Processing of projection area data generated by computed tomography scanners

Abstract

An output data set is generated that includes the projection area data at the desired / target imaging angle. The input data set that includes the projection area data captured / acquired at the desired / target imaging angle and the projection area data captured / acquired at one or more further predetermined imaging angles is processed to generate the output data set.

Description

Technical Field

[0001] The present invention relates to the field of computed tomography (CT), and more particularly to the processing of projection area data generated during a CT scan procedure.

Background Art

[0002] CT imaging has become an essential part of the medical imaging process to assist in the evaluation and diagnosis of patients / subjects.

[0003] Conventional CT scanners include an X-ray generator mounted facing one or more integrating detectors on a rotating gantry. The X-ray generator rotates around the examination area located between the X-ray generator and the one or more detectors, and emits (at least) radiation X-rays that traverse the examination area, as well as the subject and / or object placed in the examination area. The one or more detectors detect the radiation traversing the examination area and generate a signal (projection area data, or simply known as projection data) indicating the examination area, as well as the subject and / or object placed therein. The projection area data refers to raw detector data and is used to form a sinogram. The latter is a visual representation of the projection area data captured by the detector.

[0004] Typically, a reconstructor is further used to process the projection area data to reconstruct a volumetric image of the subject or object. That is, to reconstruct the image area data. The volumetric image is composed of a plurality of cross-sectional image slices. Each cross-sectional image slice is generated from the projection area data through a process of tomographic reconstruction, such as the application of a filtered back-projection algorithm. The reconstructed image data is, in fact, the inverse Radon transform of the raw projection area data.

Summary of the Invention

Problems to be Solved by the Invention

[0005] There is a continuing demand to improve the operation of CT scanners, and in particular to improve the quality of images generated by CT scanners. [Means for solving the problem]

[0006] According to an embodiment of one aspect of the present invention, a computer implementation method for processing projection area data generated by a CT scanner is provided.

[0007] This computer implementation method includes the steps of: acquiring a first input dataset containing projection region data generated by a CT scanner at a desired imaging angle, wherein the CT scanner generates projection region data at different imaging angles relative to the examination area during a scan operation; acquiring at least one further input dataset, each further input dataset containing projection region data generated by the CT scanner at a corresponding at least one further imaging angle, wherein the difference between the desired imaging angle and each corresponding further imaging angle is predetermined; inputting the first input dataset and at least one further input dataset into a machine learning algorithm, wherein the machine learning algorithm processes the first input dataset and at least one further input dataset to generate an output dataset, wherein the output dataset is different from the first input dataset and contains projection region data at a desired imaging angle relative to the examination area; and processing the first input dataset and at least one further input dataset using the machine learning algorithm to generate an output dataset.

[0008] Therefore, the present invention proposes processing projection region data, i.e., raw detector data, before it is reconstructed into image data. Specifically, input projection region data acquired at multiple imaging angles relative to the inspection region is processed using a machine learning method to provide output projection region data at a single imaging angle, i.e., a desired imaging angle.

[0009] The input projection region data includes a first input dataset containing projection region data acquired at a desired imaging angle, and one or more (i.e., at least one) further input datasets containing projection region data acquired at a predetermined imaging angle relative to the desired imaging angle. This facilitates consistent input to the machine learning method.

[0010] The fundamental understanding underlying this inventive concept is that projection region data acquired at different imaging angles can provide useful spatial or additional information for processing (e.g., reducing noise) projection region data acquired at a specific, i.e., desired angle. For example, different imaging angles still image the same part / volume of the subject or examination area. In other words, more information becomes available about a particular viewpoint. This allows for an increase in the amount of data to machine learning algorithms using naturally occurring information, without adding cost, i.e., without the need for additional image acquisition.

[0011] Therefore, the concept of multi-channel input to machine learning methods is proposed. Each channel provides projection region data acquired with a different imaging technique.

[0012] The imaging angle is defined as the angle at which the radiation source of a CT scanner emits radiation within the examination area, for example, relative to the center of the examination area. Specifically, in the case of a conventional CT scanner in which the radiation source is mounted on a rotating gantry, the imaging angle is defined as the angle that the radiation source makes with the horizontal plane passing through the center of rotation of the rotating gantry.

[0013] In a specific example, each dataset contains projection region data for a specific portion of the projection volume, which is the volume of the examination area illuminated by the CT scanner during a particular instance of the scan operation. A portion of the projection volume of at least one (i.e., one or more) further input datasets at least partially overlaps with a portion of the projection volume of the first input dataset.

[0014] The steps of the computer implementation method are performed by a processing unit or processing circuit. The machine learning algorithm is hosted by the processing unit or processing circuit. At least one additional input dataset may include a single additional input dataset or multiple additional input datasets, for example, two or more additional input datasets. If the additional input dataset includes multiple additional input datasets, a predetermined angle between the desired imaging angle and each additional imaging angle is different for each additional input dataset.

[0015] It should be clear that the difference between the desired imaging angle and each further imaging angle is non-zero.

[0016] In some embodiments, the above method further includes the step of generating an output dataset by using a machine learning algorithm to reduce at least one of the noise and artifacts in the projection region data of the first input dataset, based on the first input dataset and at least one further input dataset. Thus, a machine learning method can generate an output dataset by using the first input dataset and one or more further input datasets to reduce the noise and / or artifacts in the projection region data of the first input dataset. The proposed concept is advantageous when used to reduce noise / artifacts in projection region data, because missing or erroneous data in the first input dataset can be supplemented or corrected, for example, using data in one or more further input datasets that provide information about the same volume of subjects.

[0017] In some embodiments, the above method further includes the step of performing spectral filtering of the projection region data of the first input dataset based on the first input dataset and at least one further input dataset using a machine learning algorithm. Thus, the machine learning method can generate an output dataset by performing spectral filtering of the projection region data at a desired imaging angle using the first input dataset and one or more further input datasets.

[0018] In some embodiments, at least one further input dataset includes a first further input dataset containing projection region data generated by a CT scanner at a first imaging angle, where the difference between the desired imaging angle and the first imaging angle is equal to π.

[0019] In at least one embodiment, for each of the additional input datasets, the difference between the desired imaging angle and the corresponding additional imaging angle is a multiple of a first predetermined angle. The first predetermined angle is equal to the minimum change in imaging angle performed by the CT scanner during the scan operation. This embodiment utilizes the natural interrelationships within projection region data captured in close proximity to each other, i.e., at the closest available imaging angles (e.g., sequentially captured).

[0020] However, those skilled in the art will understand that any predetermined angle, such as an angle less than 0.5π, or more preferably an angle less than 0.1π, is suitable for the present invention.

[0021] The at least one additional input dataset may include a second additional input dataset containing projection region data generated by a CT scanner at a second imaging angle, wherein the difference between the desired imaging angle and the second imaging angle is a first predetermined angle, and a third additional input dataset containing projection region data generated by a CT scanner at a third imaging angle, wherein the difference between the third imaging angle and the desired imaging angle is a first predetermined angle. The third imaging angle is different from the second imaging angle. For example, if the desired imaging angle is θ, the second imaging angle may be θ+Δθ (or θ+π+Δθ), or the third imaging angle may be θ-Δθ (or θ+π-Δθ). In either case, the difference between the second / third imaging angles and the desired angle is the same, but the imaging angles are different.

[0022] The at least one further input dataset may include a fourth further input dataset containing projection region data generated by a CT scanner at a fourth imaging angle, wherein the difference between a desired imaging angle and the fourth imaging angle is a second predetermined angle, and the second predetermined angle is greater than the first predetermined angle; and a fifth further input dataset containing projection region data generated by a CT scanner at a fifth imaging angle, wherein the difference between the fifth imaging angle and the desired imaging angle is a second predetermined angle.

[0023] Preferably, the magnitude of the second predetermined angle is twice the magnitude of the first predetermined angle.

[0024] In some embodiments, at least one additional input data set includes no more than 10 additional input data sets. Limiting the maximum number of inputs to a limited amount can be useful in order to maintain, for example, a sufficiently large correlation within the projection area data from the first input data set and one or more additional input data sets to facilitate accurate processing by a machine learning algorithm.

[0025] Optionally, the CT scanner generates projection area data for each of a plurality of different portions of the projection volume for the subject. This projection volume is the volume of the examination area. Each of at least one additional input data set includes projection area data having a portion of the projection volume that at least partially overlaps a portion of the projection volume of the projection area data of the first input data set.

[0026] This approach is particularly useful for helical scan trajectories where the correlation of projection area data acquired at different angles decreases with an increase in the acquisition orbit pitch. This change can be accounted for by considering the portion of the projection volume associated with each instance of the projection area data. In a specific example, this change can be accounted for by using geometric knowledge of the helical scan trajectory, which includes, for example, knowledge of patient movement on the support or orbit pitch.

[0027] The method described above may further include, for each of at least one additional input data set, prior to inputting the first input data set and at least one additional input data set into a machine learning algorithm, processing the additional input data set to remove any portion of the projection area data of the additional input data set that corresponds to a portion of the projection volume of the projection area data of the additional input data set that does not overlap the projection volume of the first input data set.

[0028] That is, using the zero-padding approach, projection region data from one or more additional input data sets that do not show a correlation with the projection region data in the first input data set (i.e., projection region data of an additional input data set corresponding to an inspection region or a volume of a subject that was not irradiated during the collection of the projection region data of the first input data set) can be removed.

[0029] Optionally, the CT scanner includes a rotating gantry that rotates around a center of rotation, a radiation source that is rotatably supported by the rotating gantry, rotates with the rotating gantry, and emits radiation that traverses the inspection region, and a detector array that is rotatably supported by the rotating gantry, rotates with the rotating gantry, and generates projection region data responsive to the radiation emitted through the inspection region by the radiation source, and the imaging angle is the angle that the radiation source makes with respect to the horizontal plane.

[0030] The horizontal plane may pass through the center of rotation of the rotating gantry, for example, the center of the inspection region.

[0031] A computer program product including computer program code means is also proposed. When the computer program code means is executed on a computing device having a processing system, it causes the processing system to execute all steps of any of the methods described herein. A computer-readable medium may have a computer program or executable instructions embedded therein.

[0032] A device for processing projection region data generated by a CT scanner is also proposed.

[0033] The device includes a processing circuit or processing unit and a memory containing instructions, which, when executed by the processing circuit or processing unit, cause the processing circuit or processing unit to acquire a first input dataset containing projection region data generated by a CT scanner at a desired imaging angle, wherein the CT scanner generates projection region data at different imaging angles relative to the inspection area during a scan operation; acquire at least one further input dataset, each further input dataset containing projection region data generated by the CT scanner at a corresponding at least one further imaging angle, wherein the difference between the desired imaging angle and each corresponding further imaging angle is predetermined; input the first input dataset and the at least one further input dataset to a machine learning algorithm, which processes the first input dataset and the at least one further input dataset to generate an output dataset, wherein the output dataset is different from the first input dataset and contains projection region data at a desired imaging angle relative to the inspection area; and cause the machine learning algorithm to process the first input dataset and the at least one further input dataset to generate the output dataset.

[0034] A CT system is also proposed that includes the above-mentioned device and a CT scanner. This CT scanner includes a rotating gantry that rotates around a center of rotation, a radiation source rotatably supported by the rotating gantry and rotating with the rotating gantry, and emitting radiation that traverses the examination area, and a detector array rotatably supported by the rotating gantry and rotating with the rotating gantry, and generating projection area data in response to the radiation emitted by the radiation source through the examination area, wherein the imaging angle is the angle that the radiation source makes with respect to the horizontal plane.

[0035] The devices described above can perform any of the methods described herein, and vice versa. Similarly, computer program products, once executed, can perform any of the methods described herein, and vice versa. Those skilled in the art will be able to appropriately modify the devices, methods, and / or computer program products as needed.

[0036] These and other aspects of the present invention will become apparent from the embodiments described below and will be explained with reference to those embodiments. [Brief explanation of the drawing]

[0037] To gain a deeper understanding of the present invention and to more clearly illustrate how it is carried out, refer to the accompanying drawings as just one example.

[0038] [Figure 1] Figure 1 shows an imaging system including a CT scanner. [Figure 2] Figure 2 conceptually illustrates different imaging angles of the CT scanner in the imaging system. [Figure 3] Figure 3 shows a conceptual overview of this disclosure. [Figure 4] Figure 4 shows a conceptual overview of this disclosure. [Figure 5] Figure 5 shows an exemplary image generated using projection region data. [Figure 6] Figure 6 is a flowchart showing the method. [Figure 7] Figure 7 shows the processing apparatus. [Modes for carrying out the invention]

[0039] The present invention will be described with reference to the figures.

[0040] The detailed descriptions and specific examples illustrate exemplary embodiments of the apparatus, system, and method, but should be understood to be for illustrative purposes only and not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, system, and method of the present invention will be better understood from the following description, the appended claims, and the appended drawings. It should be understood that the figures are schematic diagrams only and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the figures to indicate the same or similar parts.

[0041] The present invention provides an approach for generating an output dataset containing projection region data at a desired / target imaging angle. A machine learning algorithm processes an input dataset containing projection region data captured / acquired at a desired / target imaging angle and projection region data captured / acquired at one or more further predetermined imaging angles to generate an output dataset.

[0042] Using this embodiment, noise / artifact reduction can be performed in the projection region space of CT projection region data.

[0043] Figure 1 shows an imaging system, particularly a CT imaging system 100, in which embodiments of the present invention may be employed.

[0044] The CT imaging system 100 includes a CT scanner 101 and a processing interface 111, the processing interface 111 which processes the data generated by the CT scanner 101 and uses that data to perform actions.

[0045] A CT scanner 101 typically 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 examination area about a longitudinal axis or the z-axis.

[0046] The patient support 120, such as a couch, supports an object or subject, such as a human patient, within the examination area. The support 120 moves the object or subject for loading, scanning, and / or unloading.

[0047] 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 that traverses the inspection area 106.

[0048] A detector array 110 that reacts to radiation defines the arc range for angles opposite the radiation source 108 across the inspection area 106. The detector array 110 includes one or more rows of detectors extending along the z-axis, which detect radiation crossing the inspection area 106 and generate projection area data indicating it. For example, the projection area data is cone beam projection data.

[0049] Therefore, the CT scanner 101 can generate / capture projection area data. As the rotating gantry 104 rotates around the examination area 106, the imaging angle of the CT scanner 101 relative to the examination area changes. Conceptually, this can be understood as the direction from which radiation is emitted from the radiation source 108 changes as the rotating gantry 104 rotates. Projection area data is generated by the CT scanner 101 as each of the multiple imaging angles of the CT scanner 101 during the CT scan procedure. Therefore, multiple datasets are generated, and each dataset contains projection area data acquired at different points in time. Temporarily adjacent datasets are generated at different imaging angles.

[0050] The patient support 120 moves along the longitudinal axis z during the CT scan procedure. This can be done in steps. The rotating gantry 104 makes a full 2π rotation between each movement of the patient support 120. This process is known as the “circular scan trajectory” approach or the “stop-and-shoot” method.

[0051] Another imaging approach is the "helical scan trajectory" approach, or helical CT scan. In this approach, projection area data is acquired during the continuous rotation of the gantry and the simultaneous translation of the patient support.

[0052] "Projection volume" refers to the entire volume of the examination area imaged during a CT scan procedure. Each dataset generated by the CT scanner contains projection area data captured at a specific imaging angle and represents a specific portion of the projection volume (for example, the illuminated portion of the projection volume). Typically, a portion of the projection volume captured at a specific imaging angle is conical, or cone-shaped.

[0053] A general-purpose computing system or computer functions as an operator console 112 and includes input devices 114 such as a mouse and keyboard, and output devices 116 such as a display monitor. The console 112 allows the operator to control the operation of the system 100.

[0054] The reconstruction device 118 processes the projection area data and reconstructs the volumetric image data. This data is displayed via one or more display monitors of the output device 116.

[0055] The reconstruction device 118 may employ filtered back projection (FBP) reconstruction, low-noise reconstruction algorithms for the image region and / or projection region (such as iterative reconstruction), and / or other algorithms. It is understood that the reconstruction device 118 can be implemented via at least one processor that executes computer-readable instructions encoded or embedded in a computer-readable storage medium such as physical memory or other non-temporary media. Furthermore, one or more processors may also execute computer-readable instructions carried by carriers, signals, and other temporary or non-temporary media.

[0056] The present invention relates to the processing of projection region data generated by a CT scanner 101 by a device such as a processing unit 119. Those skilled in the art will understand that this projection region data can be obtained directly from the CT scanner, or it can be obtained through one or more other circuit elements such as memory or buffers (not shown) that temporarily, semi-permanently, and / or permanently store the projection region data.

[0057] Figure 2 conceptually shows a cross-sectional view of the rotating gantry 104 to illustrate different imaging angles of the CT scanner 101, as shown on a horizontal plane passing through the rotating isocenter.

[0058] In particular, Figure 2 illustrates how a CT scanner acquires projection region data at multiple different imaging angles relative to the examination area. Figure 2 shows examples of radiation source positions 201, 202, 203, 204, 205, and 206 as the rotating gantry rotates. Those skilled in the art will understand that these positions are merely exemplary and that the radiation source may be positioned at other angles depending on the capabilities of the CT scanner and the CT scanning procedure. Projection region data generated at a corresponding position means that corresponding projection region data was generated at different imaging angles.

[0059] Figure 2 also shows how the imaging angle changes for each position of the radiation source (indicated by arrows). In the illustrated example, the CT scanner 101 rotates around a rotation center 210, for example, an axis passing through the center of the CT scanner entering / exiting the paper plane. Therefore, in the illustrated example, the imaging angle is the angle around the rotation center of the CT scanner 101. In other words, the imaging angle represents the position of the radiation source around the cross-section of the rotating gantry.

[0060] Specifically, in this cross-sectional view, the imaging angle is the angle between the direction of the center of the radiation emission provided by the radiation source and the horizontal plane 220 passing through the rotation center 210. Typically, the imaging angle is the angle between the radiation source, the imaginary line passing through the rotation center 210, and the horizontal plane 220.

[0061] Typically, when the radiation source is located at a specific position (such as position 201), the radiation-sensitive detector array 110 is positioned at the opposite position in the cross-sectional view (such as position 206). As mentioned above, the radiation-sensitive detector array 110 defines the arc range for the angle opposite to the radiation source 108.

[0062] The present invention proposes generating extended, improved, or modified projection region data for a desired imaging angle (such as angle θ) using projection region data from the desired imaging angle and one or more further imaging angles. The difference between this desired imaging angle and each further imaging angle is predetermined.

[0063] Figures 3 and 4 conceptually illustrate the outline 300 adopted in various embodiments. One approach involves using a machine learning algorithm 320 to process several input datasets 310, which include projection region data, to generate an output dataset 330. This allows the machine learning algorithm to operate in the projection region.

[0064] The output dataset 330 includes processed or modified projection region data at a desired imaging angle. The input dataset 310 includes a first input dataset 311 containing projection region data at a desired imaging angle, and one or more further input datasets 312, 313, 314 (a single further input dataset 312 as shown in Figure 3, or two or more further input datasets 312, 313, 314 as shown in Figure 4). Each further input dataset includes projection region data at a further imaging angle, and the angle between the desired imaging angle and the further imaging angle is predetermined.

[0065] Each input dataset contains projection area data captured by a CT scanner. This data may have been pre-processed, for example, to exclude certain predetermined frequencies. The output dataset contains projection area data processed by a machine learning algorithm.

[0066] Therefore, rather than simply processing projection region data at only the desired imaging angle to generate an output dataset, the input to the machine learning algorithm is augmented / complemented by using projection region data acquired at one or more different angles. In other words, the machine learning algorithm receives "original" projection region data at the desired imaging angle and one or more other imaging angles as input, and generates processed and / or modified projection region data at the desired imaging angle as output.

[0067] As will be explained in more detail below, when the machine learning algorithm 320 corrects the projection region data at a desired imaging angle, it improves the image properties of any image subsequently generated using the projection region data at the desired imaging angle by using projection region data acquired at other imaging angles. These imaging properties include, for example, the amount of noise, the number of artifacts, resolution, uniformity, signal-to-noise ratio, and / or contrast level.

[0068] In particular, machine learning algorithms perform noise reduction and / or artifact reduction on projected region data.

[0069] In some embodiments, a machine learning algorithm performs spectral filtering of the projection region data at a desired imaging angle. This process can be improved, for example, by using projection region data at further imaging angles to preserve elements.

[0070] For example, a machine learning algorithm receives low-dose projection region data and outputs simulated high-dose projection region data, i.e., low-noise projection region data with a noise level equivalent to that of the high-dose projection region data.

[0071] In a particularly advantageous embodiment, for at least one of one or more additional input datasets, the difference between the additional imaging angle and the desired imaging angle is π, i.e., 180°. In this embodiment, it is recognized that at least a portion of the radiation emitted from the radiation source at either of these two imaging angles, i.e., the radiation, undergoes the same attenuation from the elements placed in the inspection area. However, the portions of the volume irradiated at both imaging angles overlap at least partially. Therefore, the projection region data generated at both of these angles contains similar information that can be used to extend the projection region data acquired at either of these angles.

[0072] As an exemplary embodiment, further referring again to Figure 2, the first input dataset 311 includes projection region data acquired when the radiation source is at position 201, and the first further input dataset 312 includes projection region data acquired when the radiation source is at position 206.

[0073] As a specific example of the work, one or more additional input datasets include, for example, only a single additional input dataset 312 having projection region data acquired at an additional imaging angle, as shown in Figure 3, where the difference between this additional imaging angle and the desired imaging angle is π. In other embodiments, one or more additional input datasets include, for example, two or more additional input datasets, as shown in Figure 4, where at least one of them is similar to the single additional input dataset described above.

[0074] In some embodiments, for at least one of the one or more further input datasets 312, 313, 314, the difference between the further imaging angle and the desired imaging angle is a multiple of a given angle Δθ. That is, the difference is equal to k·Δθ. Thus, projected image data of one or more further input datasets are acquired at imaging angles θ±Δθ, θ±2Δθ, ..., θ±k·Δθ.

[0075] In some embodiments, it is preferable that one or more additional input datasets consist of 20 or fewer additional input datasets (so, for example, k is 20 or less), more preferably 10 or fewer additional input datasets (so, for example, k is 10 or less), and even more preferably 5 or fewer additional input databases (so, for example, k is 5 or less). This allows for a sufficiently large interrelationship in the corresponding projection region data, thereby improving the performance of the machine learning algorithm.

[0076] In some embodiments, one or more additional input datasets consist of at least four datasets, namely a first input dataset, a second input dataset, a third input dataset, and a fourth input dataset. The first and second input datasets each contain projection region data acquired at imaging angles θ+Δθ and θ-Δθ, respectively, where θ is a desired imaging angle and Δθ is a predetermined angle. The third and fourth input datasets each contain projection region data acquired at imaging angles θ+2·Δθ and θ-2·Δθ, respectively.

[0077] For example, a given angle Δθ is equal to the minimum change in the imaging angle performed by the CT scanner during a scan operation. For example, this minimum change is defined by the CT scanner itself, i.e., defined, for example, by a pre-scheduled scan operation and / or a given scan operation. Thus, a given angle Δθ is equal to the fixed rotation or a predetermined angle increment of the CT scanner's rotation during the scan operation.

[0078] As an exemplary embodiment, further referring again to Figure 2, a first input dataset 311 includes projection region data acquired when the radiation source 108 is at position 201, a first further input dataset 312 includes projection region data acquired when the radiation source is at position 202, and a second further input dataset 313 includes projection region data acquired when the radiation source is at position 203.

[0079] In some embodiments, the value of Δθ is equal to π. This occurs when, during a CT scan procedure, multiple measurements are taken at a desired imaging angle and / or an imaging angle opposite to the desired imaging angle (i.e., with a difference of π), for example, between perfusion measurements.

[0080] In a preferred embodiment, at least one of the further input datasets includes projection imaging data acquired at a different imaging angle than the projection imaging data of the first input dataset.

[0081] In some embodiments, for at least one of one or more additional input datasets, the difference between the additional imaging angle and the desired imaging angle is equal to the sum of π and a multiple of a given angle Δθ. Thus, the difference between the imaging angle of the projection region data of the additional input dataset and the desired imaging angle of the projection region data of the first input dataset is π ± k·Δθ.

[0082] Naturally, any other form of a given angle can be used. For example, two or more given angles, e.g., Δθ1, Δθ2, ..., Δθ k There may be at least one of them, preferably, but not required, equal to π and / or a multiple of π.

[0083] Any combination of predetermined angles previously identified can be used. As mentioned above, this is particularly advantageous if one or more further input datasets include at least one further input dataset in which projection region data is acquired / captured at an imaging angle that is a π offset from the imaging angle of the projection region data of the first input dataset.

[0084] As shown in Figures 3 and 4, the input datasets are concatenated or otherwise grouped to form a single dataset 319 that serves as input to the machine learning method.

[0085] The total number of input datasets for the machine learning algorithms shown in Figures 3 and 4 is merely an example; any number of suitable input datasets can be used.

[0086] The machine learning algorithm processes the input dataset 310 and provides an output dataset containing processed projection region data at a desired imaging angle. The processed projection region data has improved imaging characteristics compared to the imaging data contained in the first input dataset, and / or is registered, transformed, and / or converted to a given coordinate system.

[0087] The process f(.) performed by the machine learning algorithm can be modeled as follows:

number

number

[0088] Machine learning algorithms are hosted by a processing unit.

[0089] Therefore, various embodiments of the present invention utilize machine learning algorithms to process input datasets and generate output datasets. The machine learning method can perform necessary feature registration tasks, such as mapping features of different input datasets to each other, in order to generate the output dataset.

[0090] A machine learning algorithm is any self-training algorithm that processes input data to generate or predict output data. According to embodiments of the present invention, the input data consists of an input dataset comprising a first input dataset and one or more further input datasets, and the output data consists of the aforementioned output datasets.

[0091] Suitable machine learning algorithms for use in the present invention will be apparent to those skilled in the art. Examples of suitable machine learning algorithms include decision tree algorithms and artificial neural networks. Other machine learning algorithms such as logistic regression, support vector machines, or naive Bayes models are suitable alternatives.

[0092] The structure of artificial neural networks, or simply neural networks, is inspired by the human brain. A neural network consists of layers, each containing multiple neurons. Each neuron performs mathematical operations. In particular, each neuron may contain different weightings of a single type of transformation (for example, the same type of transformation, such as a sigmoid, but with different weightings). In the input data processing process, the mathematical operations of each neuron are performed on the input data to generate a numerical output, and the output of each layer of the neural network is passed sequentially to the next layer, with the final layer providing the output.

[0093] The training methods for machine learning algorithms are well-known. Typically, these methods involve obtaining a training dataset containing training input data entries and corresponding training output data entries. An initialized machine learning algorithm is applied to each input data entry to generate predicted output data entries. The machine learning algorithm is then modified using the error between the predicted output data entries and the corresponding training output data entries. This process is repeated until the error converges and the predicted output data entries are sufficiently similar to the training output data entries (e.g., ±1%). This is generally known as supervised learning.

[0094] For example, if a machine learning algorithm is formed from a neural network, the weighting of the mathematical operations of each neuron can be modified until the error converges. Known methods for modifying neural networks include gradient descent and backpropagation algorithms.

[0095] The training input data entries correspond to example input datasets. The training output data entries correspond to example output datasets.

[0096] As an example of this work, consider a scenario in which a machine learning algorithm is trained to denoise some projection region data in order to generate denoised projection region data. In this scenario, the output dataset is denoised projection region data, and the input dataset contains projection region data at a desired imaging angle and at least one other imaging angle.

[0097] In this scenario, the training input and output data entries for the training dataset are first generated by collecting projection region data that is known to produce one or more high-quality images when processed by manual intervention / selection and / or automated quality processing methods. The projection region data should include projection region data at the desired imaging angle and each further imaging angle. Noise, such as white noise or pink noise, can be added to the projection region data to simulate low-quality projection region data. The training input data entries are generated by selecting the simulated low-quality projection region data at the desired / target imaging angle and one or more other projection region data at further imaging angles.

[0098] The "further imaging angles" used to train the machine learning network do not need to be exactly the same as the further imaging angles used in later inference, as is the case when performing the method according to the embodiment. However, to improve performance, it is preferable to use the same further imaging angles during training as the further imaging angles used during inference.

[0099] A machine learning network is, for example, a convolutional neural network (CNN) using a U-Net, ResNet, or deep feedforward neural network architecture. Therefore, a machine learning network is a CNN-based machine learning framework that includes common feedforward, encoder-decoder (U-Nets), or other common CNN architectures composed of various combinations of DenseNet or ResNet building blocks.

[0100] From the above, it will be understood that the proposed approach involves inputting projection region data acquired at different imaging angles and outputting projection region data at the target / desired imaging angle with improved characteristics. Since the input dataset contains projection region data at the target / desired imaging angle, the proposed approach can effectively improve the characteristics of this projection region data.

[0101] The number of input datasets, i.e., the number of input channels, affects the number of feature maps in the first convolutional layer. In particular, a larger number of input datasets results in a larger number of feature maps. A larger number of feature maps increases the network capacity, and in principle, makes it easier to learn more complex representations from the training data distribution.

[0102] Figure 5 illustrates the effects / impacts of various embodiments of the present invention. Figure 5 shows three images of the same portion of a patient (lower coronal section). Each image is generated by processing multiple projection region datasets, each set of projection region data at a specific imaging angle. Each set of projection region data has undergone a denoising process using a machine learning method to remove noise from the projection region dataset.

[0103] In the first image 510, each set of projection region data is processed by a machine learning method that takes projection region data acquired only at the imaging angle of the projection region dataset as input.

[0104] In the second image 520, each set of projection region data is processed according to various embodiments of the present invention. Specifically, each set of projection region data is processed by a machine learning method that receives as input a first input dataset containing projection region data acquired at an imaging angle (i.e., θ) of the projection region dataset, and a further input dataset containing projection region data acquired at an opposite imaging angle (i.e., θ+π) having a predetermined relationship with the imaging angle of the projection region dataset.

[0105] In the third image 530, each set of projection region data is processed using another embodiment of the present invention. Specifically, each set of projection region data is processed by a machine learning method that receives as input a first input dataset containing projection region data acquired at an imaging angle (i.e., θ) of the projection region dataset, and two input datasets containing projection region data at two further imaging angles (i.e., θ ± Δθ) equally spaced from angle θ. The value of Δθ is equal to the minimum change in angle performed by the CT scanner during the CT scan procedure.

[0106] As shown in Figure 5, processing specific projection region data within a projection region using projection region data acquired at two or more imaging angles improves noise reduction compared to using only projection region data acquired at the desired imaging angle. This effect is most pronounced in the second image 520, which is generated using projection region data acquired at an angle opposite to the desired angle. However, the improvement over the first image 510 is also seen in the third image 530.

[0107] As previously mentioned, each dataset generated by a CT scanner may contain projection region data for specific parts of the projection volume. This disclosure acknowledges that a portion of the projection volume in one dataset may overlap with a portion of the projection volume in at least one other dataset.

[0108] To improve the performance of the machine learning algorithm, in a preferred embodiment, a portion of the projection volume of each of one or more further input datasets partially overlaps with the projection volume of the first input dataset. That is, each projection region data of a further input dataset represents at least a portion of the projection volume represented by the projection region data of the first input dataset (i.e., its information is contained in the projection region data of the first input dataset).

[0109] Some of the projected volumes represented by specific datasets of projected region data can be easily identified based on factors such as the translation of the patient support and the rotation of the rotational gantry.

[0110] In some embodiments, before inputting the first input dataset and one or more further input datasets into the machine learning algorithm, each further input dataset is processed to remove the portion of the projected region data of the further input dataset that corresponds to the portion of the projected region data of the further input dataset that does not overlap with the projected volume of the first input dataset.

[0111] As an example, portions of the projection region data that do not correspond to the projection volume overlapping with the projection volume of the first input dataset are set to "zero," that is, they are zero-padded. Other approaches, such as replacing the value with 1 instead of 0, would be obvious to those skilled in the art.

[0112] In each further input dataset, the portion of the projection region data of that further dataset that represents a portion of the projection volume that overlaps with a portion of the projection volume of the first input dataset is identified based on metadata of the projection region dataset, such as data that identifies differences in imaging angles or changes in the position of the patient support along the z-axis (i.e., translation of the patient support).

[0113] Figure 6 shows a computer implementation method 600 for processing projection area data generated by a CT scanner. Method 600 is performed by a device such as the processing unit 119 shown in Figure 1.

[0114] This method includes step 610 of acquiring a first input dataset 605 containing projection region data generated by a CT scanner at a desired imaging angle. The CT scanner generates projection region data at different imaging angles relative to the examination area during the scanning operation. An example of a suitable CT scanner has already been described with reference to Figure 1, but other examples will be readily apparent to those skilled in the art.

[0115] Method 600 also includes step 620 of obtaining one or more further input datasets 607. Each dataset contains projection area data generated by the CT scanner at one or more further imaging angles. The difference between the above desired imaging angle and each further imaging angle is predetermined.

[0116] Further examples of input datasets have already been described and can be used in Method 600.

[0117] Method 600 also includes step 630 of inputting a first input dataset and one or more further input datasets into a machine learning algorithm. The machine learning algorithm processes the first input dataset and one or more further input datasets to generate an output dataset different from the first input dataset, which includes projection region data at a desired imaging angle relative to the inspection area.

[0118] Method 600 also includes step 640 of processing a first input dataset and a further input dataset in order to generate an output dataset.

[0119] In some embodiments of the present invention, method 600 further includes step 650 of outputting an output dataset, i.e., outputting it from the processing device. The output dataset may be provided to a reconstruction device for, for example, reconstructing one or more images from the projection region data of the output dataset, or to a memory for, for example, storing the projection region data for later processing. Other approaches and purposes of projection region data will be obvious to those skilled in the art.

[0120] Steps 630, 640, and 650 form the overall process for generating and outputting projection region data.

[0121] As a further embodiment, Figure 7 shows an example of a device 70 or processing unit, in which one or more parts of the embodiment can be used. The various operations described above utilize the functionality of device 70. For example, one or more parts of a system that processes images using a CNN can be incorporated into any element, module, application, and / or component described herein. In this regard, it should be understood that the functional blocks of the system can run on a single computer or be distributed across several computers and locations (e.g., connected via the Internet).

[0122] Device 70 includes, but is not limited to, at least one of the following: a processor, a PC, a workstation, a laptop, a PDA, a palm device, a server, storage, a cloud computing device, and a distributed processing system. Generally, with respect to the hardware architecture, device 70 includes one or more processing circuits 71, memory 72, and one or more I / O devices 73 that are communicatively coupled via a local interface. The local interface is, for example, one or more buses, or other wired or wireless connections as known in the art, but is not limited to these. The local interface may have a controller, buffers (caches), drivers, repeaters, and receivers to enable communication. Furthermore, the local interface may include addresses, control units, and / or data connections to enable proper communication between the aforementioned components.

[0123] The processing circuit 71 is a hardware device that executes software which may be stored in memory 72. The processing circuit 71 may be substantially any custom-made or commercially available processing circuit, a central processing unit (CPU), a digital signal processing circuit (DSP), or any auxiliary processing circuit from among several processing units associated with device 70, and the processing circuit 71 may also be a semiconductor-based microprocessing processor in the form of a microchip.

[0124] Memory 72 may include one or a combination of volatile memory elements (e.g., random access memory (RAM) such as dynamic random access memory (DRAM) and static random access memory (SRAM)) or non-volatile memory elements (e.g., ROM, erasable programmable read-only memory (EPROM), electronically erasable programmable read-only memory (EEPROM®), programmable read-only memory (PROM), tape, compact disk read-only memory (CD-ROM), disk, floppy disk, cartridge, cassette, etc.). Furthermore, memory 72 may incorporate electronic media, magnetic media, optical media, and / or other types of storage media. Memory 72 may have a distributed architecture in which various components are located in separate locations from each other but are accessible by the processing circuit 71.

[0125] The software in memory 72 may include one or more separate programs, each containing an ordered list of executable instructions for performing a logical function. The software in memory 72 may also include, according to a model embodiment, a suitable operating system (O / S) 75, a compiler 74, source code 73, and one or more applications 76. As shown in the figure, the application 76 includes numerous functional components for performing the features and operations of the model embodiment. The application 76 of device 70 may represent a variety of applications, computing units, logic circuits, functional units, processes, operations, virtual entities, and / or modules according to a model embodiment, but the application 76 is not intended to be limiting.

[0126] O / S75 controls the execution of other computer programs and provides scheduling, input / output control, file and data management, memory management, and communication control, as well as related services. The inventor intends that Application 76 for carrying out exemplary embodiments is applicable to any commercially available operating system.

[0127] An application 76 can be a source program, an executable program (object code), a script, or any other entity including a set of instructions to be executed. If it is a source program, the program is usually translated through a compiler 74, such as a compiler, which may or may not be included in memory 72, an assembler, an interpreter, etc., so that it can function properly in relation to the OS 75. Furthermore, an application 76 can be written as an object-oriented programming language with classes of data and methods, or as a procedural programming language with routines, subroutines, and / or functions (e.g., C, C++, C#, Pascal, BASIC, API calls, HTML, XHTML, XML, ASP scripts, JavaScript®, FORTRAN, COBOL, Perl, Java®, ADA, .NET, etc., but not limited to these).

[0128] I / O devices 73 include input devices such as, but are not limited to, a mouse, keyboard, scanner, microphone, and camera. Furthermore, I / O devices 73 include output devices such as, but are not limited to, a printer and display. Finally, I / O devices 73 further include devices that communicate both input and output, such as, but are not limited to, modulators / demodulators (for accessing remote devices, other files, devices, systems, or networks), radio frequency (RF) or other transceivers, telephone interfaces, bridges, and routers. I / O devices 73 also include components for communication over various networks, such as the Internet and intranets.

[0129] If device 70 is a PC, workstation, or intelligent device, the software in memory 72 may also include a Basic Input / Output System (BIOS). The BIOS is a set of essential software routines that initialize and test the hardware at startup, boot the OS 75, and support data transfer between hardware devices. The BIOS is stored in some type of read-only memory, such as ROM, PROM, EPROM, or EEPROM, so that it is executable when device 70 is started.

[0130] During the operation of device 70, the processing circuit 71 executes software / executable instructions stored in memory 72, exchanges data with memory 72, and generally controls the operation of device 70 according to the software / executable instructions. The application 76 and OS 75, in whole or in part, are read by the processing circuit 71, and in some cases, buffered within the processing circuit 71 before being executed.

[0131] If Application 76 is implemented in software, it should be noted that Application 76 may be stored in substantially any computer-readable medium for use in or in connection with any computer-related system or method. In the context of this document, computer-readable medium may be an electronic, magnetic, optical, or other physical device or means capable of storing or preserving a computer program for use in or in connection with a computer-related system or method.

[0132] Application 76 may be embodied in any computer-readable medium used in or in connection with an instruction execution system, apparatus, or device (such as a computer-based system, a system with processing circuits, or another system capable of fetching instructions from an instruction execution system, apparatus, or device and executing those instructions). In the context of this document, “computer-readable medium” can be any means by which a program can be stored, communicated, propagated, or transferred for use in or in connection with an instruction execution system, apparatus, or device. Computer-readable mediums include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, devices, or propagation media.

[0133] In the context of this disclosure, the I / O device 73 receives an input dataset from the medical imaging device 100. In another embodiment, the input dataset is obtained from a memory location or a unit (not shown).

[0134] The I / O device 73 may provide the output dataset to the user interface 111. The user interface 111 may provide a visual representation of the output dataset, such as displaying the image(s) corresponding to the medical imaging data included in the output dataset.

[0135] Those skilled in the art will be able to easily develop a device having a processing circuit for performing the method described herein. Thus, each step in the flowchart represents a different action performed by the device's processing circuit, which can be performed by the corresponding module of the device's processing circuit.

[0136] Therefore, embodiments utilize a device. The device can be implemented in various ways using software and / or hardware to perform a variety of necessary functions. A processor is an example of a device employing one or more microprocessors programmed using software (such as microcode) to perform the necessary functions. However, the device may be implemented regardless of the adoption of a processor, and may also be implemented as a combination of dedicated hardware for performing some functions and a processor (such as one or more programmed microprocessors and associated circuits) for performing other functions.

[0137] Examples of device components that may be used 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).

[0138] In various implementations, a processor or device 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, when executed on the processing circuit, perform the necessary functions. These storage media may be fixed within the processor or device, or they may be transportable so that one or more programs stored therein can be loaded into the processor or device.

[0139] It will be understood that the disclosed method is a computer implementation method. Therefore, the concept of a computer program is also proposed, which includes code means for implementing any described method when executed on a device equipped with processing circuits, such as a computer. Thus, different parts, lines, or blocks of code in a computer program according to one embodiment may be executed by a device or computer to perform any method described herein. In some alternative implementations, the functions shown in a block diagram or flowchart may occur in a different order than shown in the diagram. For example, two blocks shown consecutively may actually be executed substantially simultaneously, or these blocks may be executed in reverse order depending on the functions involved.

[0140] Modifications of the disclosed embodiments can be understood and implemented by those skilled in the art in carrying out the claimed invention, based on a review of the drawings, disclosures, and appended claims. In the claims, the word “includes” does not exclude other elements or steps, and singular elements do not exclude plural elements. A single processor or other unit may perform the functions of several items described in the claims. The mere fact that certain means are described in mutually different dependent claims does not mean that combinations of these means cannot be used advantageously. Where a computer program is described above, it may be stored / distributed on any suitable medium, such as optical storage media or solid-state media, supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless communication systems. Note that where the term “adapted to be” is used in the claims or description, the term “adapted to be” is intended to be equivalent to the term “configured to be.” Any reference numerals in the claims should not be construed as limiting the scope.

Claims

1. A computer implementation method for processing projection area data generated by a computed tomography (CT) scanner, wherein the computer implementation method is: A step of acquiring a first input dataset including projection region data generated by the CT scanner at a desired imaging angle, wherein the CT scanner generates and acquires projection region data at different imaging angles with respect to the examination area during a scan operation. A step of acquiring at least one further input dataset, each further input dataset comprising projection region data generated by the CT scanner at a corresponding at least one further imaging angle, wherein the difference between the desired imaging angle and each corresponding further imaging angle is predetermined. The steps include inputting the first input dataset and the at least one further input dataset into a machine learning algorithm, To generate the output dataset, the machine learning algorithm is used to process the first input dataset and the at least one further input dataset. Includes, The machine learning algorithm processes the first input dataset and the at least one further input dataset to generate an output dataset, wherein the output dataset differs from the first input dataset and includes projection region data at the desired imaging angle relative to the inspection region. A computer-aided method wherein the at least one further input dataset includes a first further input dataset containing projection region data generated by the CT scanner at a first imaging angle, the difference between the desired imaging angle and the first imaging angle is equal to π.

2. The computer implementation of claim 1, further comprising the step of using the machine learning algorithm to reduce noise and / or artifacts in the projection region data of the first input dataset based on the first input dataset and the at least one further input dataset, thereby generating the output dataset.

3. The computer implementation of claim 1, further comprising the step of using the machine learning algorithm to perform spectral filtering of the projection region data of the first input dataset based on the first input dataset and the at least one further input dataset to generate the output dataset.

4. The computer implementation method according to claim 1, wherein for each of the at least one further input dataset, the difference between the desired imaging angle and the corresponding further imaging angle is a multiple of a first predetermined angle.

5. The computer-aided method according to claim 4, wherein the first predetermined angle is equal to the minimum change in imaging angle performed by the CT scanner during a scan operation.

6. The aforementioned at least one further input dataset is: A second further input dataset comprising projection region data generated by the CT scanner at a second imaging angle, wherein the difference between the desired imaging angle and the second imaging angle is the first predetermined angle, A third further input dataset comprising projection region data generated by the CT scanner at a third imaging angle, wherein the difference between the third imaging angle and the desired imaging angle is the first predetermined angle, The computer implementation method according to claim 5, including the method described in claim 5.

7. The aforementioned at least one further input dataset is: A fourth further input dataset comprising projection region data generated by the CT scanner at a fourth imaging angle, wherein the difference between the desired imaging angle and the fourth imaging angle is a second predetermined angle, and the second predetermined angle is greater than the first predetermined angle, A fifth further input dataset comprising projection region data generated by the CT scanner at a fifth imaging angle, wherein the difference between the fifth imaging angle and the desired imaging angle is the second predetermined angle, The computer implementation method according to claim 6, including the method described in claim 6.

8. The computer implementation method according to claim 1, wherein the at least one further input dataset includes 10 or fewer further input datasets.

9. The CT scanner generates projection region data for each of several different parts of the projection volume relative to the subject, and the projection volume is the volume of the examination region. The computer implementation of claim 1, wherein each of the at least one further input datasets includes projection region data having a portion of the projection volume that at least partially overlaps with a portion of the projection volume of the projection region data of the first input dataset.

10. For each of the at least one further input datasets, before inputting the first input dataset and the at least one further input dataset into the machine learning algorithm, The computer implementation of claim 9, further comprising the step of processing the further input dataset to remove any portion of the projection region data of the further input dataset that corresponds to the portion of the projection region data of the further input dataset that does not overlap the projection volume of the first input dataset.

11. The aforementioned CT scanner is A rotating gantry that rotates around the center of rotation, A radiation source is rotatably supported by the rotating gantry, rotates with the rotating gantry, and emits radiation that traverses the inspection area. A detector array rotatably supported by the rotating gantry, which rotates with the rotating gantry and generates projection area data in response to radiation emitted by the radiation source through the inspection area, Includes, The imaging angle is the angle that the radiation source makes with respect to the horizontal plane. The computer implementation method according to claim 1.

12. A non-temporary computer-readable medium for storing executable instructions, wherein, when the executable instructions are executed by a processing circuit, the processing circuit causes the processing circuit to perform the method according to any one of claims 1 to 11.

13. A computer program comprising computer program code means, wherein the computer program code means, when executed on a computing device having a processing system, causes the processing system to perform all steps of the method according to any one of claims 1 to 11.

14. A device for processing projection area data generated by a computed tomography (CT) scanner, the device comprising a processing circuit and a memory containing instructions, When the aforementioned instruction is executed by the processing circuit, the processing circuit will: The process involves acquiring a first input dataset containing projection region data generated at a desired imaging angle by the CT scanner, wherein the CT scanner generates and acquires projection region data at different imaging angles relative to the examination area during the scan operation. Acquiring at least one further input dataset, each further input dataset comprising projection region data generated by the CT scanner at a corresponding at least one further imaging angle, wherein the difference between the desired imaging angle and each corresponding further imaging angle is predetermined. Inputting the first input dataset and the at least one further input dataset into a machine learning algorithm, To generate the output dataset, the machine learning algorithm is used to process the first input dataset and the at least one further input dataset, The machine learning algorithm processes the first input dataset and the at least one further input dataset to generate an output dataset, wherein the output dataset differs from the first input dataset and includes projection region data at the desired imaging angle relative to the inspection region. The device wherein the at least one further input dataset includes a first further input dataset containing projection region data generated by the CT scanner at a first imaging angle, the difference between the desired imaging angle and the first imaging angle is equal to π.

Images