X-ray CT Image Reconstruction Method

Abstract

A method for image reconstruction of CT projection data aims to reduce motion artifacts in reconstructed images caused by anatomical object motion. An embodiment of the method reduces motion artifacts by limiting the range of data used to reconstruct each slice. More specifically, a subrange of projection data corresponding to each slice is selected. This subrange is selected based on determining one or more subwindows of visibility of the anatomical object or event of interest within the projection data sequence. An event may be a specific phase of the motion cycle of the anatomical object. A structure may be a specific portion of the anatomical object of interest. Both techniques result in the reduction of motion artifacts within a single slice by limiting the data range and focusing on the most clinically relevant data.

Description

Technical Field

[0001] The present invention relates to a method for processing data as part of an X-ray CT image reconstruction process.

Background Art

[0002] Conventional computed tomography (CT) scanners include an X-ray generator mounted on a rotating gantry facing one or more X-ray detectors. The X-ray generator rotates around an examination area between the X-ray generator and the one or more detectors and emits (typically polychromatic) radiation that traverses the examination area and the subject and / or object placed within the examination area. The one or more detectors detect the radiation traversing the examination area and generate signals indicative of the examination area and the subject and / or object placed therein. For each angular position of the X-ray beam with respect to the examination area, data is collected from an array of detectors. As the scan progresses, the platform may axially move the scan target within the examination area. Data collected from the detectors at a series of angular positions and axial displacements is referred to as projection data. Projection data refers to raw detector data and, optionally, can be used to form a projection sinogram. A projection sinogram is a visual representation of the projection data acquired by the detectors.

[0003] Typically, a reconstructor is further used to process the projection data and reconstruct a three-dimensional image of the subject or object. The three-dimensional image is composed of a plurality of cross-sectional image slices generated from the projection data by, for example, applying a filtered back projection algorithm in a tomographic reconstruction process. The reconstructed image data is, in effect, the inverse Radon transform of the raw projection data.

[0004] Even when the subject appears to be stationary, various organs are moving naturally. These organs include the heart and lungs, as well as various organs that are indirectly moved due to pressure from the lungs or heart.

[0005] Visceral movement during CT image acquisition can lead to blurring and motion artifacts in tomographic reconstruction. Therefore, it is beneficial to limit the length of the time window in which projection data for a given image reconstruction is acquired. State-of-the-art CT scanners can image entire organs (e.g., the heart) in a short time. [Overview of the Initiative] [Problems that the invention aims to solve]

[0006] For example, it is possible to image the entire heart during a portion of the heartbeat. Such image acquisition is usually performed with minimal movement during the resting phase of the heart. However, some tomographic slices, typically the slices at the beginning and end of acquisition, may be degraded by motion. [Means for solving the problem]

[0007] The present invention is defined by the claims.

[0008] According to an example relating to one aspect of the present invention, a processing device is provided for use in X-ray computed tomography (CT) reconstruction. The processing device is adapted to receive input projection data comprising a plurality of data subsets, each subset corresponding to a different slice of the body being scanned. Each data subset comprises data corresponding to a defined range of acquisition angles and a defined acquisition period. The processing device is further adapted to identify one or more time or angle sub-windows in the input projection data that, when reconstructed, are associated with the visibility of a desired anatomical feature or event in that sub-window. The processing device is further adapted to select a time or angle sub-range of data smaller than the full range of each slice from the data subset of each slice, the selection of the sub-range being based on at least one of the one or more identified time or angle sub-windows in the projection data. The processing device is further adapted to generate a data output showing the selected sub-range for each slice.

[0009] Embodiments of the present invention are based on processing a received projection dataset to select a smaller sub-range of data from the data of each slice of the body being scanned. The sub-range is selected based on information about one or more windows of visibility of an anatomical feature or event, and, optionally, also based on a sub-range of data selected in a spatially adjacent slice. An anatomical event may be, for example, a specific motion phase of an organ or structure that undergoes periodic motion. By selecting only a sub-range of projection data from each slice, the period covered by the reconstructed image is shortened, and motion artifacts are reduced. By selecting a sub-range based on visibility windows, it is ensured that the selected data is the most clinically relevant. In the context of this specification, a visibility window can be described as, in other words, a portion of the CT projection dataset (measured with respect to time and acquisition angle) that contains the most information related to a given anatomical or physiological event (compared to the rest of the CT projection data not included in the window). The relevant anatomical feature or event may span the entire sub-window (temporarily or spatially), in other words, the entire sub-window may contain or be occupied by data representing an anatomical feature or event. In some cases, though not always, a large portion of the partial window may be occupied by data representing anatomical features or events.

[0010] A partial range is selected from the complete projection data range of each slice. In this context, the complete data range of a particular slice means all recorded projection data in the input projection dataset, including projection rays passing through the voxels within that particular slice.

[0011] As will be discussed later, various methods exist for determining one or more visibility subwindows within a projection dataset. A processing unit may include a window selection module adapted to identify one or more visibility windows by applying one or more algorithms. Visibility windows may be identified based on the use of external signals related to anatomical events, or based on data analysis procedures that analyze projection data to determine portions of the data containing anatomical features or events.

[0012] Data subsets from multiple different slices may be received as a single, continuous dataset or data stream (e.g., in the case of helical imaging) or as separate data packages (e.g., in the case of individual slice scans). These data subsets may overlap, for example, in the case of helical imaging. A process may exist to extract or identify the data subset corresponding to each slice from the complete input projection dataset. This may be performed by a processing unit or an external unit.

[0013] Identifying one or more temporal or angular subwindows may involve identifying each temporal or angular subwindow for each data subset (i.e., for each slice). In this case, optionally, each temporal or angular subwindow may relate to the visibility of a desired anatomical feature or event in the dataset, if reconstructed. However, this is not mandatory; for example, if a data subset does not strictly overlap with a visibility subwindow, it may represent the visibility of an area outside the subwindow.

[0014] Selecting a data subrange for each slice may be based on maximizing the temporal alignment between the subrange and at least one visibility window. For example, it may be based on maximizing the proximity between the center time of the subrange and the center time of the identified window.

[0015] Selecting a data subrange for each slice may be based on maximizing the temporal overlap between the subrange and at least one visibility window. Alternatively, it may be based on minimizing the lack of overlap.

[0016] There are various methods for determining one or more windows of visibility for an anatomical event or object.

[0017] According to one or more embodiments, the processing device may be configured to receive a representation of an output signal generated by an external sensor adapted to sense one or more physical parameters related to an anatomical event. The received signal corresponds to the sensor output during the acquisition of projection data, i.e., during a CT scan. This may be received in real time as the data is acquired, or it may be a recorded representation of the sensor output recorded over the period of data acquisition. One or more partial windows associated with the visibility of the anatomical event may be determined based on the received signal.

[0018] For example, the input signal may be a signal from an ECG sensing configuration, acquired simultaneously with the acquisition of projection data. The ECG sensor signal makes it possible to determine the cardiac phase during projection data acquisition. Based at least in part on this, one or more visibility windows for a particular cardiac phase over the data acquisition period can be determined.

[0019] Optionally, according to one or more embodiments, the processing unit may be adapted to perform preliminary image reconstruction on one or more data subsets. Based on the preliminary reconstruction, the processing unit may be adapted to determine a time or angular partial window of the input projection dataset in which anatomical features or events are visible.

[0020] An identified subwindow is a subwindow within the complete projection dataset that spans one or more of the slices that have been pre-reconstructed. The objective is to enable the identification of a specific set or range of slices that spatially correspond to (i.e., contain) an anatomical feature or event. This allows, optionally, in subsequent processing steps, to temporally align selected data subranges of the set of slices identified as spanning an anatomical feature or event.

[0021] The processing device may be adapted to subsequently apply image processing to a preliminary reconstruction in order to identify spatial boundaries, edges, or contours of anatomical features, or to detect characteristic (e.g., graphical) identifiers of anatomical events.

[0022] According to one or more embodiments, a selected angular subrange from each slice may have a predetermined angular width / range, and selecting a data subrange from a given data subset includes determining the temporal alignment of the subrange from (or to) the data subset.

[0023] For example, all selected subranges may have an angular width of x degrees, where x is a number smaller than the total angular range of each data subset (e.g., 180 degrees). The predetermined width of a subrange may be, for example, 150 degrees out of 180 degrees. The selection process may simply involve choosing where to align the subranges angularly or temporally within the full data range of each subset. For example, they may be temporally aligned with one of the identified visibility windows of an anatomical object or event. As an example, a subrange may be temporally aligned with the ECG measurement time point corresponding to the anatomical event in question.

[0024] According to one or more embodiments, determining the temporal alignment of a data subrange can be based on signals generated by an external sensor.

[0025] According to a further group of embodiments, selecting a data partial range for one or more of the data subsets includes determining an angular width of the data partial range based on one of one or more determined time or angular partial windows.

[0026] This can be determined, for example, based on the time width or angular width of at least one partial window, for example, a partial window that temporally overlaps the data subset or is temporally closest to the subset. This can be determined based on the duration of the visibility period of an anatomical feature or event within the data.

[0027] When a visibility partial window is determined based on an input signal from an external sensor, the angular width can be determined based on the time width of the partial window.

[0028] When the partial window is determined based on a preliminary reconstruction, the angular width can be determined based on the spatial width or boundary of the partial window.

[0029] According to one or more embodiments, the input projection data can represent an anatomical object of interest, the object of interest undergoes periodic motion, and the anatomical event corresponds to a phase of the periodic motion. By way of example, the object of interest is the heart or a part thereof, and the anatomical event is a phase of the cardiac cycle.

[0030] According to one or more groups of embodiments, selecting a data partial range for each slice can be performed at least partially based on the data partial ranges selected for at least a portion of the other slices. By way of example, this can be done with the goal of achieving a relative temporal overlap or proximity of the partial ranges selected for each slice.

[0031] In some examples, this can be done in an iterative process.

[0032] In some embodiments, the data subrange of each slice may be determined partially on maximizing the temporal overlap or temporal proximity of the angular subranges of different slices. It may also be based on minimizing the lack of overlap. This helps to minimize motion artifacts between slices.

[0033] For example, if input projection data is acquired by helical scanning, subsets of the projection data will overlap. Therefore, some degree of temporal and angular overlap can be achieved within a selected data sub-range of each slice. The overlapping region or range may be selected to coincide with the visibility region or range of an anatomical object or event, i.e., the visibility window, or it may be selected to be temporally aligned with a specific anatomical event (e.g., a specific point in a kinetic cycle).

[0034] According to the advantageous embodiments, the processing unit is Perform a first image reconstruction for each slice using the complete data range of the data subset of each slice, and derive the first image for each slice. Perform a second image reconstruction for each slice using an extracted subrange of the data subset of each slice, and derive the second image for each slice. Apply low-pass filtering to each first image, apply high-pass filtering to each second image, and generate a combined image for each slice that includes the filtered combination of the first and second images. It can be adapted to generate an output showing the combined image of each slice, and to perform the following actions.

[0035] Images reconstructed from partial data have improved temporal resolution but contain limiting angle artifacts. Conversely, images reconstructed from complete data have lower temporal resolution but do not contain limiting angle artifacts. Limiting angle artifacts typically have significantly lower frequencies (i.e., a larger spatial range) than motion artifacts. This allows both types of images to be combined using a frequency splitting method, which involves adding the low-frequency portion of the full-data reconstructed image to the high-frequency portion of the partial-data reconstructed image. In this way, artifact types in each image are substantially filtered out, and the filtered image components are compensated for by adding the other filtered image data. Thus, temporal resolution is improved and limiting angle artifacts are significantly reduced.

[0036] According to one or more embodiments, the processing device may be adapted to perform the step of extracting a selected data subrange for each slice from a data subset of the slices. The processing device may be further adapted to generate a reconstructed image of each slice by performing image reconstruction of each slice using only the extracted data subranges of the data subset of each slice. A data output representing the resulting image may be generated.

[0037] An example relating to a further aspect of the present invention provides a computer-implemented method used for X-ray computed tomography (CT) reconstruction, the method being Receiving input projection data containing multiple data subsets, each data subset corresponding to a different slice of the body being scanned, and each data subset containing data corresponding to the acquisition angle range and acquisition period, If reconstructed, this involves identifying one or more time or angular subwindows within the input projection data associated with the visibility of the desired anatomical feature or event in the input projection data, The selection of a time or angular subrange of data from a subset of data for each slice, which is smaller than the full range of each slice, and the selection of the subrange is based on at least one of one or more identified time or angular subwindows within the projection data. This includes generating data output that shows the selected subrange for each slice.

[0038] An example relating to a further aspect of the present invention is a computer program product including computer program code, which is executable on a processor or computer. When the processor or computer is operably coupled to a source of input projection data, such as a CT imaging device or a data store, the code is configured to cause the processor to execute a method outlined above or according to any example or embodiment described below, or a method according to any claim of this application.

[0039] The above and other aspects of the present invention will be described and made clear with reference to the embodiments described below. [Brief explanation of the drawing]

[0040] For a better understanding of the present invention, and to more clearly illustrate how the present invention can be implemented, please refer to the following illustrative schematic diagrams. [Figure 1] Figure 1 shows an exemplary processing apparatus according to one or more embodiments, operably coupled with a CT imaging device. [Figure 2] Figure 2 shows processing steps performed by one or more processing devices according to one embodiment. [Figure 3] Figure 3 schematically shows (a portion of) an exemplary CT projection dataset, which is shown with respect to acquisition time and acquisition angle. [Figure 4]Figure 4 schematically shows several data subsets of the complete projection data sequence, with respect to acquisition time and acquisition angle. [Figure 5] Figure 5 shows the identification of a visibility subwindow for a given anatomical event based on input signals related to the anatomical event, and further shows the selection of a subrange of one exemplary slice data subset based on the identified visibility subwindow. [Figure 6] Figure 6 shows an exemplary selection of data subranges for multiple data subsets based on a common visibility subwindow. [Figure 7] Figure 7 shows an exemplary selection of data subranges for each slice, which is at least partially based on achieving temporal overlap between the selected data subranges for different slices. [Modes for carrying out the invention]

[0041] The present invention will be described with reference to the drawings.

[0042] While the detailed descriptions and specific examples illustrate exemplary embodiments of the apparatus, systems, and methods, it should be understood that these are for illustrative purposes only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, systems, and methods of the invention will be better understood from the following description, the appended claims, and the accompanying drawings. It should be understood that the drawings are for illustrative purposes only and are not drawn to scale. Also, it should be understood that the same reference numerals are used throughout the drawings to point to the same or similar parts.

[0043] The present invention provides a method for reconstructing CT projection data, aimed at reducing motion artifacts in reconstructed images caused by the movement of anatomical objects. Embodiments of the present invention are based on reducing motion artifacts by limiting the range of data used to reconstruct each slice. More specifically, a sub-range of projection data corresponding to each slice is selected. This sub-range is selected based on determining one or more sub-windows of visibility of the anatomical object or event of interest within the projection data sequence. The event may be a particular phase of the motion cycle of an anatomical object. The structure may be a particular part of the anatomical object of interest. Both methods result in a reduction of motion artifacts within a single slice by limiting the data range and focusing on the most clinically relevant data.

[0044] As a non-limiting example, the selected sub-range of projection data for each slice may be chosen to be as close as possible in time to a specific cardiac phase, such as quiescence, an anatomical event. Additionally, or alternatively, the sub-range of projection data selected for one or more particular slices may be chosen to be as close as possible in time to adjacent slices to improve local consistency. Additionally, or alternatively, in some examples, the selection of data sub-ranges may be influenced by the anatomical structures depicted in the slices. For example, multiple slices with a common anatomical structure (such as coronary arteries) may be reconstructed from temporally overlapping, joined, or adjacent projection data. In contrast, for another slice that does not contain that particular anatomical structure, a temporal discontinuity between the projection data sub-range of this slice and the data sub-range containing that anatomical structure may be acceptable.

[0045] Figure 1 schematically shows an example of an X-ray CT imaging system 100. This system includes a CT imaging device 150, such as a computed tomography (CT) scanner, and further includes processing elements and control components 140, 132, and 134 operably coupled to the imaging device.

[0046] The imaging apparatus 150 generally includes a fixed gantry 102 and a rotating gantry 104. The rotating gantry 104 is rotatably supported by the fixed gantry 102 and rotates around the inspection area about the longitudinal axis or z axis.

[0047] The patient support 120, such as a bed, supports an object or subject, such as a human patient, within the examination area. The support 120 is configured to move the object in an axial direction (defined by the z-axis) within the examination area in order to scan different slices of the object.

[0048] A radiation source 108, such as an X-ray tube, is rotatably supported by a rotating gantry 104. The radiation source 108 rotates with the rotating gantry 104, emitting radiation across the inspection area 106.

[0049] The radiation-sensitive detector array 110 defines an angular arc that spans the inspection area 106 and faces the radiation source 108. The detector array 110 includes one or more rows of detectors that extend along the z-axis and detect radiation traversing the inspection area 106, generating projection data indicating the radiation. Because the detector array includes rows of detectors along the z-axis, this means that for each angular position of the gantry, multiple X-ray measurements are acquired at a series of positions along the z-axis. This makes it possible to reconstruct multiple different slices along the z-axis using projection data from a single rotation or a portion of a rotation.

[0050] The axial distance of the subject covered by the scan is called the scan range. The width of the X-ray beam projected by the radiation source to pass through the subject is called beam collimation. The cross-sectional area of ​​a given imaging slice of the body is called the field of view (FOV).

[0051] The imaging device 150 is operably coupled to a processing unit 140 which includes one or more processing unit elements. The processing unit is configured to receive projection data acquired by the imaging device 150 and is adapted to perform one or more processing operations on the projection data.

[0052] In some examples, the processing unit may be further adapted to perform control functions, in which case the processing unit is adapted to transmit control commands to the imaging device 150 and to control one or more operating parameters of the imaging device. For example, system 100 may include one or more user input devices 132, such as a mouse or keyboard, to enable the input of user control commands. The system may further include a user output device 134, such as a display monitor, to convey system status information to the user or to display reconstructed images. User input / output devices may be operably coupled to the processing unit. However, in another example, a separate control console may be provided to control the operation of the imaging device 150, in which case the processing unit 140 shown in Figure 1 is solely for processing projection data.

[0053] The processing unit 140 includes a data selection module 142, which is adapted to perform the function of selecting a sub-range of projection data corresponding to each slice of the body being scanned. This function will be described in detail later. Optionally, the processing unit may further include a reconstruction module 144 for reconstructing stereoscopic image data based on the selected sub-ranges of data extracted for each slice of the body being scanned. Alternatively, the reconstruction module may be provided by a separate processing unit that operably communicates with the processing unit. The processing unit may be configured to receive the selected sub-ranges of projection data for each slice and to reconstruct image data for each slice based on the extracted sub-ranges of data. In either case, the reconstruction module 144 may employ filtered back projection (FBP) reconstruction, noise reduction reconstruction algorithms (e.g., iterative reconstruction) (image domain and / or projection domain), and / or other algorithms. It should be understood that the reconstruction module 144 may be implemented by a microprocessor that executes computer-readable instructions encoded or embedded on a computer-readable storage medium such as physical memory and other non-temporary media. Additionally, or alternatively, a microprocessor can execute computer-readable instructions carried by carrier waves, signals, and other transient (or non-transient) media.

[0054] The imaging device 150 can perform scans according to any CT imaging protocol. An example of a scan protocol is a helical or spiral scan. In a spiral CT scan, the radiation source 108 and detector 110 move along a spiral path relative to the object being scanned. A typical implementation involves moving the patient bed 120 within the scanner's bore 106 while the gantry 104 rotates. Compared to acquiring individual slices, using spiral CT can improve resolution for a given radiation dose. Most modern hospitals now use spiral CT scanners. The helical CT beam trajectory is characterized by its pitch, which is equal to the table 120 feed distance along the scan range (z-axis) over one rotation of the gantry divided by section collimation. A pitch greater than 1 means that the axial travel distance over a full rotation is greater than the beam width at the center of rotation.

[0055] When the pitch is high, the radiation dose for a given axial field of view (FOV) (i.e., a given slice) is reduced compared to conventional CT (essentially, each slice is reconstructed using projection data acquired over a reduced angular range). However, at high pitches, there is a trade-off between noise and longitudinal resolution.

[0056] To minimize radiation dose and improve temporal resolution, the pitch for one imaging mode is 2 (or a similar value), meaning that at the center of rotation, the projection data used to reconstruct each slice corresponds to data acquired over a rotation range of only 180°. Note that the relationship between pitch and the angular range available for reconstruction also depends on the position of a given voxel within a given slice. For voxels located at the center of rotation, pitch 2 corresponds to a data angular range of 180°, while for voxels far from the center of rotation, pitch 2 corresponds to data of less than 180°.

[0057] When attempting to image specific moving anatomical objects, blurring and motion artifacts can occur in tomographic reconstruction. Depending on the scan mode used, the temporal resolution of a particular image slice (or FOV) is limited by the time required to complete the rotational angle range used to acquire data for each slice. For example, in helical imaging, the temporal resolution of a particular slice is limited by the rotation speed of the gantry. As an example, when a single-source high-pitch mode is used, the temporal resolution of a given slice may be limited to the temporal range of data acquired over half a rotation. Embodiments of the present invention aim to improve temporal resolution based on selecting only a partial range of projection data corresponding to each slice (or FOV). This limits the time interval required to acquire the projection data used for a single slice. This limits the magnitude of motion artifacts within a single slice.

[0058] One aspect of the present invention provides only the processing unit 140. The processing unit is operably connected to the CT imaging assembly 150 while in use to receive projection data and includes inputs / outputs that optionally output control commands to the CT imaging assembly. In a further aspect, a CT imaging system including both the processing unit 140 and the imaging assembly 150 may be provided.

[0059] The following description relates to the characteristics of the processing apparatus. The processing steps performed by the processing apparatus also provide a method that is implemented by a computer, and this computer-implemented method is a further aspect of the present invention.

[0060] Figure 2 shows an overview of exemplary processing steps performed by one or more embodiments of the processing apparatus 140.

[0061] The processing unit 140 is adapted to acquire input projection data containing multiple data subsets 202. Each subset corresponds to a different slice through the body being scanned. Each data subset contains data corresponding to a specific acquisition angle range and a specific acquisition period. The data subsets may overlap in time within the complete projection dataset. In other words, data subsets of axially adjacent slices may share a common portion of projection data. This is the case, for example, if the projection data was acquired in a spiral or helical scan mode with a pitch greater than 1. However, in other cases, such as when the projection data was acquired in individual slice acquisition mode, the projection data subsets may not overlap with each other. The projection data may be received in real time from the CT scanner 150 or from a data store that records previously acquired data. The projection data may be received in the form of a continuous data stream or projection sequence, which contains projection data spanning the total scan range (total scan interval) (along the z-axis) of the object being scanned and includes multiple angular rotations. The processing unit may be adapted to perform the step of separating (or tagging or labeling) subsets of received projection data corresponding to each different slice. Alternatively, this may be performed outside the processing unit, and by the time the data arrives at the processing unit, the separation or labeling of multiple different data subsets may have already been performed.

[0062] The processing unit 140 is further adapted to identify one or more time or angular subwindows in the input projection data associated with the visibility of a desired anatomical feature or event in the input projection data 204 (if reconstructed). Various methods exist to do this, and windows of object visibility in the data may be detected based on input from an external sensor reflecting characteristics related to the anatomical event (or a recording of the external sensor output over a period corresponding to the acquisition of projection data), or based on an analysis of the projection data itself. In some examples, each slice may have its own visibility subwindow. Each slice may have its own unique subwindow, or some subwindows may be used for multiple slices (e.g., axially and temporally adjacent slices). This may occur, for example, when subsets of data from different slices partially or completely overlap. A detailed explanation of this functionality is provided below.

[0063] The processing unit 140 is further adapted to select a time or angular subrange of data (smaller than the full range of each slice) from a subset of data for each slice, and the selection of the subrange is based on at least one of one or more identified time or angular subwindows in the projection data. For example, it may be based on maximizing the temporal overlap or temporal proximity with one of the identified subwindows (e.g., the temporally closest window).

[0064] The processing unit 140 is further adapted to generate a data output indicating a selected subrange for each slice. For example, the processing unit may have inputs / outputs, and the data output is transmitted to a reconstruction module via the inputs / outputs. In some examples, the processing unit may include a reconstruction module, and the processing unit is adapted to perform a further step of generating reconstructed image data based on the selected data subrange for each slice, for example using a slice-by-slice filtered back projection. This may be compiled, for example, to form reconstructed stereoscopic image data.

[0065] For illustrative purposes and to facilitate the following explanation, Figure 3 shows T=t final A schematic representation of the complete input projection data sequence 302, spanning the total scan interval of -t0 and including multiple full angular rotations of the radiation source and detector, is shown. The figure illustrates a scan sequence containing five full rotations. However, in practice, a scan sequence may contain more than five rotations or fewer than five rotations. For example, in some cases, very high-pitch scans using wide detectors may use fewer than one rotation to image the entirety of a particular organ (e.g., the heart).

[0066] Figure 4 schematically shows examples of multiple projection data subsets 306a to 306e. Each data subset corresponds to a different slice of the body being scanned. Each data subset contains a portion of the projection data from the complete projection data sequence 302. In particular, each dataset contains data corresponding to a specific acquisition angle (rotational position of the radiation source and detector relative to the body) range or span and a specific acquisition time range relative to the body being scanned. In this example, the data subsets overlap temporally and angularly. For example, the projection data sequence 302 may be a data sequence acquired using a spiral imaging protocol with a pitch greater than 1. However, the overlap of data subsets is not an essential feature, and the technical effects of the present invention do not depend on it.

[0067] There are various options for performing the determination of one or more partial windows of visibility for an anatomical object or event.

[0068] According to at least one set of embodiments, the determination of at least some of the partial windows of one or more partial windows of visibility may be identified based on external sensor inputs related to anatomical features or events.

[0069] For example, the processing unit 140 may be configured to receive a representation of an output signal generated by an external sensor adapted to sense one or more physical parameters related to an anatomical event. The received signal corresponds to the sensor output during the acquisition of projection data, i.e., during a CT scan. This may be received in real time as the data is acquired, or it may be a recorded representation of the sensor output recorded over the period of data acquisition. One or more subwindows associated with the visibility of the anatomical event may be determined based on the input signal. For example, the processing unit may be adapted to determine a time window in which the anatomical event occurs based on the received signal.

[0070] For example, the input signal may be a signal from an ECG sensing configuration, acquired simultaneously with the acquisition of projection data. The ECG sensor signal makes it possible to determine the cardiac phase during projection data acquisition. Based on this, a visibility window for a specific cardiac phase can be determined for each slice.

[0071] ECG sensors are just one example. Other examples include the use of sensor outputs from heart rate monitors, which can also detect cardiac phases. Another example may involve the use of outputs from ultrasound sensors or ultrasound monitoring devices, which may have broader applications for sensing phases of movement of various possible anatomical objects. For example, lung movement, or the movement of organs with exercise cycles resulting from pressure application by lung movement (e.g., the liver), may be tracked. Yet another example may involve outputs from PPG sensors, which can be used to track the respiratory cycle of the lungs based on changes in blood oxygen or volume pulse wave.

[0072] The processing unit can use the input indicating the sensor output to determine the time window in which a given anatomical event (e.g., a specific cardiac phase) occurs.

[0073] In combination with, or instead of, the use of external sensor input, in further embodiments, the processing unit may be adapted to determine one or more partial windows in the received projection data in which a particular anatomical object of interest is visible. This may include the processing unit 140 performing a preliminary image reconstruction on one or more data subsets (of different slices) and, based on the preliminary reconstruction, determining a time or angular partial window of the input projection data in which an anatomical feature or event is visible. This partial window may span multiple different slices (i.e., multiple different data subsets may be included within its range). The processing unit may perform a first-pass full image reconstruction in which a preliminary reconstruction of the complete projection data sequence is obtained. This may include, for example, generating a reconstruction for each individual slice. The preliminary reconstruction may, for example, use the complete data range of each slice.

[0074] Image processing or analysis may be applied to the reconstructed image data obtained from the preliminary reconstruction. This may include, as a non-limiting example, the application of any one or more of the following: shape matching algorithms, edge detection algorithms, and image segmentation processes (e.g., model-based segmentation). These may be used to identify the boundaries or contours of anatomical objects of interest within the reconstructed image data. This allows for the definition of a region of interest within the data, e.g., a 3D region of interest. Based on this, a time window or angular window of the acquisition sequence in which the object of identification exists may be identified. This time window or angular window may then be used as at least one of the visibility sub-windows used to select the data sub-range for each slice.

[0075] Image processing algorithms used to identify objects of interest within the data may be stored in the processing unit's local memory or retrieved by the processing unit through communication with a remote data store, for example, using inputs / outputs or a communication module (e.g., a wireless communication module).

[0076] When one or more partial windows of visibility for anatomical objects or events within a projection dataset are identified, a partial range of the projection data subset for each slice is selected. This selection is based on at least one of the one or more identified visibility windows.

[0077] The method for determining the subrange differs from embodiment to embodiment. In at least one set of embodiments, the temporal (and angular) width or range of the subrange selected from each data subset is fixed, and determining the subrange involves only determining the position or alignment of the subrange within the full data range for a given slice data subset. For example, if the full projection data subset of each slice covers 180°, each selected data subrange may have a fixed angular width of 150°. Naturally, if the full projection sequence 302 is acquired at a constant table feed rate, this also corresponds to a constant temporal width of the data subrange. Determining the temporal alignment of the selected subrange within each slice data subset may be performed based on knowledge of the temporal range covered by one or more visibility subwindows of an anatomical object or event. The subrange alignment may be selected to maximize the temporal overlap between the selected data subrange and at least one visibility window, or, if overlap is not possible, to maximize the temporal proximity to at least one visibility window.

[0078] An example is schematically shown in Figure 5. This figure shows an example of a projection data subset 306 corresponding to data from a single slice of the object being imaged. Figure 5(a) shows the data subset 306. Figure 5(b) shows an exemplary sensor signal 308 generated by an external sensor adapted to sense one or more physical parameters related to an anatomical event. For example, the sensor signal may be an ECG sensor signal. Figure 5(b) shows an exemplary partial window 312 corresponding to a specific anatomical event, e.g., one phase of the cardiac cycle. The partial window has a time span Δt in the total projection sequence. The time span covers a specific time portion of the data subset 306 and a specific angle acquisition range Δθ. Figure 5(c) shows the selection of a specific partial range 314 of the data subset 306, selected based on an identified visibility time partial window 312 of the anatomical event of interest. In this example, the partial range is selected to overlap with the visibility partial window 312, and the partial window is centered within the time span of the partial range 314.

[0079] For example, a selected data subrange 314 for a particular slice may have a fixed time width. The temporal alignment of the data subrange 314 within the full width of the data subset 306 is selected such that the time center point of the subrange 314 is closest to the time center point of the subwindow 312.

[0080] In some cases, a data subset of a given slice may not overlap with any of the anatomical events or one or more identified visibility sub-windows of an object. For example, a data subset of a given slice may correspond to data of an anatomical event recorded most recently, such as data spanning 180° before and after the cardiac phase of the subject. In this example, the sub-range of the data subset may be selected so as to be smaller than the entire range of the data subset, and to have the temporal alignment within the data subset that maximizes the temporal proximity to the nearest visibility window. For example, for a slice recorded over 180° chronologically preceding the anatomical event of the subject, the last 150° of acquisition may be selected as the data sub-range to be used for reconstruction.

[0081] Figure 6 shows a further example. This figure shows an exemplary signal 308 received from an external sensor measuring physiological parameters related to an anatomical event of interest. Based on the signal, an exemplary sub-window 312 may be identified, and the sub-window is associated with the visibility within the corresponding time portion of the projection data sequence of the anatomical event. The anatomical event may be, for example, a phase of the motion cycle of an object such as the heart. Figure 6 also shows an example of multiple projection data subsets 306, each containing data corresponding to a different slice of the body. In this example, the data subsets overlap, but this is not mandatory. Figure 6 shows an example of a selected data sub-range 314 for each of the data subsets. Each sub-range is selected to maximize the overlap with the identified visibility window 312.

[0082] In the example in Figure 6, all selected subranges have the same width, but this is not mandatory. In some embodiments, the width of the selected data subrange can be adjusted independently for each slice data subset. This can be done, for example, based on the width of an identified visibility subwindow. For example, each subrange may have a width selected based on maximizing its temporal overlap with the visibility window, provided that the width is kept lower than a predetermined maximum value.

[0083] According to certain embodiments, selecting a data subrange for each slice may be performed in part on the data subranges selected for at least a portion of the other slices. For example, selecting a data subrange for each slice may be performed in part on maximizing the temporal overlap or alignment of the subranges selected for each slice. This ensures spatial consistency between slices. This could mean, for example, for each slice, performing a first substep to determine whether a strict temporal overlap between the data subranges of that slice and adjacent slices is possible, and if so, performing a second substep to select the data subrange of the given slice so that it temporally overlaps with the data subrange of an adjacent slice. If not, an alternative substep is performed to identify a subrange that maximizes the overlap with the subranges of adjacent slices. An example of this technique is schematically shown in Figure 7 and will be described in more detail in the following paragraphs.

[0084] As described above, determining one or more visibility subwindows for anatomical objects in projection data may involve performing preliminary image reconstruction on one or more data subsets (of one or more slices). This may be preferable to perform on all slices. Based on the preliminary reconstruction, one or more visibility subwindows in the input projection dataset in which a given anatomical feature or event is visible may be identified.

[0085] This exemplary technique is shown in Figure 7, which shows a preliminary image reconstruction 402 generated based on a data subset 306a–306f of a set of six slices. Image processing is applied to the initial reconstruction 402 to identify the target 3D anatomical region of interest (ROI) 404. This may be performed, for example, based on image segmentation, such as model-based segmentation. As an example, the region of interest 404 in this example corresponds to the aortic valve annulus.

[0086] Next, a time-part window of the input projection data sequence can be identified, and its boundary is aligned with the boundary of the identified region of interest 404.

[0087] The data sub-range 314 of each data subset 306 is selected based on the initial reconstruction 402 and on the attempt to maximize the temporal overlap between selected sub-ranges of slice data subset 306 (in this case, subsets 306b to 306e) that are included in (or overlap with) the region of interest sub-window 404.

[0088] In some examples, the width (and temporal alignment) of each selected data sub-range 314 may be constructed based on the identified visibility sub-window.

[0089] In some examples, the data subrange 314 selected for each slice may be chosen based on the fact that the data subrange is temporally closest to the data subrange selected for adjacent slices within the region of interest. This can be done by an iterative or recursive process over the entire set of data subsets. This has the effect of minimizing motion artifacts between adjacent slices during reconstruction.

[0090] In addition to positioning the data subset range 314, the width of the data subset range may also be adjusted based on the identified anatomical visibility subset window. In some examples, the width of each data subset can be adjusted to match adjacent slices. This helps minimize motion artifacts between slices.

[0091] Additionally, or alternatively, the width of one or more subranges 314 of the data subset 306 may be determined based on minimizing overlap with subwindows of data containing specific artifacts identified in the first-pass complete data reconstruction. The width may further be influenced by the appearance of the identified motion artifacts, or one or more derived properties related to the determined severity / magnitude of the motion artifacts.

[0092] For each slice containing data subsets 306b–306e located within the identified region of interest 404, the selected data sub-range may be chosen to maximize the temporal overlap or temporal proximity between adjacent slices. For data subsets of slices 306a and 306f located outside the region of interest 404, non-temporal overlap is acceptable, as artifacts within these (anatomically less relevant) regions are not of significant importance.

[0093] According to one or more embodiments, the time width of the data subrange 314 selected for each data subset 306 may be configured based on the relative time proximity of the data subset to the nearest sub-window, or the degree of time overlap of the data subset with the nearest sub-window. Maximum and minimum time widths or angular widths of the subrange may be set, and the widths may be selected in proportion to the degree of overlap. For example, a data subset that completely overlaps a sub-window may use the maximum subrange width, a subset that does not overlap at all may use the minimum width, and a subset that partially overlaps may use a width equal to the extent of overlap, as long as it is higher than the minimum and lower than the maximum.

[0094] As described above, according to one or more embodiments, after selecting a data subrange for each data subset, image reconstruction may be applied to reconstruct each slice based on the selected data subrange for each slice.

[0095] In some cases, artifacts that may be caused by the reduced data range used for each slice can be compensated for using compensation techniques. Compensation techniques may include neural network-based artifact compensation methods or the application of compensation methods based on the use of advanced reconstruction techniques such as iterative reconstruction. In short, iterative reconstruction is a more advanced reconstruction method than standard filtered backprojection. Iterative reconstruction is based on explicitly modeling the physics of projection and using this to iteratively reconstruct images, forward-project them, relate the forward-projected data to the actual raw data, and then continue with the next iteration of reconstruction. For example, an example of one method is outlined in the paper, Motion estimation and correction in cardiac CT angiography images using convolutional neural networks. Comput Med Imaging Graph. 2019 9 / 76:101640 by Lossau Nee Elss T et al.

[0096] Additionally, or alternatively, one or more post-processing steps can be applied to the reconstructed image data to reduce noise or artifacts. For example, filtering can be applied to the reconstructed image data.

[0097] For example, according to one advantageous set of embodiments, the processing device may be adapted to perform a filtering procedure as part of an image reconstruction process.

[0098] In particular, the processing device may be adapted to perform a first image reconstruction of each slice using the full data range of the data subset of each slice, thereby deriving a first image of each slice.

[0099] The processing unit may be further adapted to perform a second image reconstruction of each slice using an identified subrange of the data subset of each slice, thereby deriving a second image of each slice.

[0100] The processing unit may be further adapted to apply low-pass filtering to each of the first images and high-pass filtering to each of the second images. A combined image containing the filtered combinations of the first and second images is then generated for each slice. Subsequently, an output showing the combined image for each slice is generated.

[0101] This procedure is based on the frequency splitting technique. Images reconstructed from partial data have improved temporal resolution but suffer from limited-angle artifacts. Conversely, images reconstructed from complete data have lower temporal resolution but lack limited-angle artifacts. Limited-angle artifacts typically have significantly lower frequencies (i.e., a larger spatial range) than motion artifacts. By adding the low-frequency portion of the complete data reconstruction image to the high-frequency portion of the partial data reconstruction image, different artifact types in each image are effectively filtered out, and the filtered image components are compensated for by adding other filtered image data. Thus, temporal resolution is improved and limited-angle artifacts are significantly reduced.

[0102] As described above, one possible advantageous application of embodiments of the present invention is cardiac CT imaging. For example, CT scans are often performed as part of routine management and monitoring of pacemakers. For example, it is often necessary to check the condition of the pacing lead. In some cases, it may be necessary to withdraw the pacing lead, for example, for replacement. In some cases, the pacing lead to be withdrawn may have grown into the wall of the superior vena cava (SVC). This adhesion may be difficult to detect with conventional CT imaging because the lead seen in the reconstructed image is blurred due to movement artifacts. The true location of the lead may be unclear.

[0103] By applying embodiments of the present invention, motion artifacts can be reduced by narrowing the range of data used in each slice, thereby improving temporal resolution, reducing blurring, and enhancing lead localization. Furthermore, by using the above method to achieve consistency between adjacent slices, leads can be consistently tracked within the 3D anatomical region of interest.

[0104] An example relating to a further aspect of the present invention provides a computer-implemented method used for X-ray computed tomography reconstruction. This method may include performing the above-described steps with reference to the processing unit 140 shown in Figure 1, as outlined in Figure 2.

[0105] Specifically, the method includes receiving input projection data containing multiple data subsets (202), each data subset corresponding to a different slice of the body to be scanned, and each data subset containing data corresponding to a defined acquisition angle range and a defined acquisition period.

[0106] The method further includes identifying one or more time or angular subwindows in the input projection data associated with the visibility of the desired anatomical feature or event in the input projection data (if reconstructed) (204).

[0107] The method further involves selecting a time or angular subrange of data (smaller than the full range of each slice) from a subset of data for each slice (206), the selection of the subrange being based on at least one of one or more identified time or angular subwindows in the projection data.

[0108] The method further includes generating a data output that shows the selected subrange for each slice.

[0109] An example relating to a further aspect of the present invention is a computer program product including computer program code, which is executable on a processor or computer. When the processor or computer is operably coupled to a source of input projection data, such as a CT imaging device or a data store, the code is configured to cause the processor to execute a method outlined above or according to any example or embodiment described below, or a method according to any claim of this application.

[0110] The above embodiments of the present invention utilize a processing apparatus. The processing apparatus may generally comprise a single processor or multiple processors. The processors may reside in a single housing device, structure, or unit, or they may be distributed across multiple different devices, structures, or units. Therefore, reference to a processing apparatus adapted or configured to perform a particular step or task may correspond to any one or more of the multiple processing apparatus elements performing that step or task, either individually or in combination. Those skilled in the art will understand how to implement such a distributed processing apparatus.

[0111] To perform the various required functions, one or more processors in a processing unit can be implemented in diverse ways using software and / or hardware. Typically, a processor uses one or more microprocessors that can be programmed using software (e.g., microcode) to perform the required functions. A processor can be implemented as a combination of dedicated hardware for performing some functions, and one or more programmed microprocessors and accompanying circuitry for performing other functions.

[0112] Examples of circuits that may be employed in various embodiments of this disclosure include, but are not limited to, conventional microprocessors, application-specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs).

[0113] In various implementations, a processor may be associated with one or more storage media, such as volatile and non-volatile computer memory, including RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that perform the required functions when running on one or more processors and / or controllers. Various storage media may be fixed within the processor or controller, or they may be portable, with one or more programs stored on the storage media being loaded into the processor.

[0114] Based on the drawings, disclosures, and accompanying claims, modifications of the disclosed embodiments can be understood and implemented by those skilled in the art. In the claims, the terms “equip” and “include” are not precluding other elements or steps, and singular elements are not precluding plural elements.

[0115] A single processor or other unit may perform the functions of multiple items described in the claims.

[0116] The fact that multiple means are described in different dependent claims does not necessarily mean that combinations of these means cannot be suitably used.

[0117] Computer programs may be stored and / or distributed on suitable media such as optical storage media or solid-state media supplied together with or as part of other hardware, or they may be distributed in other forms such as via the Internet or other wired or wireless telecommunications systems.

[0118] Where the term "adapted" is used in the claims or specification, it should be understood to be equivalent to the term "configured."

[0119] No reference numeral in the claims should be construed as limiting its scope.

Claims

1. A processing apparatus used for X-ray computed tomography (CT) reconstruction, wherein the processing apparatus is The acquisition of input CT projection data containing multiple data subsets, wherein each data subset corresponds to a different slice of the body being scanned, and each data subset contains data corresponding to a defined acquisition angle range and a defined acquisition time range. When reconstructed, this involves identifying one or more time or angular subwindows in the input CT projection data associated with the visibility of the desired anatomical feature or event in the data subset, Selecting a time or angular subrange of the input CT projection data from the data subset of each slice, which is smaller than the complete range of each slice, wherein the selection of the time or angular subrange is based on at least one of the identified time or angular subwindows in the input CT projection data. A processing device that generates a data output for each slice indicating the selected time or angular subrange, Perform a first image reconstruction of each slice using the complete data range of the data subset of each slice, and derive a first image of each slice. Perform a second image reconstruction of each slice using the identified time or angular subrange of the data subset of each slice, and derive a second image of each slice. Applying low-pass filtering to each of the first images and high-pass filtering to each of the second images, and generating a combined image for each slice that includes the filtered combination of the first and second images, A processing apparatus that further performs the following: generating an output showing the combined image of each slice.

2. The aforementioned processing apparatus is The system receives signals generated by external sensors that sense one or more physical parameters related to anatomical events, and the signals correspond to the sensor outputs during the acquisition of the input CT projection data. The apparatus according to claim 1, which determines one or more time or angular partial windows associated with the visibility of the anatomical feature or event, at least partially based on the received signal.

3. The apparatus according to claim 1 or 2, wherein determining one or more time or angular subwindows in the input CT projection data includes performing preliminary image reconstruction on one or more of the data subsets, and determining one or more time or angular subwindows in the input CT projection data in which the anatomical features or events are visible, based on the preliminary image reconstruction.

4. The processing apparatus according to any one of claims 1 to 3, wherein the selected angular subrange of each data subset has a predetermined angular width, and selecting the time or angular subrange of the data subset includes determining the alignment of the time or angular subrange within the entire width of the data subset.

5. The processing apparatus according to claim 4, dependent on claim 2, wherein determining the alignment of the aforementioned time or angular subrange is performed based on a signal generated by the external sensor.

6. The apparatus according to any one of claims 1 to 5, wherein selecting the time or angular subrange for one or more of the data subsets includes determining the angular width of the angular subrange based on one or more of the determined time or angular subwindows.

7. The processing apparatus according to any one of claims 1 to 6, wherein the input CT projection data represents an anatomical object of interest, the anatomical object of interest undergoes periodic motion, and the anatomical events correspond to phases of the periodic motion.

8. The apparatus according to any one of claims 1 to 7, wherein the selection of the time or angular partial range of each slice is performed on a basis in part of the time or angular partial range selected for at least a portion of the other slices.

9. The apparatus according to any one of claims 1 to 8, wherein the selection of the time or angular subrange of each data subset is based in part on maximizing the temporal overlap or temporal proximity of the time or angular subrange selected for each data subset, and in part on maximizing the temporal overlap or temporal proximity of the time or angular subrange selected for adjacent slices.

10. The apparatus according to any one of claims 1 to 9, further comprising extracting the time or angular subrange selected for each slice from the data subset of each slice.

11. The apparatus according to claim 10, which reconstructs an image of each slice using only the extracted time or angular subrange of each of the data subsets of each slice.

12. A method implemented in a computer used for X-ray computed tomography (CT) image reconstruction, wherein the method is Receiving input CT projection data containing multiple data subsets, each data subset corresponding to a different slice of the body being scanned, and each data subset containing data corresponding to a defined acquisition angle range and a defined acquisition period, When reconstructed, this involves identifying one or more time or angular subwindows in the input CT projection data associated with the visibility of the desired anatomical feature or event in the data subset, Selecting a time or angular subrange of data from the data subset of each slice that is smaller than the complete range of each slice, wherein the selection of the time or angular subrange is based on at least one of the one or more identified time or angular subwindows in the input CT projection data. A method comprising generating a data output for each slice that indicates the selected time or angular subrange, The steps include: performing a first image reconstruction of each slice using the complete data range of the data subset of each slice to derive a first image of each slice; The steps include performing a second image reconstruction of each slice using the identified time or angular subrange of the data subset of each slice to derive a second image of each slice, The steps include applying low-pass filtering to each of the first images, applying high-pass filtering to each of the second images, and generating a combined image for each slice that includes the filtered combination of the first and second images, A method further comprising the step of generating an output showing the combined image of each slice.

13. A computer program comprising computer program code executable on a processor, wherein the computer program code causes the processor to execute the method according to claim 12.

14. A computed tomography (CT) scanner comprising the processing apparatus described in any one of claims 1 to 11.

Images