodiment of the present application.
  • Figure 5B is a schematic diagram of the energy spectrum data model provided by the embodiment of the present application.
  • Figure 6 is a flow chart of the material model training method provided by the embodiment of the present application.
  • Figure 7 is an architectural schematic diagram of the material model provided by the embodiment of the present application.
  • Figure 8A is a schematic structural diagram of an energy spectrum image processing device provided by an embodiment of the present application.
  • Figure 8B is a schematic structural diagram of another energy spectrum image processing device provided by an embodiment of the present application.
  • Figure 9 is a schematic structural diagram of a C-arm CT system provided by an embodiment of the present application.
  • Figure 10A is a perspective view of a CT system provided by an embodiment of the present application.
  • Figure 10B is a left view of the CT system provided by the embodiment of the present application.
  • Figure 11 is a structural block diagram of a computer device provided by an embodiment of the present application.
  • FIG. 1 is a flow chart of an energy spectrum image processing method provided by an embodiment of the present application.
  • the technical solution of this embodiment is suitable for determining the material image corresponding to the energy spectrum image.
  • This method can be executed by the energy spectrum image processing device provided in the embodiment of the present application.
  • the device can be implemented in the form of software and/or hardware, and configured and applied in the processor of the computer device.
  • the method specifically includes the following steps:
  • the measured energy spectrum data includes projection data frames of at least two levels of ray energy.
  • the measured energy spectrum data includes at least one first projection data frame corresponding to the first X-ray energy and at least one second projection data frame corresponding to the second X-ray energy, and the at least one first projection data frame
  • the first X-ray energy is higher than the second X-ray energy corresponding to the projection angle and the detector element position respectively corresponding to the at least one first projection data frame.
  • the energy spectrum data can be measured through dual sources and dual detectors (Fig. 2A), fast (or slow) speed tube voltage switching (Fig. 2B), line beam filtering (Fig. 2C), and photon counting detectors (Fig. 2D ) or double-layer detector ( Figure 2E) or a combination of multiple methods.
  • the initial material image sequence is obtained based on the trained material model, specifically: determining the energy spectrum image sequence corresponding to the measured energy spectrum data; inputting the energy spectrum image sequence into the trained material model to obtain the initial material image sequence, see Figure 3.
  • the trained material model has a generalization prior in the image space and can change the energy spectrum image sequence to its corresponding initial image sequence.
  • the initial material image sequence determined based on the measured energy spectrum data generally has high image quality, but the local quantitative accuracy may be insufficient, and the material properties of each sample may not be accurately described. , so the material data corresponding to the initial material image sequence also has the problem that local quantitative accuracy may be insufficient.
  • the material data includes material-based data corresponding to the initial material image in the initial image sequence, and the detector element position and projection angle corresponding to each data in the material-based data.
  • the material data includes at least two material-based data.
  • each initial material image in the initial material image sequence is forward projected along the corresponding one or more projection angles to obtain one or more material data frames; it is determined that it belongs to the desired material All projection angles of the image, material data frames corresponding to all projection angles, and detector probe positions corresponding to all material data frames are used to obtain the material data of the desired material image.
  • the trained energy spectrum data model in this embodiment has prior knowledge describing the data collection process in the data space, so the material data can be synthesized into corresponding energy spectrum data.
  • the difference between the estimated energy spectrum data and the measured energy spectrum data is within the set error range, it means that the accuracy of the estimated energy spectrum data is very close to the measured energy spectrum data, and therefore is used to determine the estimated energy spectrum.
  • the material data of the data has a high accuracy, and the quantitative accuracy of the initial material image sequence used to determine the material data is within the user's acceptable range, so the initial image sequence is used as the desired material image sequence.
  • the quality of the estimated energy spectrum data predicted by the trained energy spectrum data model based on the material data may be lower than the expected energy spectrum data quality. For this reason, this embodiment uses measured energy spectrum data.
  • the material data is adjusted based on the difference between the estimated energy spectrum data and the measured energy spectrum data, and then the material image sequence corresponding to the adjusted material data is used as the expected material image sequence.
  • the estimated energy spectrum data frame included in the estimated energy spectrum data corresponds one-to-one to the measured energy spectrum data frame included in the measured energy spectrum data. Determine the error between each estimated energy spectrum data frame and each corresponding measured energy spectrum data frame, then calculate the sum of all errors, adjust the material data based on the sum of all errors, and then convert the adjusted material data corresponding to The material image sequence serves as the desired material image sequence.
  • the measured energy spectrum data is original projection data and carries accurate material information
  • the difference between the measured energy spectrum data and the estimated energy spectrum data can be used as a benchmark to accurately guide the adjustment of material data.
  • this embodiment only adjusts the material data based on the difference between the estimated energy spectrum data and the measured energy spectrum data, but does not adjust the network parameters of any model. That is, during the model use stage, the network parameters of all models remain unchanged. Change.
  • Figures 4A and 4B are quantitative experimental results determined based on physical phantom experiments provided by the embodiments of the present application.
  • the experimental phantom contains 4 test tubes of iodine solutions with different concentrations and 2 test tubes with calcium chloride solutions of different concentrations.
  • the energy spectrum CT experimental platform scanned it at 100kV.
  • the photon technology detector uses two energy spectrum thresholds to collect rays passing through the experimental phantom to obtain measured energy spectrum data, which includes projection data corresponding to high energy and projection data corresponding to low energy.
  • the technology of the present application produces a corresponding iodine-based image.
  • concentrations of the four iodine solution test tubes determined by the iodine-based image were compared with the gold standard (concentration color change diagram). The comparison results showed that the concentrations of the four iodine solution test tubes were quantitatively accurate. This experiment illustrates that the energy spectrum image processing method described in the embodiment of the present application has high substance quantification accuracy.
  • the technical solution of the energy spectrum image processing method determines the initial material image sequence corresponding to the measured energy spectrum data; inputs the material data corresponding to the initial material image sequence into the trained energy spectrum data model to obtain the estimated energy spectrum data ; If the difference between the estimated energy spectrum data and the measured energy spectrum data is within the set error range, the initial material image sequence is used as the expected material image sequence; if the difference between the estimated energy spectrum data and the measured energy spectrum data is not within the set error range within, the material data is adjusted based on the difference, and the material image sequence corresponding to the adjusted material data is used as the expected material image sequence.
  • This material data adjustment method can ensure the speed and effectiveness of material data adjustment; Estimated energy spectrum data and measurement The amount of calculation required to determine the difference before the energy spectrum data and adjust the material data is relatively small, so it can meet the requirements of clinical image processing speed.
  • Figure 5A is a flow chart of a method for training an energy spectrum data model according to yet another embodiment of the present application.
  • the model training method described in the embodiment of this application is applied before the energy spectrum image processing method described in the previous embodiment. Specific method steps include:
  • the material data includes at least two material-based data and the detector element position and projection angle corresponding to each data in the material-based data.
  • the specific architecture of the energy spectrum data model is shown in Figure 5B, which includes 4 fully connected layers.
  • This energy spectrum data model can model the complete physical process of data collection. Specifically, it learns the physical prior knowledge of the data space describing the data collection process from large-scale training data, so as to synthesize the material data into its corresponding energy data. spectral data.
  • the energy spectrum data spatial mean square error is used as the training loss function to adjust the network parameters of the energy spectrum data model.
  • Figure 6 is a flow chart of a material model training method provided by yet another embodiment of the present application.
  • the training methods of this material model include:
  • the initial energy spectrum image sequence and the initial material image sequence are used as training samples.
  • the initial energy spectrum image sequence is input into the material model to obtain an estimated material image sequence, and the network parameters of the material model are adjusted according to the difference between the initial material image sequence and the estimated material image sequence output by the material model until the initial material image sequence is consistent with the current Differences between estimated material image sequences are within a set error range.
  • the image space mean square error is used as the training loss function.
  • the material model learns a generalization prior in the image space from large-scale training data, thereby transforming the energy spectrum image into its corresponding initial material image sequence.
  • the network structure of the material model is shown in Figure 7.
  • the neural network has a total of 24 convolutional neural network layers.
  • the 24 convolutional neural network layers include three types of convolutional network layers. These convolutional network layers All parameters in are learnable.
  • the first type of convolutional network layer uses a 3 ⁇ 3 convolution kernel with an interval of 1, which is marked as “Conv(3 ⁇ 3)S1” in Figure 7, followed by the batch normalization operation (Bnorm) and the rectified linear unit ( ReLu) activation function.
  • the second type of convolutional network layer uses a 3 ⁇ 3 convolution kernel with an interval of 2 and is labeled “Conv(3 ⁇ 3)S2” in Figure 7, followed by Nnorm and ReLu.
  • the third type of convolutional network layer uses a 1 ⁇ 1 convolution kernel with an interval of 1, and is marked as “Conv(1 ⁇ 1)S1” in Figure 7, followed by a linear activation function. All convolutional layers have corresponding learnable bias terms. Each convolutional layer keeps the input and output of the layer to the same spatial dimension.
  • the sampling layer uses a 2 ⁇ 2 convolution kernel and is labeled Up-sample 2 ⁇ 2 in Figure 7. All sampling layers use bilinear interpolation algorithms. Shortcut links (skip+concatenate, black solid arrow in Figure 7) are used to facilitate the network training process.
  • the convolution kernel in the network parameters uses Glorot uniformly distributed random numbers as initialization, and the bias term uses 0 as the initialization value. The remaining parameter settings and initialization values are all in the default mode.
  • FIG. 8A is a structural block diagram of an energy spectrum image processing device provided by yet another embodiment of the present application.
  • the device is used to execute the energy spectrum image processing method provided in any of the above embodiments, and the device can be implemented as software or hardware.
  • the device includes:
  • the material image module 11 is used to determine the initial material image sequence corresponding to the measured energy spectrum data
  • the energy spectrum data estimation module 12 is used to input the material data corresponding to the initial material image sequence into the trained energy spectrum data model to obtain estimated energy spectrum data;
  • the first output module 13 is configured to use the initial material image sequence as the expected material image sequence if the difference between the estimated energy spectrum data and the measured energy spectrum data is within a set error range;
  • the second output module 14 is configured to adjust the material data according to the difference if the difference between the estimated energy spectrum data and the measured energy spectrum data is not within the set error range, and correspond the adjusted material data to The material image sequence is used as the desired material image sequence.
  • the material image module 11 is used to determine the energy spectrum image sequence corresponding to the measured energy spectrum data; input the energy spectrum image sequence into the trained material model to obtain an initial material image sequence.
  • the energy spectrum data estimation module 11 is used to determine one or more projection angles corresponding to each initial material image in the initial material image sequence; and project each initial material image in the initial material image sequence along the corresponding one or Perform forward projection on multiple projection angles to obtain one or more material data frames; determine all projection angles belonging to the desired material image, the material data frames corresponding to all projection angles, and the detector probe positions corresponding to all material data frames. To obtain the material data of the desired material image.
  • the measured energy spectrum data includes projection data frames of at least two levels of ray energy; the measured energy spectrum data corresponds to at least two material-based data.
  • the device also includes a model training module 10, which is used to input a set amount of material data into the energy spectrum data model to obtain estimated energy spectrum data;
  • the difference between the measured energy spectrum data and the estimated energy spectrum data repeatedly adjust the network parameters of the energy spectrum data model, and minimize the difference between the measured energy spectrum data and the adjusted network parameters according to the back propagation algorithm.
  • the difference between the estimated spectral data output by the spectral data model is the difference between the estimated spectral data model.
  • the material data includes material-based data corresponding to the initial material image in the initial image sequence, and the detector element position and projection angle corresponding to each data in the material-based data.
  • the energy spectrum data can be obtained through one or more of dual sources and dual detectors, fast tube voltage switching, slow tube voltage switching, line beam filtering, photon counting detectors, and double layer detectors. combination to obtain.
  • the technical solution of the energy spectrum image processing device determines the initial material image sequence corresponding to the measured energy spectrum data; inputs the material data corresponding to the initial material image sequence into the trained energy spectrum data model to obtain the estimated energy spectrum data ; If the difference between the estimated energy spectrum data and the measured energy spectrum data is within the set error range, the initial material image sequence is used as the expected material image sequence; if the difference between the estimated energy spectrum data and the measured energy spectrum data is not within the set error range within, the material data is adjusted according to the difference, and the material image sequence corresponding to the adjusted material data is used as the expected material image sequence.
  • This material data adjustment method can ensure the speed and effectiveness of material data adjustment; estimated energy spectrum data and measurement
  • the calculation amount of determining the difference between energy spectrum data and adjusting the material data is relatively small, so it can meet the requirements of clinical image processing speed.
  • the energy spectrum image processing device provided by the embodiments of this application can execute the energy spectrum image processing method provided by any embodiment of this application, and has functional modules and beneficial effects corresponding to the execution method.
  • FIG. 9 is a schematic structural diagram of a C-arm CT system provided by yet another embodiment of the present application.
  • the system includes a frame 1211, a detector 1212, a bed board 1214, an X-ray tube 1215, a C-arm drive shaft 1216, a rotating shaft 1217 and a base 1219.
  • the X-ray tube 1215 and the detector 1212 are equipped at both ends of the C-shaped frame 1211, and the center connecting line between the two is perpendicular to the central axis of rotation 1218.
  • the C-shaped frame 1211 drives the detector and the X-ray tube to rotate around the central axis of rotation 1218, so that the image data of the patient 1213 on the bed is captured at different projection angles through the detector and the X-ray tube.
  • the current, voltage, exposure time, etc. of the X-ray tube 1215 are controlled by the X-ray generator 123.
  • the projection data collected by the detector 1212 is transmitted to the computer device 2 by the communication system 126.
  • the frame 1211 is connected to the C-arm drive shaft 1216. Its power is provided by rotating shaft 1217.
  • the base 1219 is responsible for bearing the load.
  • the C-arm control unit 121 controls the rotation speed, angle, position, etc. of the frame 1211.
  • the spindle control unit 122 is connected to the base 1219 and provides power support for the entire C-arm system.
  • the X-ray generator 123 controls the current, voltage and exposure time of the X-ray tube 1215.
  • Data acquisition system 124 coordinates rack 1211, detector 1212, and X-ray generator 1215, and collects collected data.
  • the bed board control system 125 controls the position and movement speed of the bed board 1214 to achieve different scanning trajectories for the patient 1213.
  • the communication system 126 connects the C-arm control unit 121 , the spindle control unit 122 , the X-ray generator 124 , the data acquisition system 124 and the bed control system 125 , and transmits the collected projection data to the memory of the computer system 2 .
  • FIGS 10A and 10B show a schematic structural diagram of yet another CT system.
  • This CT system is a spiral CT.
  • its frame 1211 is annular.
  • the detector 1212 and the X-ray tube 1215 are both arranged on the frame and relatively distributed.
  • the bed board 1214 is controlled by the bed board controller 125
  • the frame drives the detector 1212 and the X-ray tube 1215 to move around the bed board 1214.
  • Figure 11 is a structural block diagram of a computer device provided by another embodiment of the present application.
  • the computer device 2 includes a processor 201, a memory 202, an input device 203 and an output device 204; the number of processors 201 in the device It can be one or more.
  • one processor 201 is taken as an example; the processor 201, memory 202, input device 203 and output device 204 in the device can be connected through a bus or other means.
  • the processor 201, the memory 202, the input device 203 and the output device 204 are connected through a bus. For example.
  • the memory 202 can be used to store software programs, computer-executable programs and modules, such as program instructions/modules corresponding to the energy spectrum image processing method in the embodiments of the present application (for example, the material image module 11, first output module 13 and second output module 14).
  • the processor 201 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 202, that is, implementing the above energy spectrum image processing method.
  • the memory 202 may mainly include a stored program area and a stored data area, where the stored program area may store an operating system and an application program required for at least one function; the stored data area may store data created based on the use of the terminal, etc.
  • the memory 202 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device.
  • memory 202 may further include memory located remotely relative to processor 201, and these remote memories may be connected to the device through a network. Examples of the above-mentioned networks include but are not limited to the Internet, intranets, local area networks, mobile communication networks and combinations thereof.
  • the input device 203 may be used to receive input numeric or character information and generate key signal inputs related to user settings and functional control of the device.
  • the output device 204 may include a display device such as a display screen, for example, a display screen of a user terminal.
  • Embodiments of the present application also provide a storage medium containing computer-executable instructions.
  • the computer-executable instructions when executed by a computer processor, are used to perform an energy spectrum image processing method.
  • the method includes:
  • the initial material image sequence is used as the expected material image sequence
  • the material data is adjusted according to the difference, and the material image sequence corresponding to the adjusted material data is used as the expected material image sequence.
  • the embodiments of the present application provide a storage medium containing computer-executable instructions.
  • the computer-executable instructions are not limited to the method operations described above, and can also execute the energy spectrum image processing method provided by any embodiment of the application. related operations.
  • the present application can be implemented with the help of software and necessary general hardware. Of course, it can also be implemented with hardware, but in many cases the former is a better implementation. . Based on this understanding, the technical solution of the present application can be embodied in the form of a software product in essence or that contributes to the existing technology.
  • the computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk.
  • ROM Read-Only Memory
  • RAM Random Access Memory
  • FLASH flash memory
  • hard disk or optical disk etc., including a number of instructions to make a computer device (which can be Personal computer, server, or network device, etc.) executes the energy spectrum image processing method described in various embodiments of this application.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Apparatus For Radiation Diagnosis (AREA)
  • Investigating, Analyzing Materials By Fluorescence Or Luminescence (AREA)
  • Analysing Materials By The Use Of Radiation (AREA)

Abstract

一种能谱图像处理方法、装置、计算机设备及存储介质，该方法包括：确定测量能谱数据对应的初始材料图像序列(S101)，将初始材料图像序列对应的材料数据输入己训练的能谱数据模型得到估计能谱数据(S102)；如果估计能谱数据与测量能谱数据的差异在设定误差范围内，则将初始材料图像序列作为期望材料图像序列(S103)；反之，则根据该差异调整材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列(S104)。解决了现有能谱图像处理方法存在物质分解定量精度较低的问题。

Description

能谱图像处理方法、装置、计算机设备及存储介质
本申请要求2022年08月18日递交的申请号为202210990678.3、发明名称为“能谱图像处理方法、装置、计算机设备及存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请实施例涉及医学图像处理领域，尤其涉及一种能谱图像处理方法、装置、计算机设备及存储介质。
背景技术
传统CT(Computed Tomography，即电子计算机断层扫描)成像中，具有不同物质组成的解剖结构可能表现出同样的CT衰减值，因此限制其区分不同物质组成的能力。通过采集两个或多个X线能谱数据集的能谱CT技术，可用于解决传统CT成像中这一长期存在的技术局限。能谱CT涉及诸多研究方向，其中数据采集技术和定量物质分解技术是最为主要的研究方向。
关于定量物质分解技术，模型驱动的物质分解方法主要分为不直接的物质分解方法和直接的物质分解方法。1)不直接物质分解方法。这类方法中的第一个子类别首先进行投影空间物质分解，而后对材料基线积分数据进行解析或者迭代图像重建。由于物质分解是一个不稳定过程，这类方法对噪声比较敏感，特别是在低剂量条件下，难以获得精准的物质分解。这类方法中的第二个子类别首先对能谱数据进行迭代重建，而后进行图像空间物质分解。这类方法相对简单，但无法消除线束硬化伪影和金属伪影等。2)直接物质分解方法。直接从能谱数据中重建材料基图像方法，对数据采集物理过程进行建模，并引入关于基材料图像的先验知识，提高物质分解的定量准确度。然而，非线性多能X线测量模型使得直接物质分解问题的病态程度很高，尤其是低剂量数据采集条件下。另外，对数据采集物理过程进行完整、精确的建模往往是十分困难的，由此可能在材料基图像中引入误差。近年来，数据驱动的方法开始应用于能谱CT物质分解，其中包括图像空间分解网络、投影空间分解网络、投影-图像端对端分解网络。这些方法通过匹配的训练样本有监督地学习输入输出数据空间之间的变换，近似在给定能谱数据下，关于材料基图像的后验分布函数，从而求解物质分解逆问题。一旦训练结束，在预测阶段，相较于模型驱动的直接物质分解方法，这类方法的计算复杂度更低。
本申请人在实现本申请实施例的过程中发现，现有数据驱动能谱成像技术过于依赖寻找输入数据与输出数据之间的统计相关性，因此其物质分解定量精度有待提高。
发明内容
本申请实施例提供了一种能谱图像处理方法、装置、计算机设备及存储介质，解决了现有能谱图像处理方法存在物质分解定量精度较低的问题。
第一方面，本申请实施例提供了一种能谱图像处理方法，包括：
确定测量能谱数据对应的初始材料图像序列；
将所述初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；
如果所述估计能谱数据与所述测量能谱数据的差异在设定误差范围内，则将所述初始材料图像序列作为期望材料图像序列；
如果所述估计能谱数据与所述测量能谱数据的差异未在设定误差范围内，则根据该差异调整所述材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
第二方面，本申请实施例还提供了一种能谱图像处理方法，包括：
材料图像模块，用于确定测量能谱数据对应的初始材料图像序列；
能谱数据估计模块，将所述初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；
第一输出模块，用于如果所述估计能谱数据与所述测量能谱数据的差异在设定误差范围内，则将所述初始材料图像序列作为期望材料图像序列；
第二输出模块，用于如果所述估计能谱数据与所述测量能谱数据的差异未在设定误差范围内，则根据该差异调整所述材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
第三方面，本申请实施例还提供了一种计算机设备，所述计算机设备包括：
一个或多个处理器；
存储装置，用于存储一个或多个程序；
当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如任意实施例所述的能谱图像处理方法。
第四方面，一种包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行如任意实施例所述的能谱图像处理方法。
本申请实施例提供的能谱图像处理方法的技术方案，确定测量能谱数据对应的初始材料图像序列；将初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；如果估计能谱数据与测量能谱数据的差异在设定误差范围内，则将初始材料图像序列作为期望材料图像序列；如果估计能谱数据与测量能谱数据的差异未在设定误差范围内，则根据该差异调整材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列，该材料数据调整方式可保证材料数据调整的速度和有效性；估计能谱数据与测量能谱数据之间的差异确定、材料数据调整的计算量均比较小，因此 可满足临床图像处理速度的要求。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例描述中所需要使用的附图做一简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本申请实施例提供的能谱图像处理方法的流程图；
图2A是本申请实施例提供的双源双探测器获取能谱数据的示意图；
图2B是本申请实施例提供的快速管电压切换方法获取能谱数据的示意图；
图2C是本申请实施例提供的线束过滤方法获取能谱数据的示意图；
图2D是本申请实施例提供的光子计数探测器获取能谱数据的示意图；
图2E是本申请实施例提供的双层探测器获取能谱数据的示意图；
图3是本申请实施例提供的能谱图像处理方法的流程示意图；
图4A是本申请实施例提供的体模示意图；
图4B是本申请实施例提供的实验结果示意图；
图5A是本申请实施例提供的能谱数据模型训练方法的流程图；
图5B是本申请实施例提供的能谱数据模型的示意图；
图6是本申请实施例提供的材料模型训练方法的流程图；
图7是本申请实施例提供的材料模型的架构示意图；
图8A是本申请实施例提供的能谱图像处理装置的结构示意图；
图8B是本申请实施例提供的又一能谱图像处理装置的结构示意图；
图9是本申请实施例提供的C臂CT系统的结构示意图；
图10A是本申请实施例提供的CT系统的立体图；
图10B是本申请实施例提供的CT系统的左视图；
图11是本申请实施例提供的计算机设备的结构框图。
具体实施方式
为使本申请的目的、技术方案和优点更加清楚，以下将参照本申请实施例中的附图，通过实施方式清楚、完整地描述本申请的技术方案，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
实施例
图1是本申请实施例提供的能谱图像处理方法的流程图。本实施例的技术方案适用于确定能谱图像对应的材料图像的情况。该方法可以由本申请实施例提供的能谱图像处理装置来执行，该装置可以采用软件和/或硬件的方式实现，并配置在计算机设备的处理器中应用。该方法具体包括如下步骤：
S101、确定测量能谱数据对应的初始材料图像序列。
在一个实施例中，测量能谱数据包括至少两级射线能量的投影数据帧。示例性的，测量能谱数据包括对应于第一X射线能量的至少一个第一投影数据帧和对应于第二X射线能量的至少一个第二投影数据帧，以及该至少一个第一投影数据帧和该至少一个第一投影数据帧分别对应的投影角度和探测器探元位置，第一X射线能量高于第二X射线能量。
在一个实施例中，测量能谱数据可通过双源双探测器(图2A)、快(或慢)速管电压切换(图2B)、线束过滤(图2C)、光子计数探测器(图2D)或双层探测器(图2E)中的一种方式或多种方式的组合来获取。
在一个实施例中，初始材料图像序列基于已训练的材料模型得到，具体为：确定测量能谱数据对应的能谱图像序列；将该能谱图像序列输入已训练的材料模型以得到初始材料图像序列，参见图3。该实施例中，已训练的材料模型具有图像空间的泛化先验，可将能谱图像序列变化为其相应的初始图像序列。
需要说明的是，该步骤中，基于测量能谱数据确定的初始材料图像序列，一般具有较高的图像质量，但局部定量精度可能有所不足，并且可能无法准确地描述每个样本的材料性质，因此该初始材料图像序列对应的材料数据也存在局部定量精度可能不足的问题。
S102、将初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据。
其中，材料数据包括初始图像序列中的初始材料图像对应的材料基数据，以及材料基数据中的每个数据对应的探测器探元位置和投影角度。在一个实施例中，材料数据包括至少两个材料基数据。
如图3所示，初始材料图像序列得到后，将初始材料图像序列中各初始材料图像沿对应的一个或多个投影角度进行前向投影以得到一个或多个材料数据帧；确定属于期望材料图像的所有投影角度，所有投影角度对应的材料数据帧以及所有材料数据帧分别对应的探测器探元位置以得到期望材料图像的材料数据。
本实施例中已训练的能谱数据模型具有在数据空间描述数据采集过程的先验知识，因此可将材料数据合成对应的能谱数据。
S103、如果估计能谱数据与测量能谱数据的差异在设定误差范围内，则将初始材料图像序列作为期望材料图像序列。
可以理解的是，如果估计能谱数据与测量能谱数据之间的差异在设定误差范围内，则说明该估计能谱数据的准确度非常接近测量能谱数据，因此用于确定估计能谱数据的材料数据具有较高的准确性，用于确定材料数据的初始材料图像序列的定量精度在用户的可接受范围内，因此将初始图像序列作为期望材料图像序列。
S104、如果估计能谱数据与测量能谱数据的差异未在设定误差范围内，则根据该差异调整材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
由于材料数据的局部定量精度可能不足，因此已训练的能谱数据模型基于该材料数据预测的估计能谱数据的质量可能低于期望的能谱数据质量，为此本实施例以测量能谱数据为基准来衡量估计能谱数据的准确性，并根据估计能谱数据与测量能谱数据之间的差异调整材料数据，然后将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
在一个实施例中，估计能谱数据包括的估计能谱数据帧与测量能谱数据包括的测量能谱数据帧一一对应。确定各估计能谱数据帧与对应的各测量能谱数据帧之间的误差，然后计算所有误差的总和，并根据该所有误差的总和对材料数据进行调整，然后将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
由于测量能谱数据为原始投影数据，其携带有准确的物质信息，因此以测量能谱数据与估计能谱数据的差值为基准可以准确地指导材料数据的调整。
需要说明的是，本实施例仅根据估计能谱数据与测量能谱数据之间的差异调整材料数据，但并不调整任何模型的网络参数，即在模型使用阶段，所有模型的网络参数保持不变。
图4A和图4B为本申请实施例提供的基于物理体模实验确定的定量实验结果。如图4A所示，该实验体模包含4个不同浓度的碘溶液试管和2个不同浓度氯化钙溶液试管。能谱CT实验平台对其在100kV下扫描。光子技术探测器使用两个能谱阈值采集穿过实验体模的射线，以得到测量能谱数据，该测量能谱数据包括对应高能量的投影数据和对应低能量的投影数据。如图4B所示，本申请技术产生相应的碘基图像。将该碘基图像确定的4个碘溶液试管浓度(碘基图像上标注的浓度数据)并与金标准(浓度颜色变化示意图)进行比较，比较结果表明，四个碘溶液试管浓度均定量准确。该实验说明了本申请实施例所述的能谱图像处理方法具有较高的物质定量准确性。
本申请实施例提供的能谱图像处理方法的技术方案，确定测量能谱数据对应的初始材料图像序列；将初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；如果估计能谱数据与测量能谱数据的差异在设定误差范围内，则将初始材料图像序列作为期望材料图像序列；如果估计能谱数据与测量能谱数据的差异未在设定误差范围内，则根据该差异调整材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列，该材料数据调整方式可保证材料数据调整的速度和有效性； 估计能谱数据与测量能谱数据之前的差异确定、材料数据调整的计算量均比较小，因此可满足临床图像处理速度的要求。
图5A是本申请又一实施例提供能谱数据模型训练方法的流程图。本申请实施例所述的模型训练方法应用于前述实施例所述的能谱图像处理方法之前。具体方法步骤包括：
S201、将设定数量的材料数据输入能谱数据模型以得到估计能谱数据。
其中，如图5A所示，材料数据包括至少两个材料基数据以及材料基数据中的每个数据对应的探测器探元位置和投影角度。
获取设定数量患者的测量能谱数据以及每个能谱数据对应的材料数据，并将该测量能谱数据和材料数据作为训练样本。在能谱数据模型训练时，将材料数据输入能谱数据模型中，以得到估计能谱数据。
S202、根据测量能谱数据与估计能谱数据之间的差异，反复调整能谱数据模型的网络参数，根据反向传播算法最小化测量能谱数据与网络参数调整后的能谱数据模型输出的估计能谱数据之间的差异。
在一个实施例中，能谱数据模型的具体架构如图5B所示，其包括4层全连接层。
确定测量能谱数据与估计能谱数据之间的差异，并根据该差异反复调整能谱数据模型的网络参数，直至测量能谱数据与网络参数调整后的能谱数据模型输出的估计能谱数据之间的差异在设定阈值范围内。可以理解的是，当测量能谱数据与当前的估计能谱数据之间的差异在设定阈值范围内，则认为二者近似相同，模型训练结束。该能谱数据模型可以建模完整的数据采集物理过程，具体为，从大规模的训练数据中学习到数据空间描述数据采集过程的物理先验知识，以此将材料数据合成为其对应的能谱数据。
在上述模型训练过程中，采用能谱数据空间均方误差作为训练损失函数调整能谱数据模型的网络参数。
图6是本申请又一实施例提供的材料模型训练方法的流程图。该材料模型的训练方法包括：
S301、将能谱图像序列输入材料模型以得到估计材料图像序列。
获取设定数量患者的能谱图像序列以及能谱图像序列对应的材料图像序列，以得到设定数量的样本。将能谱图像序列输入材料模型以得到估计材料序列。
S302、根据材料图像序列真值与估计材料图像序列之间的差异，反复调整材料模型的网络参数，根据反向传播算法最小化材料图像序列真值与网络参数调整后的材料模型输出的估计材料图像序列之间的差异。
将初始能谱图像序列和初始材料图像序列作为训练样本。将该初始能谱图像序列输入材料模型以得到估计材料图像序列，根据初始材料图像序列和材料模型输出的估计材料图像序列之间的差异调整材料模型的网络参数，直至初始材料图像序列与当前的估计 材料图像序列之间的差异在设定误差范围内。该模型训练过程中，使用图像空间均方误差作为训练损失函数。该实施例中，材料模型从大规模训练数据中学习到图像空间的泛化先验，以此将能谱图像变换为其相应的初始材料图像序列。
在一个实施例中，材料模型的网络结构如图7所示，该神经网络共有24层卷积神经网络层，该24层卷积神经网络层包括三类卷积网络层，这些卷积网络层中的参数均为可学习的。第一类卷积网络层使用3×3卷积核，间隔为1，在图7中标记为“Conv(3×3)S1”，之后为批归一化操作(Bnorm)和整流线性单元(ReLu)激活函数。第二类卷积网络层使用3×3卷积核，间隔为2，并在图7中标记为“Conv(3×3)S2”，之后为Nnorm和ReLu。第三类卷积网络层使用1×1卷积核，间隔为1，并在图7中标记为“Conv(1×1)S1”，之后为线性激活函数。所有卷积层都有相应的可学偏差项。每个卷积层均保持该层的输入输出为同样的空间维度。采样层使用2×2卷积核，并在图7中标记为Up-sample 2×2。所有采样层使用双线性插值算法。捷径链接(skip+concatenate，图7中黑色实箭头)用以促进网络训练过程。网络参数中的卷积核使用Glorot均匀分布随机数作为初始化，偏差项使用0作为初始化值。其余参数设置与初始化值均选择默认方式。
需要说明的是，本申请技术方案中对患者影像数据的获取、存储、使用和处理等均符合国家法律法规的相关规定。
图8A是本申请又一实施例提供的能谱图像处理装置的结构框图。该装置用于执行上述任意实施例所提供的能谱图像处理方法，该装置可选为软件或硬件实现。该装置包括：
材料图像模块11，用于确定测量能谱数据对应的初始材料图像序列；
能谱数据估计模块12，用于将所述初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；
第一输出模块13，用于如果所述估计能谱数据与所述测量能谱数据的差异在设定误差范围内，则将所述初始材料图像序列作为期望材料图像序列；
第二输出模块14，用于如果所述估计能谱数据与所述测量能谱数据的差异未在设定误差范围内，则根据该差异调整所述材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
可选地，材料图像模块11用于确定所述测量能谱数据对应的能谱图像序列；将所述能谱图像序列输入已训练的材料模型以得到初始材料图像序列。
可选地，能谱数据估计模块11用于确定所述初始材料图像序列中各初始材料图像对应的一个或多个投影角度；将所述初始材料图像序列中各初始材料图像沿对应的一个或多个投影角度进行前向投影以得到一个或多个材料数据帧；确定属于期望材料图像的所有投影角度、所有投影角度分别对应的材料数据帧以及所有材料数据帧分别对应的探测 器探元位置以得到所述期望材料图像的材料数据。
可选地，所述测量能谱数据包括至少两级射线能量的投影数据帧；所述测量能谱数据对应至少两个材料基数据。
可选地，如图8B所示，该装置还包括模型训练模块10，该模型训练模块用于将设定数量的材料数据输入能谱数据模型以得到估计能谱数据；
根据所述测量能谱数据与所述估计能谱数据之间的差异，反复调整所述能谱数据模型的网络参数，根据反向传播算法最小化所述测量能谱数据与网络参数调整后的能谱数据模型输出的估计能谱数据之间的差异。
可选地，所述材料数据包括所述初始图像序列中的初始材料图像对应的材料基数据，以及所述材料基数据中的每个数据对应的探测器探元位置和投影角度。
可选地，所述能谱数据可通过双源双探测器、快速管电压切换、慢速管电压切换、线束过滤、光子计数探测器、双层探测器中的一种方式或多种方式的组合来获取。
本申请实施例提供的能谱图像处理装置的技术方案，确定测量能谱数据对应的初始材料图像序列；将初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；如果估计能谱数据与测量能谱数据的差异在设定误差范围内，则将初始材料图像序列作为期望材料图像序列；如果估计能谱数据与测量能谱数据的差异未在设定误差范围内，则根据该差异调整材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列，该材料数据调整方式可保证材料数据调整的速度和有效性；估计能谱数据与测量能谱数据之间的差异确定、材料数据调整的计算量均比较小，因此可满足临床图像处理速度的要求。
本申请实施例所提供的能谱图像处理装置可执行本申请任意实施例所提供的能谱图像处理方法，具备执行方法相应的功能模块和有益效果。
图9是本申请又一实施例提供的C臂CT系统的结构示意图。该系统包括机架1211、探测器1212、床板1214、X线光管1215、C臂驱动轴1216、转轴1217以及基座1219。X线光管1215和探测器1212装备在C型机架1211的两端，两者中心连接线与其旋转中轴线1218垂直。C型机架1211带动探测器和X线光管绕旋转中轴线1218旋转，从而通过探测器和X线光管在不同的投影角度拍摄床板上的患者1213的影像数据。X线光管1215由X线发生器123控制其电流、电压和曝光时间等，探测器1212采集到的投影数据由通信系统126传输给计算机设备2，机架1211与C臂驱动轴1216相连，其动力由转轴1217提供。基座1219负责承重。C臂控制单元121控制机架1211的旋转速度、角度、位置等。主轴控制单元122连接基座1219，并为整个C臂系统提供电力支持。X线发生器123控制X线光管1215的电流、电压和曝光时间。数据采集系统124协调机架1211、探测器1212和X线发生器1215，并收集采集到的数据。床板控制系统125控 制床板1214的位置和运动速度，以实现对患者1213的不同扫描轨道。通信系统126连接C臂控制单元121、主轴控制单元122、X线发生器124、数据采集系统124和床板控制系统125，并将采集到的投影数据传输给计算机系统2的存储器。
图10A和图10B示出了又一CT系统的结构示意图。该CT系统为螺旋CT，相较于C臂CT，其机架1211为环形，探测器1212和X线光管1215均设置于机架上，且相对分布，床板1214在床板控制器125的控制下进出机架孔径，机架带动探测器1212和X线光管1215绕床板1214运动。
图11为本申请又一实施例提供的计算机设备的结构框图，如图11所示，该计算机设备2包括处理器201、存储器202、输入装置203以及输出装置204；设备中处理器201的数量可以是一个或多个，图11中以一个处理器201为例；设备中的处理器201、存储器202、输入装置203以及输出装置204可以通过总线或其他方式连接，图11中以通过总线连接为例。
存储器202作为一种计算机可读存储介质，可用于存储软件程序、计算机可执行程序以及模块，如本申请实施例中的能谱图像处理方法对应的程序指令/模块(例如，材料图像模块11、第一输出模块13以及第二输出模块14)。处理器201通过运行存储在存储器202中的软件程序、指令以及模块，从而执行设备的各种功能应用以及数据处理，即实现上述的能谱图像处理方法。
存储器202可主要包括存储程序区和存储数据区，其中，存储程序区可存储操作系统、至少一个功能所需的应用程序；存储数据区可存储根据终端的使用所创建的数据等。此外，存储器202可以包括高速随机存取存储器，还可以包括非易失性存储器，例如至少一个磁盘存储器件、闪存器件、或其他非易失性固态存储器件。在一些实例中，存储器202可进一步包括相对于处理器201远程设置的存储器，这些远程存储器可以通过网络连接至设备。上述网络的实例包括但不限于互联网、企业内部网、局域网、移动通信网及其组合。
输入装置203可用于接收输入的数字或字符信息，以及产生与设备的用户设置以及功能控制有关的键信号输入。
输出装置204可包括显示屏等显示设备，例如，用户终端的显示屏。
本申请实施例还提供了一种包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行一种能谱图像处理方法，该方法包括：
确定测量能谱数据对应的初始材料图像序列；
将所述初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；
如果所述估计能谱数据与所述测量能谱数据的差异在设定误差范围内，则将所述初始材料图像序列作为期望材料图像序列；
如果所述估计能谱数据与所述测量能谱数据的差异未在设定误差范围内，则根据该差异调整所述材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
当然，本申请实施例所提供的一种包含计算机可执行指令的存储介质，其计算机可执行指令不限于如上所述的方法操作，还可以执行本申请任意实施例所提供的能谱图像处理方法中的相关操作。
通过以上关于实施方式的描述，所属领域的技术人员可以清楚地了解到，本申请可借助软件及必需的通用硬件来实现，当然也可以通过硬件实现，但很多情况下前者是更佳的实施方式。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分可以以软件产品的形式体现出来，该计算机软件产品可以存储在计算机可读存储介质中，如计算机的软盘、只读存储器(Read-Only Memory，简称ROM)、随机存取存储器(Random Access Memory，简称RAM)、闪存(FLASH)、硬盘或光盘等，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本申请各个实施例所述的能谱图像处理方法。
值得注意的是，上述能谱图像处理装置的实施例中，所包括的各个单元和模块只是按照功能逻辑进行划分的，但并不局限于上述的划分，只要能够实现相应的功能即可；另外，各功能单元的具体名称也只是为了便于相互区分，并不用于限制本申请的保护范围。
注意，上述仅为本申请的较佳实施例及所运用技术原理。本领域技术人员会理解，本申请不限于这里所述的特定实施例，对本领域技术人员来说能够进行各种明显的变化、重新调整和替代而不会脱离本申请的保护范围。因此，虽然通过以上实施例对本申请进行了较为详细的说明，但是本申请不仅仅限于以上实施例，在不脱离本申请构思的情况下，还可以包括更多其他等效实施例，而本申请的范围由所附的权利要求范围决定。

Claims (10)

  1. 一种能谱图像处理方法，包括：
    确定测量能谱数据对应的初始材料图像序列；
    将所述初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；
    如果所述估计能谱数据与所述测量能谱数据的差异在设定误差范围内，则将所述初始材料图像序列作为期望材料图像序列；
    如果所述估计能谱数据与所述测量能谱数据的差异未在设定误差范围内，则根据该差异调整所述材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
  2. 根据权利要求1所述的方法，其中，所述确定测量能谱数据对应的初始材料图像序列包括：
    确定所述测量能谱数据对应的能谱图像序列；
    将所述能谱图像序列输入已训练的材料模型以得到初始材料图像序列。
  3. 根据权利要求1所述的方法，其中，根据以下步骤确定所述初始材料图像序列对应的材料数据：
    确定所述初始材料图像序列中各初始材料图像对应的一个或多个投影角度；
    将所述初始材料图像序列中各初始材料图像沿对应的一个或多个投影角度进行前向投影以得到一个或多个材料数据帧；
    确定属于期望材料图像的所有投影角度、所有投影角度分别对应的材料数据帧以及所有材料数据帧分别对应的探测器探元位置以得到所述期望材料图像的材料数据。
  4. 根据权利要求1所述的方法，其中，所述测量能谱数据包括至少两级射线能量的投影数据帧，所述测量能谱数据对应至少两个材料基数据。
  5. 根据权利要求1所述的方法，其中，通过以下步骤完成所述能谱数据模型的训练，包括：
    将设定数量的材料数据输入能谱数据模型以得到估计能谱数据；
    根据所述测量能谱数据与所述估计能谱数据之间的差异，反复调整所述能谱数据模型的网络参数，根据反向传播算法最小化所述测量能谱数据与网络参数调整后的能谱数据模型输出的估计能谱数据之间的差异；
    通过以下步骤完成所述材料模型的训练，包括：
    将所述能谱图像序列输入材料模型以得到估计材料图像序列；
    根据所述材料图像序列真值与所述估计材料图像序列之间的差异，反复调整所述材料模型的网络参数，根据反向传播算法最小化所述材料图像序列真值与网络参数调整后的材料模型输出的估计材料图像序列之间的差异。
  6. 根据权利要求1所述的方法，其中，所述材料数据包括所述初始图像序列中的初 始材料图像对应的材料基数据，以及所述材料基数据中的每个数据对应的探测器探元位置和投影角度。
  7. 根据权利要求1-6任一所述的方法，其中，所述能谱数据可通过双源双探测器、快速管电压切换、慢速管电压切换、线束过滤、光子计数探测器、双层探测器中的一种方式或多种方式的组合来获取。
  8. 一种能谱图像处理装置，包括：
    材料图像模块，用于确定测量能谱数据对应的初始材料图像序列；
    能谱数据估计模块，将所述初始材料图像序列对应的材料数据输入已训练的能谱数据模型以得到估计能谱数据；
    第一输出模块，用于如果所述估计能谱数据与所述测量能谱数据的差异在设定误差范围内，则将所述初始材料图像序列作为期望材料图像序列；
    第二输出模块，用于如果所述估计能谱数据与所述测量能谱数据的差异未在设定误差范围内，则根据该差异调整所述材料数据，并将调整后的材料数据对应的材料图像序列作为期望材料图像序列。
  9. 一种计算机设备，所述计算机设备包括：
    一个或多个处理器；
    存储装置，用于存储一个或多个程序；
    当所述一个或多个程序被所述一个或多个处理器执行，使得所述一个或多个处理器实现如权利要求1-7中任一所述的能谱图像处理方法。
  10. 一种包含计算机可执行指令的存储介质，所述计算机可执行指令在由计算机处理器执行时用于执行如权利要求1-7中任一所述的能谱图像处理方法。
PCT/CN2022/129561 2022-08-18 2022-11-03 能谱图像处理方法、装置、计算机设备及存储介质 Ceased WO2024036760A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210990678.3 2022-08-18
CN202210990678.3A CN117671046A (zh) 2022-08-18 2022-08-18 能谱图像处理方法、装置、计算机设备及存储介质

Publications (1)

Publication Number Publication Date
WO2024036760A1 true WO2024036760A1 (zh) 2024-02-22

Family

ID=89940495

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/129561 Ceased WO2024036760A1 (zh) 2022-08-18 2022-11-03 能谱图像处理方法、装置、计算机设备及存储介质

Country Status (2)

Country Link
CN (1) CN117671046A (zh)
WO (1) WO2024036760A1 (zh)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101138501A (zh) * 2006-09-08 2008-03-12 通用电气公司 用于生成目标的多光谱图像的方法和系统
US20180038970A1 (en) * 2016-08-05 2018-02-08 Tsinghua University Methods and apparatuses for reconstructing incident energy spectrum for a detector
CN109916933A (zh) * 2019-01-04 2019-06-21 中国人民解放军战略支援部队信息工程大学 基于卷积神经网络的x射线计算机断层成像能谱估计方法
CN110084864A (zh) * 2019-04-09 2019-08-02 南京航空航天大学 一种基于能谱ct的电子密度图像重建方法
CN110675467A (zh) * 2019-09-30 2020-01-10 中国科学院深圳先进技术研究院 能谱ct的材料分解方法、装置、存储介质及电子设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101138501A (zh) * 2006-09-08 2008-03-12 通用电气公司 用于生成目标的多光谱图像的方法和系统
US20180038970A1 (en) * 2016-08-05 2018-02-08 Tsinghua University Methods and apparatuses for reconstructing incident energy spectrum for a detector
CN109916933A (zh) * 2019-01-04 2019-06-21 中国人民解放军战略支援部队信息工程大学 基于卷积神经网络的x射线计算机断层成像能谱估计方法
CN110084864A (zh) * 2019-04-09 2019-08-02 南京航空航天大学 一种基于能谱ct的电子密度图像重建方法
CN110675467A (zh) * 2019-09-30 2020-01-10 中国科学院深圳先进技术研究院 能谱ct的材料分解方法、装置、存储介质及电子设备

Also Published As

Publication number Publication date
CN117671046A (zh) 2024-03-08

Similar Documents

Publication Publication Date Title
JP5280450B2 (ja) Ｘ線ｃｔ画像形成方法及びそれを用いたｘ線ｃｔ装置
CN100398067C (zh) X射线计算机断层摄影装置和图象质量模拟装置
US7801264B2 (en) Method for calibrating a dual -spectral computed tomography (CT) system
CN110866959B (zh) 图像重建方法、系统、装置及存储介质
JP2017209498A (ja) Ｘ線コンピュータ断層撮影装置及び医用画像処理装置
WO2010038536A1 (ja) Ｘ線ｃｔ装置
JP3732568B2 (ja) Ｘ線コンピュータ断層撮影装置
JPH0838467A (ja) 検出器チャンネル利得較正係数を求める方法
US12111433B2 (en) Counting response and beam hardening calibration method for a full size photon-counting CT system
US20230263492A1 (en) Image processing apparatus, image processing method, and computer-readable medium
US20240070862A1 (en) Medical information processing method and medical information processing apparatus
JP2004097778A (ja) Ｘ線コンピュータ断層撮影装置
JP2001309915A (ja) Ｃｔイメージング・システムの検出器のための高次主減衰補正
JP2004321587A (ja) Ｘ線ｃｔ画像撮影方法およびｘ線ｃｔ装置
JP4817736B2 (ja) Ｘ線コンピュータ断層撮影システム及びｘ線コンピュータ断層撮影システムのデータ補正方法
US6934353B2 (en) Methods, apparatus, and computer readable mediums for performing perfusion studies
JPWO2015020072A1 (ja) Ｘ線ｃｔ装置および補正処理装置
JP2000083945A (ja) 断層写真画像を作る装置と方法
JP2024039013A (ja) 画像収集条件の決定方法及びｘ線診断装置
US12198234B2 (en) Systems and methods for motion detection in medical images
JP2002034970A (ja) マルチ・スライスｃｔ走査の螺旋再構成の方法及び装置
JP2024032518A (ja) 画像処理装置、撮影システム、画像処理方法、及びプログラム
CN115192052A (zh) 医用图像处理装置以及医用图像处理方法
WO2024036760A1 (zh) 能谱图像处理方法、装置、计算机设备及存储介质
US10192328B2 (en) Method for statistical weights design in iterative reconstruction algorithms

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22955532

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22955532

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 22955532

Country of ref document: EP

Kind code of ref document: A1

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 28/08/2025)