System for reconstructing an image of an object
By using projection data in the CT system to generate and modify the basic image as the starting image of the iterative reconstruction algorithm, the problem of computational complexity of the iterative reconstruction algorithm is solved, and faster image reconstruction and reduced computing resources are achieved.
Patent Information
- Application Number
- CN201980078285.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Priority Date
- 2018-11-28
- Filing Date
- 2019-11-19
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2039-11-19
AI Technical Summary
In medical imaging, especially in CT systems, the computational complexity of iterative reconstruction algorithms leads to high computational costs and requires reduced computing resources and time.
By providing projection data, basic image generation unit and modification unit, the modified image is generated as the starting image of the iterative reconstruction algorithm, and the basic image is modified using the image value of the voxel, reducing the calculation steps of iterative reconstruction.
The iterative reconstruction algorithm is realized faster convergence, reducing the computing time and resources required to reconstruct images, and improving the efficiency of image reconstruction.
Smart Images

Figure CN113168721B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to a system, a method and a computer program for reconstructing an image of an object. Background Art
[0002] In medical imaging, in particular in medical imaging using CT systems, iterative reconstruction algorithms are often used to reconstruct images of patients. These iterative reconstruction algorithms involve a large computational complexity. Therefore, in order to reconstruct an image using an iterative reconstruction algorithm, the computational cost (i.e., the necessary computational time and the necessary computational resources) is quite high. Therefore, it would be advantageous to provide a system that allows reducing the computational cost when reconstructing an image using an iterative reconstruction algorithm.
[0003] The article “A hybrid metal artifact reduction algorithm for x-ray CT” by Zhang Yanbo et al. (Medical Physics, AIP, Vol. 10, No. 4, pp. 41910-41910) discloses creating a good prior image so that forward projection can well replace projection data affected by metal.
[0004] The article "Suppression of metal streak artifacts in CT using MAP reconstruction procedure" by Lemmens C. et al. (the 2006 IEEE Nuclear Science Symposium Conference Record, pp. 3431-3437) discloses a reconstruction process that reduces streak artifacts and can improve the diagnostic value of CT. The process begins with MAP reconstruction using an iterative reconstruction algorithm and multimodal priors.
[0005] The article "Acceleration of ML iterative algorithms for CT by the use of fast start images" by Kevin M. Brown et al. (the proceedings of SPIE, Vol. 8313, pp. 831339) discloses a strategy for accelerating the maximum likelihood (ML) iterative reconstruction algorithm for CT by selecting a start image that is closer to the solution of the ML algorithm than the commonly used filtered backprojection image.
[0006] The article "Penalized maximum likelihood reconstruction x-ray differential phase-contrast tomography" by Brendel Bernhard et al. (Medical Physics, Vol. 43, pp. 188-194) discloses deriving a penalized maximum likelihood reconstruction algorithm to directly reconstruct attenuation, phase and scatter images from measured detector values of differential phase contrast acquisitions. Summary of the invention
[0007] It is an object of the present invention to provide a system, a method and a computer program allowing to reduce the computational costs when reconstructing an image of an object using an iterative reconstruction algorithm.
[0008] In a first aspect of the present invention, a system for reconstructing an image of an object is proposed, wherein the system comprises: a) a projection data providing unit for providing projection data of the object, wherein the projection data has been acquired by a CT system; b) a base image generating unit for generating a base image based on the projection data; c) a modification unit for modifying the base image to generate a modified image, wherein, in order to modify the base image, the modification unit is suitable for modifying the image values of voxels of the base image based on the image values of the voxels; d) an image reconstruction unit for reconstructing the image of the object using an iterative reconstruction algorithm based on the projection data, the iterative reconstruction algorithm using the modified image as a starting image for the iterative reconstruction.
[0009] Since the base image generation unit generates the base image based on the projection data, a base image based on the same data as the image that should be reconstructed can be generated. Therefore, the base image gives a good first approximation of the image that should be reconstructed. In addition, since the modification unit is suitable for modifying the base image, wherein, during the modification, the image values of the voxels of the base image are modified based on the image values of the corresponding voxels, the base image can be modified such as to form an optimal starting image for the selected iterative reconstruction. For example, if the selected iterative reconstruction converges only slowly in the presence of strong gradients, the image values of the base image can be modified so that strong gradients in the image values of the modified image (i.e., the starting image) can be avoided where possible. Therefore, since the image reconstruction unit for reconstructing the image of the object is suitable for reconstructing the image based on the projection data using an iterative reconstruction algorithm that uses the modified image as the starting image for iterative reconstruction, faster convergence of the iterative reconstruction can be achieved, so that the computing time and computing resources required to reconstruct the image are reduced.
[0010] The projection data providing unit is suitable for providing the projection data of the object collected during the CT scanning of the object. The data providing unit can, for example, be directly connected to a CT system for collecting projection data, and can directly provide the projection data collected by the CT system. In addition, the data providing unit can also be part of the CT system for collecting projection data. Alternatively, the data providing unit can be separated from the CT system, and / or can be connected to a storage unit that, for example, stores the projection data of the object that has been collected by the CT system. In addition, the data providing unit itself can be configured as a storage unit for storing the projection data of the object. The object can be any object scanned by the CT system, such as a human, an animal, or an inanimate object (such as a suitcase). In a preferred embodiment, the object is a patient, and the CT system is a conventional medical CT system.
[0011] The base image generation unit is adapted to generate a base image based on the projection data. Preferably, the base image generation unit is adapted to reconstruct the base image using the projection data and any suitable reconstruction algorithm that allows a fast and computationally inexpensive reconstruction of the image. Preferably, the base image generation unit is adapted to generate the base image by applying a filtered backprojection algorithm to the projection data. The base image may be a 2D or 3D image. Furthermore, the base image comprises a plurality of voxels, i.e. volume elements, each comprising an image value representing an attenuation characteristic of a portion of the object corresponding to the respective voxel. The term "voxel" is also used in the context of the present invention to describe a 2D pixel, i.e. a picture element. Thus, if the base image is a 3D image, the term "voxel" refers to a volume element, and if the base image is a 2D image, the term "voxel" refers to a picture element. The image value may be, for example, a grayscale value based on the Hounsfield scale. However, other scales for representing the attenuation characteristic of an object in a reconstructed image of the object may also be contemplated.
[0012] The modification unit is adapted to modify the base image to generate a modified image. The modified image is generated by modifying image values of voxels of the base image, wherein the image values are modified based on the image values of the respective voxels. Preferably, each voxel of the base image that is located in a region of interest within the base image and comprises image values that are part of a plurality of predetermined image values is modified. The region of interest may for example refer to a certain structure of an object (e.g. an anatomical structure such as a patient's head, the contents of a suitcase) or the entire image. The modification unit may be adapted to modify the image values based on the image values of the voxels, for example by using a mathematical function that links the image values of the voxels to the image values of the voxels themselves. The modification of the image values may for example refer to adding or subtracting a predetermined value from the image values, wherein the predetermined value may depend on the image value. Furthermore, more complex modifications that follow more complex mathematical functions according to the image values of the respective voxels of the base image may be implemented. Preferably, the image values assigned to the modifications of the voxels are unique in a mathematical sense, i.e. a specific image value is modified only in a predetermined way, wherein the same image values are modified identically.
[0013] Preferably, the modification unit is adapted to modify the image value of the voxel by assigning a predetermined modified image value to the voxel, wherein the assigned predetermined modified image value depends on the image value of the voxel. For example, the modification unit may be adapted to assign the predetermined modified image value to the voxel based on the image value of the voxel by using a list that links the image value of the voxel to the predetermined modified image value. Alternatively, the link can be provided as a function that links the predetermined modified image value to the image value of the voxel. The link between the image value of the voxel and the predetermined modified image value may be selected based on prior knowledge of the characteristics of an iterative reconstruction algorithm for reconstructing the projection data of the object. For example, if it is known that the iterative reconstruction algorithm converges only very slowly in the presence of strong gradients or strong local extrema, the modification unit may be adapted to analyze the image values of the base image to find such strong gradients or extrema, and provide the predetermined modified image value to these image values of the base image, so that the strong gradients or extrema are smoothed. In another example, if it is known that the iterative reconstruction algorithm converges only slowly in the case where the starting image is too far from the reconstructed image, prior knowledge about the structure or characteristics of the scanned object may be used to identify image values of the base image that belong to certain structures of the object. Predetermined modified image values may then be assigned to voxels comprising image values belonging to respective structures, such that typical properties of these structures of the object are reflected more clearly and the modified image more accurately represents the imaged object.
[0014] The image reconstruction unit is suitable for reconstructing an image of the object based on the provided projection data and an iterative reconstruction algorithm, wherein the modified image is used as a starting image for iterative reconstruction. During the iterative reconstruction, forward and backward projections based on the projection data and the image data are usually performed during each iteration step. During the first iteration step, the starting image of the iterative reconstruction is used as input for the first forward projection and the first backward projection. In a preferred embodiment, the iterative reconstruction algorithm is a maximum likelihood reconstruction algorithm. An example of a suitable maximum likelihood reconstruction can be found in the article "Statistical image reconstruction methods for transmission tomography" by JA Fessler (Handbook of Medical Imaging, volume 2, chapter 1, pages 1 to 70 (2000)).
[0015] The projection data includes positioning scan projection data acquired during the positioning scan of the object and diagnostic projection data acquired during the diagnostic scan of the object, wherein the image reconstruction unit is adapted to reconstruct the image of the object using the iterative reconstruction algorithm based on the positioning scan projection data and the diagnostic scan projection data. Positioning scan refers to a scan of the object acquired before the diagnostic scan. Typically, the positioning scan is acquired in a very fast manner and has a signal-to-noise ratio lower than that of the diagnostic image. The projection data acquired during the positioning scan may refer to 2D positioning scan projection data or 3D positioning scan projection data. For example, 2D positioning scan projection data may be acquired by moving the table of the CT system through the CT system without rotating the gantry of the CT system and thus without rotating the detector of the CT system. In addition, such 2D projection data may be acquired using a small collimation of the radiation source or using a large collimation of the radiation source, i.e., by irradiating only part or all of the detector of the CT system. For example, 3D positioning scan projection data may be acquired by using a spiral acquisition mode, wherein, in this case, the positioning scan refers to a spiral positioning scan. A scout scan may be used to plan a diagnostic scan, for example, to determine a region of interest in a scout image reconstructed based on projection data from the scout scan, wherein the diagnostic scan is then limited to the region of interest determined in the scout scan. Other functions of the scout scan may also be contemplated, such as determining the position of the object, radiation dose calculation of the radiation dose to be provided to the object during the diagnostic scan, determining or optimizing image settings for the diagnostic scan, etc. A diagnostic scan refers to a scan of the object with image settings that allow providing a diagnosis of the object based on a diagnostic image reconstructed from projection data acquired during the diagnostic scan. These image settings may, for example, refer to a good signal-to-noise ratio, a high resolution, a high contrast between certain structures of the object, etc.
[0016] The image reconstruction unit is suitable for reconstructing the image of the object using an iterative reconstruction algorithm based on the positioning scan projection data and the diagnostic scan projection data. Since the image reconstruction unit is suitable for reconstructing the image of the object not only based on the diagnostic projection data but also based on the positioning scan projection data, the iterative reconstruction can be based on a larger data base. For example, in the axial scan using a cone beam CT system, the diagnostic projection data obtained is incomplete in a mathematical sense. This incompleteness can lead to artifacts in the reconstructed image, especially in human body regions with high gradients along or parallel to the rotation axis. Although the iterative algorithm can minimize these artifacts, an inappropriate number of iterations is usually required to obtain sufficient image quality. Including the positioning scan projection data can partially or completely reduce the incompleteness. Therefore, providing additional data from the positioning scan can promote the convergence of the iterative reconstruction algorithm. In addition, a reconstructed image with higher image quality (i.e., with fewer artifacts) can be received from the reconstruction.
[0017] For example, the scout scan projection data can be used as additional input to the forward and reverse iterations of the iterative reconstruction algorithm during each iteration step. In the iterative algorithm, forward projection and reverse projection are performed in each iteration step. Preferably, these forward projections and reverse projections take into account both diagnostic projection data and scout scan projection data. For example, in order to update the value of a particular voxel during the iteration step, the diagnostic projection data associated with the ray passing through the voxel is taken into account for these forward projections and reverse projections, wherein, if there is scout scan projection data for the particular voxel, the scout scan projection data is also taken into account.
[0018] In an embodiment, during the iterative reconstruction, the scout scan projection data receives a lower weight than the diagnostic projection data. Since the scout scan projection data typically provides a lower signal-to-noise ratio, providing a lower weight to the scout scan projection data during iterative reconstruction ensures that for areas of the object where there is diagnostic projection data with good coverage and a good signal-to-noise ratio, the iterative reconstruction algorithm will cause the reconstruction of the image to be primarily based on the diagnostic projection data. However, in areas with low coverage of diagnostic projection data, such as at the edges of the diagnostic scan area or in areas where diagnostic projection data is truncated, even low-weighted scout scan projection data will have an impact on the reconstruction of the image. For areas with high gradients along or parallel to the axis of rotation of the CT system (with which the projection data is acquired), the inclusion of the scout scan projection data can reduce artifacts typically caused by these gradients even when scout scan data with a low weight is used.
[0019] In an embodiment, the locator scan projection data is collected during the spiral locator scan. If the locator scan projection data is collected during the spiral locator scan, a complete set of projection data of the object is provided by the locator scan projection data. Therefore, a 3D image can be reconstructed from the locator scan projection data. In a preferred embodiment, the basic image generation unit is suitable for generating the basic image based on the locator scan projection data. Since the locator scan projection data must be reconstructed into a locator scan image for planning the acquisition of diagnostic images, if the basic image generation unit is suitable for generating a basic image based on the locator scan projection data, it is not necessary to reconstruct another image (i.e., the basic image) for subsequent iterative reconstruction. Therefore, no additional computational cost is generated for providing a basic image. In this embodiment, the locator scan projection data refers to, for example, 3D locator scan projection data collected during the spiral locator scan.
[0020] Preferably, the image reconstruction unit is further adapted to register the modified image with the initial reconstruction of the diagnostic projection data, and use the registered modified image as the starting image. Since the positioning scan projection data and the diagnostic projection data are collected during different scans, movement of the object (e.g., the patient) may have occurred between scans. In order to ensure that the modified image is as close as possible to the final reconstructed image, it is preferred that such movement is compensated if necessary. This can be done during the registration of the modified image with the initial reconstruction of the diagnostic projection data. The initial reconstruction can, for example, refer to an image after only a few iterations of an iterative algorithm (e.g., after the first iteration). The iteration for providing the initial reconstruction can start from any starting image (e.g., from a starting image comprising only one image value). Alternatively, the initial reconstruction can be based on another reconstruction algorithm, for example, based on a filtered backprojection algorithm applied to the diagnostic projection data. Such an initial reconstruction can be only a very rough estimate of the object. Any suitable known registration algorithm can be used to perform the registration between the modified image and the initial reconstruction. For example, during registration, highly attenuated objects or regions with high contrast can be identified in the modified image and in the initial reconstruction, and the registration can be based on these highly attenuated objects, such as bones or metal objects. For example, an example of a suitable registration algorithm is described in the article "Evaluation of 4D-CT Lung Registration" by S. Kabus et al. (MICCAI 2009, Lecture Notes in Computer Science, Vol. 5761, pp. 747 to 754 (2009)). In addition, the image reconstruction unit can also be adapted to use the registered modified image to register the positioning scan projection data with the diagnostic scan projection data, and apply the iterative algorithm to the registered positioning scan projection data and the diagnostic projection data. For example, if the registration between the modified image and the initial reconstruction of the diagnostic projection data is known, the registration between the unmodified base image and the diagnostic projection data can also be determined. Therefore, in addition, the registration of the positioning scan projection data and the diagnostic projection data is also known. Thus, reconstruction of an image using an iterative reconstruction algorithm may be based on the registered scout scan projection data and diagnostic projection data, compensating for movement of the patient that may have occurred between the acquisition of the scout scan projection data and the diagnostic projection data.
[0021] In an embodiment, the modification unit is adapted to determine whether the image value of a voxel lies within a predetermined image value range and to modify the image value of the voxel based on the result of the determination. The predetermined image value range may be defined by providing two image value thresholds. Furthermore, the predetermined image value range may also be defined by one image value threshold, wherein all image values below or above the image value threshold fall within the image value range. Preferably, more than one image value range may be predetermined, wherein the predetermined image value ranges do not overlap. For example, the predetermined image value ranges may be selected such that an upper boundary threshold of one image value range refers to a lower image value threshold of another image value range, wherein in this case it is also determined to which image value range the image value threshold itself belongs. The image value of the voxel may then be modified in one way based on the determination that the image value of the voxel lies within the corresponding predetermined image value range, wherein the image value of the voxel may be modified in another way if the image value of the voxel lies within another corresponding predetermined image value range. If the predetermined image value range is defined by two image value thresholds, the modification unit is adapted to determine whether the image value of the voxel lies within the predetermined image value range by comparing the image value of the voxel with the two image value thresholds of the predetermined image value range. In particular, the modification unit is adapted to determine whether the image value of the voxel is greater than a lower image value threshold of a predetermined image value range and less than a higher image value threshold of the predetermined image value range, wherein then the image value of the voxel lies within the predetermined image value range. If the image value range is determined by only one threshold so that all image values greater than or less than the image value threshold lie within the predetermined image value range, the modification unit is adapted to determine whether the image value of the voxel lies within the predetermined image value range by comparing the image value of the voxel with the image value threshold, and if the image value of the voxel is greater than or less than the image value threshold of the predetermined image value range, the modification unit is adapted to determine whether the image value of the voxel lies within the predetermined image value range, determining that the image value of the voxel lies within the predetermined image value range.
[0022] The modification unit is then adapted to modify the image value of the voxel based on the result of the determination. For example, if the result of the determination is that the image value of the voxel falls within a predetermined image value range, the modification unit may be adapted to modify the image value of the voxel. Preferably, if the result of the determination is that the image value of the voxel falls within the predetermined image value range, the modification unit is adapted to modify the image value of the voxel by assigning a predetermined modified image value to the voxel. In another example, if the result of the determination is that the image value of the voxel does not fall within the predetermined image value range, the modification unit may be adapted not to modify the image value of the voxel, preferably not to assign a modified image value to the voxel.
[0023] In a preferred embodiment, the predetermined image value range is based on characteristics of the object. The characteristics of the imaged object may, for example, refer to structural characteristics of the imaged object or compositional characteristics of the imaged object. For example, if the object is a patient, the characteristics of the imaged object may refer to certain tissues expected in the imaged region of the object, such as bone tissue, muscle tissue or organ tissue. Therefore, it is possible to modify only the image values of the base image that are known to be within a specific image value range (e.g., an image value range that is known to cause problems during iterative reconstruction). In addition, if the image value range is based on characteristics of the imaged object, the already known characteristics of the imaged object may be taken into account when determining the image value range that may cause problems during iterative reconstruction.
[0024] In a preferred embodiment, the modification unit is adapted to modify the image values of the voxels of the base image further based on the characteristics of the image values of all voxels in the region of interest in the base image that are located within the predetermined image value range. The characteristics of the image values of the base image that are located within the predetermined image value range may, for example, refer to the distribution of the image values of the base image that are located within the predetermined image value range, the average value of the image values of the base image that are located within the predetermined image value range or other statistical analysis values that can be acquired based on the image values of the base image that are located within the predetermined image value range. For example, if the characteristics of the image values of the base image that are within the predetermined image value range show a substantially uniform distribution, the image values of the voxels that are located within this range may be modified to a constant predetermined modified image value. Preferably, in this case, the modification unit is adapted to modify the image value of the voxel by assigning a predetermined modified image to the voxel, for example, the average image value of all image values of the base image that are within the predetermined image value range. Alternatively, for example, if the characteristic of the image values of the base image within the predetermined image value range shows that substantially all image values of the base image within the predetermined image value range comprise the same image value (e.g. image values located within the upper third of the predetermined image value range), the modification unit may be adapted to assign the predetermined modified image value to all voxels of image values within the predetermined image value range that refer to this particular image value. Other embodiments are also contemplated, in which, for example, the modified image value shows a more complex dependency on the characteristic of the image values of the base image within the predetermined image value range.
[0025] In a preferred embodiment, the modification unit is adapted to modify the image value of each voxel of the base image within the region of interest including within the predetermined image value range by assigning a predetermined modified image value to the voxel. The predetermined modified image value may then, for example, be a value determined based on prior knowledge about the properties of the object or based on properties of the image values of the base image within the predetermined image value range. If the object is a patient, the predetermined modified image value may, for example, refer to a known average image value expected for soft tissue.
[0026] In an embodiment, the modification unit is adapted to modify the image values of the voxels of the base image within the region of interest by assigning a first modified image value to each voxel of the region of interest comprising an image value within a first image value range corresponding to vacuum or air and by assigning a second modified image value to each voxel of the region of interest comprising an image value within a second image value range corresponding to soft tissue. The first image value range corresponding to vacuum or air may, for example, be defined as comprising all voxels of image values below -530HU, and the second image value range corresponding to soft tissue may, for example, be defined as comprising all voxels of image values between -530HU and 560HU. The first modified image value may then correspond to an image value for vacuum or air defined as -1000HU. The second modified image value may be defined by a known typical image value of soft tissue, for example organ tissue or muscle tissue, and may, for example, be selected as 25HU. The inventors have found that in such an embodiment, the convergence of the iterative reconstruction algorithm may be particularly improved.
[0027] In another aspect of the present invention, a method for reconstructing an image of an object is proposed, wherein the method comprises the following steps: a) providing projection data of the object, wherein the projection data has been acquired by a CT system; b) generating a base image based on the projection data; c) modifying the base image to generate a modified image, wherein, in order to modify the base image, the image values of the voxels of the base image are modified based on the image values of the voxels; d) reconstructing the image of the object, wherein the image is reconstructed using an iterative reconstruction algorithm based on the projection data, and the iterative reconstruction algorithm uses the modified image as a starting image for the iterative reconstruction.
[0028] In another aspect of the present invention, a computer program for reconstructing an image of an object is proposed, wherein the computer program includes a program code module, which is used to cause the system to perform the steps of the method according to the other aspect of the present invention when the computer program is executed by a computer controlling the system according to the one aspect of the present invention.
[0029] It shall be understood that the system for reconstructing an image of an object according to one aspect of the invention, the method according to another aspect of the invention and the computer program according to another aspect of the invention have similar and / or identical preferred embodiments.
[0030] It should be understood that a preferred embodiment of the present invention can also be any combination of the above embodiments and corresponding embodiments.
[0031] These and other aspects of the invention will be apparent from and elucidated with reference to the embodiments described hereinafter. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] In the following figures:
[0033] Figure 1 Schematically and exemplarily, an embodiment of a system for providing an image of an object is shown,
[0034] Figure 2 shows exemplary results of image reconstruction according to an embodiment of the system,
[0035] Figure 3 shows an exemplary result of image reconstruction according to another embodiment of the system, and
[0036] Figure 4 A flow chart exemplarily illustrating an embodiment of a method for providing an image of an object is shown. DETAILED DESCRIPTION
[0037] Figure 1 An embodiment of a system for reconstructing an image of an object is schematically and exemplarily shown. In this embodiment, the system 100 is suitable for reconstructing an image of a patient 150 lying on a support device 151 (such as a patient table). In other embodiments, the object can also be an animal or an inanimate object. The system 100 includes a projection data providing unit 110, which provides projection data of the patient 150 acquired by a CT system 160 during a CT scan of the patient 150. In addition, the system 100 includes a base image generation unit 120, a modification unit 130 and an image reconstruction unit 140.
[0038] In a first embodiment, the data providing unit 110 is adapted to provide diagnostic projection data acquired during a diagnostic scan of the object as projection data, wherein no scout scan is performed before the diagnostic scan of the patient 150. The base image generating unit 120 is then adapted to generate a base image based on the provided projection data, which refers to the diagnostic projection data, by e.g. applying a filtered backprojection algorithm to the projection data.
[0039] In this embodiment, the modification unit 130 is adapted to assign a predetermined modified image value to a voxel of the base image based on the image value of the voxel. In the embodiment described herein, a first image value range and a second image value range are determined. The first image value range is determined to include all voxels of the base image including image values that are expected to represent air, and the second image value range is determined to include all voxels of the base image including image values that are expected to represent soft tissue (e.g., organ tissue or muscle tissue) of the patient 150. Therefore, for the first image value range, an upper image value threshold of -530HU is selected, wherein the first image value range is defined as including all voxels of the base image including image values below the selected upper image value threshold. For the second image value range, in this embodiment, a lower image value threshold of -530HU and an upper threshold of 560HU are selected. Furthermore, in this embodiment, the modification unit 130 is adapted to provide the first modified image value and the second modified image value, respectively, to all voxels of the base image including image values within the first image value range and image values within the second image value range. As a representation for air, a first modified image value of -1000HU is assigned by the modification unit 130 to each voxel of the base image that includes an image value within the first image value range. As a representation for soft tissue, a second modified image value of 25HU is assigned by the modification unit 130 to each voxel of the base image that includes an image value within the second image value range. Thus, voxels of the base image that include image values that are neither within the first image value range nor within the second image value range remain unchanged.
[0040] In this embodiment, the modification unit 130 modifies the base image according to the above-mentioned rules to generate a modified image. In this case, the modified image refers to an image in which voxels including image values relating to air or soft tissue respectively include representative constant values. The image reconstruction unit 140 then uses an iterative reconstruction algorithm (preferably a maximum likelihood reconstruction algorithm) to reconstruct an image of the patient 150 based on the projection data provided by the projection data providing unit 110, wherein the modified image is used as a starting image for iterative reconstruction. The inventors have found that the modification of the base image described above results in particularly fast and accurate reconstruction of a CT image using iterative reconstruction (preferably, maximum likelihood reconstruction). Figure 2 The result of such a reconstruction using the modified image as the starting image for the iterative reconstruction algorithm generated according to the embodiments described above is shown in .
[0041] Figure 2 The image 210 in the upper left corner of shows a conventional starting image for a maximum likelihood iterative reconstruction algorithm, wherein the conventional starting image is acquired, for example, by using a filtered backprojection algorithm on the projection data. The image 210 is shown at a level of 350 HU and a window of 70 HU. Figure 2The lower left image 220 shows a reconstructed image, where the image is iteratively reconstructed using the conventional starting image 210. The image 220 after 100 iterations is also shown at a level of 35HU and a window of 70HU. Figure 2 The image 230 in the upper right corner shows a modified image generated according to the embodiments described above. It can be seen that compared to the conventional starting image 210, the modified image 230 shows a clear delineation between different structures of the patient (ie, between air, soft tissue and bone). Figure 2 The image 240 in the lower right corner of shows the reconstruction result based on the image 230 as the starting image, wherein the maximum likelihood iterative reconstruction is also used for the reconstruction. It can be seen that the image 220 and the image 240 are almost identical, wherein the image 240 reconstructed based on the modified image 230 as the starting image only requires 20 iterations for reconstruction. Therefore, using a modified starting image as represented by the modified image 230, for example, can greatly reduce the computational workload required to reconstruct a CT image.
[0042] In an embodiment of the present invention as described above, the projection data providing unit 110 is adapted to provide projection data of an object, which includes scout scan projection data acquired during a spiral scout scan of the patient 150 and diagnostic projection data acquired during a diagnostic scan of the patient 150. In this embodiment, the base image generating unit 120 may be adapted to generate a base image based on the scout scan projection data. Then, the modifying unit 130 may modify the generated base image based on the scout scan projection data, as already described above. Figure 3 The effect of using a base image generated based on the scout scan projection data is shown in FIG.
[0043] Figure 3 310 shows a conventionally reconstructed axial CT scan image, while image 320 shows a reconstructed image reconstructed based on the same projection data as image 310, except that the image reconstructed from the scout scan projection data is used as the starting image. In order to clearly illustrate the effect of providing a base image generated based on the scout scan projection data, image 310 is reconstructed without modifying the base image, so that the base image generated from the scout scan projection data is directly used as the starting image for iterative reconstruction. In a comparison between image 310 and image 320, it can be seen that image 320 based on reconstruction using the scout scan image as the starting image shows fewer artifacts. Therefore, generating a base image based on the scout scan projection data can lead to more accurate reconstruction results.
[0044] Furthermore, in addition to or as an alternative to generating a base image based on the scout scan projection data, the image reconstruction unit 140 can be adapted to reconstruct an image of the patient 150 based on the scout scan projection data and the diagnostic projection data. In this embodiment, in order to reconstruct the image, the image reconstruction unit 140 is also adapted to first register the scout scan projection data with the diagnostic projection data, because for the patient, movement between the acquisition of the scout scan projection data and the diagnostic projection data is possible. In other embodiments where there may be no movement, for example, if the object is a suitcase, the registration can be omitted. In order to register the scout scan projection data and the diagnostic projection data, the image reconstruction unit 140 can be adapted to use an already determined modified image determined based on the scout scan projection data, i.e., a base image reconstructed based on the scout scan projection data, or if the modified image is not based on the scout scan projection data, it can be adapted to reconstruct the scout scan image based on the scout scan projection data using, for example, a filtered backprojection algorithm. In addition, the image reconstruction unit 140 can be adapted to provide an initial reconstructed image based on the diagnostic projection data, wherein the initial reconstructed image can be reconstructed by applying a filtered backprojection algorithm to the diagnostic projection data, or by applying several iterations of an iterative algorithm to the diagnostic projection image based on an arbitrary starting image. The image reconstruction unit 140 can then determine image structures based on the localization scan projection data, such as highly attenuated structures caused by bones or metal in the initial reconstructed image and in the reconstructed image. Based on the determined image structures, the two images, i.e., the reconstructed image based on the localization scan projection data, can then be registered. For example, the modified image is translated and / or rotated to correspond to the initial reconstructed image. The localization scan projection data can then be registered to the diagnostic projection data accordingly.
[0045] The image reconstruction unit 140 is then adapted to reconstruct an image based on the registered positioning scan projection data and diagnostic projection data. For example, during the forward and reverse projections in each iteration step, the positioning scan projection data may be used in addition to the diagnostic projection data. For example, in order to update the value of a particular voxel during an iteration step (i.e., during forward and reverse projections), the data associated with the ray passing through the voxel is considered, wherein, if there is positioning scan projection data for the voxel, the positioning scan projection data is included in addition to the diagnostic projection data. Furthermore, in this embodiment, the image reconstruction unit 140 is adapted to provide weights to the projection data during each iteration step, wherein the positioning scan projection data is weighted lower than the diagnostic projection data. Reconstructing an image based on the above-described embodiment results in a reconstructed image showing fewer image artifacts.
[0046] Figure 4A flow chart exemplarily illustrating an embodiment of a method for reconstructing an image of an object is shown. The method 400 comprises a first step 410 of providing projection data of a patient 150 that has been acquired during a CT scan using a CT system 160. The projection data is provided by a projection data providing unit 110 that is connected to the CT system 160. In a second step 420, a base image is generated based on the provided projection data. The base image is then modified in a step 430 to generate a modified image. The modification comprises assigning modified image values to voxels of the base image based on the image values of the voxels. In a final step 440, an image of the patient 150 is reconstructed based on the projection data, wherein an iterative reconstruction algorithm is used using the modified image as a starting image for iterative reconstruction.
[0047] In CT image reconstruction, iterative reconstruction algorithms can be very important for certain applications (e.g., where very low patient radiation doses are necessary). For such low-dose images or in order to minimize cone beam artifacts in axial scans, maximum likelihood image reconstruction algorithms are particularly suitable. However, since these maximum likelihood image reconstruction algorithms have a large computational complexity, i.e., require a large computational effort, it is advantageous to obtain the result after the minimum possible number of iterations.
[0048] In the present invention, it is proposed, for example, to obtain a specifically modified image as a starting image for iterative reconstruction of CT projection data, wherein the base image on which the modification is based is obtained, for example, by using a filtered backprojection algorithm. In addition, an algorithm is proposed to modify the base image to provide a modified image, wherein the base image is modified based on a carefully selected image value threshold (i.e., image value range). Preferably, a set of thresholds (i.e., a set of image value ranges) is selected based on an absorption coefficient (i.e., image value). For example, a threshold, i.e., an image value range, can be selected for distinguishing between vacuum / air, soft tissue and bone. In addition, it is proposed to select a modified absorption coefficient, i.e., a modified image value, which represents the image value of soft tissue. In addition, a base image suitable for use in iterative reconstruction is selected. The base image is then modified based on the threshold (i.e., image value range) selected above. For example, each voxel of the base image associated with vacuum / air (i.e. comprising image values within the first image value range) is set to -1000HU, wherein each voxel of the base image comprising image values associated with soft tissue is set to a modified image value representing soft tissue, wherein, in addition, all voxels outside the selected image value range are associated with bone and remain unchanged. Such a modified image is then used as a starting image for iterative reconstruction.
[0049] In an embodiment of the present invention, in addition to the diagnostic projection data, the scout scan projection data is also collected during the scout scan of the object. Then the iterative reconstruction algorithm for reconstructing the CT image is adjusted so that it contains the scout scan projection data from the scout scan. This scout scan can be performed in different ways. For example, the scout scan projection data can be obtained on a linear segment using a small collimation, wherein, in this case, the scout scan projection data is collected when the patient table moves but there is no rotation of the rack of the CT system. In addition, the scout scan projection data can be obtained on a linear segment using a large collimation, wherein, in this case, the entire detector of the CT system is irradiated in each view. In another example, the scout scan projection data can be obtained during a spiral scout scan. Each of these exemplary acquisitions is suitable for use with the present invention as proposed herein. However, the acquisition of scout scan projection data using large collimation or in a spiral scout scan allows the acquisition of a complete scout scan projection data set. For example, the scout scan projection data set using the scout scan projection data collected with large collimation can be completed by combining with the corresponding diagnostic projection data from the diagnostic scan, while the scout scan projection data obtained during the spiral scout scan alone produces a complete scout scan projection data set.
[0050] The iterative reconstruction algorithm used to reconstruct the CT image calculates the forward projection and the reverse projection in each iterative step. The positioning scan projection data collected during the positioning scan can then be used as additional projection data in these forward iterations and reverse iterations. In addition, the noise model can be included in the iterative reconstruction algorithm, which is preferably a maximum likelihood reconstruction algorithm that ensures that the positioning scan projection data receives a lower weight than the diagnostic projection data, because the positioning scan projection data is usually collected using a lower X-ray dose. However, for areas with high density (such as areas including bones) that usually cause cone beam artifacts, even if the positioning scan projection data receives a smaller weight than the diagnostic scan projection data, the inclusion of the positioning scan projection data can also make a difference.
[0051] In addition, if the acquired scout scan projection data produces a complete projection data set (e.g., scout scan projection data acquired during a spiral scout scan), the scout scan projection data can also be used to generate a base image. For example, a base image can be generated by providing a filtered back-projection algorithm to the scout scan projection data. In this case, in order to alleviate the problems caused by the patient's motion between the acquisition of the scout scan and the diagnostic scan, it is advantageous to provide the registration between the scout scan projection data and the diagnostic scan projection data as part of the iterative algorithm. Registration can, for example, be included in identifying bones and / or other high-contrast parts in a base image or a modified image based on the scout scan projection data and in the initial reconstruction of the diagnostic projection data. Based on this registration, the base image or the modified image can be scaled and translated so that it best fits the initial reconstruction. The registered base image or the registered modified image is then used as a starting image for iterative reconstruction.
[0052] Although in the above-described embodiments the object to be imaged by the CT system is a patient, in other embodiments the object may be an animal or an inanimate object such as a suitcase. In the latter case, the modified image value or the predetermined image value range may be defined based on characteristics of the inanimate object (e.g., based on the expected contents of the suitcase). In this case, the image value range may be defined, for example, based on an image value representing vacuum / air, an image value representing, for example, clothing, and an image value representing, for example, a metal object.
[0053] Although in the above described embodiments two image value ranges are determined, in other embodiments only one image value range may be determined, or more than two image value ranges may be determined, for example if the imaged object comprises more than two structures comprising different expected attenuation characteristics. Furthermore, the determination of the image value ranges may be omitted and the base image may be modified based on, for example, a function linking the image values of the voxels of the base image to a modification of the image value or to a modified image value, wherein the modification unit is then adapted to modify the base image by modifying the image values of the voxels based on the provided function. The function may also be provided as a list for linking the image values of the voxels of the base image to a modification of said image value or to a modified image value, which is stored in the storage unit and provided to the modification unit.
[0054] Although in the above described embodiments the scout scan projection data is acquired during a helical scout scan, in other embodiments the scout scan projection data may also be acquired during other scout scan modes. The scout scan projection data may then refer to 2D or 3D scout scan projection data.
[0055] Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims.
[0056] In the claims, the word "comprising" does not exclude other elements or steps, and the word "a" or "an" does not exclude a plurality.
[0057] A single unit or device may fulfill the functions of several items recited in the claims.The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.
[0058] The processes of generating a base image or providing projection data as performed by one or several units or devices may be performed by any other number of units or devices. These processes and / or operations of the system may be implemented as program code modules of a computer program and / or dedicated hardware.
[0059] The computer program may be stored / distributed on suitable media, such as optical storage media or solid-state storage media provided together with or as part of other hardware, but the computer program may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunications systems.
[0060] Any reference signs in the claims should not be construed as limiting the scope.
[0061] The present invention relates to providing a system that allows reducing computational costs when using an iterative reconstruction algorithm. The system comprises: a providing unit for providing CT projection data; a base image generating unit for generating a base image based on the projection data; a modifying unit for generating a modified image, wherein the image values of voxels of the base image are modified based on the image values of the voxels; and an image reconstruction unit for reconstructing an image using an iterative reconstruction algorithm, the iterative reconstruction algorithm using the modified image as a starting image. Since the modifying unit is adapted to modify the base image, the base image can be modified such as to form an optimal starting image for a selected iterative reconstruction, so that a faster convergence of the iterative reconstruction can be achieved.
Claims
1. A system for reconstructing an image of an object (150), wherein: The system comprises: a projection data providing unit (110) for providing projection data of the object (150), wherein the projection data has been acquired by a CT system (160), a basic image generating unit (120), configured to generate a basic image based on the projection data, a modification unit (130) for modifying the base image to generate a modified image, wherein, in order to modify the base image, the modification unit (130) is adapted to modify the image values of the voxels based on the image values of the voxels of the base image, an image reconstruction unit (140) for reconstructing an image of the object (150) based on the projection data using an iterative reconstruction algorithm, the iterative reconstruction algorithm using the modified image as a starting image for iterative reconstruction, wherein the projection data comprises scout scan projection data acquired during a scout scan of the object (150) and diagnostic projection data acquired during a diagnostic scan of the object (150), wherein the image reconstruction unit (140) is adapted to reconstruct the image of the object (150) based on the scout scan projection data and the diagnostic projection data using the iterative reconstruction algorithm, The invention is characterized in that the image reconstruction unit (140) is adapted such that during the iterative reconstruction the scout scan projection data receives a lower weight than the diagnostic projection data.
2. The system according to claim 1, wherein: The scout scan projection data is acquired during a helical scout scan.
3. The system according to claim 2, wherein: The base image generation unit is adapted to generate the base image based on the scout scan projection data.
4. The system according to claim 3, wherein: The image reconstruction unit (140) is further adapted to register the modified image with an initial reconstruction of the diagnostic projection data and to use the registered modified image as a starting image.
5. The system according to any one of claims 1 to 4, wherein: The base image generation unit (120) is adapted to generate the base image by applying a filtered backprojection algorithm to the projection data.
6. The system according to any one of claims 1 to 4, wherein: The modification unit (130) is adapted to determine whether an image value of a voxel lies within a predetermined image value range and to modify the image value of the voxel based on a result of the determination.
7. The system according to claim 6, wherein: The predetermined range of image values is based on characteristics of the object (150).
8. The system according to claim 6, wherein: The modification unit is adapted to modify the image values of the voxels of the base image further based on a characteristic of the image values of all voxels in the region of interest in the base image that are within the predetermined image value range.
9. The system according to claim 7, wherein: The modification unit is adapted to modify the image value of each voxel of the base image within the region of interest comprising an image value within the predetermined image value range by assigning a predetermined modified image value to the voxel.
10. The system according to any one of claims 1 to 4, wherein: The modification unit is suitable for modifying the image value of each voxel of the base image within the region of interest by assigning a first modified image value to each voxel of the region of interest including an image value within a first image value range corresponding to vacuum or air, and assigning a second modified image value to each voxel of the region of interest including an image value within a second image value range corresponding to soft tissue.
11. The system according to any one of claims 1 to 4, wherein: The iterative reconstruction algorithm is a maximum likelihood reconstruction algorithm.
12. A method for reconstructing an image of an object (150), wherein: The method comprises the following steps: providing (410) projection data of the object, wherein the projection data has been acquired by a CT system (160), wherein the projection data includes scout scan projection data acquired during a scout scan of the object (150) and diagnostic projection data acquired during a diagnostic scan of the object (150), generating (420) a base image based on the projection data, modifying (430) the base image to generate a modified image, wherein modified image values are assigned to voxels based on image values of the voxels of the base image, Reconstruct (440) an image of the object (150), wherein the image is reconstructed based on the projection data using an iterative reconstruction algorithm, the projection data including the localization scan projection data and the diagnostic projection data, the iterative reconstruction algorithm using the modified image as a starting image for iterative reconstruction, characterized in that during the iterative reconstruction, the localization scan projection data receives a lower weight than the diagnostic projection data.
13. A computer program for reconstructing an image of an object (150), wherein: The computer program comprises program code means for causing the system to perform the steps of the method according to claim 12 when the computer program is executed by a computer controlling the system according to claim 1 .
Citation Information
Patent Citations
Computed tomography (CT) iteration reconstruction method for asymmetrical detector
CN102376097A
Tomography synthesis image reconstruction method and system
CN106373165A
Image iteration reconstruction method and device
CN106683144A
CT (computed tomography) dose modulation method and device, CT scanning method and CT system
CN107928694A
Method And Apparatus For Motion Correction In CT Imaging
US20170340287A1
