CN104182932B - CT (Computed Tomography) device, CT image system and CT image generation method - Google Patents

Abstract

The invention provides a CT device, a CT image system and a method for generating a CT image, which can take into account both the quality of the CT image and the efficiency of generating the CT image on the premise of ensuring safety. The CT apparatus scans a scanning area with X-rays to generate a CT image of a scanning object located in the scanning area, and includes a reference device provided at a predetermined position in the scanning area and a CT image generating device. The CT image of the scanning object is generated according to the known CT image information of the reference object device and the scanning data of the scanning area obtained through scanning. Wherein, when using iterative reconstruction to generate the CT image of the scanning object, in particular, the known CT image of the reference device and the reconstructed image and updated image corresponding to the reference device during iteration can be used to determine the iteration step size. Therefore, the number of iterations in iterative reconstruction can be effectively reduced, and the efficiency of iterative reconstruction can be improved.

Description

CT device, CT image system and CT image generation method

technical field

The invention relates to a CT device, a CT image system and a CT image generating method, in particular to a CT device, a CT image system and a CT image generating method for improving the efficiency of iterative reconstruction of a CT image.

Background technique

X-ray computed tomography (CT) technology has been widely used to examine the human body. CT images have been used as the basis for disease diagnosis for 30 years. Research on CT image reconstruction technology can reduce radiation dose and improve CT image quality. , Reducing image artifacts has always been a hot issue in research and clinical practice.

In practical applications, CT image reconstruction techniques mainly include filtered back-projection methods and iterative reconstruction methods. Among them, the filtered back-projection method is a traditional method of CT image reconstruction, and has been widely used in current CT products. However, in the filtered back projection method, the projection data of the reconstructed image is assumed to be noise-free, but in fact, noise always exists with the projection data, especially in the case of low-dose scanning, so it is difficult to Obtain high-quality CT images. However, with the development of clinical diagnosis and treatment, the breadth and depth of clinical application of CT have gradually reached unprecedented heights. Under this new situation, the industry has new and improved safety considerations and image quality for CT use high demands. This makes it difficult for the filtered back projection method to meet new requirements. Even in low-end applications, the filtered back-projection method still requires a new and more accurate back-projection method to reduce artifacts and improve image quality.

In response to the above new requirements, in high-end applications, the iterative reconstruction method has been paid attention to and researched. The iterative reconstruction method can well deal with image artifacts caused by electronic noise and other physical factors, thereby reducing the X-ray dose during inspection while ensuring image quality. In the past, due to its huge amount of calculation, the imaging speed was slow and it could not be practically applied in clinical practice. In recent years, with the rapid development of computer hardware and computing science, it has become possible to apply iterative reconstruction methods to actual products, and as society pays more and more attention to medical health, the impact of X-ray radiation in CT diagnosis on human health is becoming more and more serious. With more and more people's attention, low X-ray radiation dose has become the future trend of CT development. Therefore, the iterative reconstruction method has attracted more and more attention and is a current research hotspot.

The iterative reconstruction process mainly includes multiple iterations of projection and back projection. In this process, through each round of iterations, the obtained image will gradually approach the ideal image, which can ensure good image resolution under low noise conditions. However, it often requires many iterations and takes a long time, which has become a major bottleneck in clinical application.

Therefore, in the current CT field, both the filtered back projection method and the iterative reconstruction method have their own technical problems, and it is urgent to take into account both CT image quality and CT image generation efficiency under the premise of ensuring safety.

Contents of the invention

Based on the above background, the object of the present invention is to provide a CT device, a CT image system and a CT image generation method, which can take into account both CT image quality and CT image generation efficiency under the premise of ensuring safety.

In order to achieve the above object, the present invention relates to a CT device, which scans the scanning area by X-rays to generate a CT image of the scanning object located in the scanning area, and is characterized in that it has: a reference object device, which is arranged in the scanning area a predetermined position in the scanning area; and a CT image generation device, which generates a CT image of the scanning object according to the known CT image information of the reference device and the scanning data of the scanning area obtained by scanning.

According to the CT device of the present invention, a reference object device with known CT image information (such as a given material) is added at a specified position in the scanning area near the scanning object (such as a human body), and the CT image of the scanning object is generated using The known CT image information of the reference device can take into account both CT image quality and CT image generation efficiency under the premise of ensuring safety.

In the above CT device, the CT image generation device may use the known CT image information of the reference device to iteratively reconstruct the scan data of the scan region obtained by scanning, thereby generating the A CT image of the scanned object.

Therefore, in the iterative reconstruction method, the CT image of the scanning object is generated by referring to the CT image information known by the reference object device, which can effectively reduce the number of iterations in the iterative reconstruction and improve the efficiency of the iterative reconstruction. Since the iterative reconstruction method has the characteristics of high security and high image quality, the efficiency of the iterative reconstruction method is further improved, thereby greatly improving the practicability of the iterative reconstruction method.

In the above-mentioned CT device, in the iterative reconstruction, the CT image generation device may be based on the known CT image of the reference device and the reference object where the reference device is located in the scan area The current reconstructed image and updated image of the region determine the step size used in this iteration.

In iterative reconstruction, the step size setting has a great influence on the convergence speed of iterative reconstruction. By using the known CT image of the reference device and the corresponding reconstructed image and updated image of the reference device in the iteration to adaptively determine the iteration step size, the step size can be determined more appropriately, thereby greatly accelerating the process of iterative convergence, Improve the efficiency of iterative reconstruction methods.

In the above CT apparatus, in the iterative reconstruction, when the difference between the scan data of the scan region and the projection data of the current reconstructed image of the scan region is smaller than a first predetermined threshold, the The above-mentioned CT image generation device uses a random step size within a specified range in this iteration.

In the above CT device, in the iterative reconstruction, if the step size used in the previous iteration is smaller than the second predetermined threshold, the CT image generation device uses a step size within a predetermined range in this iteration. random step size.

Here, aiming at the problem of local optimum that may occur in the process of using the reference device to determine the iterative step size, the random step size within the specified range is used to jump out of the local optimum, preventing oscillation and convergence within the local optimum range Slow or unable to converge, thereby improving the accuracy and efficiency of iterative convergence.

In the above-mentioned CT apparatus, in the iterative reconstruction, when the current reconstructed image of the reference object area reaches the specified quality, or the quality of the reconstructed image of the reference object area is lower than that of the current iteration In the case of lower than in the previous iteration, the CT image generation device stops the iteration.

In the prior art, sometimes the iteration termination condition cannot always be satisfied, resulting in the failure of the iterative reconstruction convergence determination. In this regard, setting the iteration termination condition based on the reconstructed image corresponding to the reference object device can solve the above-mentioned problems in the prior art, and accurately and reliably perform iterative reconstruction convergence determination.

In the above CT apparatus, when the difference between the known CT image of the reference object device and the current reconstructed image of the reference object area is smaller than a third predetermined threshold, the CT image generates The device determines that the current reconstructed image of the reference object area has reached the specified quality, and the difference between the known CT image of the reference object device and the reconstructed image of the reference object area is greater than the previous one in this iteration In the case of an iteration, the CT image generation device determines that the quality of the reconstructed image of the reference object region is lower in the current iteration than in the previous iteration.

Here, specific iteration termination conditions based on the reference device are provided. According to the above iteration termination condition, iterative reconstruction convergence determination can be accurately and reliably performed.

In the above-mentioned CT device, in the iterative reconstruction, the CT image generation device may further perform regularization filtering, and the regularization filtering process is to reconstruct the scanning data of the scanning area and the scanning area The residual image corresponding to the difference between the projection data of the image is subjected to a regularization filtering process, and in the regularization filtering process, the CT image generation device determines a regularization according to the noise of the reconstructed image of the reference object area. The filter parameters used in the filtering.

In the prior art, when regularized filtering is performed in iterative reconstruction, filter parameters are set according to the noise intensity of the image, but due to errors in the obtained noise intensity, appropriate filter parameters cannot be set. In this regard, by using the noise of the reconstructed image corresponding to the reference object device, the problems in the prior art can be solved, and the parameters of the filter are adaptively set in the regularization filtering process.

In the above-mentioned CT device, the reference object device may be integrally or separately arranged in an annular area surrounding the scanning object, and in the iterative reconstruction, the CT image generation device compares the original image to The pixel values of the outer regions of the annular region are set to 0.

When the reference object device is arranged in a ring around the scanning object, it can be considered that there is no scanning object in the area outside the ring area. Therefore, by setting the pixel value of the area outside the annular area to 0, the initial image can be made closer to the final result image, thereby speeding up the process of convergence and improving the efficiency of iterative reconstruction.

In the above CT device, the reference device may be one of an integral ring, an integral rectangle, multiple separate rectangles, and multiple separate circles, uniformly surrounding the scanning object configuration.

Here, several preferred configuration forms of the reference device are specifically listed. In this way, it is possible to more conveniently grasp the arrangement position of the reference object device in the scanning area, so as to better grasp the CT image information of the reference object device.

In addition, in order to achieve the object of the present invention, the present invention also relates to a CT image system characterized by comprising the above-mentioned CT device and a CT image output device for outputting a CT image of the scanning object generated by the CT image generating device. .

According to the CT image system of the present invention, a reference object device with known CT image information (such as a given material) is added at a specified position in the scanning area near the scanning object (such as a human body), and when generating a CT image of the scanning object By using the CT image information known by the reference device, the quality of the CT image and the efficiency of CT image generation can be taken into account while ensuring safety. As a result, CT images of higher quality can be output (for example, displayed) at a faster speed.

In addition, in order to achieve the purpose of the present invention, the present invention also relates to a method for generating a CT image, which generates a CT image of a scanning object located in the scanning area according to the scanning data obtained by scanning the scanning area with X-rays, which It is characterized in that the scan data is iteratively reconstructed using known CT image information of a reference device set at a specified position in the scan area, thereby generating a CT image of the scan object; in the In iterative reconstruction, the image obtained by back-projecting the scanned data is used as the initial reconstructed image, and the following steps (1) to (3) are repeated until the iteration stop condition is satisfied: (1) The scanned data based on the scanned area The difference between the projection data and the reconstructed image of the scanning area is obtained to obtain an updated image of the scanning area; (2) According to the known CT image of the reference device and the reference in the scanning area The reconstructed image and the updated image of the reference object area where the object device is located determine the step size used in this iteration; (3) according to the reconstructed image and updated image of the scanned area, use the determined step size to obtain the A new reconstructed image of the scanning area; when the iteration stop condition is satisfied, a CT image of the scanning object is generated according to the current reconstructed image of the scanning area.

According to the CT image generating method of the present invention, when generating a CT image of a scanning object (such as a human body) based on an iterative reconstruction method, the CT image information set at a specified position in the scanning area near the scanning object is known (such as a given Material) reference device, especially by using the known CT image of the reference device and the corresponding reconstructed image and updated image of the reference device in the iteration to adaptively determine the iteration step size. Therefore, the number of iterations in iterative reconstruction can be effectively reduced, and the efficiency of iterative reconstruction can be improved. Since the iterative reconstruction method has the characteristics of high security and high image quality, the efficiency of the iterative reconstruction method is further improved, thereby greatly improving the practicability of the iterative reconstruction method.

According to the CT apparatus, CT image system and CT image generation method of the present invention, both CT image quality and CT image generation efficiency can be taken into consideration under the premise of ensuring safety. Wherein, the present invention is not limited to the forms listed above. For example, in the CT image generation method of the present invention, not only the above-mentioned step size adaptive determination method based on the reference device can be used, but also the above-mentioned random step setting method, the reference device-based The iterative reconstruction convergence judgment method, the adaptive regularization filter parameter setting method based on the reference device, the iterative initialization image correction method based on the reference device, etc. Moreover, each step in the CT image generating method of the present invention can also be implemented as a functional module.

Description of drawings

Figure 1 is a structural diagram of a CT image system.

FIG. 2A is a schematic diagram of an example of the arrangement position of the reference object device.

Fig. 2B is a schematic diagram of several configurations of the reference device.

Figure 3 is a diagram of the basic steps of iterative reconstruction.

FIG. 4A is a schematic diagram of an iterative process when a fixed step size is used in the past and the step size is relatively large.

FIG. 4B is a schematic diagram of an iterative process when the step size is small when a fixed step size is used in the past.

FIG. 4C is a schematic diagram of an iterative process using an adaptive step size.

FIG. 4D is a schematic diagram of calculating an adaptive step size based on a reference device.

FIG. 4E is a comparison diagram of experimental results in the case of a fixed step size and an adaptive step size.

Figure 5 is a comparison diagram of the experimental results in the case of adaptive step size and adaptive + random step size.

Fig. 6 is a schematic diagram of the convergence determination of iterative reconstruction based on the reference device.

Fig. 7 is a schematic diagram of parameter setting of an adaptive regularization filter based on a reference object device.

Figure 8 depicts an iterative initialization image correction process based on a reference device.

FIG. 9 is a flowchart showing a CT image generation method of the present invention.

FIG. 10 is a flowchart showing an example of the CT image generation method of the present invention.

detailed description

First, the configuration of the CT image system according to the present invention will be described. Figure 1 is a structural diagram of a CT image system. As shown in FIG. 1 , the CT image system mainly includes a CT device 1 and a CT image output device 2 . The CT apparatus 1 uses, for example, a conventional X-ray scanner to scan a scanning area with X-rays, and generates a CT image of a scanning object located in the scanning area. Here, the scanning object is, for example, a human body or the like. The CT image output device 2 outputs a CT image of a scanning target generated by the CT device 1 . Here, the CT image output device 2 is typically a CT image display device, and displays the CT image of the scanning object generated by the CT device 1 on a screen. Of course, the CT image output device is not limited to the CT image display device, and may also be a data transmission interface for sending the CT image generated by the CT device 1 through a network, a printer for printing the CT image generated by the CT device 1, and the like.

The characteristic structure of the CT apparatus 1 involved in the present invention mainly includes a reference object device 11 and a CT image generating device 12 . The reference object device 11 is set at a specified position in the scanning area, and the reference object device 11 will be described in detail later. The CT image generating device 12 is realized by, for example, a general-purpose processor or a dedicated integrated circuit, and generates a CT image of the scanning object according to the known CT image information of the reference device 11 and the scanning data of the scanning area obtained through scanning.

Hereinafter, the reference object device 11 proposed by the present invention will be described in detail. The present invention adds a reference object device 11 in the CT image system. FIG. 2A is a schematic diagram of an example of the arrangement position of the reference object device. As shown in FIG. 2A , in the conventional X-ray scanner of the CT apparatus 1 , the rotation track 201 is the rotation track of the X-ray source 202 and the detector 203 . As an example of the reference object device 11 , a reference object 204 is installed around a body scanning area 205 to be scanned. Here, the reference material can be made of any solid high-purity material, such as non-metallic materials such as silicon, organic materials such as synthetic polymer materials, or metal materials such as iron and copper, and the higher the purity and consistency of the material, the better. This helps the CT image value corresponding to the reference object to be a constant, and the CT image information such as the CT image value corresponding to the reference object manufactured by each material can be obtained through prior experimental testing and is known. Fig. 2B is a schematic diagram of several configurations of the reference device. As shown in FIG. 2B , the reference objects 204 distributed around the human body scanning area (the elliptical area near the center of the figure) can be in the form of an integral ring, an integral rectangle, multiple separate rectangles, or multiple separate circles. A certain type is uniformly arranged around the scanning area of the human body as the scanning object. Here, the shape and position of the reference object 204 are not limited, but the position of the reference object 204 in the scanning area needs to be known, so as to obtain the pixel area corresponding to the reference object in the CT image (also referred to as the reference object area). Here, preferably, the reference objects are evenly distributed around the scanning area of the human body as the scanning object, which will be illustrated in a circle as a schematic diagram in the present application.

As an embodiment of the present invention, the CT image generation device 12 is implemented by a processor, mainly including a general-purpose processor unit for basic program and parameter control and an iterative reconstruction processing unit dedicated to iterative reconstruction. The CT image generation device 12 uses known CT image information of the reference device 11 to iteratively reconstruct the scan data of the scan area obtained by scanning, thereby generating a CT image of the scan object.

Figure 3 is a diagram of the basic steps of iterative reconstruction. The following is a brief description. For example, the CT device 1 uses a basic X-ray scanner to actually scan the scanning area (step 301) to obtain the actual scan data S, and the actual scan data S undergoes a filtered back-projection process (step 309) to obtain an initial iterative reconstruction image Projecting the initial image (step 308) to obtain calculated projection data Then for S and Calculate the difference (step 302) to the projection residual ΔS, then back-project the projection residual ΔS (step 303) to obtain the residual image ΔI, and perform regularized filtering on the residual image ΔI (step 304) to obtain an updated image ΔUI, and then judge whether all pixels of ΔUI are approximately 0 (step 306), if not approximately 0, then accumulate and update (step 307) to reconstruct the image, and accumulate and update is to weight the updated image ΔUI with a step size (also called relaxation factor) After α, it is added to the reconstructed image obtained by the previous cycle iteration, that is,

(Formula 1)

The initial reconstructed image of the first iteration is Reconstruct the image again Perform projection to obtain calculated projection data for the next round of iterative process until all pixels of the updated image ΔUI are approximately 0. At this time, the reconstructed image That is, the final reconstructed image for the entire iterative reconstruction.

In iterative reconstruction, the iterative process can be considered as the process of minimizing the objective function J. In order to solve I, the objective function J is minimized, that is

(Formula 2)

Among them, A is the system matrix, I is the CT image, S is the actual scan data, and P(I) is the image prior information item, which is reflected by the regularized filter in the iterative process. To minimize and solve the above objective function, the gradient descent method is generally used to obtain

(Formula 3)

Where α is the step size (relaxation factor) in the cumulative update process. The size setting of the step size (relaxation factor) has a great influence on the convergence speed of iterative reconstruction, which will be described in detail below.

FIG. 4A is a schematic diagram of an iterative process when a fixed step size is used in the past and the step size is relatively large. As shown in Figure 4A, if a larger step size (relaxation factor) α is set, when the value of the objective function J approaches the minimum value I _∞ or encounters a local minimum value, the update process will cause the objective function value to be at the minimum value or It oscillates near the local minimum, and cannot quickly converge to the minimum I _∞ even after many iterations.

FIG. 4B is a schematic diagram of an iterative process when the step size is small when a fixed step size is used in the past. As shown in Fig. 4B, when a small step size (relaxation factor) α is set, each iteration updates very little, the iterative process is very slow, and it also takes many iterations to converge to the minimum value I _∞ .

Therefore, if the size of the step size (relaxation factor) α can be adaptively adjusted according to the quality of the reconstructed image in the iterative process, the process of iterative convergence will be greatly accelerated. FIG. 4C is a schematic diagram of an iterative process using an adaptive step size. As shown in FIG. 4C , compared with the situation in FIG. 4A and FIG. 4B , the iterative convergence process is greatly accelerated.

In order to adaptively adjust the size of the step size (relaxation factor) α according to the quality of the reconstructed image in the iterative process, the adaptive step size (relaxation factor) is calculated according to the reference device 11 in the present invention. In iterative reconstruction, the CT image generation device 12 determines the current reconstruction image and updated image of the reference object area where the reference object device 11 is located in the scanning area based on the known CT image of the reference object device 11, and determines the current reconstruction image used in this iteration. the step size. As a specific example, as shown in Fig. 4D, for a given iterative round, a step size (relaxation factor) α is selected, so that under this step size (relaxation factor) α, the reference object area in the reconstructed image is accumulating After updating, the square of the difference between the CT image values corresponding to the reference material is the smallest, which can be expressed as

(Formula 4)

Wherein I _r is the known CT image of the reference object device 11, which represents the reference object image area, and its value is a constant, that is, the CT image value corresponding to the reference object material, is the reconstructed image of the current reference object image area, and ΔUI _r is the updated image of the current reference object image area.

FIG. 4E is a comparison diagram of experimental results in the case of a fixed step size and an adaptive step size. In the simulation experiment shown in Figure 4E, the iteration termination condition is set to be less than 50 residuals, it can be seen that using the adaptive step size (relaxation factor) method can effectively improve the convergence speed of iterations.

As mentioned above, by using the known CT image of the reference device 11 and the corresponding reconstructed image and updated image of the reference device 11 in the iteration to determine the iteration step size, the step size can be determined more appropriately, thereby greatly speeding up the iteration The process of convergence improves the efficiency of the iterative reconstruction method.

Hereinafter, a modified example of the above-mentioned adaptive step size determination based on the reference object device will be described in detail. In the process of determining the above-mentioned adaptive step size, the image of the reference object region is only a local area in the entire CT image, therefore, the optimal step size (relaxation factor) based on this region may make the iterative process fall into a local optimum, that is, although The reference object area has been reconstructed to a very high accuracy, but the whole image reconstruction result is still not optimal, which leads to stop or slow iteration. In this case, using a random step size can effectively jump out of the local optimum.

As a case of using a random step to jump out of the local optimum, in iterative reconstruction, when the residual error is small but the iteration termination condition has not been met, the random step is used to jump out of the local optimum. That is, when the difference between the scan data of the scan region and the projection data of the current reconstructed image of the scan region (that is, the residual) is smaller than the first prescribed threshold T1, the CT image generation device 12 uses the prescribed A random step in the range. Here, the above-mentioned residual may use a projection residual ΔS, or may use a residual image ΔI. The first specified threshold T1 can generally be set to 5% of the residual error of the first round of iterations or experimentally determined according to specific implementation conditions. In addition, the random step size is a random number controlled within a certain range, and the range of this random number is determined according to actual experiments.

As another case of using a random step size to jump out of the local optimum, when the step size is small in iterative reconstruction but the residual has not yet met the iteration termination condition, the random step size is used to jump out of the local optimum. That is, when the step size used in the previous iteration is smaller than the second predetermined threshold, the CT image generation device 11 uses a random step size within a predetermined range in the current iteration. Here, the second prescribed threshold can generally be set to 5% of the first-round step size (relaxation factor) or experimentally determined according to specific implementation conditions. In addition, the random step size is a random number controlled within a certain range, and the range of this random number is determined according to actual experiments.

Figure 5 is a comparison diagram of the experimental results in the case of adaptive step size and adaptive + random step size. As shown in Figure 5, in the case of low residual error (corresponding to the first case above), using the random step size can make the iteration converge to the set value quickly when the subsequent iteration of the adaptive step size converges slowly. Iteration termination target (iteration termination condition met). Likewise, in the case of a low step size (corresponding to the second case above), a similar effect can also be obtained. That is, for the problem of local optimum that may occur in the process of using the reference object device 11 to determine the iterative step size, by using a random step size to jump out of the local optimum in the above situation, it is possible to prevent oscillation in the local optimum range Convergence is slow or unable to converge, thereby improving the accuracy and efficiency of iterative convergence.

Hereinafter, another embodiment of the present invention will be described in detail, that is, iterative reconstruction convergence determination based on a reference object device.

In the traditional iterative reconstruction, as shown in Figure 3, the termination condition of the iteration is often based on whether the updated image is approximately 0, or whether the projection residual is approximately 0, but due to the projection and back-projection models is an approximate model, the update image and projection residual may not always be approximately 0, causing the method to sometimes fail. Therefore, the present invention uses the reference device 11 to propose a convergence determination method based on the reference device. That is, in iterative reconstruction, when the current reconstructed image of the reference object area reaches the specified quality, or the quality of the reconstructed image of the reference object area in this iteration is lower than in the previous iteration, CT image generation The means 12 stops iterating.

As an example of a specific iteration termination condition based on the reference device, when the difference between the known CT image of the reference device 11 and the current reconstructed image of the reference region is less than a third specified threshold, the CT image is generated The device 12 judges that the current reconstructed image of the reference object area has reached the specified quality. In addition, when the difference between the known CT image of the reference object device 11 and the reconstructed image of the reference object area is larger in this iteration than in the previous iteration, the CT image generation device 12 determines that the reconstruction of the reference object area The quality of the images in this iteration is lower than in the previous iteration. An example of specific iteration termination conditions based on the reference object device will be described below with reference to the accompanying drawings. Fig. 6 is a schematic diagram of the iterative reconstruction convergence determination based on the reference device. As shown in Figure 6, when the square of the difference between the reference object area and the CT image value corresponding to the reference object material in the reconstructed image is less than the threshold T _stop , or the square of the difference in this iteration is greater than the square of the difference in the previous iteration , then stop the iteration, and the formula can be expressed as

or

(Formula 5)

Among them, T _stop can be specified by the user according to the required image quality level. Here, the difference between the known CT image of the reference object device 11 and the reconstructed image of the reference object area is not limited to the square of the difference between the CT image values corresponding to the reference object area and the reference object material in the reconstructed image, and may also be It can be represented by other appropriate values such as the absolute value of the difference, and at this time, a corresponding threshold can be appropriately set.

In the above-mentioned iterative reconstruction convergence determination based on the reference device, since the CT image of the reference device 11 is known, it can be more accurately and reliably compared with the conventional case of using an approximate model based on updated images or residuals. Iterative reconstruction convergence determination.

Hereinafter, another embodiment of the present invention will be described in detail, that is, the regularization filter parameter is adaptively set based on the reference object device.

In iterative reconstruction, regularized filtering corresponds to prior information items in the objective function. For example, adjacent pixels in the image often have similar pixel values. Such prior information is fused into the objective function through regularized filtering. Some smoothing filters or edge-preserving complex filters are often used, such as Gaussian smoothing filter, bilateral filter, Geman filter, etc. The parameters in these filters are often set according to the noise intensity of the image. However, in the iterative process, the noise intensity of each round of iteration results is different. In practical applications, a fixed parameter is generally set or adjusted according to a part of the image with better consistency, such as the estimated noise intensity of a certain organ tissue area. , but the organ and tissue regions in the reconstructed image may not have complete consistency, and there are certain errors in the estimated noise intensity.

In view of the above-mentioned problems in the prior art, an adaptive regularization filter parameter setting method based on a reference object device is proposed here. That is, the CT image generation device 12 also performs regularization filtering processing, and performs regularization filtering processing on the residual image corresponding to the difference between the scan data of the scan area and the projection data of the reconstructed image of the scan area. During the processing, the CT image generator 12 determines filter parameters used for regularization filtering based on the noise of the reconstructed image of the reference object region. Fig. 7 is a schematic diagram of parameter setting of an adaptive regularization filter based on a reference object device. As shown in Figure 7, calculate the standard deviation SD _noise of the noise in the reference object area in the reconstructed image (step 701), and then set the parameters of the regularization filter according to this standard deviation (step 702), and the specific parameter setting depends on different filtering For example, for the Gaussian smoothing filter, the variance in the Gaussian filter is proportional to the standard deviation SD _noise , that is, stronger noise requires greater smoothing strength. The calculation method of SD _noise is as follows:

(Formula 6)

Among them, mean is the mean function, is the reference object area pixels in the image, and N is the total number of reference object area pixels in all images.

In the above implementation of adaptively setting regularization filter parameters based on reference device, using known reference device 11 to set regularization filter parameters can be more appropriate than setting based on empirical estimation in the prior art Set the regularization filter parameters accordingly. Especially in the case of a high degree of uniformity of the reference object arrangement 11, the regularization filter parameters can be set with high precision.

Hereinafter, another embodiment of the present invention is described in detail, that is, correcting an iterative initialization image based on a reference object device.

Figure 8 depicts an iterative initialization image correction process based on a reference device. In iterative reconstruction, the initial image often uses the traditional filtered back-projection result image, but when the actual projection data is incomplete, as shown in the left figure in Fig. 8, the obtained filtered back-projection image will have very big difference.

Considering that the iterative process is the optimization process of the objective function, if the initial state of the iterated image is closer to the final image, the iterative process will be accelerated. Therefore, in the case where the reference object device 11 is integrally or separately arranged in an annular area surrounding the scanning object, in iterative reconstruction, the CT image generation device 12 uses the data of the area outside the annular area in the initial image The pixel value is set to 0. That is, as shown in FIG. 8, in the area 801 outside the reference object device 802, since they are all air, the values of the final result image must be all 0, therefore, in the initialization image, set the pixels of the area 801 outside the reference object device The value is 0. This can make the initial image closer to the final result image, thereby speeding up the convergence process.

Hereinafter, the CT image generation method according to the present invention will be specifically described. The CT image generating method according to the present invention can be executed by the CT apparatus 1 , more specifically, can be executed by the CT image generating device 12 of the CT apparatus 1 . FIG. 9 is a flowchart showing a CT image generation method of the present invention. As shown in FIG. 9 , the CT image generation method according to the present invention generates a CT image of a scan object located in the scan area based on scan data obtained by scanning the scan area with X-rays. Wherein, the scan data is iteratively reconstructed by using the known CT image information of the reference device 12 set at a predetermined position in the scan area in the CT device 1 , thereby generating a CT image of the scan object.

In the iterative reconstruction, first in step S1, the image obtained by back-projecting the scan data is used as an initial reconstructed image. Then, in step S2, an updated image of the scanned area is obtained based on the difference between the scanned data of the scanned area and the projection data of the reconstructed image of the scanned area. In step S3, the step size used in this iteration is determined according to the known CT image of the reference object device 11 and the reconstructed image and updated image of the reference object area where the reference object device 11 is located in the scanning area. In step S4, according to the reconstructed image and updated image of the scanned area in the last iteration, a new reconstructed image of the scanned area is obtained by using the determined step size. In step S5, it is judged whether the iteration termination condition is satisfied, and if the iteration termination condition is not satisfied, steps S2-S4 are repeatedly executed. When the iteration stop condition is satisfied, a CT image of the scanning object is generated based on the current reconstructed image of the scanning area and the process ends.

According to the CT image generation method of the present invention, when using iterative reconstruction to generate a CT image of a scanning object (such as a human body), the CT image information at a specified position in the scanning area near the scanning object is known (such as a given Material) reference device, especially using the known CT image of the reference device and the corresponding reconstructed image and updated image of the reference device in the iteration to determine the iteration step size. Therefore, the number of iterations in iterative reconstruction can be effectively reduced, and the efficiency of iterative reconstruction can be improved. Since the iterative reconstruction method has the characteristics of high security and high image quality, the efficiency of the iterative reconstruction method is further improved, thereby greatly improving the practicability of the iterative reconstruction method.

In the CT image generation method of the present invention, not only the step size adaptive determination method based on the reference object device in the above-mentioned embodiment can be adopted, but also the random step-size setting method in the above-mentioned modified example, the above-mentioned In other embodiments, the iterative reconstruction convergence determination method based on the reference device, the adaptive regularization filter parameter setting method based on the reference device, the iterative initialization image correction method based on the reference device, etc. Hereinafter, an example of the CT image generation method of the present invention that combines the above-described embodiments and modified examples will be specifically described.

FIG. 10 is a flowchart showing an example of the CT image generation method of the present invention. First, in step 901, the filtered back-projection initialization image is corrected by using the iterative initialization image correction method based on the reference object device in the above-mentioned embodiment. Then in step 902, projection is performed to obtain projection data. In step 903, the projection residual is further calculated. In step 904, back-projection is performed on the projection residual to obtain a residual image. In step 905, the residual image is subjected to regularization filtering using the adaptive regularization filtering parameter set based on the reference object device in the above-mentioned embodiment to obtain an updated image. Then in step 906, it is judged whether the cumulative sum of the projection residuals is smaller than the threshold T1. If less, use the random adaptive step size (relaxation factor) in the above modification in step 908 . Otherwise, in step 907, the adaptive step size is estimated based on the reference device according to the above-mentioned embodiment. Then in step 909, weight the updated image with the random adaptive step size obtained in step 908 or the adaptive step size obtained in step 907, and accumulate and update to the reconstructed image. In step 910, the noise mean square value and standard deviation of pixels corresponding to the reference object area in the current reconstructed image are calculated. Here, the standard deviation calculated in step 910 is used to adaptively adjust the parameters of the regularization filter in the next iteration, and the mean square value calculated in step 910 is used to judge the convergence condition. In step 911, if the iteration termination condition based on the reference object device in the above-mentioned embodiment is met, then stop the iteration to obtain the final result; otherwise, return to step 902 for the next iteration.

In the above-mentioned example of the CT image generation method, it is needless to say that further modifications can be made based on the respective embodiments and modified examples of the CT apparatus 1 already described in this specification. For example, in step 906, it is also possible to judge whether the step size is smaller than the second specified threshold as mentioned above, and use a random step size to jump out of the local optimum when the step size is small in iterative reconstruction but the residual error has not yet met the iteration termination condition . In addition, in step 911 , it can be judged that the convergence condition is satisfied according to the fact that the current reconstructed image of the reference object region reaches a specified quality or according to that the quality of the reconstructed image of the reference object region in this iteration is lower than that in the previous iteration.

The specific embodiments of the present invention have been described above with reference to the drawings. Wherein, the specific implementation manners described above are only specific examples of the present invention, and are used for understanding the present invention, and are not used to limit the scope of the present invention. Those skilled in the art can make various modifications, combinations and rational omission of elements to the specific implementation based on the technical idea of the present invention, and the resulting modes are also included in the scope of the present invention.

Claims (10)

1. a kind of CT devices, are scanned by X-ray to scanning area, generate the sweep object in the scanning area CT images, it is characterised in that possess：

Reference substance device, the assigned position being arranged in the scanning area；And

CT video generation devices, according to the known CT image informations of the reference substance device and by scanning the institute for obtaining The scan data of scanning area is stated, the CT images of the sweep object are generated,

The CT video generation devices utilize the known CT image informations of the reference substance device, to the institute obtained by scanning The scan data for stating scanning area is iterated reconstruction, thus generates the CT images of the sweep object,

In the iterative approximation, the CT video generation devices according to the known CT images of the reference substance device and The current reconstruction image and more new images of the reference object area that reference substance device described in the scanning area is located, determines this Step-length used in secondary iteration.

2. CT devices as claimed in claim 1, it is characterised in that

In the iterative approximation, in scan data and the current reconstruction image of the scanning area of the scanning area , less than in the case of the first defined threshold, the CT video generation devices are used in current iteration for difference between data for projection Arbitrary width in prescribed limit.

3. CT devices as claimed in claim 1, it is characterised in that

In the iterative approximation, less than in the case of the second defined threshold, the CT schemes the step-length used in last iteration As arbitrary width of the generating means used in current iteration in prescribed limit.

4. CT devices as claimed in claim 1, it is characterised in that

In the iterative approximation, in the case where the current reconstruction image of the reference object area reaches definite quality, or In the case that quality described in person with reference to the reconstruction image of object area is less than in last iteration in current iteration, the CT images Generating means stop iteration.

5. CT devices as claimed in claim 4, it is characterised in that

Difference between the known CT images of the reference substance device and the current reconstruction image of the reference object area In the case of less than the 3rd defined threshold, the CT video generation devices are judged as the current reconstruction figure of the reference object area As reaching definite quality,

Difference between the known CT images of the reference substance device and the reconstruction image of the reference object area is at this In the case of being more than in iteration in last iteration, the CT video generation devices are judged as the reconstruction figure of the reference object area The quality of picture is less than in once iteration in current iteration.

6. CT devices as claimed in claim 1, it is characterised in that

In the iterative approximation, the CT video generation devices also carry out regularization Filtering Processing, the regularization Filtering Processing Be the scan data to the scanning area and the scanning area reconstruction image data for projection between difference corresponding to Residual image carry out the process of regularization filtering,

In the regularization Filtering Processing, the CT video generation devices are made an uproar according to the reconstruction image of the reference object area Sound, determines the filter parameter used in regularization filtering.

7. CT devices as claimed in claim 1, it is characterised in that

It is disposed around in the annular region of the sweep object to reference substance device one or split,

In the iterative approximation, the CT video generation devices are by area more more outward than the annular region in initial pictures The pixel value in domain is set to 0.

8. CT devices as claimed in claim 1, it is characterised in that

Certain in annular, the rectangle of one, multiple rectangles of split, multiple circles of split that the reference substance device is integrated One kind, evenly around sweep object configuration.

9. a kind of CT picture systems, it is characterised in that possess：

CT devices any one of claim 1～8；And

CT image output devices, export the CT images of the sweep object generated by the CT video generation devices.

10. a kind of CT image generating methods, the scan data according to obtained from being scanned to scanning area by X-ray, are given birth to Into the CT images of the sweep object in the scanning area, it is characterised in that

Using the known CT image informations of the reference substance device at the assigned position being arranged in the scanning area, to described Scan data is iterated reconstruction, thus generates the CT images of the sweep object,

In the iterative approximation, so that image obtained from back projection is carried out to scan data as initial reconstruction image, instead Following step (1)～(3) are performed again until meeting iteration stopping condition：

(1) difference between scan data based on the scanning area and the data for projection of the reconstruction image of the scanning area It is different, obtain the more new images of the scanning area；

(2) it is located according to reference substance device described in the known CT images and the scanning area of the reference substance device Reference object area reconstruction image and more new images, determine current iteration used in step-length；

(3) according to the reconstruction image and more new images of the scanning area, the step-length for being determined is utilized, obtains the scanning The new reconstruction image in region；

When iteration stopping condition is met, according to the current reconstruction image of the scanning area, the CT figures of sweep object are generated Picture.