X-ray CT apparatus and processor for image processing

Abstract

The present invention addresses the problem of providing an X-ray CT device and a processor for image processing, which prevent excessive dynamic correction or deformation during dynamic correction and reconstruction of the X-ray CT device, in particular, maintain image continuity in the body axis direction, and perform dynamic correction and reconstruction under optimal conditions corresponding to image reconstruction conditions. An image processing processor of an X-ray CT apparatus according to the present invention is provided with a dynamic estimation model for acquiring dynamic information, the dynamic estimation model including regularization terms that can be independently adjusted with respect to the continuity of a spatial domain and a time domain. The dynamic estimation model is configured so as to be able to adjust the weight of the regularization term or the position of the control point. The dynamic information acquisition unit automatically adjusts the dynamic estimation model or a control point parameter calculation method using the dynamic estimation model on the basis of an FOV or an image reconstruction interval, which is an image reconstruction condition. As a result, it is possible to satisfactorily maintain image continuity in one cross-section and inter-slice image continuity in the body axis direction.

Description

Technical Field

[0001] The present invention relates to an X-ray CT apparatus and a processor for processing transmission X-ray data obtained by the X-ray CT apparatus, and more particularly to a dynamic correction reconstruction technique for CT images. Background Art

[0002] In an X-ray CT apparatus, a subject is disposed in an opening of a scanner equipped with an X-ray source and an X-ray detector, and imaging is performed while rotating the scanner. An image of a cross-section of the subject is reconstructed using X-ray projection data acquired at each angle of the scanner. At this time, imaging is performed while moving the position of the subject in the body axis direction relative to the scanner, thereby obtaining images of a plurality of cross-sections along the body axis direction.

[0003] When imaging a dynamic part of a subject such as the chest using a CT apparatus, in order to reduce artifacts caused by movement, dynamic correction reconstruction that detects dynamic information of the subject during imaging and uses the dynamic information for image reconstruction is widely employed.

[0004] In the detection of dynamic information, partial reconstruction (PAR) images (hereinafter referred to as PAR images) reconstructed using less than 180 degrees of transmission X-ray data (hereinafter referred to as projection data) acquired at positions that are time-aligned with respect to the image reconstruction position as the target are used (for example, Patent Document 1). Specifically, two PAR images constituting a pair of PAR images are images reconstructed from data projected from directions in which the subject is inverted by 180 degrees. If the subject does not move during the 180-degree movement of the scanner, they become the same image, but there are changes caused by heartbeat or breathing movements during this period. In dynamic correction reconstruction, the change amount of each pixel of such PAR images is calculated as dynamic information, and the projection data is corrected based on the dynamic information to reconstruct an image.

[0005] In cardiac imaging targeting the heart, in order to minimize the influence of heartbeat as much as possible, electrocardiogram-synchronized imaging is performed in which the image reconstruction position as the target is made to coincide with a specific phase of the cardiac phase. Therefore, the dynamic information obtained from the PAR images is also information corresponding to a specific phase. On the other hand, in chest imaging without electrocardiogram synchronization (non-synchronized imaging), etc., the central phases in the images reconstructed for each cross-section are different, and thus there is a problem that the continuity of the images is significantly reduced particularly in the body axis direction.

[0006] Patent Document 1: U.S. Patent Publication No. 2018 / 0005414

[0007] In the technology described in Patent Document 1, as a dynamic estimation model when using PAR images to obtain dynamic information, a free-form deformation model based on a 4D B-spline function that also takes into account the time information when acquiring PAR images is used, and by using multiple pairs of PAR images, deterioration of image continuity caused by inconsistent reconstruction center phases is prevented. However, in this technology, since multiple pairs of PAR images are used in the reconstruction of one cross-section, there is a problem of high computational cost. Also, in this technology, the continuity of the images in the body axis direction is not considered.

[0008] Moreover, in asynchronous photography, usually the table speed is faster than that in electrocardiogram-synchronized photography, resulting in sparse scanning, and the data available for PAR image production becomes less. Therefore, there is a problem of reduced dynamic correction accuracy.

[0009] Furthermore, in the case of reconstructing images from data continuously acquired along the body axis direction, sometimes the image reconstruction conditions such as slice thickness or FOV are changed according to the imaging position, but in the conventional dynamic correction reconstruction technology, the dynamic correction accuracy depends greatly on the image reconstruction conditions. Summary of the Invention

[0010] An object of the present invention is to maintain image continuity in the body axis direction and perform optimal dynamic correction reconstruction without being affected by image reconstruction conditions, especially in photography using an X-ray CT device.

[0011] As a dynamic estimation model for obtaining dynamic information, the present invention introduces a dynamic estimation model in which the continuity in the spatial domain and the temporal domain can be independently controlled. By independently controlling the continuity in the spatial domain and the temporal domain, it is possible to well maintain the image continuity within one cross-section and the image continuity between the slices in the body axis direction.

[0012] Moreover, the present invention is configured to be able to dynamically adjust the dynamic estimation model or the calculation method using the dynamic estimation model in response to changes in image reconstruction conditions.

[0013] That is, the first aspect of the present invention is the following X-ray CT device.

[0014] An X-ray CT device includes: an imaging unit having a scanner on which an X-ray source and an X-ray detector that rotate around a subject are mounted, and a moving mechanism that relatively moves the position of the scanner in the body axis direction of the subject to acquire transmission X-ray data (hereinafter, also referred to as projection data) that is different in terms of the angle and the position in the body axis direction with respect to the subject; and an arithmetic unit that generates a tomographic image of the subject using the projection data acquired by the imaging unit.

[0015] The arithmetic unit includes: a partial reconstruction image generation unit that generates a pair of partial reconstruction images at the facing position using projection data; a dynamic information acquisition unit that applies a dynamic estimation model to the pair of partial reconstruction images to acquire dynamic information of the subject being scanned; and a dynamic correction reconstruction unit that reconstructs a tomographic image using the dynamic information and projection data acquired within an angular range of 180 degrees or more. The dynamic estimation model includes a first regularization term that can be independently adjusted and maintains the spatial continuity of the image, and a second regularization term that maintains the temporal continuity. The dynamic information acquisition unit automatically changes at least one of the dynamic estimation model and the calculation method of dynamic estimation using the dynamic estimation model according to the image reconstruction conditions.

[0016] Moreover, a second aspect of the present invention is a processor for image processing, characterized by: generating a pair of partial reconstruction images at the facing position using projection data; applying a dynamic estimation model including a first regularization term that can be independently adjusted and maintains the spatial continuity of the image, and a second regularization term that maintains the temporal continuity, to the pair of partial reconstruction images to acquire dynamic information of the subject being scanned; and reconstructing a tomographic image using the dynamic information and projection data acquired within an angular range of 180 degrees or more.

[0017] Advantages of the Invention

[0018] According to the present invention, by having a dynamic estimation model including a first regularization term that can be independently adjusted and maintains the spatial continuity of the image, and a second regularization term that maintains the temporal continuity, it is possible to adjust while maintaining the spatial and temporal continuity of the image corresponding to the set image reconstruction conditions or changes thereto, thereby improving the accuracy of dynamic estimation. And, according to the present invention, it is possible to dynamically change the dynamic estimation process using the dynamic estimation model. Even when a change in the image reconstruction conditions occurs during image reconstruction, it is possible to maintain the continuity of the image especially in the body axis direction, prevent deformation caused by excessive dynamic correction, and provide an image helpful for diagnosis. Brief Description of the Drawings

[0019] Figure 1 It is a diagram showing the overall outline of an X-ray CT apparatus to which the present invention is applied.

[0020] Figure 2 It is a functional block diagram of an X-ray CT apparatus based on an embodiment of the present invention.

[0021] Figure 3 It is a functional block diagram of an embodiment of a processor included in the X-ray CT apparatus.

[0022] Figure 4 It is a diagram showing the outline of the process of dynamic correction reconstruction common to the embodiments of the present invention.

[0023] Figure 5 It is a diagram illustrating dynamic correction reconstruction.

[0024] Figure 6 It is a diagram illustrating the generation of a partially reconstructed image.

[0025] Figure 7 It is a diagram illustrating the concept of the dynamic estimation model of Embodiment 1.

[0026] Figure 8 It is a diagram illustrating the processing of Embodiment 2 and showing the relationship between the FOV and the control point interval.

[0027] Figure 9 It is a diagram illustrating the processing of Embodiment 2 and showing the relationship between the FOV and the control point position.

[0028] Figure 10 It is a diagram showing the processing flow of Embodiment 3.

[0029] Figure 11 It is a diagram illustrating the processing of Embodiment 3.

[0030] Symbol Explanation

[0031] 1 - X - ray CT apparatus, 10 - imaging unit, 30 - arithmetic unit, 200 - processor, 310 - image reconstruction unit, 311 - PAR image generation unit, 312 - dynamic correction reconstruction unit, 330 - dynamic information acquisition unit, 331 - motion vector calculation unit, 332 - dynamic estimation model adjustment unit. Specific Embodiment

[0032] Hereinafter, embodiments of the X - ray CT apparatus of the present invention and the processor mainly responsible for image processing will be described. In addition, in the present embodiment, the processor refers to a general - purpose computer having a general - purpose CPU (Central Processing Unit) or GPU (Graphic Processing Unit) and a memory, and hardware including programmable ICs such as ASIC (Application Specific Integrated Circuit), FPGA (Field - Programmable Gate Array), and CPLD (Complex Programmable Logic Device), and any one or combination of these is collectively referred to as the processor.

[0033] First, the overall structure of the X - ray CT apparatus to which the present invention is applied will be described.

[0034] AsFigure 1 As shown in Figure 1 , the X-ray CT apparatus 1 includes: an imaging unit 10 having a gantry 100 and a table device 101 for taking tomographic images and fluoroscopic images of a subject 3; and an operation unit 20 for operating and controlling the imaging unit 10.

[0035] As Figure 2 shown in Figure 2 , the gantry 100 includes: an X-ray generating device 102 that generates X-rays irradiated onto the subject 3; a collimator device 104 that narrows the X-ray beam generated by the X-ray generating device 102; an X-ray detection device 103 that detects the X-rays transmitted through the subject; a scanner 108 for mounting them; a high-voltage generating device 105 that applies a high voltage to the X-ray generating device 102; a data collection device 106 that collects the transmitted X-ray data (referred to as projection data) obtained from the X-ray detection device 103; and a drive device 107 that rotates the scanner around the subject 3. Although not shown, the X-ray generating device 102 includes an X-ray tube, and a predetermined amount of X-rays are irradiated onto the subject 3 by passing a predetermined tube current through the X-ray tube.

[0036] The operation unit 20 includes: a processor 200 that functions as a central control device for controlling the various devices built in the gantry; and an input / output device 210 that functions as a user interface for interacting between the user and the processor 200. An arithmetic unit 30 that performs various operations such as image reconstruction on the projection data collected by the data collection device 106 is mounted in the processor 200. However, a processor that processes the data from the X-ray CT apparatus independently of the X-ray CT apparatus and different from the processor 200 can also function as the arithmetic unit 30, and such an independent processor is also included in the present invention. The functions of the processor 200 are realized by the processor 200 reading and executing a program that describes an arithmetic algorithm or a processing sequence of control, but a part of the operations or processes performed by the arithmetic unit 30 can also be performed using a PLD (programmable logic device) such as an ASIC or an FPGA.

[0037] The input / output device 210 is composed of an input device 212 used by an operator to input imaging conditions and the like, a display device 211 for displaying data such as imaging images or a GUI, and a storage device 213 for storing imaging required data such as programs or device parameters.

[0038] The arithmetic unit 30 includes: an image reconstruction unit 310 that performs back-projection processing on the projection data obtained by the data collection device 106 to create a tomographic image; and a dynamic information acquisition unit 330 that detects the dynamics of the subject during scanning. And, as Figure 3As shown, in addition to the function of creating a tomographic image using transmission X-ray data within an angular range of 180 degrees or more, the image reconstruction unit 310 also has a function of creating a partial reconstruction image (PAR image) using transmission X-ray data within an angular range of less than 180 degrees (PAR image generation unit 311), and a function of performing dynamic correction reconstruction using the dynamic information obtained by the dynamic information acquisition unit 303 (dynamic correction reconstruction unit 312). The dynamic information acquisition unit 330 has a function of calculating a movement vector (also referred to as a displacement vector) representing the dynamics of the subject using a pair of PAR images produced by the PAR image generation unit 311, that is, PAR images with different projection data acquisition angular ranges and a specified dynamic estimation model (movement vector calculation unit 331), and a function of optimizing the dynamic estimation model according to the reconstruction conditions (dynamic estimation model adjustment unit 332).

[0039] The processor 200 operates as a processor that implements the functions of the arithmetic unit 30 as described above, and controls the imaging unit 10 (X-ray generating device 102, X-ray detecting device 103, high-voltage generating device 105, collimator device 104, table device 101, drive device 107, data collection device 106), the input / output device 210, and the arithmetic unit 30 according to an operation instruction from an operator via the input device 212. Under the control of the processor 200, these units operate to perform reconstruction of a CT image, display or storage of the reconstructed CT image, and so on.

[0040] Reference Figure 4 will describe an outline of the processing for photographing and performing dynamic correction on a part with dynamics as an object in the X-ray CT apparatus having the above structure.

[0041] <S1>

[0042] Place the subject 3 on the table device 101, perform positioning photography, and after setting the photography range (the region in the body axis direction of the subject), as the scanner 108 rotates, the imaging unit 10 starts scanning the photography range. The X-ray detecting device 103 in the data collection device 106 collects projection data obtained at each rotation angle and sends it to the processor 200.

[0043] <S2>

[0044] The user sets the image reconstruction conditions for the projection data collected in the photography step S1. The image reconstruction conditions include, for example, the type or parameters of the reconstruction filter used for reconstruction, the image reconstruction interval, the effective field of view (FOV: Field of View), the applicability / inapplicability of the dynamic correction function, and other conditions.

[0045] <S3>

[0046] The operation unit 30 generates tomographic images using the projection data collected by the data collection device 106 according to the image reconstruction conditions set by the user. When the dynamic correction function is not applicable, the image reconstruction unit 310 performs image reconstruction by methods such as filtered backprojection according to the set image reconstruction conditions. On the other hand, when the application of dynamic correction is selected as an image reconstruction condition, the PAR image generation unit 311 generates PAR images at positions symmetric to each other with respect to the target reconstruction center.

[0047] <S4>

[0048] The dynamic information acquisition unit 330 performs processing to estimate the dynamics of the subject through non-rigid registration using the PAR images, that is, acquires dynamic information. The dynamic estimation (non-rigid registration) based on the dynamic information acquisition unit 330 is an operation using a dynamic estimation model (B-spline function) mainly performed by the motion vector calculation unit 331, and as dynamic information, an MVF (Motion Vector Field) that maps the motion vectors of each pixel (control point) is obtained.

[0049] The dynamic information acquisition unit 330 dynamically adjusts the dynamic estimation model, parameters, etc. or calculation methods used for this operation according to the image reconstruction conditions at the time when this operation is applied, in order to maintain the spatial and temporal image continuity between images (processing of the dynamic estimation model adjustment unit 332). The adjustment of the dynamic estimation model will be described in detail in the embodiments described later.

[0050] <S5>

[0051] The dynamic correction reconstruction unit 312 uses the dynamic information (MVF) acquired by the dynamic information acquisition unit 330 to perform image reconstruction of the obtained projection data within a range of 180 degrees or more. The method of dynamic correction image reconstruction is the same as the known method. As Figure 5 shown, the dynamic correction reconstruction unit 312 estimates the magnitude or direction of the dynamics of the subject when each projection data 500 for image reconstruction is acquired based on the dynamic information (MVF) 520 obtained from the above PAR image pair 510, and while correcting the image at the target reconstruction position using the estimation result, performs filtered backprojection to reconstruct the tomographic image 530.

[0052] The reconstruction of the tomographic image (the above steps S3 to S5) is executed at the image reconstruction interval set as the image reconstruction condition, and a plurality of 2D tomographic image data or 3D tomographic image data are obtained along the body axis direction. The image reconstruction unit 310 causes the tomographic image data to be displayed as an image in a prescribed display form on the input / output device 210 (display device 211).

[0053] Based on the structure and outline of the processing of the X-ray CT apparatus described above, hereinafter, an embodiment of the dynamic correction reconstruction performed by the processor 200, particularly the adjustment of the dynamic correction model, will be described. In addition, in the following embodiments, as an example, the case of non-synchronously photographing a plurality of cross-sectional images in the body axis direction with respect to a dynamic part such as the chest will be described.

[0054] <Embodiment 1>

[0055] The feature of this embodiment is that, as the dynamic estimation model, a free-form deformation model (FFD) based on a 3D B-spline function is used, and a model modified in consideration of image continuity is used. Regarding the continuity of the image, information in the spatial domain and information in the temporal domain are independently introduced, respectively, and continuity within the tomographic image and continuity of the image in the body axis direction are achieved.

[0056] Hereinafter, the processing of this embodiment, particularly the processing of dynamic estimation ( Figure 4 S3 and S4) will be mainly described. In the following description, reference will be made as needed to Figure 4 .

[0057] <S1, S2>

[0058] The imaging unit 10 scans the imaging range and collects transmission X-ray data at each angle of the scanner. The user's setting of the image reconstruction conditions is received via the input / output device 210.

[0059] <Processing of PAR Image Generation Unit: S3>

[0060] If the dynamic correction function is selected and applied as the image reconstruction condition, the PAR image generation unit 311 generates two partial reconstruction images (PAR images) from the projection data collected by the data collection device 106 by the filtered backprojection method. The two PAR images are local images generated using projection data with an angular range of less than 180 degrees obtained at positions facing each other with the target reconstruction image position as the center. The target reconstruction image position is the position where the scanner is located at a specified rotation angle (for example, the position of 0 degrees where the X-ray source is directly above the subject), and the user can set the image reconstruction conditions or change the default values.

[0061] The relationship between the rotation angle of the scanner, the projection data, and the pair of PAR images is shown in Figure 6 . The projection data 500 at the center of the drawing is the projection data collected in the angular range 500A of 180 degrees or more of the scanner. By back-projecting the projection data 501 and 502 collected in the same angular ranges 501A and 502 that are directly opposite with respect to the image reconstruction center position 500C in the projection data 500, the following is obtained asFigure 5 The partial images 511 and 512 shown on the right side. Since these use projection data within an angular range less than 180 degrees, they are not complete cross-sectional images, but each pixel has position information capable of estimating dynamic information.

[0062] In addition, the PAR image may not be a two-dimensional image but a three-dimensional image.

[0063] When the PAR image generation unit 311 captures a plurality of cross-sections along the body axis direction, it generates a PAR image for each cross-section.

[0064] <Processing of the motion vector calculation unit: S4>

[0065] Using the pair of PAR images generated by the PAR image generation unit 311, the dynamic information acquisition unit 330 estimates the dynamics of the subject during the movement of the scanner from the position of 501A to the position of 502A. Before performing the dynamic estimation, if necessary, filtering for noise reduction may be performed on the two PAR images. By performing the noise reduction process first, the accuracy of the subsequent dynamic estimation process can be improved. As the filter, a smoothing filter such as a bilateral filter can be used, and by adjusting its parameters, the degree of noise reduction can be adjusted. When the projection data ( Figure 6 of 511 and 512) of the two PAR images is data at positions symmetric about the rotation angle 0 degrees of the scanner, usually, the tube current conditions are the same and the degree of noise is also the same. However, when the tube current conditions are different, the parameters of the filters applicable to each can be made different according to the magnitude of the tube current when acquiring the projection data.

[0066] In the dynamic estimation, non-rigid registration is performed between the pixel of interest in the local image 511 and the pixel of the local image 512 corresponding to the pixel of interest, and the motion vector (MVF: Motion Vector Field) between the images is calculated. Non-rigid registration is a process of calculating the MVF using a dynamic estimation model (function). In the present embodiment, as the dynamic estimation model, a free-form deformation (FFD) model based on a 3D B-spline function is used. The FFD model based on the 3D B-spline function is represented, for example, by the following equation (1).

[0067] [Equation 1]

[0068]

[0069] In the equation, T t,j represents FFD, x is an arbitrary voxel, t is the time point of interest, j is the control point of interest, B is the third-order tensor product of the cubic B-spline, d is the interval of the control points in the spatial region, Θ t,jis the displacement vector (3D) of the control point at time t and position j, and Θ is a set of displacement vectors (3D) of the control points representing the dynamic relationship between the reference time point and each time point (MVF: Motion Vector Field).

[0070] The dynamic estimation process amounts to calculating the change parameter Θ (control point parameter) of the control points of this dynamic estimation model. Then, the control point parameter Θ can be obtained by minimizing the dissimilarity (Equation (2) below) of the SSD (sum of squared differences) between the PAR image based on the reference time point and the PAR image at an angle opposite to the angle of the reference time point.

[0071] [Equation 2]

[0072]

[0073] In the equation, D(Θ) is the dissimilarity based on SSD, and P target is the PAR image of the reference time point, and P source is the PAR image at an angle (time point) 180 degrees from the reference time point.

[0074] Since this convergence calculation for minimization is an ill-posed problem with multiple transformation parameters, a regularization term is introduced in order to solve it effectively and robustly. The regularization term is introduced as a term that penalizes the differences in the control point parameters. However, in this embodiment, as shown in Equations (3) and (4), the regularization term in the spatial domain (the first regularization term) and the regularization term in the temporal domain (the second regularization term) are introduced as independent regularization terms. Equation (3) is the regularization term R1(Θ) that penalizes the differences in the control point parameters that are spatially adjacent, and is a regularization term for suppressing excessive deformation within the image of interest. Equation (4) is the regularization term R2(Θ) that penalizes the differences in the control point parameters that are temporally adjacent, and is a term for suppressing excessive deformation from the images adjacent to the image of interest. Images that are temporally adjacent refer to the image of interest and the images before and after it (before and after in time, i.e., before and after in the body axis direction).

[0075] [Equation 3]

[0076]

[0077] [Equation 4]

[0078]

[0079] If these regularization terms are used, the cost function of the final control point parameter is represented by Equation (5) below.

[0080] [Equation 5]

[0081]

[0082] In the formula, λ1 and λ2 are the weights of two regularization terms R1(Θ) and R2(Θ), respectively.

[0083] By introducing such two regularization terms, the motion vector calculation unit 331 can stably solve the optimal solution that minimizes formula (2), thereby obtaining the motion vector Θ.

[0084] The concept of the dynamic estimation model into which the regularization term is introduced is shown in Figure 7 . As shown in the figure, when reconstructing a plurality of tomographic images, the (t - 1)-th image 701, the t-th image 702, and the (t + 1)-th image 703 at a prescribed image reconstruction interval in the body axis direction, when the point at the center of the t-th image 702 is set as the control point of interest, in this dynamic estimation model, the parameters of the control point of interest are determined by considering the information of the control points adjacent within the image 702 (spatially adjacent information) and the information of the control points of the images 701 and 703 corresponding to the control point of interest of the image 702 (temporally adjacent information).

[0085] Therefore, the continuity of the image depends on the interval of the control points within the image and the interval of the control points between adjacent images. And it also depends on the magnitudes of the weights of the above two regularization terms. The dynamic estimation model of the present embodiment is characterized in that the control points or the weights of the regularization terms are set to be adjustable according to image reconstruction conditions such as FOV or image reconstruction interval, thereby enabling optimization of the image continuity.

[0086] Specific examples of the adjustment will be described in the embodiments to be described later. For example, the adjustment of the control points changes the number of pixels or the position (coordinates) on the image that determines the interval of the control points according to the image reconstruction conditions (such as FOV). For the weights of the respective regularization terms, for example, default prescribed values are preset for standard image reconstruction conditions (FOV or image reconstruction interval), and when the FOV or image reconstruction interval set by the user is different from the standard FOV or image reconstruction interval, the weight λ1 of the regularization term R1(Θ) in the spatial domain or the weight λ2 of the regularization term R2(Θ) in the temporal domain is adjusted. The adjustment of the weights corresponding to the image reconstruction conditions is not only performed when the user sets, but can also be dynamically performed corresponding to the change of the image reconstruction conditions during shooting.

[0087] <Dynamic correction reconstruction process: S5>

[0088] The dynamic correction reconstruction unit 312 performs image reconstruction using the dynamic estimation result of the above dynamic model. As Figure 5As shown, in dynamic correction reconstruction, the MVF 520 calculated by the dynamic information acquisition unit 330 using the PAR image pair 510 estimates the size or direction of the subject's motion when acquiring each projection data for image reconstruction, and while correcting the image at the target reconstruction position using the estimation result, filter-corrected backprojection is performed to reconstruct the tomographic image 530.

[0089] As described above, in the X-ray CT apparatus according to the present embodiment, by introducing regularization terms in the spatial domain and time domain that can be independently adjusted as a dynamic estimation model when performing dynamic estimation processing in the arithmetic unit for dynamic correction reconstruction, the frequency of occurrence of unnatural deformation can be reduced, and in particular, a reconstructed image that maintains image continuity in the body axis direction can be provided.

[0090] Moreover, according to the present embodiment, regarding dynamic estimation, only a pair of PAR images is used for each section, so the amount of calculation is reduced, and the processing time or memory usage can be decreased.

[0091] Next, based on the above-described Embodiment 1, a specific embodiment of further adjusting the dynamic estimation model according to the image reconstruction conditions will be described. The adjustment of the dynamic estimation model includes adjustment of the formula for calculating the control parameter (displacement vector) between the above-described images. Specifically, there are adjustments of the weights of the two regularization terms (the regularization term in the spatial domain and the regularization term in the time domain) in the calculation formula, adjustment of the control points, adjustment of the calculation method, etc., and they are adjusted in a manner corresponding to the image reconstruction conditions.

[0092] <Embodiment 2>

[0093] This embodiment describes the adjustment of the dynamic estimation model corresponding to the FOV in the image reconstruction conditions. The processing flow is the same as the Figure 4 shown flow, but if the image reconstruction conditions are set in S2 of the Figure 4 flow, the dynamic estimation model adjustment unit 332 automatically adjusts the number of pixels between control points (the number of pixels between the control point of interest and the next control point) according to the set FOV.

[0094] Specifically, when the FOV (FOV adj ) set by the user is different from the reference FOV (FOV base ), the dynamic estimation model adjustment unit 332 uses the number of pixels between control points of FOV base (INT base ) as a reference to adjust the number of pixels between control points (INT adj ) so that the control point interval (mm) is equal to the control point interval (mm) of FOV base . That is, the number of pixels between control points calculated by the following formula is used as the set FOVadj Number of pixels between control points (INT adj ).

[0095] [Equation 6]

[0096]

[0097] As an example, taking the reference FOV base to be 300 mm, the case where the set FOV adj is 1 / 3 of 300 mm (100 mm) is shown in Figure 8 . In the case of the image with FOV base being 300 mm shown in the left figure, if the control point interval is 25 mm (INT base : 42 pixels), then in the image with FOV adj = 100 mm in the right figure, the number of pixels for the same control point interval is 3 times (126). By adjusting based on the above formula (6), it is possible to prevent excessive deformation caused by the control point interval on the actual image becoming too small. And, it is possible to perform the same dynamic estimation as in the reference FOV base without unnecessarily increasing the number of control points. Additionally, in the example of Figure 8 , it is set to cover the size of the subject range scanned in the reference FOV base , but the reference FOV base can also be set to different values for each imaging part. For example, the chest is set to FOV = 300 mm, the heart is set to FOV = 150 mm, etc. as reference values. In this case, it is possible that the subsequently set FOV adj is larger than the reference FOV base , but as long as the number of pixels between control points is slightly adjusted according to formula (6) to keep the control point interval (mm) the same.

[0098] According to this embodiment, even when the user sets a FOV different from the reference FOV, by adjusting the number of pixels between control points according to the FOV, it is possible to prevent the degree of deformation from changing depending on the FOV, and eliminate excessive deformation or insufficient deformation. That is, it is possible to improve the dynamic estimation accuracy and the effectiveness of dynamic correction reconstruction.

[0099] In addition, in the above description, the case where the image reconstruction conditions set by the user are different from the reference FOV set as the default condition for image reconstruction has been described. However, for example, when observing an image of both lungs with a relatively large FOV (300 mm), a concerning lesion such as a nodule is found. Therefore, it can also be applied to a case where the image reconstruction is changed by the user, such as magnifying the image to FOV 150 mm for magnified image reconstruction. In this case, as the image reconstruction conditions change, the content of the dynamic estimation process is dynamically changed, thereby maintaining the continuity with the previous dynamic estimation process and preventing excessive deformation and the like accompanying the condition change.

[0100] <Example of Modification 1 of Embodiment 2>

[0101] In the above Embodiment 2, the number of pixels between control points during dynamic estimation was adjusted according to the image reconstruction conditions (FOV) set by the user. However, in this modification, the adjustment of the image position of the fixed control points is performed regardless of the FOV. That is, even when the FOV is changed, the coordinates of the control points are controlled so that the position of the control points relative to the center of the image remains unchanged. In this example, the case where the FOV is changed to a smaller FOV during image reconstruction from the initially set FOV is described.

[0102] As Figure 9 shown, when the coordinates of the control point of interest at a certain point in time during the dynamic estimation process are set to (px bf , py bf ) and the reconstruction center coordinates are set to (cx bf , cy bf ), if the FOV (FOV bf ) at this point in time is changed to FOV Figure 9 as shown in the right figure of af , the dynamic estimation model adjustment unit 332 receives this change. Then, for the next reconstructed image, the coordinates of the control point of interest are determined according to the following formula.

[0103] [Equation 7]

[0104]

[0105] Through this process, even if the reconstructed FOV changes, the coordinates of each control point are always the same in the image coordinates, that is, they are fixed. Therefore, it is possible to prevent fluctuations in the dynamic estimation accuracy caused by changes in the FOV, and thus it is possible to achieve dynamic estimation with constant accuracy.

[0106] <Example of Modification 2 of Embodiment 2>

[0107] In the above-described Embodiment 2 and Modification 1, the number of pixels between control points or the coordinates of the control points in dynamic estimation were adjusted according to the FOV. However, instead of adjusting the control points, the weight λ1 of the regularization term R1(Θ) in the spatial domain in the aforementioned formula (5) for calculating the displacement vector may be changed (adjustment of the calculation formula for dynamic estimation). For example, when setting a value less than the reference FOV relative to the default λ set for the reference FOV 1base the weight λ1 is set such that λ1 > λ 1base . By increasing the weight of the regularization term in the spatial domain, it is possible to increase the penalty for excessive changes (dynamic estimation) in the spatial region caused by reducing the FOV, thereby improving the accuracy of dynamic estimation. And conversely, when setting a FOV greater than the reference FOV, the weight λ1 is set such that λ1 < λ 1base . In this case, it is also possible to prevent omission of fine dynamics due to an increase in the FOV and improve the accuracy of dynamic estimation.

[0108] In addition, the method of this Modification 2 can also be used in combination with the method of Embodiment 2, i.e., adjustment of the number of pixels between control points, or the method of Modification 1, i.e., fixing of the control point coordinates.

[0109] According to this modification, similar to Embodiment 2 or its Modification 1, it is possible to prevent excessive dynamic estimation or a decrease in the accuracy of dynamic estimation as the FOV changes, and improve the accuracy of dynamic estimation.

[0110] <Embodiment 3>

[0111] In this embodiment, the same dynamic estimation model as in Embodiments 1 and 2 is also used. However, in Embodiment 2 and its modifications, the control points of the dynamic estimation model or the weights of the regularization terms were adjusted as the FOV changed. In this embodiment, the method for calculating the control parameters between images is dynamically changed so as not to be affected by the change in the image reconstruction interval.

[0112] When reconstructing tomographic images at a prescribed image reconstruction interval in the body axis direction, the process of dynamic estimation is to determine the control point parameters of the image adjacent to the reference image from the reference image. Figure 7 The image reconstruction intervals of the images 701 to 703 shown are set as image reconstruction conditions. However, when the image reconstruction interval is changed, the control point intervals in the body axis direction become different.

[0113] In the present embodiment, the control parameter calculation method is adjusted so that even if the image reconstruction interval changes, it is the same as when performing dynamic estimation at a constant control point interval. Therefore, in the present embodiment, it is assumed that there are virtual tomographic images between adjacent images, and the dynamic estimation between the virtual tomographs is calculated. That is, the temporary control point parameters are calculated for the virtual tomographic images in sequence, and when the position of the virtual tomographic image reaches the position of the adjacent tomographic image, the temporary control point parameters calculated for the virtual tomographic image are used as the control point parameters of the adjacent tomographic image.

[0114] Hereinafter, with reference to Figure 10 and Figure 11 , the processing of the present embodiment will be specifically described.

[0115] Figure 10 FIG. is a diagram showing the processing of the present embodiment. As shown in the figure, in the present embodiment, after automatically setting the control point positions first (S21), a PAR image is generated for the virtual tomographic image (S22).

[0116] For the control points in the tomographic image, the automatic setting of the control point positions is performed by the method of Embodiment 2 or its modification example 1, and the control point interval (interval based on the number of pixels) or the coordinates of the control points are set according to the FOV set as the image reconstruction condition. That is, when the FOV is smaller than the preset FOV, the number of pixels between the control points is changed so that the distance (mm) between the control points in the real space is constant (the method of Embodiment 2). Or, control is also performed on the image after the FOV change so that the position of the control points relative to the image center is the same as before the FOV change (the method of Modification Example 1).

[0117] Regarding the body axis direction, as Figure 11 shown, it is assumed that the virtual tomographic image 7011 is located at a position of a predetermined interval d relative to the tomographic image 701 of interest. This interval d is a value smaller than the set image reconstruction interval D, and this interval d becomes the control point interval in the body axis direction in the present embodiment. For example, when the standard image reconstruction interval D is set to 5 mm, the interval d is a fraction (e.g., 1 / 10) thereof, and in order to perform a more detailed diagnosis, a value that can cope with a small image reconstruction interval such as 0.625 mm is set. This value d can be preset to a predetermined value, or can be set according to the ratio to the set image reconstruction interval. Also, it can be configured to be set by the user.

[0118] If the position of the virtual tomographic image 7011 is determined, the PAR image generation unit 311 reconstructs a PAR image according to the reconstruction center position of the virtual tomographic image from a pair of projection data obtained within a predetermined angular range.

[0119] Next, the dynamic information acquisition unit 330 uses the PAR image to calculate the control point parameters for the virtual tomographic image using the dynamic estimation model (Formula (5)), and acquires dynamic information (S23). The operation unit 30 sequentially calculates the control point parameters for the virtual tomographic images 7012, etc., while changing the position of the virtual tomographic image along the body axis direction until the PAR image generation (S22) and dynamic information acquisition (S23) reach the image reconstruction position of the adjacent tomographic image 702. When the position of the virtual tomographic image reaches the image reconstruction position (S24), the control point parameters calculated for the virtual tomographic image are used as the control point parameters of the adjacent image.

[0120] The dynamic correction reconstruction unit 312 reconstructs the adjacent tomographic image 702 using the MVF formed by the control point parameters.

[0121] Thereafter, the processes of S22 to S24 are performed in the same manner until the image reconstruction condition (image reconstruction interval) is changed.

[0122] According to this embodiment, virtual tomographic images are assumed at a constant interval between adjacent images, and temporary control point parameters are calculated sequentially, so that the same result as the result of calculating the control point parameters at a constant image reconstruction interval regardless of the image reconstruction condition can be obtained, and stable dynamic estimation can be performed. Furthermore, even when the image reconstruction interval is large, dynamic estimation that fully considers the continuity of the body axis direction can be performed.

[0123] According to the method of this embodiment, since a PAR image is generated for each virtual tomographic image, the amount of calculation increases. However, since only the PAR image is generated without reconstructing the virtual tomographic image itself, an excessive calculation load can be avoided.

[0124] In the present embodiment, the method of the first embodiment, that is, the method of adjusting the weight λ2 of the regularization term in the time domain of the dynamic estimation model (Formula (5)) according to the change in the image reconstruction interval, can also be used in combination.

[0125] In the above, the embodiment of the processing of the operation unit 30 (processor) of the X-ray CT apparatus of the present invention is mainly described, but as long as there is no technical contradiction, the processing described as each embodiment and modification example can also be appropriately combined, and such combination is also included in the present invention. In addition, the present invention is effectively applied to asynchronous imaging, but can also be applied to synchronous imaging in the same manner.

[0126] Furthermore, changes or additions to known structures included in the X-ray CT apparatus and the processor of the above-described embodiment are also included in the present invention within the scope not changing the gist of the present invention.

Claims

1. An X-ray CT apparatus, characterized in that, Comprising: An imaging unit having a scanner on which an X-ray source and an X-ray detector that rotate around a subject are mounted, and a moving mechanism that relatively moves the position of the scanner in the body axis direction of the subject with respect to the subject, and acquiring transmission X-ray data that is different in terms of the angle and the position in the body axis direction with respect to the subject; and a calculation unit that generates a tomographic image of the subject using the transmission X-ray data acquired by the imaging unit, The calculation unit comprising: A partial reconstruction image generation unit that generates a pair of partial reconstruction images at facing positions using the transmission X-ray data; a dynamic information acquisition unit that applies a dynamic estimation model to the pair of partial reconstruction images to acquire dynamic information of the subject during scanning; and a dynamic correction reconstruction unit that reconstructs a tomographic image using the dynamic information and the transmission X-ray data acquired within an angular range of 180 degrees or more, The dynamic estimation model includes a first regularization term that can be independently adjusted and maintains the spatial continuity of the image and a second regularization term that maintains the temporal continuity, The dynamic information acquisition unit automatically changes at least one of the dynamic estimation model and the calculation method of the dynamic estimation using the dynamic estimation model according to the image reconstruction conditions.

2. The X-ray CT apparatus according to claim 1, wherein The dynamic information acquisition unit includes a dynamic estimation model adjustment unit that adjusts the dynamic estimation model according to the image reconstruction conditions.

3. The X-ray CT apparatus according to claim 2, wherein The dynamic estimation model adjustment unit adjusts at least one of the weight of the first regularization term and the weight of the second regularization term according to the image reconstruction conditions.

4. The X-ray CT apparatus according to claim 3, wherein The dynamic estimation model adjustment unit adjusts the weight of the first regularization term according to the FOV set as the image reconstruction condition.

5. The X-ray CT apparatus according to claim 3, wherein The dynamic estimation model adjustment unit adjusts the weight of the second regularization term according to the image reconstruction interval set as the image reconstruction condition.

6. The X-ray CT apparatus according to claim 3, wherein The dynamic estimation model adjustment unit adjusts the weight of the regularization term according to the image reconstruction conditions and adjusts the control points of the dynamic estimation model.

7. The X-ray CT apparatus according to claim 2, wherein The dynamic estimation model adjustment unit adjusts the control points of the dynamic estimation model according to the FOV set as the image reconstruction condition.

8. The X-ray CT apparatus according to claim 7, wherein The dynamic estimation model adjustment unit adjusts the number of pixels between adjacent control points of the dynamic estimation model according to the FOV.

9. The X-ray CT apparatus according to claim 7, wherein The dynamic estimation model adjustment unit adjusts the coordinates of the control points according to the FOV so that the position of the control points relative to the image center in the image space is constant.

10. The X-ray CT apparatus according to claim 1, wherein: in the calculation of the control point parameters using the dynamic estimation model by the dynamic information acquisition unit, the calculation method is dynamically adjusted according to the image reconstruction interval set as an image reconstruction condition.

11. The X-ray CT apparatus according to claim 10, wherein: the dynamic information acquisition unit sets one or more virtual cross-sections at a constant interval between the cross-section of interest and the cross-section adjacent in the body axis direction of the subject, performs the calculation of the control point parameters using the partial reconstruction images in the virtual cross-sections until the virtual cross-section position becomes the position of the adjacent cross-section, and uses the control point parameters calculated in the virtual cross-section when the virtual cross-section position reaches the position of the adjacent cross-section as the control point parameters of the adjacent cross-section.

12. A processor for image processing that processes the transmitted X-ray data acquired by an X-ray CT apparatus, wherein the processor: generates a pair of partial reconstruction images at the facing positions using the transmitted X-ray data; applies a dynamic estimation model including a first regularization term that can be independently adjusted and maintains the spatial continuity of the image and a second regularization term that maintains the temporal continuity to the pair of partial reconstruction images to obtain the dynamic information of the subject during scanning; and reconstructs a tomographic image using the dynamic information and the transmitted X-ray data acquired within an angular range of 180 degrees or more.

13. The processor according to claim 12, wherein: in the acquisition of the dynamic information, at least one of the dynamic estimation model for acquiring the dynamic information and the calculation method using the dynamic estimation model is dynamically changed according to the image reconstruction conditions set during dynamic correction reconstruction.

Images