Motion compensation apparatus and motion compensation method

Abstract

The motion compensation apparatus of the embodiment includes an acquisition unit, a determination unit, a first reconstruction unit, an estimation unit, and a second reconstruction unit. The acquisition unit acquires projection data scanned by a computed tomography imaging system from an imaging subject. The determination unit determines a target time point for reconstructing a motion-corrected image of the imaging subject. The first reconstruction unit reconstructs a first group of feature-enhanced partial images and a second group of feature-enhanced partial images based on a first set of projection data and a second set of projection data related to both sides of the determined target time point, respectively. The estimation unit estimates a motion vector field at the determined target time point based on the reconstructed first group of feature-enhanced partial images and the reconstructed second group of feature-enhanced partial images. The second reconstruction unit performs motion compensation on the acquired projection data based on the estimated motion vector field, thereby reconstructing an image of the imaging subject as the motion-corrected image at the determined target time point.

Description

Reference to relevant applications

[0001] This application enjoys the benefits of priority to U.S. Patent Application No. 19 / 088,507, filed March 24, 2025, and Japanese Patent Application No. 2026-036541, filed March 6, 2026, the entire contents of which are incorporated herein by reference. Technical Field

[0002] The embodiments disclosed in this specification and accompanying drawings relate to motion compensation devices and motion compensation methods. Background Technology

[0003] Motion artifacts, especially when motion is difficult to estimate, are one of the biggest technical problems in computed tomography (CT) imaging. In medical imaging, the main types of motion are generally respiratory motion and cardiac motion. Respiratory motion involves the movement of the lungs and surrounding structures during respiration, while cardiac motion refers to the contraction and relaxation of the heart and associated blood vessels.

[0004] The heart is constantly in motion, exhibiting both regular and irregular movements, making cardiac CT scans particularly challenging. Consequently, generating accurate images of the heart requires highly sophisticated imaging techniques. Inadequate motion compensation often results in image distortion or blurring, impairing the diagnostic value of the scan.

[0005] Typically, to mitigate the effects of motion artifacts induced by cardiac motion, compensation methods are applied to restore image quality. These methods mostly involve two main stages: (1) estimating cardiac motion and (2) incorporating the estimated motion into the image reconstruction process to address motion artifacts. The performance of these motion compensation methods is primarily determined by the accuracy of the motion estimation stage.

[0006] Therefore, it is desirable to enhance current approaches to motion estimation and motion compensation. Summary of the Invention

[0007] A motion compensation apparatus for performing motion compensation in a computed tomography imaging system includes an acquisition unit, a determination unit, a first reconstruction unit, an estimation unit, and a second reconstruction unit. The acquisition unit acquires projection data scanned from an imaging object by the computed tomography imaging system. The determination unit determines a target time for reconstructing a motion-corrected image of the imaging object. The first reconstruction unit reconstructs a first group and a second group of feature-enhanced images based on a first set and a second set of projection data related to both sides of the determined target time. The estimation unit estimates a motion vector field for the determined target time based on the reconstructed first and second groups of feature-enhanced images. The second reconstruction unit performs motion compensation on the acquired projection data based on the estimated motion vector field, thereby reconstructing an image of the imaging object as the motion-corrected image at the determined target time. Attached Figure Description

[0008] This application can be better understood by referring to the description shown in a non-limiting manner with reference to the accompanying drawings.

[0009] Figure 1 A block diagram illustrating an exemplary apparatus for performing motion compensation in a computed tomography (CT) imaging system according to an embodiment of the present disclosure.

[0010] Figure 2 This is a flowchart illustrating exemplary steps for performing motion compensation in a CT imaging system according to embodiments of the present disclosure.

[0011] Figure 3 A block diagram illustrating the feature enhancement portion reconfiguration circuit of an embodiment of this disclosure.

[0012] Figure 4 This is an illustrative scenario representing an embodiment of the present disclosure of a reconstructed image of 16 feature-enhanced portions.

[0013] Figure 5 A flowchart illustrating exemplary steps for reconstructing a feature-enhanced portion of an image according to an embodiment of the present disclosure.

[0014] Figure 6 A block diagram illustrating the motion vector field estimation circuit of an embodiment of the present disclosure.

[0015] Figure 7An exemplary pair of images illustrating embodiments of the present disclosure for estimating motion vector fields.

[0016] Figure 8 A flowchart illustrating exemplary steps for estimating a motion vector field according to an embodiment of the present disclosure.

[0017] Figure 9A And 9B represents an exemplary scenario of using a neural network to estimate a motion vector field in an embodiment of this disclosure.

[0018] Figure 10 A schematic block diagram illustrating an exemplary CT imaging system capable of incorporating the techniques disclosed in this specification. Detailed Implementation

[0019] An apparatus for performing motion compensation in a computed tomography (CT) imaging system is disclosed. The apparatus includes a processing circuit configured to acquire projection data scanned from an imaging object by the CT imaging system, determine a target time for reconstructing a motion-corrected image of the imaging object, and reconstruct a first group and a second group of feature-enhanced images based on a first group and a second group of projection data, respectively. The processing circuit is further configured to estimate a motion vector field for the determined target time based on the reconstructed first group and second group of feature-enhanced images, perform motion compensation on the acquired projection data based on the estimated motion vector field, and reconstruct an image of the imaging object as the motion-corrected image for the determined target time.

[0020] In addition, a method for performing motion compensation in a computed tomography (CT) imaging system is disclosed. This method includes the following steps: acquiring projection data scanned from an imaging object by the CT imaging system; determining a target time for reconstructing a motion-corrected image of the imaging object; reconstructing a first group and a second group of feature-enhanced images based on a first group of projection data located on the opposite side of the determined target time and a second group of projection data, respectively; estimating the motion vector field of the determined target time based on the reconstructed first group and second group of feature-enhanced images; and performing motion compensation on the acquired projection data based on the estimated motion vector field to reconstruct an image of the imaging object as the motion-corrected image of the determined target time.

[0021] Additionally, an apparatus for performing motion compensation in a computed tomography (CT) imaging system is disclosed. This apparatus includes a processing circuit configured to: acquire projection data scanned from a first imaging object by the CT imaging system; determine a first target time for reconstructing a motion-corrected image of the first imaging object; reconstruct a first group and a second group of feature-enhanced images based on a first group and a second group of projection data located on the opposite side of the determined first target time, respectively; input the reconstructed first group and second group of feature-enhanced images to the input of a pre-trained neural network; infer the motion vector field of the determined first target time; perform motion compensation on the acquired projection data based on the inferred motion vector field; and reconstruct an image of the first imaging object as the motion-corrected image of the determined first target time. The pre-trained neural network is trained using a third group and a fourth group of feature-enhanced images, which are reconstructed based on the first group and the second group of projection data acquired from a second imaging object, respectively. The first and second sets of projection data obtained from the second imaging object are located on the opposite side of the second target time. The second imaging object may be the same as or different from the first imaging object.

[0022] It should be noted that the foregoing description does not specify all embodiments and / or new incremental methods of this disclosure. Of course, the foregoing description is merely a preliminary explanation of different embodiments and corresponding points of novelty. More detailed and / or possible aspects of this disclosure and its embodiments are shown in the following further described embodiments and corresponding figures.

[0023] The following disclosure provides several different implementations or embodiments for carrying out various features of the provided subject matter. To simplify this disclosure, specific examples of constituent elements and configurations are described below. Of course, these are merely examples and are not intended to be limiting.

[0024] For example, the order in which the different steps described in this specification are presented is for ease of understanding. Generally, these steps can be performed in any suitable order. Furthermore, in this specification, there are instances where different features, techniques, structures, etc., are described in different parts of this disclosure, meaning that the concepts can be implemented independently or in combination with each other. Therefore, this disclosure can be implemented and understood in many different ways.

[0025] Furthermore, in the context of this instruction manual, unless otherwise specified, the word "a" or similar words generally have the meaning of "more than one".

[0026] This disclosure provides a feature enhancement method for generating motion estimation and motion compensation of computed tomography (CT) images without motion artifacts. In this method, multiple small partial images are generated to perform motion estimation and motion compensation. These partial images are reconstructed from a smaller amount of projection data, thus achieving improved temporal resolution compared to methods that rely on large pairs of partial images. Furthermore, instead of directly using conventional partial images, this method achieves more accurate motion estimation by enhancing and extracting edge and high-frequency information from the partial images.

[0027] Figure 1 This is a block diagram illustrating an exemplary motion compensation device 100 for performing motion compensation in a CT imaging system according to an embodiment of the present disclosure. The motion compensation device 100 includes a projection data acquisition circuit 110, a feature enhancement reconstruction circuit 120, a motion vector field estimation circuit 130, and an image reconstruction circuit 140. Here, the projection data acquisition circuit 110 is an example of an acquisition unit. Furthermore, the feature enhancement reconstruction circuit 120 is an example of a determination unit and a first reconstruction unit. The motion vector field estimation circuit 130 is an example of an estimation unit. The image reconstruction circuit 140 is an example of a second reconstruction unit.

[0028] The projection data acquisition circuit 110 acquires the projection data of the imaging object and sends the acquired data to both the feature enhancement reconstruction circuit 120 and the image reconstruction circuit 140. The projection data can be the raw projection data obtained by scanning the imaging object using a CT imaging system.

[0029] In addition, the feature enhancement reconstruction circuit 120 receives input from the operator of the CT imaging system and determines the target time T based on the operator's input. r Target time T r This represents the instant at which the motion-corrected image is generated through motion estimation and motion compensation. The feature enhancement reconstruction circuit 120 uses the received projection data to reconstruct a group of two feature enhancement images (one at the target time T). r The previous feature-enhanced image group, the other is the target time T. r (Then a group of the feature-enhanced partial images). These two groups of feature-enhanced partial images are then sent to the motion vector field estimation circuit 130.

[0030] The motion vector field estimation circuit 130 estimates the target time T based on the group of received feature-enhanced partial images. rThe motion vector field is then estimated and sent to the image reconstruction circuit 140.

[0031] Image reconstruction circuit 140 uses the motion vector field received from motion vector field estimation circuit 130 and the original projection data received from projection data acquisition circuit 110 to reconstruct the target time T. r The motion-corrected image is generated by the image reconstruction circuit 140, which can generate the image by applying various motion-compensated reconstruction algorithms. By removing or mitigating motion artifacts based on the estimated motion vector field, the resulting image quality can be significantly improved.

[0032] Additionally, refer to Figures 3-8 The structure and function of the feature enhancement reconstruction circuit 120 and the motion vector field estimation circuit 130 are further explained.

[0033] Figure 2 This is a flowchart illustrating an exemplary step 200 for performing motion compensation in a CT imaging system according to an embodiment of the present disclosure. In step S210, the original projection data of the imaging object is obtained. In step S220, the target time T for reconstructing the motion-corrected image is determined. r In step S230, the reconstruction is performed at the determined target time T. r The two feature-enhanced images on either side are grouped together. In step S240, the target time T is estimated based on the group of the two feature-enhanced images. r The motion vector field. In step S250, based on the estimated motion vector field and the original projection data, the target time T is reconstructed. r The motion-corrected image.

[0034] Figure 3 This is a block diagram illustrating a feature enhancement partial reconstruction circuit 120 according to an embodiment of the present disclosure. The feature enhancement partial reconstruction circuit 120 includes a projection data receiving circuit 310, a feature enhancement circuit 320, a target time determination circuit 330, and a partial image reconstruction circuit 340.

[0035] The projection data receiving circuit 310 obtains projection data from the projection data acquisition circuit 110 and sends the obtained data to the feature enhancement circuit 320.

[0036] Feature enhancement circuit 320 generates feature-enhanced projection data by applying feature enhancement to the projection data, and sends this feature-enhanced projection data to partial image reconstruction circuit 340. Feature enhancement can be achieved through high-pass filtering, derivation operations, or other similar methods. This process extracts important features such as high-frequency information within the projection data. As a result, the edges of anatomical or physiological structures within the imaged object can be emphasized in the reconstructed partial image.

[0037] For example, in the reconstructed portion of the image, structural edges can be made more defined. Without such feature enhancement, motion is difficult to detect in areas of uniform contrast. However, motion near edges can cause significant blurring. By enhancing these features, motion within the imaged object becomes easier to detect, and identification and quantification become simpler.

[0038] The target time determination circuit 330 receives input from the operator of the CT imaging system and determines the target time T based on this input. r For example, a radiologist can manually select a specific phase of the cardiac cycle (e.g., 50%, 75%, etc.) for generating motion-corrected images. The target time determination circuit 330 can then determine the time corresponding to that specified phase as the target time T. r .

[0039] Partial image reconstruction circuit 340 receives feature enhancement projection data and target time T. r Both sides. Regarding the target time T... r On either side, two periods of equal length can be determined. For example, the length of each period can be determined to correspond to the imaging range of a half scan covering the CT imaging system. Alternatively, the length of these periods can also be determined based on a specific feature present in a portion of the image, thereby ensuring that the period is sufficient to cover the imaging range associated with that feature. For example, one period is selected to cover half of the imaging range of the feature, and another period is selected to cover the opposite half of the imaging range.

[0040] Figure 4 This illustrates an example of partial image reconstruction according to an embodiment of the present disclosure. As shown, at the target time T... r Two periods, corresponding to a 180° imaging range, are symmetrically arranged on both sides. For example, one period can be configured at the target time T. r Immediately before, another period is configured at the target time T. r Just after that. Furthermore, Figure 4The configuration shown is not limited and can also be used to maintain the relationship between each period and the target time T by using periods covering angles beyond 180°. r The time interval between them.

[0041] Furthermore, each period is further divided into multiple smaller time intervals. Figure 4 In the example shown, the period on the right is equally divided into eight intervals t0, t1, ... t7, and the period on the left is equally divided into eight intervals t8, t9, ... t15. Here, each time interval represents a shorter period of time, so it is assumed that no movement occurs within each interval, but movement may occur between intervals.

[0042] Once these time intervals are established, the corresponding feature-enhanced projection data can be determined. Using this feature-enhanced projection data, multiple parts of the image can be reconstructed at time intervals. Figure 4 In the example, the group (Par0, Par1, ... Par7) of partial images of one side is related to the target time T. r The preceding period corresponds to the group of partial images (Par8, Par9, ... Par15) of the other party and the target time T. r The subsequent period corresponds to this. Unlike previous partial images, these feature-enhanced partial images contain feature maps that emphasize the edges and high-frequency details of the imaged objects, providing a more reliable basis for motion estimation and motion compensation.

[0043] exist Figure 4 In the example shown, each portion of the image is reconstructed using only 1 / 16 of the fully rotated projection data, resulting in a significant improvement in temporal resolution. The number of portions (16 in this example) is illustrative and not limited to this. Depending on the specific application, more or fewer portions of the image can be reconstructed.

[0044] Figure 5 This is a flowchart illustrating step 500 for reconstructing a feature-enhanced portion of the image according to an embodiment of the present disclosure. In step S510, raw projection data of the imaging object is received. In step S520, feature enhancement is performed on the received raw projection data. In step S530, the target time T is determined based on input from the operator of the CT imaging system. r In step S540, feature-enhanced projection data is used to reconstruct the target time T. r The first group of partial images and the second group of partial images on both sides.

[0045] Figure 6A block diagram of a motion vector field estimation circuit 130 according to an embodiment of the present disclosure is shown. The motion vector field estimation circuit 130 includes a partial image warping circuit 610, an image pair generation circuit 620, and a motion model update circuit 630.

[0046] In embodiments of this disclosure, a four-dimensional (4D) motion model is used to characterize the motion of the imaging object. At any time t, the motion of the imaging object is represented by the motion vector field MVF(x, y, z, t). Here, (x, y, z) represents the spatial coordinates of a voxel within the imaging object.

[0047] Through this method, Figure 4 In the example shown, the motion corresponding to the 16 partial images can be obtained as 16 motion vector fields MVF(x,y,z,t0), MVF(x,y,z,t1), ..., MVF(x,y,z,t15).

[0048] The partial image warping circuit 610 warps partial images Par0, Par1, ..., Par15 based on these motion vector fields, generating 16 warped partial images. Specifically, each partial image is warped according to its corresponding motion vector field as follows.

[0049] , …, Motion within a portion of the image is corrected by warping each pixel of the portion using the new spatial location calculated using the motion vector field. The warped portion of the image is then sent to the image pair generation circuit 620.

[0050] The image pair generation circuit 620 uses the distorted partial image received from the partial image distortion circuit 610 to generate an image pair im0 and im1 for optimizing the parameters of the motion model. The generated image pair is then sent to the motion model update circuit 630.

[0051] Figure 7 This illustrates an example of a generated image pair according to an embodiment of the present disclosure. For example, the first image im0 is obtained by summing the distorted partial images Par0-Par7 as follows.

[0052] im0 = par0 + par1 + ... + par7 Similarly, the second image im1 is obtained by summing the distorted partial images Par8 to Par15 as follows.

[0053] im1 = par8 + par9 + ... + par15 These two images form an image pair used to optimize the parameters of the motion model.

[0054] The motion model update circuit 630 adjusts the parameters of the motion model based on the loss function used in the image registration between images im0 and im1, and sends the adjusted motion model to the partial image warping circuit 610. The parameters of the motion model can be refined through iterative operations of the partial image warping circuit 610, the image pair generation circuit 620, and the motion model update circuit 630. Note that the motion model update circuit 630 sets the parameters of the motion model to predefined initial values ​​for the first iteration.

[0055] Because the motion of the heart is smooth, the motion of any voxel can be approximated using a smooth fitting function. Furthermore, by using multiple small partial images, highly nonlinear motion models such as high-order polynomial functions and B-spline functions can be employed.

[0056] For example, the smooth motion of a voxel at a specific location (x0, y0, z0) can be modeled using a cubic polynomial function as follows.

[0057] MVF(t) = at 3 +bt 2 +ct+d Thus, the motion estimation problem is transformed into an optimization problem aimed at determining coefficients a, b, c, and d through image registration between images im0 and im1. There are no particular restrictions on the form of the loss function used in image registration; various types of loss functions can be used to determine the difference between im0 and im1.

[0058] When im0 and im1 are fully consistent, the coefficients of the polynomial function are determined. At this moment, the motion model update circuit 630 stops further updating of the motion model and calculates the target time T based on the determined motion model. r The estimated motion vector field is then output to the image reconstruction circuit 140. Using the estimated motion vector field and the original projection data, the image reconstruction circuit 140 reconstructs a motion-corrected image of the imaging object.

[0059] Figure 8This is a flowchart illustrating an exemplary step 800 for estimating a motion vector field according to an embodiment of the present disclosure. In step S810, a group of two feature-enhanced partial images is received. In step S820, the feature-enhanced partial images are warped based on a motion model. In step S830, an image pair (im0, im1) is generated based on the warped partial images. In step S840, image registration of the image pair (im0, im1) is performed, and the parameters of the motion model are optimized. In step S850, it is determined whether a predetermined end criterion is met. If the end criterion is not met, the motion model is updated in step S860, and the process returns to step S820. If the end criterion is met, in step S870, the motion model is determined, and the target time T is calculated based on the determined motion model. r The calculated motion vector field (4D MVF) is output as the inferred motion vector field.

[0060] For example, the termination criterion could be that the loss function used for image registration converges to a predefined threshold, or that the number of iterations in steps S820-S860 reaches a predefined value. For instance, updates could stop when the number of iterations reaches 200.

[0061] The motion vector field of the target time Tr can be estimated using a neural network. Figure 9A and Figure 9B The neural network shown is used to estimate a motion vector field in a scenario illustrating an embodiment of the present disclosure. The neural network includes, for example, one or more convolutional layers, one or more spatial transformation layers, and at least one loss layer.

[0062] Figure 9A The diagram illustrates an exemplary training step of the neural network. The convolutional layers provide the motion vector fields from time t0 to t15 to the spatial transformation layer. Based on these motion vector fields, the spatial transformation layer warps and sums the partial images Par0-Par15 to generate image pairs im0 and im1. The loss layer obtains a registration loss to quantify the difference between im0 and im1. The weights of the convolutional layers are updated based on the registration loss. Through this training process, the neural network learns the relationship between the two groups of partial images (Par0 to Par7, Par8 to Par15) on either side of the target time Tr and the target time T. r The mapping relationship between the motion vector fields. For example... Figure 9B As shown, it is possible to extract the target time T from the output of a convolutional layer trained for image reconstruction without motion. r The inferred motion vector field.

[0063] Computational efficiency can be further improved by using pre-trained neural networks. For example, the neural network can be pre-trained using data scanned from an imaging object, and then applied to motion compensation of that object or other imaging objects. More preferably, to capture as much diversity as possible, the neural network can be pre-trained using data scanned from multiple imaging objects (e.g., more than 100). This method can significantly improve processing speed.

[0064] The method described in this disclosure is particularly suitable for motion correction of the whole heart. In particular, it offers significant advantages over conventional methods that rely on large pairs of partial images, by using multiple small partial images.

[0065] First, by segmenting the projection data into smaller segments and reconstructing each part of the image with a limited amount of data, the temporal resolution becomes very high, effectively minimizing motion artifacts within each part of the image. For example, as... Figure 4 As shown, only about 20 degrees of projection data are used in the reconstruction of each image segment, resulting in a significant improvement in temporal resolution. In contrast, when reconstructing large image pairs, each image segment covers a wider data range, causing motion within some images and reducing the accuracy of subsequent image registration.

[0066] Second, by using multiple small partial images, highly nonlinear motion models such as high-order polynomials or B-spline models can be achieved. In contrast, previous methods generally relied on linear motion models, which are often inappropriate in clinical settings.

[0067] Furthermore, by using multiple small partial images, data utilization is improved. Due to motion artifacts, previous methods typically failed to fully utilize the data obtained from a complete rotation. In contrast, the method using multiple small partial images allows for the use of all available data without compromising image registration. For example, as... Figure 4 As shown, in order to reconstruct the 16 partial images, the projection data obtained from the whole rotation can be used effectively.

[0068] Furthermore, this disclosure is described in the context of cardiac CT imaging, but the concepts disclosed can also be applied to motion estimation and motion compensation in other anatomical regions (e.g., lung imaging, but not limited thereto).

[0069] Figure 10 This is a schematic block diagram of a CT imaging system (e.g., an X-ray CT apparatus or an X-ray CT scanner) according to one embodiment of this disclosure. Figure 10As shown, the gantry 1050 is illustrated from the side and includes an X-ray tube 1051, a flange 1052, and a multi-row or two-dimensional array type X-ray detector 1053. The X-ray tube 1051 and the X-ray detector 1053 are mounted radially, sandwiching a subject OBJ supported on the flange 1052, which is rotatable about the rotation axis RA. The rotation unit 1057 rotates the flange 1052 at a high speed of 0.275 seconds per revolution as the subject OBJ moves along the rotation axis RA inward or outward along the plane shown in the figure.

[0070] Hereinafter, embodiments of the CT imaging system of this disclosure will be described with reference to the accompanying drawings. Furthermore, CT imaging systems include various types of devices, such as rotating / rotational type devices in which the X-ray tube and X-ray detector rotate together around the subject being examined, and fixed / rotational type devices in which multiple detector elements are arranged in a ring or plane and only the X-ray tube rotates around the subject being examined. This disclosure can be applied to any type. Here, the rotating / rotational type, which is currently the mainstream, is illustrated.

[0071] The CT imaging system (e.g., a multi-slice X-ray CT apparatus) also includes a high-voltage generator 1059. The high-voltage generator 1059 generates a tube voltage applied to the X-ray tube 1051 via a slip ring 1058, causing the X-ray tube 1051 to generate X-rays. The X-rays irradiate the subject OBJ. A circle represents the cross-sectional area of ​​the subject OBJ. For example, the X-ray tube 1051 has a lower average X-ray energy in the first scan than the average X-ray energy in the second scan. This allows for more than two scans corresponding to different X-ray energies. An X-ray detector 1053, located on the opposite side of the X-ray tube 1051 across the subject OBJ, detects the X-rays propagating through the subject OBJ. The X-ray detector 1053 also includes various detection elements or detection units.

[0072] The CT imaging system also includes other devices for processing the detection signals from the X-ray detector 1053. The data acquisition circuit, or data acquisition system (DAS) 1054, converts the signals output from the X-ray detector 1053 of each channel into voltage signals, amplifies these signals, and then converts them into digital signals. The X-ray detector 1053 and DAS 1054 are configured to process a predetermined total number of projections per rotation (TPPR).

[0073] The aforementioned data is transmitted via a non-contact data transmitter 1055 to a preprocessing device 1056 housed in a console external to the camera gantry 1050. The preprocessing device 1056 performs specific corrections, such as sensitivity correction, related to the raw data. A storage device 1062 stores the resulting data, also known as projection data, in the stage immediately preceding the reconstruction process. The storage device 1062, along with the reconstruction device 1064, input device 1065, and display 1066, is connected to a system controller 1060 via a data / control bus 1061. The system controller 1060 controls a current regulator 1063 that limits the current to a level sufficient to drive the CT system.

[0074] In various generations of CT scanning systems, the detectors are rotated and / or fixed relative to the patient. In one embodiment, the CT system described above may be an example of a system combining third-generation and fourth-generation structures. In a third-generation system, the X-ray tube 1051 and the X-ray detector 1053 are radially mounted on an annular frame 1052, rotating around the subject OBJ as the annular frame 1052 rotates about the rotation axis RA. In a fourth-generation system, the detectors are fixedly positioned around the patient, and the X-ray tube rotates around the patient. In another embodiment, the gantry 1050 has multiple detectors configured on an annular frame 1052 supported by a C-arm and a stand.

[0075] The storage device 1062 is capable of storing measured values ​​of the radiation irradiance of the X-rays emitted by the X-ray detector 1053. Furthermore, the storage device 1062 is capable of storing a dedicated program for performing CT image reconstruction, material identification, and motion estimation and motion compensation methods, including those described in this specification.

[0076] The reconstruction device 1064 is capable of performing the methods described in this specification. Furthermore, the reconstruction device 1064 is capable of performing pre-reconstruction image processing such as volume rendering and image difference processing as needed.

[0077] The pre-reconstruction processing of the projection data performed by the pre-processing device 1056 can include, for example, corrections for detector calibration, detector nonlinearity, and polarity effects.

[0078] The post-reconstruction processing performed by the reconstruction device 1064 can include image filtering and smoothing, volume rendering, and image differencing as needed. The image reconstruction process can be performed using filtered back projection, successive approximation image reconstruction, or probabilistic image reconstruction. The reconstruction device 1064 can, for example, use memory to store projection data, reconstructed images, calibration data and parameters, and computer programs.

[0079] The reconfigurable device 1064 can include a CPU (processing circuit) that can be executed as discrete logic gates, such as an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA), or a Complex Programmable Logic Device (CPLD). The FPGA or CPLD can also be coded using VDHL, Verilog, or other hardware description languages. This code can be stored directly in the electronic memory within the FPGA or CPLD, or it can be stored as a separate electronic memory. Furthermore, the storage device 1062 can be non-volatile, such as ROM, EPROM, EEPROM, or flash memory. The storage device 1062 can also be volatile, such as static RAM or dynamic RAM, or it can be configured with a processor, such as a microcontroller or microprocessor, to manage the electronic memory and the interaction between the FPGA or CPLD and the memory.

[0080] Alternatively, the CPU within the reconfigurable device 1064 can execute a computer program containing a group of computer-readable instructions that perform the functions described herein, the program being stored in any of the aforementioned non-transitory electronic storage and / or hard disk drive, CD, DVD, flash drive, or other known storage media. Furthermore, the computer-readable instructions may also be provided by utility applications, background daemons, or components of an operating system, or combinations thereof, and can be executed in cooperation with processors such as Intel Xenon or AMD Opteron, as well as Microsoft 10, UNIX, Solaris, LINUX, Apple, macOS, and other operating systems known to those skilled in the art. Furthermore, the CPU can be installed as multiple processors operating in parallel to execute instructions.

[0081] In one embodiment, the reconstructed image can be displayed on display 1066. Display 1066 can be an LCD display, CRT display, plasma display, OLED, LED, or other displays known in the art.

[0082] The storage device 1062 can be configured as a hard disk drive, CD-ROM drive, DVD drive, flash memory drive, RAM, ROM, or other electronic storage device known in the art.

[0083] According to at least one of the embodiments described above, the accuracy of motion compensation can be further improved.

[0084] Several embodiments of the present invention have been described, but these embodiments are given by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other ways, and various omissions, substitutions, modifications, and combinations of embodiments are possible without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the scope of the invention as set forth in the claims and its equivalents.

[0085] Regarding the above-described embodiments, as an aspect of the invention and as an optional feature, the following notes are disclosed.

[0086] (1) A motion compensation device for performing motion compensation in a computed tomography imaging system, the motion compensation device comprising: The acquisition unit acquires projection data scanned from the imaging object by the computed tomography imaging system; The decision unit determines the target time for reconstructing the motion-corrected image of the imaging object; The first reconstruction unit reconstructs a first group of feature-enhanced images and a second group of feature-enhanced images based on a first group of projection data related to both sides of the determined target time and a second group of projection data, respectively. The estimation unit, based on the first group of reconstructed feature-enhanced partial images and the second group of reconstructed feature-enhanced partial images, estimates the motion vector field at the determined target time; and The second reconstruction unit performs motion compensation on the acquired projection data based on the estimated motion vector field, thereby reconstructing the image of the imaging object into the motion-corrected image at the determined target time.

[0087] (2) The motion compensation device according to (1), wherein, The first reconstruction unit reconstructs a first group of the feature-enhanced images and a second group of the feature-enhanced images by performing the following processes: By performing feature enhancement on the acquired projection data, feature-enhanced projection data is generated, in which feature information corresponding to the edges of the physiological structure of the imaging object within the acquired projection data is enhanced; The determination includes a first period prior to the determined target time and a second period following the determined target time, which has the same length as the first period. The first period is divided into a first plurality of time intervals, and the second period is divided into a second plurality of time intervals; Partial reconstruction is performed using the feature-enhanced projection data associated with each of the first plurality of time intervals, thereby generating a first plurality of partial images corresponding to different time intervals of the first plurality of time intervals, as a first group of the feature-enhanced partial images; and Partial reconstruction is performed using the feature-enhanced projection data associated with each of the second plurality of time intervals, thereby generating a second plurality of partial images corresponding to different time intervals of the second plurality of time intervals, as a second group of the feature-enhanced partial images.

[0088] (3) The motion compensation device according to (2), wherein, The first reconstruction unit enhances the feature information within the acquired projection data by performing high-pass filtering on the acquired projection data or by performing an export operation on the acquired projection data.

[0089] (4) The motion compensation device according to (2), wherein, The first reconstruction unit sets the length of the first period to a predefined length.

[0090] (5) The motion compensation device according to (3), wherein, The first reconstruction unit sets the length of the first period based on the enhanced feature information.

[0091] (6) The motion compensation device according to (1), wherein, The first reconstruction unit receives from the operator of the computed tomography imaging system an input indicating which stage of the period of motion of the object being imaged is being reconstructed after motion correction, and determines the target time based on the received input.

[0092] (7) The motion compensation device according to (2), wherein The estimating unit estimates the motion vector field in the following way: The following process is repeated until a predefined termination criterion is met: warping the first plurality of partial images based on a motion model to generate a first plurality of warped partial images; warping the second plurality of partial images based on the motion model to generate a second plurality of warped partial images; combining the first plurality of warped partial images to obtain a first image; combining the second plurality of warped partial images to obtain a second image; and updating the parameters of the motion model based on a loss function calculated based on image registration between the first image and the second image; and The output is the motion vector field derived from the updated motion model as the inferred motion vector field.

[0093] (8) The motion compensation device according to (7), wherein, The estimation unit sets the parameters of the motion model to predefined initial values ​​for the first iteration.

[0094] (9) The motion compensation device according to (7), wherein, The estimation part uses a specific motion model constructed based on a polynomial function or a B-spline function as the motion model.

[0095] (10) The motion compensation device according to (2), wherein, The estimation unit inputs the first plurality of partial images and the second plurality of partial images into the neural network to train the neural network, and obtains the motion vector field inferred by the trained neural network for the determined target time as the estimated motion vector field, thereby estimating the motion vector field.

[0096] (11) A motion compensation method for performing motion compensation in a computed tomography imaging system, the motion compensation method comprising the following steps: The step of obtaining projection data scanned from the imaging object by the computed tomography imaging system; The step of determining the target time for reconstructing the motion-corrected image of the imaging object; The steps of reconstructing a first group of feature-enhanced partial images and a second group of feature-enhanced partial images based on a first group of projection data related to both sides of the determined target time and a second group of projection data, respectively; The step of estimating the motion vector field at the determined target time based on the first group and the second group of reconstructed feature-enhanced partial images; and Based on the estimated motion vector field, motion compensation is performed on the acquired projection data to reconstruct the image of the imaging object, which is then used as the motion-corrected image at the determined target time.

[0097] (12) According to the motion compensation method described in (11), wherein, The steps of reconstructing the first group and the second group of the feature-enhanced partial image include: The step of performing feature enhancement on the acquired projection data to generate feature-enhanced projection data, wherein feature information corresponding to the edges of the physiological structure of the imaging object within the acquired projection data is enhanced; The steps of determining a first period before the determined target time and a second period after the determined target time having the same length as the first period; The steps of separating the first period into a first plurality of time intervals and separating the second period into a second plurality of time intervals; Performing partial reconstruction using the feature-enhanced projection data associated with each of the first plurality of time intervals, generating first plurality of partial images corresponding to different time intervals of the first plurality of time intervals, as a first group of the feature-enhanced partial images; and The step of performing partial reconstruction using the feature-enhanced projection data associated with each of the second plurality of time intervals to generate a second plurality of partial images corresponding to different time intervals of the second plurality of time intervals, as a second group of the feature-enhanced partial images.

[0098] (13) According to the motion compensation method described in (12), wherein, The step of generating the feature-enhanced projection data enhances the feature information within the obtained projection data by performing high-pass filtering or exporting the obtained projection data.

[0099] (14) According to the motion compensation method described in (12), wherein, In the step of determining the first period and the second period, the length of the first period is set to a predefined length.

[0100] (15) According to the motion compensation method described in (13), wherein, In the step of determining the first period and the second period, the length of the first period is set based on the enhanced feature information.

[0101] (16) According to the motion compensation method described in (11), wherein, In the step of determining the target time, input is received from the operator of the computed tomography imaging system, indicating at which stage of the period of motion of the object being imaged is the motion-corrected image reconstructed, and the target time is determined based on the received input.

[0102] (17) According to the motion compensation method described in (12), wherein, In the step of estimating the motion vector field The following process is repeated until a predefined termination criterion is met: warping the first plurality of partial images based on a motion model to generate a first plurality of warped partial images; warping the second plurality of partial images based on the motion model to generate a second plurality of warped partial images; combining the first plurality of warped partial images to obtain a first image; combining the second plurality of warped partial images to obtain a second image; and updating the parameters of the motion model based on a loss function calculated based on image registration between the first image and the second image. The motion vector field derived from the updated motion model is output as the inferred motion vector field.

[0103] (18) According to the motion compensation method described in (17), wherein, In the step of estimating the motion vector field, the parameters of the motion model are set to predefined initial values ​​for the first iteration.

[0104] (19) According to the motion compensation method described in (17), wherein, In the step of estimating the motion vector field, a specific motion model constructed based on a polynomial function or a B-spline function is used as the motion model.

[0105] (20) A motion compensation device for performing motion compensation in a computed tomography imaging system, the motion compensation device comprising: The acquisition unit acquires projection data scanned from the first imaging object by the computed tomography imaging system; The decision unit determines the first target moment for reconstructing the motion-corrected image of the first imaging object; The first reconstruction unit reconstructs a first group of feature-enhanced images and a second group of feature-enhanced images based on a first group of projection data related to both sides of the determined first target time and a second group of projection data, respectively. The estimation unit, by inputting a first group and a second group of the reconstructed feature-enhanced images into the input of a pre-trained neural network, estimates the motion vector field at the determined first target time; and The second reconstruction unit performs motion compensation on the acquired projection data based on the estimated motion vector field, thereby reconstructing the image of the first imaging object as the motion-corrected image at the determined first target time. The pre-trained neural network was trained using a third group and a fourth group of feature-enhanced images. These third and fourth groups of feature-enhanced images were reconstructed based on a first group of projection data obtained from the second imaging object and a second group of projection data obtained from the second imaging object, i.e., the first and second groups of projection data related to both sides of the second target time. The second imaging object is the same as or different from the first imaging object.

[0106] Based on the foregoing teachings, various corrections and modifications can be made to the embodiments described in this specification. Therefore, it should be understood that, within the scope of the claims, this disclosure can be implemented by methods other than those specifically described in this specification.

Claims

1. A motion compensation device for performing motion compensation in a computed tomography imaging system, the motion compensation device comprising: The acquisition unit acquires projection data scanned from the imaging object by the computed tomography imaging system; The decision unit determines the target time for reconstructing the motion-corrected image of the imaging object; The first reconstruction unit reconstructs a first group of feature-enhanced images and a second group of feature-enhanced images based on a first group of projection data related to both sides of the determined target time and a second group of projection data, respectively. The estimation unit estimates the motion vector field at the determined target time based on the first group of reconstructed feature-enhanced images and the second group of reconstructed feature-enhanced images; as well as The second reconstruction unit performs motion compensation on the acquired projection data based on the estimated motion vector field, thereby reconstructing the image of the imaging object as the motion-corrected image at the determined target time.

2. The motion compensation device according to claim 1, wherein, The first reconstruction unit reconstructs a first group of the feature-enhanced images and a second group of the feature-enhanced images by performing the following processes: Feature enhancement is performed on the acquired projection data to generate feature-enhanced projection data, in which feature information corresponding to the edges of the physiological structure of the imaging object within the acquired projection data is enhanced; The determination includes a first period prior to the determined target time and a second period following the determined target time, which has the same length as the first period. The first period is divided into a first plurality of time intervals, and the second period is divided into a second plurality of time intervals; Partial reconstruction is performed using the feature-enhanced projection data associated with each of the first plurality of time intervals, thereby generating a first plurality of partial images corresponding to different time intervals of the first plurality of time intervals, as a first group of the feature-enhanced partial images; as well as Partial reconstruction is performed using the feature-enhanced projection data associated with each of the second plurality of time intervals, thereby generating a second plurality of partial images corresponding to different time intervals of the second plurality of time intervals, as a second group of the feature-enhanced partial images.

3. The motion compensation device according to claim 2, wherein, The first reconstruction unit enhances the feature information within the acquired projection data by performing high-pass filtering on the acquired projection data or by performing an export operation on the acquired projection data.

4. The motion compensation device according to claim 2, wherein, The first reconstruction unit sets the length of the first period to a predefined length.

5. The motion compensation device according to claim 3, wherein, The first reconstruction unit sets the length of the first period based on the enhanced feature information.

6. The motion compensation device according to claim 1, wherein, The first reconstruction unit receives input from the operator of the computed tomography imaging system indicating which stage of the motion-corrected image is reconstructed during the period of motion of the object being imaged, and determines the target time based on the received input, thereby determining the target time.

7. The motion compensation device according to claim 2, wherein, The estimating unit estimates the motion vector field in the following way: Repeat the following process until a predefined end benchmark is met: warp the first plurality of partial images based on the motion model to generate the first plurality of warped partial images; The second plurality of partial images are distorted based on the motion model to generate the second plurality of distorted partial images; The first image is obtained by combining the first plurality of distorted partial images; The second plurality of distorted partial images are combined to obtain a second image; and the parameters of the motion model are updated based on a loss function calculated based on image registration between the first image and the second image; and The output is the motion vector field derived from the updated motion model as the inferred motion vector field.

8. The motion compensation device according to claim 7, wherein, The estimation unit sets the parameters of the motion model to predefined initial values ​​for the first iteration.

9. The motion compensation device according to claim 7, wherein, The estimation part uses a specific motion model constructed based on a polynomial function or a B-spline function as the motion model.

10. The motion compensation device according to claim 2, wherein, The estimation unit trains the neural network by inputting the first plurality of partial images and the second plurality of partial images into the neural network, and obtains the motion vector field inferred by the trained neural network for the determined target time as the estimated motion vector field, thereby estimating the motion vector field.

11. A motion compensation method for performing motion compensation in a computed tomography imaging system, the motion compensation method comprising the following steps: The step of obtaining projection data scanned from the imaging object by the computed tomography imaging system; The step of determining the target time for reconstructing the motion-corrected image of the imaging object; The steps of reconstructing a first group of feature-enhanced partial images and a second group of feature-enhanced partial images based on a first group of projection data related to both sides of the determined target time and a second group of projection data, respectively; The step of estimating the motion vector field at the determined target time based on the first group of reconstructed feature-enhanced partial images and the second group of reconstructed feature-enhanced partial images; as well as Based on the estimated motion vector field, motion compensation is performed on the acquired projection data to reconstruct the image of the imaging object, which is then used as the motion-corrected image at the determined target time.

12. A motion compensation device for performing motion compensation in a computed tomography imaging system, the motion compensation device comprising: The acquisition unit acquires projection data scanned from the first imaging object by the computed tomography imaging system; The decision unit determines the first target moment for reconstructing the motion-corrected image of the first imaging object; The first reconstruction unit reconstructs a first group of feature-enhanced images and a second group of feature-enhanced images based on a first group of projection data related to both sides of the determined first target time and a second group of projection data, respectively. The estimation unit estimates the motion vector field at the determined first target time by inputting the first group of reconstructed feature-enhanced images and the second group of reconstructed feature-enhanced images into the input of a pre-trained neural network. as well as The second reconstruction unit performs motion compensation on the acquired projection data based on the estimated motion vector field, thereby reconstructing the image of the first imaging object as the motion-corrected image at the determined first target time. The pre-trained neural network is trained using a third group and a fourth group of feature-enhanced partial images. These three groups are reconstructed based on a first group of projection data obtained from the second imaging object and a second group of projection data obtained from the second imaging object, i.e., the first group and the second group of projection data related to both sides of the second target time. The second imaging object is the same as or different from the first imaging object.

Images