Reducing noise in computed tomography image slices

Abstract

A computer-implemented method of reducing noise in a computed tomography (CT) image slice (1101.. i,) passing through an anatomical region is provided. The method comprises: receiving CT data (120) representative of the image slices (1101.. i,); adjusting an amount of noise in the image slice (1101.. i,) to provide a denoised image slice () having a similar amount of noise; and outputting the de-noised image slice ().

Description

TECHNICAL FIELD

[0001] The present disclosure relates to reducing noise in computed tomography (CT) image slices. Computer-implemented methods, computer program products, and systems are disclosed. BACKGROUND

[0002] CT data is often reconstructed to provide image slices through an anatomical region. The image slices are analyzed to diagnose a patient. For example, multi-slice CT images are often analyzed to investigate the presence of coronary artery disease.

[0003] Various types of CT imaging systems can be used to acquire CT data for reconstruction into image slices. These CT imaging systems include so-called "multi-slice" CT imaging systems, as well as earlier so-called "single-slice" CT imaging systems. Multi-slice CT imaging systems include a detector having multiple rows of detector pixels. The multiple rows of detector pixels are distributed along an axis of rotation of the CT imaging system, and the pixels in each row are arranged transaxially with respect to the axis of rotation or along an arc. CT data acquired by rotating an x-ray source and the multiple rows of detector pixels around an anatomical region is reconstructed to provide multiple transaxial image slices through the anatomical region. Single-slice CT imaging systems include a detector having a single row of detector pixels. CT data acquired by rotating an x-ray source and the single row of detector pixels around an anatomical region is reconstructed to provide a single transaxial image slice through the anatomical region.

[0004] Both multi-slice CT imaging systems and single-slice CT imaging systems can be operated in a so-called "helical" mode or in a so-called "step-and-shoot" mode. In the helical mode, the CT imaging system remains in the same axial position with respect to the anatomical region during image acquisition. In the step-and-shoot mode (also referred to as a "sequential" mode), the position of the CT imaging system or the anatomical region is stepped along the axis of rotation of the CT imaging system to provide multiple discrete positions of the CT imaging system with respect to the anatomical region. CT data for one or more image slices is acquired at each discrete position. The step-and-shoot mode increases the axial extent of the anatomical region from which CT data is acquired.

[0005] Both multi-slice CT imaging systems and single-slice CT imaging systems can be operated in conjunction with a gating protocol. For example, acquisition of CT data can be timed using a prospective ECG gate such that CT data is acquired only during cardiac phases in which cardiac motion is minimal. Alternatively, a retrospective ECG gate can be used to select portions of continuously acquired CT data for reconstruction such that the reconstructed CT data corresponds to cardiac phases in which cardiac motion is minimal. Such gating protocols reduce the amount of blurring in the image slices due to motion.

[0006] In addition to multi-slice imaging systems and single-slice CT imaging systems, other types of CT imaging systems can also be used to acquire CT data for reconstruction into image slices. For example, so-called helical CT imaging systems generate CT data while continuously translating the position of the CT imaging system relative to the anatomical region. The CT data represents a helical cross-section through the anatomical region. Often, the acquired CT data is interpolated prior to reconstruction of the CT data. The interpolated data can then be reconstructed to provide image slices through the anatomical region. In another example, a spectral CT imaging system can be used to acquire spectral CT data for reconstruction into image slices. In yet another example, an x-ray projection imaging system can be used to acquire CT data for reconstruction into image slices. X-ray projection imaging systems typically acquire projection data while holding the x-ray detector in a stationary position relative to the anatomical region. However, x-ray projection imaging systems can also acquire projection data while rotating their x-ray detector around the anatomical region. Projection data acquired from multiple angles around the anatomical region can be referred to as CT data. X-ray projection imaging systems include a two-dimensional array of detector pixels, so CT data for multiple image slices through the anatomical region is generated by rotating the x-ray detector of the x-ray imaging system around the rotation axis of the x-ray imaging system while holding the x-ray detector in a fixed position along the rotation axis. X-ray projection imaging systems can also operate in the step-and-shoot mode described above.

[0007] Image noise is inherently present in image slices generated by all types of CT imaging systems. Image noise presents a challenge when analyzing image slices, especially image slices acquired in low-dose imaging procedures. Accordingly, various image “denoising” methods have been developed. These image “denoising” methods include spatial domain filtering techniques employing linear or non-linear filters, deep learning algorithms, and transform domain filtering techniques (e.g., wavelet-based denoising).

[0008] However, there is still room for improvement in reducing noise in CT image slices.

[0009] Document US2019 / 311507A1 discloses an image processing method in which each slice of a plurality of slices of medical image data is adaptively processed to an estimated noise magnitude of the respective slice using a denoising mode selected according to a target noise magnitude of all slices to obtain consistently processed images across slices, different body size patients, and different reconstruction thicknesses. Cross-slice inconsistent noise level image data or processed images with a waxy / unnatural appearance due to unnecessary denoising and overcleaning of images are avoided.

[0010] Ma, Y. et al., "Low-Dose CT Image Denoising Using a Generative Adversarial Network With a Hybrid Loss Function for Noise Learning", IEEE ACCESS, USA, Vol. 8, 7 April 2020, pp. 67519-67529 discloses a noise learning generative adversarial network for low-dose CT denoising combining least square, structural similarity and LI loss. SUMMARY

[0011] According to one aspect of the disclosure, there is provided a computer- implemented method of reducing noise in a computed tomography, CT, image slice passing through an anatomical region. The method comprises: receiving CT data representing the image slice; adjusting an amount of noise in the image slice to provide a denoised image slice having a similar amount of noise; and outputting the denoised image slice.

[0012] The inventors have observed that a drawback of existing denoising techniques is that they result in image slices having different amounts of noise. The inventors have determined that this is due to differences in the amount of x-ray attenuation in the image slices. For example, in a cardiac imaging procedure, the image slices are generated through the heart at different positions along the head-foot axis of the patient (i.e. along the rotation axis of the CT imaging system). The slices intercept different parts of the patient's anatomy. For example, an image slice intercepting the upper end of the heart can intercept matter such as the aorta, and an image slice intercepting the upper end of the heart has a relatively lower amount of x-ray attenuation than an image slice intercepting the lower end of the heart, and an image slice intercepting the lower end of the heart can intercept matter such as the liver and diaphragm. These differences in the amount of x-ray attenuation in the image slices results in image slices having different amounts of noise. These differences in the amount of noise present a challenge to the physician in interpreting the image slices, as they hinder the physician's ability to distinguish between real image features and noise. Movement of the position of the patient's organs during the interval in which the CT data for the image slice is acquired also exacerbates these noise variations.

[0013] In the above method, the amount of noise in the image slice is adjusted so as to provide a denoised image slice having a similar amount of noise. This facilitates more accurate interpretation of the image slice.

[0014] Further aspects, features and advantages of the disclosure will become apparent from the following example description made with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1This is a flowchart illustrating an example of a computer-implemented method for reducing noise in CT image slices according to some aspects of this disclosure.

[0016] Figure 2 The illustration shows some aspects of this disclosure for reducing CT image slices 110. 1..i , A schematic diagram of an example of a system 200 with noise.

[0017] Figure 3 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure. 1..i The first example of image intensity values.

[0018] Figure 4 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure. 1..i The second example of image intensity values.

[0019] Figure 5 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure. 1..i The third example of image intensity values.

[0020] Figure 6 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure. 1..i The fourth example of image intensity values.

[0021] Figure 7 The illustration shows a reduction in CT image slices 110 according to some aspects of this disclosure. 1..i , A schematic diagram illustrating an example of a noise-based method. Detailed Implementation

[0022] Examples of this disclosure are provided with reference to the following description and accompanying drawings. In this specification, for purposes of explanation, numerous specific details of certain examples are set forth. References to “example,” “implementation,” or similar language in the specification mean that a feature, structure, or characteristic described in connection with the example is included in at least that example. It should also be understood that a feature described with respect to one example may also be used in another example, and for the sake of brevity, not all features need to be repeated in each example. For example, features described with respect to a computer-implemented method may be implemented in a corresponding manner in computer program products and systems.

[0023] In the following description, reference is made to a computer-implemented method for reducing noise in CT image slices passing through anatomical regions. In the referenced example, the anatomical region is the patient's heart. However, it should be understood that the method can be used alternatively to reduce noise in CT image slices passing through general anatomical regions, including CT image slices passing through anatomical regions such as the head, liver, etc.

[0024] This paper also references examples of using CT imaging systems to acquire CT data. In this regard, it should be understood that CT data can be generated by various types of imaging systems, including, for example, so-called multi-slice CT imaging systems, single-slice CT imaging systems, spiral CT imaging systems, spectral CT imaging systems, and X-ray projection imaging systems. It should also be noted that the methods disclosed herein can alternatively be used to reduce noise in image slices traversing anatomical regions in positron emission tomography (PET) and single-photon emission computed tomography (SPECT). Therefore, PET and SPECT data representing image slices can be processed in a similar manner to provide denoised image slices for these types of data.

[0025] It should be noted that the computer-implemented methods disclosed herein can be provided as a non-transitory computer-readable storage medium, the non-transitory computer-readable storage medium including computer-readable instructions stored thereon, which, when executed by at least one processor, cause the at least one processor to perform the method. In other words, the computer-implemented methods can be implemented in a computer program product. The computer program product can be provided by dedicated hardware or hardware capable of running software in conjunction with appropriate software. When provided by a processor, the functionality of the method features can be provided by a single dedicated processor, or by a single shared processor, or by multiple individual processors (some of which can be shared). The functionality of one or more of the method features can be provided, for example, by a processor shared within a networked processing architecture (e.g., client / server architecture, peer-to-peer network architecture, the Internet, or the cloud).

[0026] The explicit use of the terms "processor" or "controller" should not be construed as referring specifically to hardware capable of running software, but may implicitly include, but is not limited to, digital signal processor (DSP) hardware, read-only memory (ROM) for storing software, random access memory (RAM), non-volatile storage devices, etc. Furthermore, examples of this disclosure may take the form of a computer program product accessible from a computer-usable storage medium or a computer-readable storage medium, which provides program code for use by or in connection with a computer or any instruction execution system. For the purposes of this specification, a computer-usable storage medium or a computer-readable storage medium may be any means capable of including, storing, communicating, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device. The medium may be an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system or device or a propagation medium. Examples of computer-readable media include: semiconductor or solid-state memory, magnetic tape, removable computer disk, random access memory (RAM), read-only memory (ROM), rigid magnetic disk, and optical disk. Current examples of optical discs include: read-only optical discs "CD-ROM", rewritable optical discs "CD-R / W", Blu-ray discs, and DVDs.

[0027] It should also be noted that some of the operations performed in the computer-implemented methods disclosed herein can be implemented using artificial intelligence techniques. Suitable techniques may include machine learning techniques and deep learning techniques (e.g., neural networks). For example, one or more neural networks may be trained in a supervised manner, or in some cases in an unsupervised manner, to implement the operations performed in the computer-implemented methods disclosed herein.

[0028] As mentioned above, there is still room for improvement in reducing noise in CT image slices.

[0029] Figure 1 This is a flowchart illustrating an example of a computer-implemented method for reducing noise in CT image slices according to some aspects of this disclosure. Figure 2 The illustration shows some aspects of this disclosure for reducing CT image slices 110. 1..i , A schematic diagram of an example of a system 200 with noise in it. It should be noted that, regarding... Figure 1 The operations described in the flowchart in the diagram can also be performed by... Figure 2 The system 200 illustrated in the figure is executed by one or more processors 210. Similarly, the operations described with respect to the one or more processors 210 of system 200 can also be found in the reference. Figure 1 Execute within the described method. (See reference) Figure 1 Computer-implemented methods for reducing noise in CT image slices include: Receive S110 indicates image slice 110 1..i , CT data 120; Adjust S120 image slice 110 1..i , The amount of noise in the image is used to provide denoised image slices with similar noise levels. ;as well as Output S130 denoised image slices .

[0030] The inventors have observed that a drawback of existing denoising techniques is that they result in image slices with varying levels of noise. The inventors have determined that this is due to differences in X-ray attenuation within the image slices. For example, in a cardiac imaging procedure, image slices are generated at different locations along the patient's head-to-tail axis (i.e., along the rotation axis of the CT imaging system). The slices capture different portions of the patient's anatomy. For instance, an image slice capturing the upper part of the heart might capture material such as the aorta, and such a slice has relatively lower X-ray attenuation than one capturing the lower part, while a slice capturing the lower part might capture material such as the liver and diaphragm. These differences in X-ray attenuation within the image slices produce image slices with varying levels of noise. These differences in noise levels pose a challenge for physicians in interpreting image slices, as they hinder their ability to distinguish true image features from noise. The movement of the patient's organs during the intervals between acquiring CT data for image slices also exacerbates these noise variations.

[0031] In the method described above, the amount of noise in the image slices is adjusted to provide denoised image slices with similar noise levels. This helps to interpret the image slices more accurately.

[0032] To illustrate the point Figure 1 The problem solved by the method illustrated in the flowchart is... Figure 3 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure. 1..i , The first example of image intensity values. Figure 3 The image intensity value in the image represents an image slice 110 through the coronal plane of the heart. 1..i X-ray attenuation in the system. The rotation axis of the CT imaging system (also known as the z-axis of the imaging system) is via... Figure 3 The symbol 'z' is used for illustration. The CT image slice 110 in the illustrated example was acquired using step and capture modes. 1..i In this process, CT data from multiple slices are acquired in each of two steps, Step 1 and Step 2. (Reference) Figure 2The system 200 shown provides the position of the CT imaging system relative to the anatomical region by stepping the patient's position along the rotation axis z of the CT imaging system. 1、 Step 2 generates the representation. Figure 3 The CT image slice 110 shown 1..i CT data. At each of the discrete locations in Step 1 and Step 2, CT data of multiple image slices were acquired. Figure 3 Image slice 110 shown in the figure 1..i By reconstructing image slice 110 1..i The acquired CT data is used to generate image intensity in image slices spanning the coronal plane. CT image slice 110 1..i Extending to Figure 3 In the plotting plane, and the intensity in each slice represents the X-ray attenuation in the image slice at the depth of the coronal plane. Figure 3 Image slice 1101 in the illustration is generated towards the upper part of the heart and thus cuts off the aorta. The image slice in the central part of Step 2 cuts off the diaphragm (i.e., from...). Figure 3 The left side extends to Figure 3 (The curved boundary on the right). Figure 3 In a lower position (e.g., in image slice 110) i (In the image, slices were taken from the liver and diaphragm.)

[0033] Observe carefully Figure 3 The image intensity in the image slices illustrated in the diagram shows that the amount of noise varies between image slices. The amount of noise varies between image slices generated by the CT imaging system relative to each location in the anatomical region. For example, the noise level in the first image slice 1101 at location Step 1 is relatively higher than that in the last image slice at location Step 1. The amount of noise also varies between image slices generated by the CT imaging system relative to different locations in the anatomical region. For example, the noise level in the first image slice 1101 at location Step 1 is relatively higher than that in the last image slice 110 at location Step 2. i These noise variations pose a challenge to physicians interpreting image slices because they impair their ability to distinguish true image features from noise. The movement of patient organs during the intervals between acquiring CT data of image slices also exacerbates these noise variations.

[0034] exist Figure 4 to Figure 6 Image slice 110 in the image shown in the middle. 1..i Similar changes in noise levels can also be observed between image slices. Figure 4 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure.1..i The second example of image intensity values. Figure 5 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure. 1..i The third example of image intensity values. Figure 6 The illustration shows multiple CT image slices 110 of the heart according to some aspects of this disclosure. 1..i The fourth example of image intensity values. (Compared to...) Figure 3 Image slice 110 shown in the figure 1..i compared to, Figure 4 to Figure 6 The example shown in the figure is a CT image slice 110. 1..i It is generated at different locations in the coronal plane. For example, in... Figure 3 As illustrated in the image, between image slices generated by the imaging system at different locations relative to the anatomical region, Figure 4 to Figure 6 The amount of noise in the images shown in the diagram also varies. For example, the amount of noise in the first image slice 1101 at position Step 1 is relatively higher than the amount of noise in the last image slice at position Step 1. The amount of noise also varies between image slices generated at different locations of the imaging system relative to the anatomical region. For example, the amount of noise in the first image slice 1101 at position Step 1 is relatively higher than the amount of noise in the last image slice 110 at position Step 2. i The amount of noise in it.

[0035] The inventors have determined that these changes are due to differences in the amount of X-ray attenuation in the image slices.

[0036] Return to Figure 1 The method illustrated in the flowchart, in operation S110, receives a representation of image slice 110. 1..i CT data 120.

[0037] Generally, the CT data 120 received in operation S110 can be raw data (i.e., data not yet reconstructed into a volumetric or 3D image), or the CT data 120 received in operation S110 can be reconstructed image data (i.e., data already reconstructed into a volumetric image). CT data can represent one or more complete image slices across an anatomical region, or alternatively, CT data can represent portions of one or more image slices across an anatomical region. In the latter case, the CT data can, for example, represent only the central portion of the field of view of a CT imaging system containing the heart. This avoids the need to process CT data for entire image slices. CT data can also be referred to as volumetric data. The CT data 120 received in operation S110 can be generated by a CT imaging system, or, as described in more detail below, the CT data 120 received in operation S110 can be generated by an X-ray source and X-ray detector of an X-ray projection imaging system that rotates or steps around an anatomical region.

[0038] CT imaging systems generate CT data by rotating or stepping the X-ray source-detector arrangement around the anatomical region and acquiring X-ray attenuation data of the anatomical region from multiple rotational angles relative to the anatomical region. The CT data can then be reconstructed into a 3D image of the anatomical region. Examples of CT imaging systems that can be used to generate CT data 120 include cone-beam CT imaging systems, photon-counting CT imaging systems, dark-field CT imaging systems, and phase-contrast CT imaging systems. Figure 2 The diagram illustrates an example of a CT imaging system 220 that can be used to generate CT data 120 received in operation S110. As described above, the CT imaging system can be a multi-slice CT imaging system or a single-slice CT imaging system. The CT imaging system can operate in a circular mode, or in a step-and-shoot mode, or in a spiral mode. As an example, the CT data 120 can be generated by a CT5000 Ingenuity CT scanner sold by Best Philips Healthcare in the Netherlands.

[0039] As described above, the CT data 120 received in operation S110 can alternatively be generated by rotating or stepping an X-ray projection imaging system around the anatomical region using an X-ray source and X-ray detector. The X-ray projection imaging system may include a support arm, such as a so-called "C-arm" supporting the X-ray source and X-ray detector. Alternatively, the X-ray projection imaging system may include a support arm with a different shape than this example, such as an O-arm. Other types of X-ray projection imaging systems may be used alternatively, where the X-ray source and X-ray detector are mounted or supported in different ways. Unlike CT imaging systems, X-ray projection imaging systems typically generate X-ray attenuation data of the anatomical region, where the X-ray source and X-ray detector are in a stationary position relative to the anatomical region. In contrast to the volumetric data generated by a CT imaging system, the X-ray attenuation data can be referred to as projection data. The X-ray attenuation data generated by an X-ray projection imaging system is typically used to generate 2D images of the anatomical region. However, an X-ray projection imaging system can generate CT data (i.e., volumetric data) by rotating or stepping its X-ray source and X-ray detector around the anatomical region and acquiring projection data of the anatomical region from multiple rotational angles relative to the anatomical region. The volumetric image can then be reconstructed from the projection data acquired from the multiple rotational angles using image reconstruction techniques in a manner similar to how a volumetric image is reconstructed using X-ray attenuation data acquired from a CT imaging system. Therefore, the CT data 120 received in operation S110 can be generated by a CT imaging system, or alternatively, the CT data 120 received in operation S110 can be generated by an X-ray projection imaging system. An example of an X-ray projection imaging system that can be used to generate CT data 120 is the Azurion7 X-ray projection imaging system sold by Best Philips Healthcare, Netherlands.

[0040] In some examples described in more detail below, the CT data 120 received in operation S110 includes spectral CT data. The spectral CT data is defined in several different energy ranges DE. 1..m X-ray attenuation within the anatomical region in each energy interval. Generally, there can be two or more energy intervals; that is, m is an integer and m≥2. At this point, the spectral CT data 120 received in operation S110 can be generated by a spectral CT imaging system or a spectral X-ray projection imaging system. In the latter case, the spectral CT data can be acquired as described above by rotating or stepping the X-ray source and X-ray detector of the spectral X-ray projection imaging system around the anatomical region. More generally, the spectral CT data 120 received in operation S110 can be generated by a spectral X-ray imaging system.

[0041] The ability to generate X-ray attenuation data across multiple energy ranges distinguishes spectral X-ray imaging systems from conventional X-ray imaging systems. By processing data from multiple different energy ranges, it is possible to distinguish media with similar X-ray attenuation values ​​when measured within a single energy range (which is indistinguishable in conventional X-ray attenuation data). Examples of spectral X-ray imaging systems that can be used to generate spectral CT data 120 received in operation S110 include: cone-beam spectral X-ray imaging systems, photon-counting spectral X-ray imaging systems, dark-field spectral X-ray imaging systems, and phase-contrast spectral X-ray imaging systems. An example of a spectral CT imaging system that can be used to generate spectral CT data 120 is the Spectral CT7500 sold by Best Philips Healthcare, Netherlands.

[0042] Generally, spectral CT data 120 can be generated by a spectral X-ray imaging system with various configurations including an X-ray source and an X-ray detector. For example, the X-ray source of the spectral X-ray imaging system may include multiple monochromatic sources or one or more multicolor sources, and the X-ray detector of the spectral X-ray imaging system may include a common detector for detecting multiple different X-ray energy ranges, or each detector may detect a different X-ray energy range DE. 1..m Multiple detectors, or multilayer detectors in which X-rays with energies in different X-ray energy ranges are detected by corresponding layers, or photon counting detectors that bin detected X-ray photons into one of multiple energy ranges based on the individual energy of the detected X-ray photons. Other combinations of the aforementioned X-ray sources and detectors can also be used to provide spectral CT data 120. For example, in one configuration, an X-ray source-detector pair is mounted to the gantry at a rotational offset position about a rotation axis. In this configuration, each source-detector pair operates independently, thereby eliminating the need to sequentially switch different X-ray sources emitting X-rays in different energy ranges.

[0043] Generally, the CT data 120 received in operation S110 can be received via any form of data communication (including wired, optical, and wireless communication). As some examples, when using wired or optical communication, communication can be conducted via signals transmitted over cables or optical fibers; while when using wireless communication, communication can be conducted, for example, via radio frequency or optical signals. The CT data 120 received in operation S110 can be received from various sources. For example, the CT data 120 can be received from an imaging system (e.g., one of the imaging systems described above). Alternatively, the CT data 120 can be received from another source (e.g., a computer-readable storage medium, the Internet, or the cloud).

[0044] Return to Figure 1The method illustrated in the flowchart shows that in operation S120, image slice 110 is adjusted. 1..i The amount of noise in the image is adjusted to provide denoised image slices with similar noise levels. Generally, the adjustment operation S120 may include adjusting the amplitude of noise in the image slice, so that the denoised image slice... The noise has similar amplitude. Adjustment operation S120 can be performed by adjusting the noise amplitude while maintaining the same two-dimensional noise distribution in each image slice. The noise amplitude in the denoised image slice can be in the range of, for example, approximately ±5%, approximately ±10%, or approximately ±20%. Adjustment operation S120 can be performed to maintain the shape of the noise power spectrum in each image slice. Various techniques for performing adjustment operation S120 will be described below. In some of these techniques, image slice 110 is adjusted based on the X-ray attenuation in the image slice. 1..i The amount of noise in each image slice. In another method, image slice 110 is adjusted based on the amount of noise in the image slice. 1..i The amount of noise in each image slice. In some techniques, the amount of noise is adjusted in the raw CT data, while in others, the amount of noise is adjusted in the reconstructed CT image slices.

[0045] In operation S130, a denoised image slice is output. In this operation, the denoised image slice can be output to a display device (e.g., a monitor), or in other ways to a printer or computer-readable storage medium.

[0046] refer to Figure 3 The described operation results in denoised image slices with similar noise levels. This helps to interpret the image slices more accurately.

[0047] As described above, various techniques are envisioned for performing the adjustment operation S120. In one example, S120 adjusts image slice 110. 1..i , The operation of noise levels includes: For each image slice 110 1..i , : Estimating the noise distribution in an image slice ; Scaling factor α i The noise distribution in the estimated image slice is applied to provide a scaled noise distribution in the image slice. ; Subtract the scaled noise distribution from the image slice. To provide denoised image slices ;and Here, the scaling factor for each image slice Selected to provide denoised image slices with similar noise levels.

[0048] refer to Figure 7 To describe these operations, Figure 7 The illustration shows a reduction in CT image slices 110 according to some aspects of this disclosure. 1..i , A schematic diagram illustrating an example of a noise reduction method. Figure 7 The left side of the image shows a slice. Along the direction Figure 7 The arrow on the right, then for each image slice Noise estimation Then the scaling factor The noise distribution in each estimated image slice is applied to provide a scaled noise distribution in the image slice. Choose a scaling factor for each image slice. This provides denoised image slices with similar noise levels. It should be noted that in some cases, the scaling factor... The scaling factor can have positive values; in some cases, it can have zero values; and in others, it can have negative values. Figure 7 In the example shown in the diagram, the scaling factor for each image slice is calculated based on the X-ray attenuation in the image slice. The scaling factor can, for example, be proportional to the average X-ray attenuation in the image slice.

[0049] Various techniques have been envisioned for calculating X-ray attenuation in each image slice. For example, X-ray attenuation could be calculated based on the original CT data 120 or on reconstructed image slices, such as... Figure 7 As shown. Alternatively, X-ray attenuation can be estimated from camera image data, as described in more detail below.

[0050] In one example, the X-ray attenuation in each image slice is calculated using the following operation: Reconstructing CT data representing image slices 120; and Calculate the average X-ray attenuation in the reconstructed image slices; In another example, the X-ray attenuation in each image slice is calculated using the following method: X-ray attenuation in the image slice is estimated based on the raw CT data 120 representing the image slice; In another example, the X-ray attenuation in each image slice is calculated using the following method: The received CT data 120 is reconstructed to provide a reconstructed volumetric CT image representing image slices; and The average X-ray attenuation in each image slice is calculated based on the reconstructed volumetric CT image; In another example, the X-ray attenuation in each image slice is calculated using the following method: Receive second CT data 120 representing the anatomical region; Reconstruct the second CT data 120 to provide a second reconstructed volumetric CT image; and The average X-ray attenuation in the image slice corresponding to the image slice in the received CT data 120 is calculated based on the second reconstructed volumetric CT image.

[0051] In this example, the second reconstructed volumetric CT image may be a localization view CT image (also known as a reconnaissance scan) representing the anatomical region. The second CT data may include a relatively lower spatial resolution than CT data 120. Therefore, the reconstructed localization view image may have a relatively lower spatial resolution than image slices reconstructed from the CT data. Alternatively, the second CT data may include the same spatial resolution as CT data 120, and the second CT data may be spatially binned to a relatively lower spatial resolution before reconstruction, thereby providing a reconstructed localization view image with a relatively lower spatial resolution than image slices reconstructed from the CT data. Alternatively, the second CT data may be provided according to the aforementioned imaging procedure on the anatomical region.

[0052] In another example, the X-ray attenuation in each image slice is calculated using the following method: Receive camera image data 130 representing the anatomical region; and Camera image data 130 was used to estimate the average X-ray attenuation in the image slices.

[0053] In this example, camera image data 130 can be provided by camera 240 configured to view anatomical areas, such as Figure 2 As shown. The camera can be a visible light camera or a depth camera. The average X-ray attenuation in each image slice can be estimated using the camera image data 130 by: fitting an anatomical model representing an anatomical region to the camera image data, wherein the anatomical model includes X-ray attenuation values ​​representing one or more anatomical regions; determining the location of the image slice in the anatomical model, for example, by registering the location of the CT imaging system to the anatomical model; and determining the average X-ray attenuation in each image slice based on the X-ray attenuation values ​​in the anatomical model.

[0054] In another example, the above technique can be used to calculate the X-ray attenuation in each image slice, and a smoothing function can also be applied to the X-ray attenuation value in the image slice. The smoothed value of the X-ray attenuation in the image slice is then used to calculate the scaling factor for each image slice. This helps reduce potential anomalous shifts in noise levels between image slices.

[0055] Finally, Figure 7 On the right side, by extracting from each image slice Subtract the scaled noise distribution from the image slices. To generate denoised image slices Then output denoised image slices. .

[0056] In the alternative method, the scaling factor for each image slice is not determined based on the X-ray attenuation within the image slice. Instead, the scaling factor for each image slice can be calculated based on the amount of noise in each image slice. Therefore, in one example, the scaling factor for each image slice is determined by the following operation. : The received CT data 120 representing image slices is reconstructed to provide one or more reconstructed CT images representing the image slices; Calculate the amount of noise in each image slice based on one or more reconstructed CT images; and The scaling factor is set for each image slice based on the amount of noise in the image slice.

[0057] In this example, a single-volume CT image can be reconstructed based on CT data 120 and the amount of noise determined at the location corresponding to the image slice. Alternatively, individual image slices can be reconstructed based on CT data 120 and the amount of noise determined in each image slice.

[0058] refer to Figure 7 The example shown in the figure envisions a method for estimating the noise distribution (i.e., parameters) in an image slice. Various techniques for operating on the value of noise. In one example, the operations for estimating the noise distribution in an image slice include: Slice the image into 110 slices. 1..i , Input into the neural network; Generate an image slice 110 in response to the input. 1..i , The estimated noise distribution ;and The neural network is trained to generate an estimated noise distribution for an image slice using training data and ground truth data. The training data includes multiple image slices, and the ground truth data includes an estimated noise distribution for each image slice in the training data.

[0059] In this example, the neural network is trained to generate the estimated noise distribution for an image slice by: For each image slice in the training data: Image slices are input into a neural network; The noise distribution estimated for an image slice is generated using a neural network; and The parameters of the neural network are adjusted based on the difference between the estimated noise distribution for an image slice generated by the neural network and the corresponding estimated noise distribution for the image slice from the benchmark ground truth data; and Repeat the input, generation, and adjustment until the stopping criterion is met.

[0060] In this example, the neural network can be provided by various architectures, such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), or transformers. As mentioned above, the training process of a neural network involves tuning its parameters. Parameters (or more specifically, weights and biases) control the operation of the activation functions in the neural network. In supervised learning, the training process automatically tunes the weights and biases so that when input data is presented, the neural network accurately provides the corresponding expected output data. To this end, the value of a loss function, or error, is calculated based on the difference between the predicted output data and the expected output data. The value of the loss function can be calculated using functions such as negative log-likelihood loss, mean absolute error (or L1 norm), mean squared error, root mean square error (or L2 norm), Huber loss, or (binary) cross-entropy loss. During training, the value of the loss function is typically minimized, and training terminates when the value of the loss function meets a stopping criterion. Sometimes, training terminates when the value of the loss function meets one or more of a plurality of criteria.

[0061] Various methods are known for solving the loss minimization problem, such as gradient descent, quasi-Newton methods, etc. Various algorithms have been developed to implement these methods and their variants, including but not limited to: stochastic gradient descent (SGD), batch gradient descent, mini-batch gradient descent, Gauss-Newton method, Levenberg-Marquardt, momentum method, Adam, Nadam, Adagrad, Adadelta, RMSProp, and the Adamax "optimizer". These algorithms use the chain rule to calculate the derivative of the loss function with respect to the model parameters. This process is called backpropagation because the derivative is calculated starting from the last layer or output layer and moving towards the first layer or input layer. These derivatives tell the algorithm how the model parameters must be adjusted to minimize the error function. That is, the adjustment of the model parameters starts from the output layer and proceeds backward through the network until it reaches the input layer. In the first training iteration, the initial weights and biases are often randomized. Then the neural network predicts the output data, which is also randomized. Backpropagation is then used to adjust the weights and biases. The training process is performed iteratively by adjusting the weights and biases in each iteration. Training terminates when the error or difference between the predicted output data and the expected output data is within an acceptable range of the training data or some validation data. The neural network can then be deployed, and the trained neural network uses the trained values ​​of its parameters to predict new input data. If the training process is successful, the trained neural network accurately predicts the expected output data based on the new input data.

[0062] In an alternative technique, the operation of estimating the noise distribution in an image slice includes: For image slice 110 1..i , Each image slice in: Applying a noise filter to an image slice; and Subtract the corresponding image intensity value from the filtered image slice from the image intensity value in the image slice to provide the noise distribution in the image slice.

[0063] In this example, the noise filter can employ techniques such as total variation denoising, nonlocal mean “NLM” denoising, block matching 3D “BLM3D” denoising, or iterative techniques (e.g., the Rudin-Osher-Fatemi “ROF” model), or the Bregman algorithm.

[0064] return Figure 1 The method illustrated in the flowchart, in another example, involves adjusting the amount of noise in an S120 image slice, including: A noise filter is applied to the received CT data 120, wherein the noise filter is configured to provide denoised image slices with similar noise levels.

[0065] Therefore, in this example, the noise filter directly provides a denoised image slice, rather than as referenced above. Figure 7 That is, by subtracting the scaled noise distribution from each image slice To obtain denoised image slices. In this example, the following techniques can be used to provide denoised image slices, such as: total variation denoising, non-local means "NLM" denoising, block matching 3D "BLM3D" denoising, or iterative techniques (e.g., the Rudin-Osher-Fatemi "ROF" model), or the Bregman algorithm.

[0066] As described above, CT data 120 for performing the aforementioned operations can be acquired by operating the CT imaging system in various modes. These modes include so-called stepping and imaging modes. Thus, in one example, CT data 120 is generated by stepping the CT imaging system 220 or the anatomical region along the rotation axis A-A' of the CT imaging system to provide multiple discrete positions of the CT imaging system relative to the anatomical region. In this example, image slices 110 are acquired at each discrete position. 1..i , CT data of one or more image slices in 120.

[0067] When acquiring CT data 120 in the aforementioned stepping and imaging modes, it can also be based on one or more image slices 110 at discrete locations. 1..i , The estimated X-ray attenuation is used to adjust one or more image slices for acquiring each discrete location.110 1..i , The CT data is based on an X-ray dose of 120. This can be based on one or more image slices at discrete locations. 1..i , The X-ray dose can be adjusted by the average X-ray attenuation in the image. As an example, this can be done by adjusting one or more image slices at discrete locations. 1..i , The X-ray dose is adjusted proportionally to the average X-ray attenuation. Therefore, a relatively higher X-ray dose is used at discrete locations with relatively higher X-ray attenuation during the acquisition of CT data 120 compared to discrete locations with relatively lower X-ray attenuation. This helps reduce the difference in noise levels in the images (one or more) acquired at each discrete location, thus alleviating the burden of the aforementioned denoising methods. The adjustment of the X-ray dose contrasts with known step-and-shoot imaging procedures, where the X-ray dose, determined by factors such as exposure time, X-ray tube voltage, and X-ray tube current, is set to the same level for all steps in the imaging procedure. The X-ray dose is typically set based on factors such as the patient's weight and body size.

[0068] In this example, one or more image slices 110 at each discrete location can be estimated using a technique similar to the one described above, for example, by the following operation. 1..i , X-ray attenuation in: Receive second CT data 120 representing the anatomical region; Reconstruct the second CT data 120 to provide a second reconstructed volumetric CT image; and Calculate the image slices 110 at discrete locations based on the second reconstructed volumetric CT image. 1..i , The average X-ray attenuation in the corresponding volume region; Alternatively, in another example, one or more image slices at discrete locations can be estimated by the following operation 110 1..i , X-ray attenuation in: Receive camera image data 130 representing the anatomical region; and Use camera image data 130 to estimate one or more image slices 110 at discrete locations. 1..i , The average X-ray attenuation in the medium.

[0069] When acquiring CT data 120 in the aforementioned stepping and imaging modes, it can alternatively be based on one or more image slices 110 at discrete locations. 1..i , The estimated noise level is used to adjust one or more image slices acquired at each discrete location.110 1..i , The CT data 120 X-ray dose. In this case, one or more image slices 110 at each discrete location can be estimated by the following operation. 1..i , Noise level in: Receive second CT data 120 representing the anatomical region; Reconstruct the second CT data 120 to provide a second reconstructed volumetric CT image; and Calculate the image slices 110 at discrete locations based on the second reconstructed volumetric CT image. 1..i , The average noise level in the corresponding volume region; Alternatively, one or more image slices at discrete locations can be estimated by the following operation 110 1..i , X-ray attenuation in: Receive camera image data 130 representing the anatomical region; and Camera image data 130 was used to estimate the average X-ray attenuation in the image slices.

[0070] In another example, a computer program product is provided. The computer program product includes instructions that, when executed by one or more processors 210, cause the one or more processors to perform a reduction of computed tomography (CT) image slices 110 across an anatomical region. 1..i , A method for addressing noise in [the environment]. The method includes: Receive S110 indicates image slice 110 1..i , CT data 120; Adjust S120 image slice 110 1..i , The amount of noise in the image is used to provide denoised image slices with similar noise levels. ;as well as Output S130 denoised image slices .

[0071] In another example, a method for reducing computed tomography CT image slices 110 across anatomical regions is provided. 1..i , System 200 for noise in the system. The system includes one or more processors 210, the one or more processors 210 being configured to: Receive S110 indicates image slice 110 1..i , CT data 120; Adjust S120 image slice 110 1..i , The amount of noise in the image is used to provide denoised image slices with similar noise levels. ;as well as Output S130 denoised image slices .

[0072] Figure 2 The diagram illustrates an example of system 200. It should be noted that system 200 may also include one or more of the following: an imaging system 220 for generating CT data 120 received in operation S110, for example, Figure 4 The diagram shows a CT imaging system 220; a hospital bed 230 for supporting the patient and / or stepping the patient's position along the rotation axis z of the CT imaging system 220; a camera 240 for acquiring camera image data 130; and a display (…). Figure 2 (Not shown in the figure), it is used to display data output by one or more processors 210 (e.g., image slice 110). 1..i (e.g., denoised image slicing, etc.) and user input devices configured to receive user input, such as keyboards, mice, touchscreens, etc.

[0073] Below are some examples of this disclosure in the enumeration list: Example 1: A method for reducing computed tomography (CT) image slices across anatomical regions (110). 1..i , A computer-implemented method for dealing with noise in a computer, the method comprising: Receive (S110) indicates that the image slice (110) 1..i , CT data (120); Adjust (S120) the image slice (110) 1..i , The amount of noise in the image is used to provide denoised image slices with similar noise levels. );as well as Output (S130) the denoised image slice ( ).

[0074] Example 2, a computer-implemented method according to Example 1, wherein the adjustment (S120) of the image slice (110) 1..i , The noise levels in the data include: For the image slice (110) 1..i , Each image slice in ) Estimate the noise distribution in the image slice ( ); The scaling factor (α) i ) is applied to the estimated noise distribution in the image slice to provide a scaled noise distribution in the image slice ( ); Subtract the scaled noise distribution from the image slice ( ) to provide denoised image slices ( );and Wherein, the scaling factor for each image slice ( () was selected to provide denoised image slices with similar noise levels.

[0075] Example 3, the computer-implemented method according to Example 2, wherein estimating the noise distribution in the image slice includes: Slice the image (110) 1..i , The input is fed into the neural network; and Generate a slice of the image (110) in response to the input. 1..i , The estimated noise distribution () );and The neural network is trained to generate an estimated noise distribution for the image slice using training data and ground truth data. The training data includes multiple image slices, and the ground truth data includes an estimated noise distribution for each image slice in the training data.

[0076] Example 4: A computer-implemented method according to Example 3, wherein the neural network is trained to generate the estimated noise distribution for the image slice by: For each image slice in the training data: The image slices are input into the neural network; The neural network is used to generate an estimated noise distribution for the image slice; and The parameters of the neural network are adjusted based on the difference between the estimated noise distribution for the image slice generated by the neural network and the corresponding estimated noise distribution for the image slice from the benchmark ground truth data; and Repeat the input, generation, and adjustment until the stopping criterion is met.

[0077] Example 5, the computer-implemented method according to Example 2, wherein estimating the noise distribution in the image slice includes: For the image slice (110) 1..i , Each image slice in ) Apply a noise filter to the image slice; and The corresponding image intensity value in the filtered image slice is subtracted from the image intensity value in the image slice to provide the estimated noise distribution in the image slice.

[0078] Example 6. A computer-implemented method according to any one of Examples 2-5, wherein the scaling factor ( ) is applied to each image slice. The value is calculated based on the X-ray attenuation in the image slice.

[0079] Example 7. A computer-implemented method according to Example 6, wherein the X-ray attenuation in each image slice is calculated by one of the following operations: Reconstruct the CT data (120) representing the image slices; and Calculate the average X-ray attenuation in the reconstructed image slices; or The X-ray attenuation in the image slice is estimated based on the original CT data (120) representing the image slice; or The received CT data (120) is reconstructed to provide a reconstructed volumetric CT image representing the image slices; and The average X-ray attenuation in each image slice is calculated based on the reconstructed volumetric CT image; or Receive second CT data (120) representing the anatomical region; Reconstruct the second CT data (120) to provide a second reconstructed volumetric CT image; and The average X-ray attenuation in the image slice corresponding to the image slice in the received CT data (120) is calculated based on the second reconstructed volumetric CT image; or Receive camera image data (130) representing the anatomical region; and The camera image data (130) is used to estimate the average X-ray attenuation in the image slice.

[0080] Example 8: A computer-implemented method according to Example 2, wherein the scaling factor ( ) is applied to each image slice. The value is determined through the following operations: The received CT data representing the image slices (120) are reconstructed to provide one or more reconstructed CT images representing the image slices; The noise level in each image slice is calculated based on one or more reconstructed CT images; and The scaling factor is set for each image slice based on the amount of noise in the image slice.

[0081] Example 9. A computer-implemented method according to Example 1, wherein adjusting (S120) the amount of noise in the image slice includes: A noise filter is applied to the received CT data (120), wherein the noise filter is configured to provide the denoised image slices with similar noise levels.

[0082] Example 10: A computer-implemented method according to any of the foregoing examples, wherein the anatomical region includes the heart of the object, or the head of the object, or the liver of the patient.

[0083] Example 11, a computer-implemented method according to any of the preceding examples, wherein the CT data (120) includes spectral CT data, which defines X-ray attenuation in the anatomical region within each of a plurality of different energy ranges.

[0084] Example 12. A computer-implemented method according to any of the foregoing examples, wherein the CT data (120) is generated by: stepping the position of the CT imaging system (220) or the anatomical region along the rotation axis (A-A') of the CT imaging system to provide a plurality of discrete positions of the CT imaging system relative to the anatomical region; and acquiring images for the image slices (110) at each discrete position. 1..i , The CT data (120) of one or more image slices in the image.

[0085] Example 13. A computer-implemented method according to any of the foregoing examples, wherein denoising image slicing ( They have similar noise amplitudes.

[0086] Example 14: A computer program product including instructions that, when executed by one or more processors (210), cause the one or more processors to perform the method according to any one of Examples 1-13.

[0087] Example 15, A method for reducing computed tomography (CT) image slices across anatomical regions (110 1..i , A system (200) for handling noise, the system comprising one or more processors (210) configured to: Receive (S110) indicates that the image slice (110) 1..i , CT data (120); Adjust (S120) the image slice (110) 1..i , The amount of noise in the image is used to provide denoised image slices with similar noise levels. );as well as Output (S130) the denoised image slice ( ).

[0088] It should be understood that the examples above are merely illustrative of this disclosure and are not restrictive. Other examples are also contemplated. For instance, examples described with respect to computer-implemented methods may also be provided in a corresponding manner by a computer program product, a computer-readable storage medium, or a system. It should be understood that features described with respect to any example may be used alone or in combination with other described features, and may be used in combination with one or more features of another example, or in combination with other examples. Furthermore, equivalent and modified embodiments not described above may be employed without departing from the scope of the invention as defined in the appended claims. In the claims, the word "comprising" does not exclude other elements or operations, and the words "a" or "an" do not exclude a plurality. The fact that certain features are recited in dissimilar dependent claims does not in itself indicate that combinations of these features cannot be advantageously used. No reference numerals in the claims should be construed as limiting their scope.

Claims

1. A computer-implemented method of reducing noise in computed tomography (CT) image slices (110 1..i , ) through an anatomical region, the method comprising: receiving (S110) CT data (120) representing the image slice (110 1..i , ) adjusting (S120) an amount of noise in the image slice (110 1..i , ) to provide a de-noised image slice (120) having a similar amount of noise ) and outputting (S130) the denoised image slice (S120) ); and wherein the adjusting (S120) the amount of noise in the image slice (110 1..i , ) comprises: for each image slice (110 1..i , ) of the image slices (110 estimating a noise distribution in the image slice (1000) ) applying a scaling factor (a i ) to the estimated noise distribution in the image slice to provide a scaled noise distribution in the image slice ( ) subtracting the scaled noise distribution in the image slice from the image slice ( ) to provide a denoised image slice ( ); and wherein the scaling factor (s) for each image slice is chosen to provide denoised image slices with similar amounts of noise. ) is chosen to provide denoised image slices with similar amounts of noise.

2. The computer-implemented method of claim 1, wherein, The estimated noise distribution in the image slice ( This includes estimating the spatial distribution of noise across the image slices. ); and wherein, the noise distribution in the estimated image slice ( This includes the spatial distribution of noise across the image slices. ), and wherein the image slice ( The scaled noise distribution in ) This includes the spatial distribution of noise across the image slices. ).

3. The computer-implemented method of claim 1 or claim 2, wherein, said estimating a noise distribution in the image slice comprises: inputting the image slices (110 1..i , ) into a neural network; and Generate a slice of the image (110) in response to the input. 1..i , The estimated noise distribution () );and wherein the neural network is trained to generate an estimated noise distribution for the image slice using training data and ground truth data, the training data comprising a plurality of image slices, the ground truth data comprising, for each image slice in the training data, a corresponding estimated noise distribution for the image slice.

4. The computer-implemented method of claim 3, wherein, said neural network is trained to generate an estimated noise distribution for the image slice by: for each image slice in the training data: inputting the image slice into the neural network; generating, using the neural network, an estimated noise distribution for the image slice; and adjusting parameters of the neural network based on a difference between the estimated noise distribution for the image slice generated by the neural network and the corresponding estimated noise distribution for the image slice from the ground truth data; and repeating the inputting, the generating and the adjusting until a stopping criterion is met.

5. The computer-implemented method of claim 1, wherein, said estimating a noise distribution in the image slice comprises: for each image slice of the image slices (110 1..i , ) applying a noise filter to the image slice; and subtracting, from image intensity values in the image slice, corresponding image intensity values in the filtered image slice to provide an estimated noise distribution in the image slice.

6. The computer-implemented method of any one of claims 1-5, wherein, The scaling factor for each image slice ( The value is calculated based on the X-ray attenuation in the image slice.

7. The computer-implemented method of claim 6, wherein, said X-ray attenuation in each image slice is calculated by one of: reconstructing the CT data (120) representing the image slice; and calculating an average X-ray attenuation in the reconstructed image slice; or estimating the X-ray attenuation in the image slice from raw CT data (120) representing the image slice; or reconstructing the received CT data (120) to provide a reconstructed volumetric CT image representing the image slice; and calculating an average X-ray attenuation in each image slice from the reconstructed volumetric CT image; or receiving second CT data (120) representing the anatomical region; reconstructing the second CT data (120) to provide a second reconstructed volumetric CT image; and calculating an average X-ray attenuation in an image slice corresponding to the image slice in the received CT data (120) from the second reconstructed volumetric CT image; or receiving camera image data (130) representing the anatomical region; and using the camera image data (130) to estimate an average X-ray attenuation in the image slice.

8. The computer-implemented method of claim 1, wherein, The scaling factor for each image slice ( The value is determined through the following operations: reconstructing received CT data (120) representing the image slice to provide one or more reconstructed CT images representing the image slice; calculating a noise amount in each image slice from the one or more reconstructed CT images; and setting the scaling factor for each image slice based on the noise amount in the image slice.

9. The computer-implemented method of claim 1, wherein, said adjusting (S120) the noise amount in the image slice comprises: applying a noise filter to the received CT data (120), and wherein the noise filter is configured to provide the denoised image slice with a similar amount of noise.

10. The computer-implemented method of any preceding claim, wherein, the anatomical region comprises a heart of the subject, or a head of the subject, or a liver of the patient.

11. The computer-implemented method of any preceding claim, wherein, the CT data (120) comprises spectral CT data defining X-ray attenuation in the anatomical region within each of a plurality of different energy intervals.

12. The computer-implemented method of any preceding claim, wherein, the CT data (120) is generated by stepping a position of a CT imaging system (220) or the anatomical region along a rotation axis (A-A') of the CT imaging system to provide a plurality of discrete positions of the CT imaging system relative to the anatomical region; and acquiring the CT data (120) for one or more image slices in the image slices (110 1..i , ) at each discrete location.

13. The computer-implemented method of any preceding claim, wherein, Denoising image slicing ( They have similar noise amplitudes.

14. A computer program product comprising instructions which, when executed by one or more processors (210), cause the one or more processors to perform the method of any one of claims 1-13.

15. A method for reducing the number of computed tomography CT image slices (110) across anatomical regions. 1..i , A system (200) for handling noise, the system comprising one or more processors (210) configured to: receiving (S110) CT data (120) representing the image slice (110 1..i , ) adjusting (S120) an amount of noise in the image slice (110 1..i , ) to provide a de-noised image slice (120) having a similar amount of noise ); and outputting (S130) the denoised image slice ).

Images