Device-less motion state estimation
By extracting and fitting motion functions in 4D-CT imaging technology and defining motion states directly from projected image data, the hardware dependence and error sensitivity problems in the prior art are solved, and more accurate and efficient image reconstruction is achieved.
Patent Information
- Application Number
- CN202380032710.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-06-29
- Filing Date
- 2023-03-28
- Publication Date
- 2024-11-15
AI Technical Summary
Existing 4D-CT imaging techniques require additional hardware and complex setups when measuring respiratory and cardiac motion signals, resulting in discomfort, costly, and error-sensitive, and the inability to directly determine motion signals for the region of interest.
By receiving projected image data, selecting continuous parts to form a plate image sequence, extracting motion functions, identifying trigger points and grouping projected data to define motion states, thereby performing image reconstruction without the need for additional hardware to measure motion signals, using cycle functions Fit the motion function to smooth and regularize the motion signal.
The 4D-CT image reconstruction is achieved without the need for additional hardware to measure respiratory and cardiac motion, reducing the complexity and cost of imaging workflows and improving the accuracy of motion signals and image quality.
Smart Images

Figure CN118974773A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates generally to medical image reconstruction. In particular, but not exclusively, the present invention relates to motion compensation in computed tomography image reconstruction. Background Art
[0002] In medical imaging, techniques such as computed tomography (CT) are used to visualize patient anatomy. A CT imaging system generally includes a radiation (e.g., x-ray radiation) source mounted on a rotatable gantry relative to a detector array having one or more rows of detector pixels. The x-ray source rotates around an examination region between an x-ray tube and the detector array, and emits radiation that passes through the examination region and a subject disposed in the examination region. The detector array detects radiation passing through the examination region, and generates projection data indicating the examination region and an object or subject disposed in the examination region. The projection data is reconstructed to generate volumetric image data indicating the subject. The volumetric image data can be processed to generate one or more images including a scanned portion of the subject.
[0003] 4D-CT imaging is commonly used in radiation oncology. In 4D-CT imaging, a series of 3D volume images of a subject are acquired over a period of time. This type of imaging can support visualization of the motion of the patient's anatomical structures. This is particularly helpful when the area of interest (such as a tumor) undergoes regular and involuntary motion (e.g., breathing and cardiac motion). Based on respiratory phase-based gating, the motion of lung tumors can be tracked during respiratory time.
[0004] For 4D-CT imaging, data acquisition takes several minutes. During this time, involuntary regular movements (e.g., breathing movements) will occur. The patient cannot hold his breath during this period, and thus the patient is allowed to breathe freely or is required to follow a breathing protocol. In order to obtain sufficient image quality, the data for the corresponding motion states are combined into groups for image reconstruction. These groups are also commonly referred to as "gates" or "bins". In order to define the bins, it is necessary to measure the motion signal during image acquisition.
[0005] The most common way to measure the respiratory signal throughout image acquisition is to use a strap mounted on the patient's chest or abdomen. The strap attaches a pressure sensor to the patient. Pressure changes are caused by abdominal movements, and these pressure changes are seen as representing the respiratory signal. However, such a strap is uncomfortable for the patient.
[0006] An alternative option to using a pressure strap is to place marker blocks on the patient's torso and monitor the movement of the marker blocks with a camera-based system. However, camera-based systems require an unobstructed line of sight to the markers, which is not always available in practice.
[0007] Furthermore, both methods for measuring respiratory motion have the disadvantage of requiring additional hardware, setup, calibration, and sterilization after use. This increases the complexity of the imaging workflow and makes it expensive to use and sensitive to errors. Summary of the invention
[0008] The current invention seeks to provide a method for 4D-CT image reconstruction that does not require a separate measurement of the motion signal.
[0009] Furthermore, a method and a system for reconstructing a computed tomography image of a subject, as well as an imaging arrangement for acquiring a computed tomography image of a subject and a computer program for performing the method are provided.
[0010] A method for reconstructing a computed tomography image of a subject includes receiving projection image data of the subject, selecting consecutive portions of the projection image data, and reconstructing each of the selected portions into a volumetric slab image to form a sequence of slab images. The method also includes: for each of the slab images in the sequence, extracting a motion function from the sequence of slab images by registering the slab image with at least one other slab image separated from the slab image in the sequence by a separation parameter to obtain at least one difference metric; selecting a sequence of difference metrics from the obtained difference metrics for the sequence of slab images; and forming the motion function using the selected sequence of difference metrics. The method also includes: identifying at least one trigger point in the motion function to define a motion state; grouping the projection image data into the defined motion states; and for each defined motion state, reconstructing the grouped projection data into a volumetric image of the subject. The method is preferably implemented by a computer or by other suitable computing devices.
[0011] In one embodiment of the method, extracting the motion function from the sequence of plate images, identifying at least one trigger point and grouping the projection image data into defined motion states is performed at least partially simultaneously with receiving the projection data. This has the advantage that image reconstruction can be started during image acquisition without having to wait for the entire CT scan to be completed.
[0012] Preferably, the volumetric plate image is reconstructed at a lower resolution than the volumetric image used for the motion state. The lower image quality can be sufficient for extracting the motion function and allows for faster and more efficient reconstruction of the plate image.
[0013] According to one aspect, the motion function is formed by fitting a periodic function to a selected sequence of differential metrics. Fitting a periodic function to a sequence of differential metrics results in a smooth and regular motion function. For example, the cos^4(x) function provides a good approximation of human respiratory motion.
[0014] According to another optional aspect, forming the motion function includes analysis and correction for outliers. For example, when the subject is coughing, this may cause a substantial interruption in the motion function. Such outliers can be identified by analyzing a sequence of selected differential metrics or by analyzing the motion function initially calculated. Optionally, corresponding groups of image projection data can be removed to correct the motion function and improve the quality of the reconstructed image.
[0015] In an optional or additional option, the separation parameter has at least one preset value. Using one preset value allows for a fast and robust extraction of the motion function. When one preset value is used for the separation parameter, a single sequence of differential measures is obtained for the sequence of plates. This sequence is automatically selected as the sequence of differential measures. In another option, the separation parameter has two or more preset values. When two or more preset values are used, the best value for the separation value can be used for each plate image to select the sequence of differential measures. Optionally, when two or more preset values are used, a majority vote can be used to select the sequence of differential measures.
[0016] Registering the plate images may include applying one or more of the following: elastic deformation and rigid deformation. Preferably, a rigid deformation such as translation and / or rotation is applied. Applying a rigid deformation may have sufficient quality, require less calculation, and be more robust.
[0017] In one embodiment, the differential metric indicates the amount and / or direction of movement of the region of interest. This has the advantage of providing improved image quality for the motion state of the actual region of interest itself.
[0018] Preferably, the projection image data of the subject is received from one or more of: a computed tomography imaging apparatus, an imaging data storage device, and a database such as a dedicated imaging data storage database, a hospital database, or a patient record database.
[0019] A system for reconstructing a computed tomography image of a subject includes an input configured to receive projection image data of the subject, a plate image reconstructor, a motion extraction unit, and a further reconstruction processor. The plate image reconstructor is configured to select consecutive portions of the projection image data and reconstruct each of the selected portions into a volumetric plate image to form a sequence of plate images. The motion function extraction unit is configured to: for each of the plate images in the sequence, register the plate image with at least one other plate image separated from the plate image in the sequence by a separation parameter to obtain at least one differential metric; select a sequence of differential metrics from the differential metrics obtained for the sequence of plate images; and form a motion function using the selected sequence of differential metrics. The reconstruction processor is configured to: identify at least one trigger point in the motion function to define a motion state; group the projection data into the defined motion states; and reconstruct the grouped projection data into an image of the volume of the subject for each defined motion state. Preferably, the system also includes a display for displaying at least one of the volumetric images. Displaying the image can assist a physician in diagnosing a patient and formulating a treatment plan.
[0020] In one embodiment, the system further comprises a user interface configured to display to a user differential metrics calculated for at least one separation parameter, and preferably also configured to allow the user to select at least one of: a sequence of differential metrics and a value for the separation parameter.
[0021] The apparatus for acquiring a computed tomography image of a subject comprises a computed tomography imaging device for acquiring projection image data of the subject and the above-mentioned system for reconstructing a computed tomography image of the subject.
[0022] The computer program product comprises instructions for causing a processor to perform the above method when the computer program is executed.
[0023] The advantage of the present invention is that 4D-CT images are obtained without the need to obtain regular motion signals from the subject. Specifically, it is no longer necessary to use straps and other instruments to measure respiratory and / or cardiac motion. Such equipment may increase the complexity of the procedure and cause discomfort to the patient.
[0024] Another advantage is that the motion signal can be determined more accurately. The current invention can determine the motion directly for the anatomical region of interest. When using an external device such as a strapping device, the respiratory phase detected at the axial position being scanned may be different from the axial position where the strapping is located. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] In the following figures:
[0026] Figure 1 Schematically and exemplarily, an arrangement for acquiring a CT image of a subject is shown, the arrangement comprising a system for reconstructing a CT image of a subject.
[0027] Figure 2 Projection data acquisition and reconstruction of a volumetric slab image are schematically shown.
[0028] Figure 3 An example of a method for reconstructing a computed tomography image of a subject is schematically shown.
[0029] Figure 4 An example of method steps for extracting a motion function from a sequence of slab images is schematically shown.
[0030] Figure 5 A further example of method steps for extracting a motion function from a sequence of plate images is schematically shown.
[0031] Figure 6 An example of a user interface is schematically shown, the user interface being configured for allowing a user to select at least one value for a separation parameter. DETAILED DESCRIPTION
[0032] In the following examples, computed tomography images are acquired for a subject. Such a subject may be a human being, in particular a human patient requiring imaging for medical purposes. However, alternative subjects are also envisioned, for example, animals such as pets or livestock. The method according to the invention may also be applied to inanimate objects with intrinsic periodic motion that cannot be paused during image acquisition.
[0033] Figure 1 An imaging arrangement 100 for acquiring a computed tomography image of a subject is shown.
[0034] The imaging arrangement 100 includes an imaging device 110, such as a computed tomography (CT) scanner. The computed tomography imaging device 110 includes a fixed gantry 102 and a rotating gantry 104 rotatably supported by the fixed gantry 102. When the system is in operation, the rotating gantry 104 rotates around an examination region 106 about a longitudinal axis or z-axis. A radiation source 108, such as an x-ray tube, is supported by the rotating gantry 104 and rotates with the rotating gantry 104 and emits radiation that passes through the examination region 106. A source collimator 109 collimates the emitted radiation to form a generally fan-shaped, wedge-shaped, or cone-shaped radiation that passes through the examination region 106.
[0035] A support 118, such as a table, is provided to support an object or subject to be imaged in the examination region 106. The support 118 is movable along the z-axis to cooperate with the rotation of the rotating gantry 104 to facilitate a desired scanning trajectory, preferably a helical scanning trajectory.
[0036] The radiation sensitive detector array 112 detects radiation emitted by the radiation source 108 that traverses the examination region 106 and generates projection image data 120 indicative of the detected radiation. The illustrated radiation sensitive detector array 112 includes one or more rows of radiation sensitive photosensor pixels along the z-axis.
[0037] The imaging arrangement 100 also includes an image reconstruction system 130 that reconstructs the projection data and generates volumetric image data indicating an examination region 106 including structures of an object or subject disposed therein. One or more images may be generated from the volumetric image data. In this example, the imaging arrangement 100 also includes an optional display 140 for displaying one or more generated images.
[0038] The system 130 for reconstructing CT images comprises an input 132 for receiving projection image data 120 of a subject, a slab image reconstructor 134 , a motion extraction unit 136 and a further reconstruction processor 138 .
[0039] The input 132 is configured to receive the projection image data 120 directly from the CT imaging device 110 during or after the acquisition of the data. Alternatively, the projection data 120 may be stored by an imaging data storage device or database (e.g., a dedicated imaging data storage database, a hospital database, or a patient record database). In this option, the reconstruction system 130 may operate independently of the imaging device 110, and the input is configured to receive data from the data storage device or database.
[0040] The slab image reconstructor 134 is configured to select consecutive portions of the projection image data 120 and to reconstruct each of the selected portions into a volumetric slab image, thereby forming a sequence of slab images. The motion function extraction unit 136 is configured to: for each of the slab images in the sequence, register the slab image with at least one other slab image separated from the slab image in the sequence by a separation parameter to obtain at least one difference metric; select a sequence of difference metrics from the difference metrics obtained for the sequence of slab images; and form a motion function using the selected sequence of difference metrics.
[0041] The reconstruction processor 138 is configured to: identify at least one trigger point in the extracted motion function to define a motion state; group the projection data into the defined motion states; and reconstruct the grouped projection data into a volumetric image of the subject for each defined motion state. For respiratory motion, the trigger points may be, for example, an end expiratory state, an end inspiratory state, and a mid-expiratory state and a mid-inspiratory state. For 4D-CT images, preferably at least one trigger point and more preferably at least two trigger points are selected to show the motion of the region of interest in the patient at different motion stages.
[0042] Figure 2 The projection data acquisition and volumetric image reconstruction of a CT image are schematically shown. Data is acquired when a subject moves on a support through an imaging device. A commonly used way of acquiring data is by means of a spiral scanning trajectory. In a spiral scanning trajectory, the support moves the subject through an examination area in a continuous manner while the radiation source and the detector array rotate around the subject. Alternatively, the movement of the table can be stopped at each scanning position and then the radiation source and the detector array rotate around the subject in a fixed position. However, this step-by-step shooting method is more cumbersome to perform and is also less comfortable for the subject. In Figure 2 In FIG. 1 , the position z of the examination table is shown on the horizontal axis as a distance relative to the end of the support. The time t is shown on the vertical axis and increases in the upward direction.
[0043] Projection data is collected in a portion of the source and detector that rotates around the subject's angle. When the most prominent subject motion that needs to be compensated is respiratory motion, such a portion is preferably at least 180°. The advantage of a portion of data having at least 180° is that each portion can be selected and reconstructed into an image of sufficient quality to be displayed to the user. However, a portion with such a degree of rotation is not necessary for current reconstruction methods. When a higher timing resolution is required, for example when the most prominent subject motion that needs to be compensated is cardiac motion, the portion may also be less than 180°. Portions of projection data may be adjacent in acquisition angle, but may also partially overlap. Such portions of projection data represent data acquired for a subject at a specific examination table position and at a moment in the acquisition time. Optionally, portions of projection data may be defined using a time window within the acquisition time span or a time window covering the average position of the support.
[0044] Figure 2 A grid of volumetric image data 210 is shown divided into slices 220. As shown in FIG. Figure 1 As explained, the imaging radiation sensitive detector array has one or more rows of photosensitive pixels along the z-axis. When the detector rotates around the subject, each pixel collects projection data, and when a predefined rotation preferably greater than 180° is completed, the data can be reconstructed into a volumetric image of a 2D cross section of the subject. The image is called a slice 220. Adjacent pixels of the array each provide adjacent slices. The field of view 230 of the detector is mainly determined by the number of pixels spanning the length of the inspection area. The total coverage of the subject in the CT image 240 is increased by moving the support and collecting data at subsequent positions of the support. Preferably, multiple data slices are collected for each position of the subject support. This data collection method is called oversampled image acquisition. Collecting more slices at each bed position improves image quality, but also requires a slower support speed for the subject and increases the acquisition time. Therefore, in practice, a trade-off is made between the amount of projection data acquired and the scanning speed.
[0045] From the multiple portions of the acquired projection image data, each portion can be selected and reconstructed into a separate volumetric image. The 3D CT image is a volumetric plate image 250. The plate image is a volumetric image of the subject for a selected examination table position and a moment in the acquisition time. Figure 2 In FIG. 2 , the slab images are shown as rows of slices 220. The slabs are shown here as being adjacent in time, but alternatively, the slabs can also partially overlap, or a time interval can be left between the slabs.
[0046] The volumetric plate images form a temporal sequence 260 of 3D images. The temporal sequence begins with a first plate image S1, followed by a second plate image S2, and so on, until a final plate image S N .exist Figure 2 The total number of plate images N shown in the schematic diagram of FIG. 1 is 16, but in practice it is usually more. The total number of plate images is determined by the detector width of the CT imaging device, imaging parameters such as the rotation speed of the detector, the translation speed of the examination table, and the required field of view 240.
[0047] Continuous Plate S i and S i+D Separated in the sequence by the separation parameter D. Figure 2 In the example shown, the separation parameter D has a value of 4, but the separation parameter D can have any integer value starting from 1. However, in practice, there is an upper limit to the value that can be used for the separation parameter. i and S i+D , a spatial overlap in the transport direction z is required. This constraint determines an upper bound for D. Preferably, the region of interest for extracting the motion function should be at least partially present in both images. When all plate images overlap sufficiently, an upper bound of N-1 is possible for D in principle. However, in practice, the maximum value for D that will produce a motion function of sufficient quality for 4D-CT image reconstruction will depend on the frequency of the subject's motion. This will be substantially smaller than N, for example lower than N / 10, or even lower than N / 20.
[0048] Figure 3 An example of a method 300 for reconstructing a computed tomography image of a subject is schematically shown. The method for reconstructing a CT image begins with receiving 310 projection image data of a subject. For example, the data may be received from a computed tomography imaging device, an imaging data storage device, or a database, such as a dedicated imaging data storage database, a hospital database, or a patient record database. Next, consecutive portions of the projection image data are selected 320. In a next step, each of the selected portions of the image projection data is reconstructed 330 into a volumetric plate image to form a sequence of plate images.
[0049] The volume plate images can be reconstructed at the same resolution as the volume images for the motion state provided by the method, but this is not necessary. The sequence of plate images is used to extract the motion function. For this purpose, it is not necessary to have diagnostic image or display quality. It is even preferred to reconstruct the plate images at a lower resolution, as this allows faster and more efficient calculations.
[0050] When a sequence of plate images has been formed, a motion function is extracted 340. To extract the motion function, for each of the plate images in the sequence, the plate image is registered with at least one other plate image in the sequence. The other plate images are separated from the plate image in the sequence by a separation parameter. Each of the plate images in the sequence is registered 342 with the at least one other plate image to obtain at least one differential metric. The differential metric can characterize and / or quantify differences in the anatomical structure of the subject between the two plate images.
[0051] Plate images can be registered by applying elastic deformation and / or rigid deformation. Optionally, additional techniques such as automatic masking or intensity shearing can be used additionally during registration to improve robustness. Elastic deformation can be applied by calculating a deformation vector field. Subsequently, a differential metric can be determined by analyzing the vector field. Rigid deformation can be a translation and / or rotation. Preferably, a rigid deformation is applied. For respiratory motion, it is even more preferred to apply a rigid registration, which is optimized for translation in the anterior-posterior direction. This has the advantage of being more robust and easier to calculate. The differential metric that can be obtained from the rigid deformation is, for example, the amount of rotation, the angle of rotation, the amount of translation, the direction of translation, or a combination thereof. In the case of these examples, the differential metric can be a scalar indicating the amount of change, or a vector indicating both the amount of change and the direction of change. For extracting the motion function, the number of differential metrics can be sufficient to represent the amount of movement of the region of interest between the plate images. Having additional information such as the direction of movement may also be of interest to clinicians, and thus additional information can be provided to support diagnosis or treatment planning.
[0052] From the difference metrics obtained for the sequence of plate images, a sequence of difference metrics is selected 344. Subsequently, a motion function is formed 346 using the selected sequence of difference metrics.
[0053] The method also includes identifying 350 at least one trigger point in the motion function to define the motion state and grouping 360 the projection image data into the defined motion state. During the grouping process, the projection data can be assigned to one or more motion states. In one option, the projection data can be grouped by classifying the data into the motion state. In the case of classification, the projection data is assigned to the most suitable or closest motion state. In the optional option, the projection data can be assigned to one or more of the defined motion states by applying a smooth weighting function. In this method, a weighting function is calculated based on the defined motion state, and the weighting function is applied to the projection data for matching the data with the motion state. The advantage of using this smooth weighting function is that the amount of data used for image reconstruction is minimized while maintaining full coverage and timing resolution to the subject. Moreover, the use of a smooth weighting function allows a smooth transition between motion states, thereby avoiding artifacts.
[0054] Furthermore, for each of the defined motion states, the grouped projection data are reconstructed 370 into a volumetric image of the subject. The volumetric images of these defined motion states together form a 4D-CT image.
[0055] According to the present invention and Figure 3 The CT image reconstruction method 300 shown in the example of can be used to reconstruct projection image data after all data has been acquired. However, it is not necessary to first complete image acquisition before the method begins. Specifically, the steps of extracting 340 a motion function from a sequence of plate images, identifying 350 at least one trigger point, and grouping 360 the projection image data into defined motion states can be performed at least partially simultaneously with receiving 310 projection image data. When enough projection data has been acquired and received to form a first portion of the data, that portion may have been reconstructed into a volumetric plate image while the next consecutive portion of data is received, and so on. Extracting 340 a motion function can begin when at least the first two plate images of the sequence are available. Also, the motion function can be formed while the sequence of differential metrics is being constructed. Once an initial portion of the motion function is available, identification of trigger points and grouping of projection data into motion states can begin. Extraction of the motion function and identification of trigger points and grouping of projection data continue while the sequence of plate images continues to be formed until all data has been received and these processing steps have been completed.
[0056] Figure 4 and Figure 5 Both schematically show examples of method steps for extracting a motion function from a sequence of plate images. Figure 4 and Figure 5The embodiments of the present invention show a combination of several optional additional method steps. Although these steps are shown in combination, this is not necessary and only represents an advantageous option. The optional steps can also be implemented individually or in combination of two or more.
[0057] Figure 4 An embodiment for extracting 440 a motion function from a sequence of plate images is shown. In this embodiment, a separation parameter has a preset value. Having a preset value for D enables a sequence of difference metrics to be computed quickly and robustly from a sequence of plate images.
[0058] In this example, the differential metric is iteratively calculated according to the sequence of plates. 1+D The first image S1 is registered 441 to begin with. Based on the registration output, a difference metric is calculated 442. In this embodiment, a rigid registration is applied, and the difference metric is the shift of the anatomical structure of interest. For example, a motion signal can be extracted for respiratory motion. When the diaphragm moves upward, the difference metric will be a positive number indicating the magnitude of the movement, and when the diaphragm moves downward, the difference metric will be a negative number indicating the magnitude of the movement. In this way, both the amount of movement and the direction of movement can be captured.
[0059] In the current section S i After calculating the differential metric, the method continues 443 to the next block S i+1 , and returns to registering it 441 with the plate separated from it by D in the sequence, computing 442 the differential measure, and so on. When the final plate has been registered, a single sequence of numbers has been computed, and this is the one that is automatically selected 444 as the sequence of differential measures.
[0060] Next, a motion function is formed using the sequence of difference measures. In this example, this option is used to perform the analysis and correct the motion function for outliers.
[0061] Based on the sequence of differential metrics, an initial motion function is formed 445. The initial motion function can be the same as the sequence of differential metrics. Optionally, some mathematical operations can be applied to the sequence of differential metrics to form the initial motion function, such as subtracting a mean, inserting additional data points, integrating the sequence and / or applying a smoothing filter. Outliers in patient motion may be caused by sudden large-amplitude irregular movements (e.g., coughing). The initial motion function is analyzed to identify 446 the presence of any such deviations. If outliers are identified, the corresponding differential metrics are removed from the sequence to correct 447 these outliers. Although not shown here, it is also advantageous to remove portions of the projection data of the corresponding plate images from the final CT image reconstruction. This will improve image quality.
[0062] Using the sequence of corrected differential metrics, a motion function is formed 448. The motion function can be formed in a similar manner as described above for the initial motion function. In addition, a periodic function can also be fitted to the sequence of differential metrics to form the motion function. Fitting such a function to the corrected sequence provides a smoother and more robust motion function. Alternatively, outliers can be removed directly from the initial motion function to form the motion function.
[0063] Figure 5 Another embodiment for extracting 540 a motion function from a sequence of plate images is shown. In this embodiment, the separation parameter has two or more preset values. An advantage of using multiple values for D is that the robustness of the calculation is improved. It also has the advantage that the extraction of the motion function can be adjusted when artifacts are present in one of the plate images, or when the subject's motion pattern changes during image acquisition.
[0064] First, set 541 for the value of D. Then, in this example, Figure 4 In the same way as described in the example shown, the differential metric is iteratively calculated according to the sequence of plates. i+D With the current image S i The images are registered 542, the differential metric is calculated 543, and the next plate is selected 544. These steps are repeated until the final plate image is reached. This iterative process produces 545 a sequence of differential metrics for set values of D. When the sequence of differential metrics is completed, the next value for the separation parameter is selected 546, and the process is repeated. This process is repeated again until a sequence of differential metrics for each value of D is obtained.
[0065] In an optional option, the iteration loop can be reversed. For each value of D, each tile image Si+D With the current image S iThe registration is performed 542 , the differential metrics are calculated 543 , and the next plate is selected 544 , and the process is repeated until the final plate is reached.
[0066] Next, a sequence of differential metrics is selected 547 to be used to form 548 the motion function. One or more values for D may be selected by the user, for example using Figure 6 The interface shown, but it can also be selected automatically.
[0067] Alternatively, a majority vote of several values for the separation parameter may be used to select a sequence of differential metrics. For example, five consecutive values for D may be used, such as 1, 2, 3, 4, and 5. Each value of D will produce its own sequence of differential metrics. These sequences will be similar enough in shape to allow them to be superimposed on a selected sequence. The superposition will also include appropriate offsets for the sequences to match. For example, this may be the first sequence, where D=1, or the middle sequence, where D=3. When all superimposed values are close for a plate image in the sequence, all superimposed values will be averaged to provide the selected value for the differential metric for that position in the sequence. If one of the plate images is corrupted or has artifacts, this may result in outliers or missing data. Although outliers will exist in all sequences, they will occur at different locations in each sequence. Sequences with values for D with outliers will be ignored, and the other values will be averaged to provide the selected value for the differential metric. In this way, forming the motion function also includes automatic analysis and correction for outliers.
[0068] When the sequence of differential measures has been established 547, a motion function is formed 548 from the sequence. In this example, the motion function is formed by fitting a periodic function to the sequence of differential measures, in particular the function cos^4(x), where x corresponds to the sequence position number. The function is suitable for human breathing patterns.
[0069] Figure 6 An example of a user interface is schematically shown, the user interface being configured to allow a user to select at least one value for a separation parameter. The user interface is configured to display a differential metric calculated for at least one value of D. In this example, the differential metric is displayed as a differential metric plot 610. The horizontal axis of the plot shows the number of plates. In this example, 180 plate images are selected and reconstructed. The vertical axis of the plot shows values for D, which increase from 1 at the top to a value of 10 at the bottom. The differential metric plot 610 displays the value of the differential metric in grayscale. More advantageously than grayscale, color coding can be used for the differential metric plot 610.
[0070] The differential metric chart 610 is superimposed with a separation parameter selector 620. In this example, the separation parameter selector is a straight horizontal line that allows the user to select a single value by moving the selector up or down. Alternatively, the user may be allowed to draw a slanted line or curve to select multiple values for D. When the separation parameter selector 620 is positioned to the user's satisfaction, a confirmation button 630 is used to select a sequence of differential metrics accordingly.
[0071] In another option, selector 620 may have two straight horizontal lines or a rectangular box to select a smaller sub-range of values for D. This range may then be used to apply a Figure 5 A majority voting scheme is described to compose a sequence of difference metrics.
[0072] In addition to the differential metric graph 610, the user interface 600 also displays a motion function graph 640. The motion function graph visualizes a sequence 650 of selected differential metrics and two possible motion functions 660 and 670 as a function of the number of plate images in the sequence. In an advantageous option, the sequence of differential metrics is displayed in real time as the parameter selector 620 moves across the differential metric graph. This can help the user select the best sequence of metrics. A confirmation button can be used to confirm the selection for forming the motion function. Figure 6 In the example of , two options for forming a motion function are shown in addition to the sequence of metrics 650. In the first option 660, as shown in solid lines, the motion function is formed by integrating the sequence of differential metrics. In the second option 670, as shown in dashed lines, the motion function is formed by integrating the sequence of differential metrics and subtracting the mean. Both options allow for easy identification of trigger points corresponding to the breathing phase of the subject.
[0073] Any method steps disclosed herein may be recorded in the form of a computer program, which includes instructions that cause the processor to perform these method steps when executed on a processor. Instructions may be stored on a computer program product. The computer program product may be provided by dedicated hardware and hardware capable of executing software in association with appropriate software. When provided by a processor, these functions may be provided by a single dedicated processor, a single shared processor, or multiple separate processors, some of which may be shared. In addition, embodiments of the present invention may take the form of a computer program product accessible from a computer-available or computer-readable storage medium, which provides program code used by or in conjunction with a computer or any instruction execution system. For the purposes of this description, a computer-available or computer-readable storage medium may be any device that may include, store, communicate, propagate, or transmit a program to be used by or in conjunction with an instruction execution system, device, or apparatus. The medium may be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, or device or equipment, or a propagation medium. Examples of computer-readable media include semiconductor or solid-state memory, magnetic tape, removable computer disk, random access memory "RAM", read-only memory "ROM", rigid disk, and optical disk. Current examples of optical disks include compact disk - read only memory "CD-ROM", compact disk - read / write "CD-R / W", Blu-Ray TM and DVD. Examples of communication media are the Internet or other wired or wireless telecommunication systems.
[0074] By studying the drawings, the specification and the claims, those skilled in the art can understand and implement other variations to the disclosed embodiments in the process of practicing the claimed invention.It should be noted that the various embodiments can be combined to achieve further advantageous effects.
[0075] In the claims, the word "comprising" does not exclude other elements or steps, and the singular article does not exclude a plurality.
[0076] A single unit or device may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.
[0077] Any reference signs in the claims should not be construed as limiting the scope.
Claims
1. A method (300) for reconstructing a computed tomography image of a subject, the method comprising: - receiving (310) projection image data of the subject; - selecting (320) consecutive portions of said projection image data; - reconstructing (330) each of the selected portions into a volumetric plate image to form a sequence of plate images; - extracting (340, 440, 540) a motion function from the sequence of plate images by: - for each of the patch images in the sequence, registering (342, 442, 543) the patch image with at least one other patch image separated from the patch image in the sequence by a separation parameter (D) to obtain at least one difference metric; - selecting (344, 444, 547) a sequence of difference measures from the difference measures obtained for the sequence of plate images; as well as - forming (346, 448, 548) a motion function using the selected sequence of said difference metrics; The method further comprises: - identifying (350) at least one trigger point in the motion function to define a motion state; - grouping (360) said projection image data into said defined motion states; and - For each of the defined motion states, reconstructing (370) the grouped projection data into a volumetric image of the subject.
2. The method according to claim 1, characterized in that Extracting (340, 440, 540) the motion function from the sequence of slab images, identifying (350) the at least one trigger point, and grouping (360) the projection image data into the defined motion states is performed at least in part simultaneously with receiving (320) the projection data.
3. The method according to claim 1 or 2, characterized in that: The volume slab image is reconstructed at a lower resolution than the volume image for the motion state.
4. The method according to any one of claims 1 to 3, characterized in that The motion function is formed by fitting a periodic function to the selected sequence of the difference metrics.
5. The method according to any one of claims 1 to 3, characterized in that: Forming (445, 446, 447, 448, 547, 548) the motion function includes analyzing and correcting for outliers.
6. The method according to any one of claims 1 to 5, characterized in that The separation parameter (D) has at least one preset value.
7. The method according to claim 6, characterized in that The separation parameter (D) has two or more preset values, and a majority vote is used to select (547) the sequence of difference metrics.
8. The method according to any one of claims 1 to 7, characterized in that Registering the plate images includes applying one or more of: an elastic deformation and a rigid deformation.
9. The method according to any one of claims 1 to 8, characterized in that The differential metric indicates an amount and / or a direction of movement of the region of interest.
10. The method according to any one of claims 1 to 9, characterized in that The projection image data of the subject is received from one or more of: a computed tomography imaging device, an imaging data storage device, and a database such as a dedicated imaging data storage database, a hospital database, or a patient record database.
11. A system (130) for reconstructing a computed tomography image of a subject, the system (130) comprising: - an input (132) configured to receive projection image data (120) of the subject; a slab image reconstructor (134), the slab image reconstructor (134) being configured to: - selecting consecutive portions of said projection image data; - reconstructing each of the selected portions into a volumetric plate image to form a sequence of plate images; - a motion function extraction unit (136), the motion function extraction unit (136) being configured to: - for each of said patch images in said sequence, registering said patch image with at least one other patch image separated from said patch image in said sequence by a separation parameter (D) to obtain at least one difference metric; - selecting a sequence of difference measures from said difference measures obtained for said sequence of plate images; as well as - forming a motion function using the selected sequence of said difference metrics; The system further comprises: - a reconstruction processor (136) configured to: - identifying at least one trigger point in the motion function to define a motion state; - grouping said projection data into said defined motion states; as well as - for each of the defined motion states, reconstructing the grouped projection data into a volumetric image of the subject.
12. The system according to claim 11, characterized in that The system also includes a user interface (600) configured to display differential metrics (610) calculated for at least one separation parameter (D) and to allow a user to select (620, 630) at least one of: a sequence of differential metrics and at least one value for the separation parameter (D).
13. The system according to claim 11 or 12, characterized in that The system also includes a display (140) for displaying at least one volumetric image of the subject.
14. An imaging device (100) for acquiring a computed tomography image of a subject, comprising: -Computed tomography imaging equipment (110); - A system (130) according to any one of claims 11-13, wherein the system (130) is used to reconstruct the computed tomography image of the subject.
15. A computer program product comprising instructions for causing a processor to perform the method according to any one of claims 1 to 10 when the computer program is executed.
