Image denoising method and radiotherapy system

Abstract

This application discloses an image denoising method and a radiotherapy system, relating to the field of medical technology, for optimizing the denoising effect of projected images and improving the quality of image reconstruction. The method includes: acquiring an original projected image; performing a logarithmic transformation on the pixel values ​​of the original projected image to obtain a logarithmic projected image; denoising the human tissue pixel region in the logarithmic projected image to obtain an initial denoised image; and denoising the low photon number pixel region in the initial denoised image to obtain a target denoised image.

Description

Technical Field

[0001] This application relates to the field of medical technology, and in particular to an image denoising method and a radiotherapy system. Background Technology

[0002] In cone-beam computer tomography (CBCT), the noise in the projected and reconstructed images is inversely proportional to the number of photons received by the detector. When scanning large patients, with low exposure doses, or when the gantry is at a high attenuation angle, the noise in the corresponding areas increases sharply due to the reduced photon count. Common noise reduction methods (such as Gaussian filtering and bilateral filtering) are difficult to effectively denoise these low-photon-count areas, easily forming noise stripes in the reconstructed images and obscuring the boundaries of normal tissue structures. Summary of the Invention

[0003] The purpose of this application is to provide an image denoising method and a radiotherapy system for optimizing the denoising effect of projected images and improving the quality of image reconstruction.

[0004] In a first aspect, an image denoising method is provided, comprising the following steps: acquiring an original projected image; performing a logarithmic transformation on the pixel values ​​of the original projected image to obtain a logarithmic projected image; denoising the human tissue pixel region in the logarithmic projected image to obtain an initial denoised image; and denoising the low photon number pixel region in the initial denoised image to obtain a target denoised image.

[0005] It is understood that the image denoising method provided in this application, based on logarithmic transformation and preliminary denoising of the original image, further denoising is performed on low photon number pixel regions, thereby achieving differentiated denoising processing for pixels with different levels of noise interference, thus improving the clarity and accuracy of the reconstructed image.

[0006] In some embodiments, denoising the human tissue pixel region in a logarithmic projection image to obtain an initial denoised image includes: segmenting different human tissue regions in the logarithmic projection image using image segmentation techniques to obtain multiple segmented regions; calculating the gray-level centroid of each segmented region for each of the multiple segmented regions; determining a noise estimation sub-region in the segmented region centered on the gray-level centroid; the noise estimation sub-region being a region where the ray attenuation is greater than an attenuation threshold; determining a corresponding relaxation factor based on the noise parameters of the noise estimation sub-region; the relaxation factor being used to adjust the denoising intensity of the segmented region; and performing denoising processing on the segmented region based on the relaxation factor to obtain the initial denoised image.

[0007] In some embodiments, determining the corresponding relaxation factor based on the noise parameters of the noise estimation sub-region includes: substituting the noise parameters of the noise estimation sub-region into the relaxation factor mapping function to calculate the corresponding relaxation factor.

[0008] In some embodiments, the relaxation factor mapping function is a mapping function obtained by fitting a sample of the correspondence between noise parameters and relaxation factors; wherein, the sample of the correspondence between noise parameters and relaxation factors is constructed based on the noise parameters and the relaxation factors corresponding to the noise parameters of multiple sets of historical reconstructed images; the multiple sets of historical reconstructed images are obtained by selecting multiple sets of different radiation doses and performing multi-angle scanning on calibration phantoms of different diameters.

[0009] In some embodiments, before denoising the low-photon-count pixel regions in the initial denoised image, the method further includes: segmenting the original projected image based on a low-photon-count threshold to determine the low-photon-count pixel regions in the original projected image; wherein, the low-photon-count pixel regions are regions composed of pixels in the original projected image whose photon count is less than the low-photon-count threshold; and determining the low-photon-count pixel regions in the initial denoised image based on the low-photon-count pixel regions in the original projected image.

[0010] In some embodiments, denoising the low-photon-count pixel region in the initial denoised image includes: performing segmented weighted denoising processing on a first pixel region in the low-photon-count pixel region; the first pixel region is a region composed of pixels with a photon count greater than or equal to an extremely low photon-count threshold; performing pixel value replacement processing on a second pixel region in the low-photon-count pixel region; the second pixel region is a region composed of pixels with a photon count less than an extremely low photon-count threshold.

[0011] In some embodiments, segmented weighted denoising processing is performed on a first pixel region in a low photon number pixel region, including: dividing the photon number of the first pixel region into multi-level photon number intervals according to at least one secondary photon number threshold; the secondary photon number threshold is less than a low photon number threshold and greater than an extremely low photon number threshold; filtering the initial denoised image with filter kernels of different sizes to obtain filtered images with different denoising strengths; the size of the filter kernel is negatively correlated with the photon number of the photon number interval; for the pixel to be processed in any photon number interval of the multi-level photon number interval, weighted processing is performed based on the target pixel value in the image corresponding to the photon number interval to obtain the weighted denoised pixel value of the pixel to be processed; wherein, the target pixel value is the pixel value of the pixel at the same position as the pixel to be processed in the image corresponding to the photon number interval; the image corresponding to the photon number interval includes the initial denoised image and / or the filtered image.

[0012] In some embodiments, weighted processing is performed on the target pixel value in the filtered image corresponding to the photon number interval to obtain the weighted denoised pixel value of the pixel to be processed, including: determining the weight of the target pixel value based on the photon number of the pixel to be processed and the end value of the photon number interval; and performing weighted calculation on the target pixel value based on the weight to obtain the weighted denoised pixel value of the pixel to be processed.

[0013] In some embodiments, the pixel value replacement process includes any one of the following: replacing the photon count of the pixels in the second pixel region with an extremely low photon count threshold, then performing a logarithmic transformation on the replaced pixels to obtain transformed pixels, and replacing the original pixel value in the second pixel region with the pixel value of the transformed pixels; or using an interpolation method to complete and replace the pixel value of the pixels in the second pixel region.

[0014] Secondly, a radiotherapy system is provided, comprising: an X-ray tube, a detector, a rotating gantry, and a processing device; the X-ray tube is used to emit X-rays to irradiate the object to be scanned; the detector is disposed opposite to the X-ray tube and is used to receive the X-rays passing through the object to be scanned and generate a raw projection image, and send the raw projection image to the processing device; the rotating gantry is used to drive the X-ray tube and the detector to rotate synchronously and to acquire multi-angle projection data of the object to be scanned; the processing device is communicatively connected to the detector and is configured to acquire the raw projection image data acquired by the detector and perform any of the optional image denoising methods described in the first aspect above.

[0015] Thirdly, an image denoising device is also provided, comprising: an acquisition unit and a processing unit; the acquisition unit is used to acquire an original projected image; the processing unit is used to perform logarithmic transformation on the pixel values ​​of the original projected image to obtain a logarithmic projected image; to denoise the human tissue pixel region in the logarithmic projected image to obtain an initial denoised image; and to denoise the low photon number pixel region in the initial denoised image to obtain a target denoised image.

[0016] In some embodiments, the processing unit is specifically used to segment different human tissue regions in a logarithmic projection image using image segmentation techniques to obtain multiple segmented regions; for each segmented region, the gray-level centroid of the segmented region is calculated; a noise estimation sub-region is determined in the segmented region with the gray-level centroid as the center; the noise estimation sub-region is a region where the ray attenuation is greater than the attenuation threshold; a corresponding relaxation factor is determined based on the noise parameters of the noise estimation sub-region; the relaxation factor is used to adjust the noise reduction intensity of the segmented region; and the segmented region is denoised based on the relaxation factor to obtain an initial denoised image.

[0017] In some embodiments, the processing unit is specifically used to substitute the noise parameters of the noise estimation sub-region into the relaxation factor mapping function to calculate the corresponding relaxation factor.

[0018] In some embodiments, the relaxation factor mapping function is a mapping function obtained by fitting a sample of the correspondence between noise parameters and relaxation factors; wherein, the sample of the correspondence between noise parameters and relaxation factors is constructed based on the noise parameters and the relaxation factors corresponding to the noise parameters of multiple sets of historical reconstructed images; the multiple sets of historical reconstructed images are obtained by selecting multiple sets of different radiation doses and performing multi-angle scanning on calibration phantoms of different diameters.

[0019] In some embodiments, the processing unit is specifically configured to segment the original projected image based on a low photon number threshold to determine a low photon number pixel region in the original projected image; wherein, the low photon number pixel region is a region composed of pixels in the original projected image whose photon number is less than the low photon number threshold; and to determine a low photon number pixel region in the initial denoised image based on the low photon number pixel region in the original projected image.

[0020] In some embodiments, the processing unit is specifically configured to perform segmented weighted noise reduction processing on a first pixel region in a low photon number pixel region; the first pixel region is a region composed of pixels with a photon number greater than or equal to an extremely low photon number threshold; and to perform pixel value replacement processing on a second pixel region in a low photon number pixel region; the second pixel region is a region composed of pixels with a photon number less than an extremely low photon number threshold.

[0021] In some embodiments, the processing unit is specifically configured to divide the photon count of the first pixel region into multi-level photon count intervals based on at least one secondary photon count threshold; the secondary photon count threshold is less than a low photon count threshold and greater than an extremely low photon count threshold; the initial denoised image is filtered using filter kernels of different sizes to obtain filtered images with different denoising strengths; the size of the filter kernel is negatively correlated with the photon count of the photon count interval; for the pixel to be processed in any photon count interval of the multi-level photon count interval, weighted processing is performed based on the target pixel value in the image corresponding to the photon count interval to obtain the weighted denoised pixel value of the pixel to be processed; wherein, the target pixel value is the pixel value of the pixel at the same position as the pixel to be processed in the image corresponding to the photon count interval; the image corresponding to the photon count interval includes the initial denoised image and / or the filtered image.

[0022] In some embodiments, the processing unit is specifically used to determine the weight of the target pixel value based on the photon count of the pixel to be processed and the end value of the photon count range; and to perform a weighted calculation on the target pixel value based on the weight to obtain the weighted and denoised pixel value of the pixel to be processed.

[0023] In some embodiments, the processing unit, specifically for pixel value replacement processing, includes any one of the following: replacing the photon count of the pixels in the second pixel region with an extremely low photon count threshold, then performing a logarithmic transformation on the replaced pixels to obtain transformed pixels, and replacing the original pixel value in the second pixel region with the pixel value of the transformed pixels; and using an interpolation method to complete and replace the pixel value of the pixels in the second pixel region.

[0024] Fourthly, an electronic device is also provided, comprising: a processor and a memory configured to store processor-executable instructions; wherein the processor is configured to execute the instructions to implement any of the optional image denoising methods of the first aspect described above.

[0025] Fifthly, this application provides a computer-readable storage medium storing instructions that, when executed by a device, enable the device to perform any of the optional image denoising methods described in the first aspect.

[0026] In a sixth aspect, this application provides a computer program product including computer instructions that, when executed on a processor of a device, enable the device to perform any of the optional image noise reduction methods described in the first aspect above. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the structure of a radiotherapy system provided in an embodiment of this application; Figure 2 A flowchart of an image noise reduction method provided in an embodiment of this application; Figure 3 A flowchart illustrating another image noise reduction method provided in this application embodiment; Figure 4 A schematic diagram of image segmentation provided for an embodiment of this application; Figure 5 A flowchart illustrating yet another image noise reduction method provided in this application embodiment; Figure 6 A flowchart illustrating yet another image noise reduction method provided in this application embodiment; Figure 7 This is a schematic diagram of the structure of an image noise reduction device provided in an embodiment of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0029] Figure reference numerals: radiotherapy system 100, radiation source 101, detector 102, rotating gantry 103, processing equipment 104. Detailed Implementation

[0030] In the embodiments of this application, the terms "first," "second," "third," "fourth," "fifth," and "sixth" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," "third," "fourth," "fifth," and "sixth" may explicitly or implicitly include one or more of that feature.

[0031] In embodiments of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0032] "A and / or B" includes the following three combinations: A only, B only, and a combination of A and B.

[0033] In the embodiments of this application, "parallel," "perpendicular," and "equal" include the described situation and situations similar to the described situation, where the range of similarity is within an acceptable deviation range, which is determined by those skilled in the art taking into account the measurement under discussion and the error associated with the measurement of a particular quantity (i.e., the limitations of the measurement system). For example, "parallel" includes absolute parallelism and approximate parallelism, where the acceptable deviation range for approximate parallelism can be, for example, a deviation within 5°; "perpendicular" includes absolute perpendicularity and approximate perpendicularity, where the acceptable deviation range for approximate perpendicularity can also be, for example, a deviation within 5°. "Equal" includes absolute equality and approximate equality, where the acceptable deviation range for approximate equality can be, for example, a difference between the two equals being less than or equal to 5% of either one.

[0034] The noise in the projected image (reconstructed image) of CBCT imaging is inversely proportional to the number of photons reaching the detector. When scanning large patients with low exposure doses or when the gantry is located at a high attenuation angle, the noise will increase sharply when the number of photons in the projected image is below a certain threshold. This makes it impossible for ordinary noise reduction methods (such as Gaussian filtering, bilateral filtering, etc.) to effectively remove noise, resulting in severe noise stripes in the reconstructed image that obscure the boundaries of normal tissue structures.

[0035] Penalized Weighted Least-Squares (PWLS) is based on a noise statistical model. In the fidelity term of the loss function, it applies less weight to low-photon pixels, thus effectively avoiding noise stripes introduced by low-photon pixels in the reconstructed image. However, this method relies on iterative solving of the loss function, which is inefficient; and the design of the regularization term (smoothing term) often fails to preserve the sharp boundaries of objects while effectively denoising.

[0036] The 3D Block Matching 3D (BM3D) denoising method is based on similar block matching. It uses two-dimensional similar blocks to construct three-dimensional data. In the three-dimensional frequency domain, the object's structural information becomes a low-frequency component, thus effectively removing high-frequency noise. It is a recognized powerful edge-preserving denoising method. However, this method does not consider the photon statistical characteristics of CBCT imaging and cannot effectively remove noise from pixels with a sharp increase in low photon counts. Furthermore, since CBCT projection covers a large area of ​​the patient's scan, the noise in different regions is not consistent. Using the same relaxation factor may result in incomplete noise removal or a smooth transition.

[0037] To address the aforementioned technical problems, this application provides an image denoising method and a radiotherapy system that can optimize the denoising effect of projected images and improve the quality of image reconstruction. The image denoising method is as follows: acquiring an original projected image; performing a logarithmic transformation on the pixel values ​​of the original projected image to obtain a logarithmic projected image; denoising the human tissue pixel region in the logarithmic projected image to obtain an initial denoised image; and denoising the low photon number pixel region in the initial denoised image to obtain a target denoised image.

[0038] The present application will now be described in further detail with reference to the accompanying drawings and specific embodiments. It is to be understood that the specific embodiments described herein are merely illustrative of the present application and not intended to limit the scope of the application.

[0039] Figure 1 This is a schematic diagram of the structure of a radiotherapy system provided in an embodiment of this application, as shown below. Figure 1 As shown, the radiotherapy system 100 mainly includes: a radiation source 101, a detector 102, a rotating gantry 103, and a processing device 104.

[0040] The X-ray source 101 (e.g., an X-ray tube) is mounted on a rotating gantry 103 as an X-ray generating device to generate and emit a high-energy X-ray beam to penetrate the object to be scanned placed on the treatment table. Detectors 102 are positioned opposite the X-ray tube on either side of the rotating gantry. Detectors 102 receive the X-rays after attenuation by the tissue of the object being scanned, and convert the optical signal into a digital signal through photoelectric conversion and analog-to-digital conversion, thereby generating raw projection image data. This data contains anatomical information about the object being scanned. Detectors 102 transmit the generated raw projection image data to a processing device 104 connected to them in real time. The rotating gantry 103 has sufficient mechanical strength and rotational accuracy to support the X-ray source 101 and detector 102 and allow them to rotate synchronously around the object being scanned at an angle of 360 degrees or greater.

[0041] In some embodiments, during rotation, the X-ray source 101 continuously or pulses, and the detector 102 synchronously acquires data, thereby realizing the acquisition of multi-angle projection data of the object to be scanned, providing a data basis for subsequent image-guided radiotherapy or cone-beam CT image reconstruction.

[0042] In some embodiments, the processing device 104 establishes a wired and / or wireless communication connection with the detector 102 to receive and store the raw projection image data acquired by the detector 102.

[0043] For example, the processing device 104 may typically be a computer, server, or dedicated image processing workstation with high-performance computing capabilities.

[0044] In some embodiments, the processing device 104 is used to implement the image noise reduction method provided in the embodiments of this application.

[0045] It should be noted that the application scenarios of the embodiments in this application are not limited. The system architecture and business scenarios described in the embodiments of this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of medical technology and the emergence of new business scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0046] The image noise reduction method provided in the embodiments of this application will now be described in conjunction with the accompanying drawings.

[0047] Figure 2 A flowchart of an image denoising method provided in an embodiment of this application is shown below. Figure 2 As shown, the method includes the following steps: S101. Obtain the original projected image.

[0048] In some embodiments, the original projection data is, for example, two-dimensional projection data containing noise acquired by a CBCT scanning device. The original projection data of the object to be scanned is acquired by the CBCT scanning device to obtain the original projection image, which contains noise caused by radiation dose limits.

[0049] S102. Perform a logarithmic transformation on the pixel values ​​of the original projected image to obtain a logarithmic projected image.

[0050] In some embodiments, a logarithmic transformation is performed on each pixel value of the original projected image, and the logarithmically transformed image is used as a logarithmically projected image. Through logarithmic transformation, the exponentially decaying ray signal can be converted into a linearly decaying signal that is easy to process.

[0051] S103. Denoise the human tissue pixel region in the logarithmic projection image to obtain the initial denoised image.

[0052] In some embodiments, an initial denoising strategy is used to denoise the pixel region corresponding to human tissue in the logarithmic projection image to obtain an initial denoised image, thereby initially suppressing statistical noise in the uniform tissue region.

[0053] S104. Denoise the low-photon-count pixel region in the initial denoised image to obtain the target denoised image.

[0054] In some embodiments, a further denoising strategy is employed to enhance denoising processing for low-photon-count pixel regions in the initial denoised image, thereby obtaining the final target denoised image, thus focusing on denoising regions with extremely low signal-to-noise ratios in the image.

[0055] It is understood that, based on logarithmic transformation and preliminary noise reduction of the original image, this application further performs noise reduction processing on low photon number pixel regions, thereby achieving differentiated noise reduction processing for pixels with different levels of noise interference, thus improving the clarity and accuracy of the reconstructed image.

[0056] Figure 3 A flowchart of another image denoising method provided in the embodiments of this application is shown below. Figure 3 As shown, the above S103 can be implemented in the following steps: S201. Image segmentation technology is used to segment different human tissue regions in the logarithmic projection image to obtain multiple segmented regions.

[0057] In some embodiments, region segmentation can be achieved through multi-dimensional image feature analysis: First, the outer contour of the human body is extracted from the logarithmic projection image, and the gray-scale amplitude distribution curve is calculated along the long axis of the human body to reflect the changes in the equivalent thickness of the tissue. Then, the local fluctuation characteristics of the gray-scale curve are analyzed. The obvious gray-scale fluctuations in the chest caused by the lung and heart structures, the smooth high gray-scale areas in the abdomen and pelvis caused by solid organs, and the contours and gray-scale features of the head and neck due to their near-circular shape and small thickness are used to comprehensively analyze the three criteria of outer contour shape, thickness amplitude, and gray-scale trend changes, and the head, neck, chest, abdomen, and pelvis are segmented from top to bottom.

[0058] In some embodiments, deep learning technology can be used to achieve automated segmentation. A training dataset containing a large number of log-projected images and their pixel-level region annotations is pre-constructed, and a semantic segmentation network (such as the U-Net architecture) is designed. Through end-to-end training, the network learns the mapping relationship from the input image to the region classification probability map. In practical applications, the log-projected image to be processed is directly input into the trained artificial intelligence network, and the network inference outputs pixel-level segmentation results in one go, thereby dividing multiple tissue regions such as head, neck, chest, abdomen, and pelvis.

[0059] S202. For each segmented region among multiple segmented regions, calculate the gray-scale centroid of the segmented region.

[0060] In some embodiments, for each segmented region among the multiple segmented regions, the gray-scale centroid coordinates of that region are calculated. The gray-scale centroid is a weighted center with pixel gray values ​​as the weight. It is obtained by multiplying the coordinates of each pixel in the region by the gray value of that pixel, summing the weights, and then dividing by the sum of the gray values ​​of all line bundles in the region.

[0061] It is understandable that, since the gray value in a logarithmic projection image is proportional to the degree of ray attenuation, the gray centroid of the segmented region represents the location of the most severe ray attenuation within the human tissue of each segmented region.

[0062] S203. Using the gray-scale centroid as the center, determine the noise estimation sub-region in the segmented region.

[0063] The noise estimation sub-region is the region where the ray attenuation is greater than the attenuation threshold.

[0064] In some embodiments, an initial rectangular or circular region is defined centered on the grayscale centroid. A preset attenuation threshold is used to filter out all pixels with grayscale values ​​higher than this threshold within the initial region. This set of high-attenuation pixels is then determined as the final noise estimation sub-region. If the number of selected pixels is too small, the initial region is expanded or the threshold is adjusted to ensure the sub-region has a sufficient sample size. This sub-region corresponds to areas with significant ray attenuation and noise, and can be used for subsequent noise feature analysis.

[0065] Near the centroid mentioned above, pixel regions that meet the high attenuation criteria are further selected. By setting an attenuation threshold, it is ensured that the selected sub-regions are located entirely in areas with sufficiently high ray attenuation, avoiding the inclusion of pixels from low attenuation regions (such as air or thin tissue) that could lead to noise underestimation.

[0066] For example, such as Figure 4 As shown, Figure 4 In the diagram, solid boxes represent segmented regions, dots represent the gray-level centroids of each segmented region, and dashed boxes represent noise estimation sub-regions within the segmented regions.

[0067] S204. Determine the corresponding relaxation factor based on the noise parameters of the noise estimation sub-region.

[0068] The relaxation factor is used to adjust the noise reduction intensity of the segmented region.

[0069] In some embodiments, noise estimation is performed on multiple noise estimation sub-regions, and the noise is substituted into the noise-BM3D relaxation factor mapping function to obtain the relaxation factor of each region.

[0070] As one possible implementation, determining the corresponding relaxation factor based on the noise parameters of the noise estimation sub-region can be achieved by substituting the noise parameters of the noise estimation sub-region into the relaxation factor mapping function, thereby calculating the corresponding relaxation factor.

[0071] In some embodiments, the relaxation factor mapping function is a mapping function obtained by fitting a sample of the correspondence between noise parameters and relaxation factors; wherein, the sample of the correspondence between noise parameters and relaxation factors is constructed based on the noise parameters and the relaxation factors corresponding to the noise parameters of multiple sets of historical reconstructed images; the multiple sets of historical reconstructed images are obtained by selecting multiple sets of different radiation doses and performing multi-angle scanning on calibration phantoms of different diameters.

[0072] In the specific implementation process, in order to achieve the matching between noise parameters and relaxation factors, this embodiment adopts a pre-calibration method to determine the relaxation factor mapping function. This mapping function is a correspondence obtained based on fitting a large amount of experimental data. The specific calibration and storage process is as follows: First, multiple calibration phantoms with different diameters (e.g., 20cm, 30cm, 40cm) are prepared. By adjusting the tube voltage (kV) and tube current (mA) of the scanning equipment, different radiation dose conditions that may be encountered in clinical practice are simulated. N sets of representative projection data are then collected to ensure that the reconstructed images based on each set of data have differentiated noise levels. Next, for each set of data, considering the consistency of noise distribution in each projection image under the same scanning angle, noise parameters characterizing the noise level of that set of data are estimated by calculating the local variance or standard deviation of the projection domain. Subsequently, the parameter optimization stage begins: for each set of data, the operator manually adjusts the relaxation factor in the BM3D denoising algorithm and reconstructs the image after denoising. The noise value is measured by selecting a region of interest in the phantom region of the reconstructed image until the value reaches the preset ideal target noise level. The corresponding relaxation factor at this point is recorded as the optimal relaxation factor for that set of data.

[0073] Through the above steps, the projection noise parameters and their corresponding optimal relaxation factors in each experimental group are recorded sequentially, forming N sets of discrete data points. Finally, a curve fitting algorithm (such as polynomial fitting or exponential fitting) is used to process these data points, establishing a continuous mapping function from the projection noise parameters to the relaxation factors, and the coefficients of this function are saved in the system. During actual scanning, the system only needs to call this function to quickly match a suitable relaxation factor based on the real-time calculated noise parameters.

[0074] S205. Denoising is performed on the segmented region based on the relaxation factor to obtain the initial denoised image.

[0075] In some embodiments, different relaxation factors are used to denoise each segmented region, achieving complete denoising of each segmented region while preserving the structural boundaries of the object itself. First, appropriate relaxation factors are assigned to each segmented region. Then, a preset iterative denoising algorithm is used to process each segmented region, such as using a total variation-based iterative denoising model to complete iterative denoising and obtain an initial denoised image.

[0076] For example, for tissue regions with significant noise but simple structure, a larger relaxation factor (e.g., 0.8-1.0) is used to accelerate noise reduction convergence; for edge or textured regions, a smaller relaxation factor (e.g., 0.2-0.5) is used to avoid excessive smoothing that leads to blurred boundaries.

[0077] Through the above process, the relaxation factor was used to reduce noise in different segmented regions, thus completing the initial noise reduction process of the original projected image.

[0078] In some embodiments, to ensure that further denoising of low-photon-count regions can be accurately located and maintain spatial consistency with the original data, the original projected image is first used for region segmentation before denoising the initial denoised image. Specifically, since the initial denoised image is a pre-processed image, its pixel values ​​may have changed. If the segmentation of low-photon-count regions is directly based on the photon count in the initial denoised image, the segmentation result may deviate from the original physical location. Therefore, this application adopts the following process: First, the original projected image is segmented based on a preset low-photon-count threshold, and all pixels in the original projected image with photon counts less than the low-photon-count threshold are marked to form low-photon-count pixel regions in the original projected image; then, according to the one-to-one correspondence between pixels in the original projected image and the initial denoised image, the low-photon-count pixel regions determined in the original projected image are mapped onto the initial denoised image, thereby determining the low-photon-count pixel regions in the initial denoised image. These low-photon-count pixel regions need to undergo subsequent denoising processing.

[0079] In this way, the area targeted by the noise reduction process strictly corresponds to the original low photon number area, avoiding misjudgment of the area caused by image preprocessing, and providing an accurate spatial range for subsequent segmented weighted noise reduction or pixel value replacement.

[0080] To more precisely distinguish the noise interference of different degrees in the low photon number region and avoid image artifacts caused by direct denoising, this application denoises the low photon number pixel region in the initial denoised image by performing differentiated processing on different regions: on the one hand, segmented weighted denoising processing is performed on the first pixel region in the low photon number pixel region; on the other hand, pixel value replacement processing is performed on the second pixel region in the low photon number pixel region.

[0081] The first pixel region consists of pixels in the low-photon pixel region whose photon count is greater than or equal to the extremely low photon count threshold. In this region, a small portion of effective projection data is still retained, and weighted noise reduction can be directly performed based on the different photon counts of the pixels. The second pixel region consists of pixels in the low-photon pixel region whose photon count is less than the extremely low photon count threshold. In this region, the effective projection data has been completely covered by electronic noise. At this time, the projection data measured in this region is not the projection data obtained from the real scanning rays, but is formed by noise. If noise reduction is performed directly, it will show bright stripes or bright spots in the subsequently obtained reconstructed image, affecting the quality of the reconstructed image.

[0082] In practice, each pixel in the low-photon pixel region of the initial denoised image is traversed, and the number of photons in the pixel is compared with a preset extremely low photon number threshold. If the number of photons in the pixel is greater than or equal to the extremely low photon number threshold, it is classified as the first pixel region. If the number of photons in the pixel is less than the extremely low photon number threshold, it is classified as the second pixel region.

[0083] In some embodiments, the extremely low photon number threshold can be calibrated based on the electronic noise level of the projection device. For example, the extremely low photon number threshold can be set as the critical photon number value when the ratio of the projection data signal strength to the electronic noise intensity (signal-to-noise ratio) is less than 1:1.

[0084] Furthermore, this application provides a specific process for performing segmented weighted noise reduction processing on the first pixel region in the low photon number pixel region, such as... Figure 5 As shown, it includes the following steps: S301. Divide the photon count of the first pixel region into multi-level photon count intervals based on at least one secondary photon count threshold.

[0085] Among them, the secondary photon number thresholds are all less than the low photon number thresholds and greater than the extremely low photon number thresholds.

[0086] For example, at least one secondary photon number threshold includes a primary photon number threshold M1 and a secondary photon number threshold M2, and satisfies: low photon number threshold > M1 > M2 > low photon number threshold, thereby dividing the first pixel region into three photon number intervals. These thresholds can be predetermined based on the system's noise model or experimental calibration.

[0087] S302. The initial denoised image is filtered using filter kernels of different sizes to obtain filtered images with different denoising strengths.

[0088] The size of the filter kernel is negatively correlated with the number of photons in the photon number range.

[0089] For example, assuming three photon count intervals are defined, three different sizes of filter kernels are generated. For the interval with a relatively high photon count, a small 3×3 filter kernel is used for slight noise reduction; for the interval with a medium photon count, a medium 5×5 filter kernel is used for moderate noise reduction; and for the interval with a low photon count but still above the extremely low photon count threshold, a large 7×7 filter kernel is used for strong noise reduction.

[0090] It is understandable that the larger the filter kernel size, the more neighboring pixels participate in the calculation, and the greater the noise reduction effect.

[0091] S303. For any photon number interval in the multi-level photon number interval, perform weighted processing based on the target pixel value in the image corresponding to the photon number interval to obtain the weighted and denoised pixel value of the pixel to be processed.

[0092] The target pixel value is the pixel value of the pixel at the same position as the pixel to be processed in the image corresponding to the photon number interval.

[0093] The images corresponding to the photon number range include the initial denoised image and / or the filtered image.

[0094] In some embodiments, the image corresponding to a photon number range is determined based on the size of the end value of the photon number range. Specifically, the larger the end value of the photon number range, the smaller the noise reduction effect of the image corresponding to the photon number range, and the larger the end value of the photon number range, the greater the noise reduction effect of the image corresponding to the photon number range, so as to balance the noise level and detail preservation of different photon number regions.

[0095] It should be noted that the filtered image is the image obtained by denoising the initial denoised image using filter kernels of different sizes. Therefore, the denoising strength of the initial denoised image is smaller than that of any filtered image.

[0096] For example, suppose there are two photon number intervals, and the two photon number intervals are [ , ]and[ , ],in > > Each interval corresponds to two images. Therefore, for two photon number intervals, at least three images are needed: the initial denoised image and two filtered images with different denoising strengths. Assuming these two filtered images with different denoising strengths include a lightly filtered image and a heavily filtered image, for a larger photon number interval... , ], select the initial denoised image and the lightly filtered image as the images corresponding to this photon number range; for smaller photon number ranges [ , If so, then the lightly denoised image and the heavily filtered image are selected as the images corresponding to this photon number range.

[0097] Furthermore, such as Figure 6 As shown, S303 can be implemented as follows: S401-S402: S401. Determine the weight of the target pixel value based on the photon count of the pixel to be processed and the endpoints of the photon count range.

[0098] In some embodiments, the weight of the target pixel value is determined based on the ratio of the difference between the photon number of the pixel to be processed and the lower end of the photon number interval to the interpolation difference between the upper end and the lower end of the photon number interval.

[0099] For example, if the aforementioned first pixel value range is divided into two photon count ranges, that is, there is one and only one secondary photon count threshold, assuming the two photon count ranges are respectively [ , ]and[ , ], For low photon number threshold, This is the secondary photon number threshold. If the threshold is for extremely low photon count, then using this secondary photon count, low photon count threshold, and extremely low photon count threshold, two weights are obtained through the following relationship:

[0100]

[0101] in, The number of photons in the pixel to be processed. For low photon number threshold, For extremely low photon number threshold, The first weight is used to assign photon counts to photons within the range of [[]. , The target pixel values ​​corresponding to the pixels to be processed are weighted and processed. The second weight is used to determine the photon number within the range []. , The target pixel values ​​corresponding to the pixels to be processed are weighted.

[0102] S402. The target pixel value is weighted and calculated based on the weight to obtain the weighted and denoised pixel value of the pixel to be processed.

[0103] In some embodiments, the filtered image is weighted and fused using the calculated weight values ​​to obtain the weighted and denoised pixel values ​​of the pixels to be processed.

[0104] For example, for the two photon number intervals mentioned above, for photon numbers in [... , The pixel value after weighted noise reduction. The following relationship must be satisfied:

[0105] in, The pixel value of the target pixel in the initial denoised image. The pixel value of the target pixel in the first filtered image is denoised. The denoising strength of the initial denoised image is less than that of the first filtered image.

[0106] For photon number in [ , The pixel value after weighted noise reduction. The following relationship must be satisfied:

[0107] in, Let be the pixel value of the target pixel in the first filtered image. The pixel value of the target pixel in the second filtered image is given, and the noise reduction effect of the first filtered image is less than that of the second filtered image.

[0108] Since the photon count in the second pixel region is below the extremely low photon count threshold, its original projection data is completely covered by electronic noise and has no practical physical meaning. Directly performing conventional noise reduction or logarithmic transformation on this region will amplify the noise, resulting in bright spots or stripe artifacts in the reconstructed image. Therefore, this application provides two pixel value replacement processing methods to repair the data in this region. In practical applications, one method can be selected based on system configuration or image quality requirements.

[0109] As a possible pixel value replacement method, the photon count of pixels in the second pixel region is replaced with an extremely low photon count threshold. Then, a logarithmic transformation is performed on the replaced pixels to obtain transformed pixels. The transformed pixel values ​​replace the original pixel values ​​in the second pixel region. In a specific implementation, each pixel in the second pixel region is first traversed, and its original photon count is directly replaced with a preset constant. For example, this constant can be set to an extremely low photon count threshold. After completing the photon count replacement, a logarithmic transformation is performed on the updated second pixel region to obtain transformed pixel values. The calculated transformed pixel values ​​then replace the original pixel values ​​in the second pixel region.

[0110] As another possible pixel replacement method, an interpolation method is used to complete and replace the pixel values ​​of pixels in the second pixel region. In specific implementation, a preset interpolation algorithm is used to estimate the pixel value of each pixel in the second pixel region. The interpolation algorithm includes linear interpolation, bilinear interpolation, or distance-based inverse weighted interpolation. Finally, the calculated interpolation result is assigned to the corresponding pixel in the second pixel region.

[0111] The foregoing mainly describes the solutions provided by the embodiments of this application from a methodological perspective. To achieve the above functions, the image denoising apparatus includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0112] This application embodiment can, based on the above method, exemplarily divide the image denoising device into functional modules. For example, the image denoising device may include functional modules corresponding to each functional division, or two or more functions may be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division; in actual implementation, there may be other division methods.

[0113] Figure 7 This is a schematic diagram of an image noise reduction device provided in an embodiment of this application. (Refer to...) Figure 7 The image denoising device 700 includes: an acquisition unit 701 and a processing unit 702; the acquisition unit 701 is used to acquire an original projected image; the processing unit 702 is used to perform logarithmic transformation on the pixel values ​​of the original projected image to obtain a logarithmic projected image; to denoise the human tissue pixel region in the logarithmic projected image to obtain an initial denoised image; and to denoise the low photon number pixel region in the initial denoised image to obtain a target denoised image.

[0114] In some embodiments, the processing unit 702 is specifically used to segment different human tissue regions in a logarithmic projection image using image segmentation techniques to obtain multiple segmented regions; for each segmented region, calculate the gray-level centroid of the segmented region; determine a noise estimation sub-region in the segmented region with the gray-level centroid as the center; the noise estimation sub-region is a region where the ray attenuation is greater than the attenuation threshold; determine the corresponding relaxation factor based on the noise parameters of the noise estimation sub-region; the relaxation factor is used to adjust the noise reduction intensity of the segmented region; and perform noise reduction processing on the segmented region based on the relaxation factor to obtain an initial denoised image.

[0115] In some embodiments, the processing unit 702 is specifically used to substitute the noise parameters of the noise estimation sub-region into the relaxation factor mapping function to calculate the corresponding relaxation factor.

[0116] In some embodiments, the relaxation factor mapping function is a mapping function obtained by fitting a sample of the correspondence between noise parameters and relaxation factors; wherein, the sample of the correspondence between noise parameters and relaxation factors is constructed based on the noise parameters and the relaxation factors corresponding to the noise parameters of multiple sets of historical reconstructed images; the multiple sets of historical reconstructed images are obtained by selecting multiple sets of different radiation doses and performing multi-angle scanning on calibration phantoms of different diameters.

[0117] In some embodiments, the processing unit 702 is specifically used to segment the original projected image based on a low photon number threshold to determine a low photon number pixel region in the original projected image; wherein, the low photon number pixel region is a region composed of pixels in the original projected image whose photon number is less than the low photon number threshold; Based on the low photon number pixel regions in the original projected image, the low photon number pixel regions in the initial denoised image are determined.

[0118] In some embodiments, the processing unit 702 is specifically used to perform segmented weighted noise reduction processing on a first pixel region in a low photon number pixel region; the first pixel region is a region composed of pixels with a photon number greater than or equal to an extremely low photon number threshold; and to perform pixel value replacement processing on a second pixel region in a low photon number pixel region; the second pixel region is a region composed of pixels with a photon number less than an extremely low photon number threshold.

[0119] In some embodiments, the processing unit 702 is specifically configured to divide the photon count of the first pixel region into multi-level photon count intervals based on at least one secondary photon count threshold; the secondary photon count threshold is less than a low photon count threshold and greater than an extremely low photon count threshold; the initial denoised image is filtered using filter kernels of different sizes to obtain filtered images with different denoising strengths; the size of the filter kernel is negatively correlated with the photon count of the photon count interval; for the pixel to be processed in any photon count interval of the multi-level photon count interval, weighted processing is performed based on the target pixel value in the image corresponding to the photon count interval to obtain the weighted denoised pixel value of the pixel to be processed; wherein, the target pixel value is the pixel value of the pixel at the same position as the pixel to be processed in the image corresponding to the photon count interval; the image corresponding to the photon count interval includes the initial denoised image and / or the filtered image.

[0120] In some embodiments, the processing unit 702 is specifically used to determine the weight of the target pixel value based on the photon count of the pixel to be processed and the end value of the photon count range; and to perform a weighted calculation on the target pixel value based on the weight to obtain the weighted and denoised pixel value of the pixel to be processed.

[0121] In some embodiments, the processing unit 702 is specifically used for pixel value replacement processing including any one of the following: replacing the photon number of the pixel in the second pixel region with an extremely low photon number threshold, then performing a logarithmic transformation on the replaced pixel to obtain a transformed pixel, replacing the original pixel value in the second pixel region with the pixel value of the transformed pixel; and using an interpolation method to complete and replace the pixel value of the pixel in the second pixel region.

[0122] Figure 8 This is a schematic diagram of the structure of an electronic device 1000 provided in an embodiment of this application. Figure 8 As shown, the electronic device 1000 includes, but is not limited to, a processor 1001 and a memory 1002.

[0123] The memory 1002 described above is used to store the executable instructions of the processor 1001. It is understood that the processor 1001 is configured to execute instructions to implement the image noise reduction method in the above embodiments.

[0124] It should be noted that those skilled in the art will understand that Figure 8 The structure of the electronic device 1000 shown herein does not constitute a limitation on the electronic device 1000; the electronic device 1000 may include, but is not limited to, other electronic devices. Figure 8 This may indicate more or fewer components, or combinations of certain components, or different component arrangements.

[0125] The processor 1001 is the control center of the electronic device 1000. It connects to various parts of the electronic device 1000 via various interfaces and lines. By running or executing software programs and / or modules stored in the memory 1002, and by calling data stored in the memory 1002, it performs various functions and processes data of the electronic device 1000, thereby providing overall monitoring of the electronic device 1000. The processor 1001 may include one or more processing units. Optionally, the processor 1001 may integrate an application processor and a modem processor. The application processor mainly handles the operating system, user interface, and applications, while the modem processor mainly handles wireless communication. It is understood that the modem processor may not be integrated into the processor 1001.

[0126] The memory 1002 can be used to store software programs and various data. The memory 1002 may primarily include a program storage area and a data storage area. The program storage area may store the operating system, application programs required by at least one functional module (such as a determination unit, processing unit, etc.), etc. Furthermore, the memory 1002 may include high-speed random access memory, and may also include non-volatile memory, such as at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0127] In an exemplary embodiment, a computer-readable storage medium including instructions is also provided, such as a memory 1002 including instructions, which can be executed by a processor 1001 of an electronic device 1000 to implement the methods in the above embodiments.

[0128] In actual implementation, Figure 7 Both the acquisition unit 701 and the processing unit 702 can be derived from... Figure 8 The processor 1001 calls the computer program stored in the memory 1002 to implement the process. The specific execution process can be found in the description of the method section in the previous embodiment, and will not be repeated here.

[0129] Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, such as a read-only memory (ROM), random access memory (RAM), CD-ROM, magnetic tape, floppy disk, and optical data storage device.

[0130] In an exemplary embodiment, this application also provides a computer program product including one or more instructions, which can be executed by the processor 1001 of the electronic device 1000 to perform the methods described above.

[0131] It should be noted that when one or more instructions in the computer-readable storage medium or computer program product are executed by the processor of the electronic device 1000, they implement the various processes of the above method embodiments and achieve the same technical effect as the above method. To avoid repetition, they will not be described again here.

[0132] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0133] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0134] The units described as separate components may or may not be physically separate. A component shown as a unit can be one or more physical units; that is, it can be located in one place or distributed in multiple different locations. Some or all of the classified units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0135] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0136] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solution of the embodiments of this application, essentially, or the part that contributes to the prior art, or a complete or partial classification of the technical solution, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0137] In the description of the embodiments of this application, specific features, structures, materials or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0138] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An image denoising method, characterized in that, The method includes the following steps: Obtain the original projected image; The pixel values ​​of the original projected image are logarithmically transformed to obtain a logarithmic projected image; The human tissue pixel region in the logarithmic projection image is denoised to obtain an initial denoised image; The low-photon-count pixel region in the initial denoised image is denoised to obtain the target denoised image.

2. The method according to claim 1, characterized in that, The step of denoising the human tissue pixel region in the logarithmic projection image to obtain an initial denoised image includes: Image segmentation technology is used to segment different human tissue regions in the logarithmic projection image to obtain multiple segmented regions; For each of the plurality of segmented regions, calculate the gray-scale centroid of the segmented region; Using the grayscale centroid as the center, a noise estimation sub-region is determined within the segmented region; the noise estimation sub-region is a region where the ray attenuation is greater than the attenuation threshold. Based on the noise parameters of the noise estimation sub-region, a corresponding relaxation factor is determined; the relaxation factor is used to adjust the noise reduction intensity of the segmented region. The segmented region is denoised based on the relaxation factor to obtain the initial denoised image.

3. The method according to claim 2, characterized in that, The step of determining the corresponding relaxation factor based on the noise parameters of the noise estimation sub-region includes: The noise parameters of the noise estimation sub-region are substituted into the relaxation factor mapping function to calculate the corresponding relaxation factor.

4. The method according to claim 3, characterized in that, The relaxation factor mapping function is a mapping function obtained by fitting a sample of the correspondence between noise parameters and relaxation factors; wherein, the sample of the correspondence between noise parameters and relaxation factors is constructed based on the noise parameters of multiple sets of historical reconstructed images and the relaxation factors corresponding to the noise parameters; the multiple sets of historical reconstructed images are obtained by selecting multiple sets of different radiation doses and performing multi-angle scanning on calibration phantoms of different diameters.

5. The method according to claim 1, characterized in that, Before denoising the low-photon-count pixel regions in the initial denoised image, the method further includes: The original projected image is segmented based on a low photon number threshold to determine low photon number pixel regions in the original projected image; wherein, the low photon number pixel regions are regions composed of pixels in the original projected image whose photon number is less than the low photon number threshold; Based on the low photon number pixel regions in the original projected image, the low photon number pixel regions in the initial denoised image are determined.

6. The method according to claim 1, characterized in that, The noise reduction of low-photon-count pixel regions in the initial noise-reduced image includes: The first pixel region in the low photon number pixel region is subjected to segmented weighted noise reduction processing; the first pixel region is a region composed of pixels with a photon number greater than or equal to the extremely low photon number threshold; The second pixel region in the low photon number pixel region is subjected to pixel value replacement processing; the second pixel region is a region composed of pixels whose photon number is less than the extremely low photon number threshold.

7. The method according to claim 6, characterized in that, The segmented weighted noise reduction processing of the first pixel region in the low photon number pixel region includes: The photon count of the first pixel region is divided into multiple photon count intervals based on at least one secondary photon count threshold; the secondary photon count threshold is less than the low photon count threshold and greater than the extremely low photon count threshold. The initial denoised image is filtered using filter kernels of different sizes to obtain filtered images with different denoising strengths; the size of the filter kernel is negatively correlated with the photon number range. For any photon number interval in the multi-level photon number intervals, a weighted processing is performed based on the target pixel value in the image corresponding to the photon number interval to obtain the weighted denoised pixel value of the pixel to be processed; wherein, the target pixel value is the pixel value of the pixel at the same position as the pixel to be processed in the image corresponding to the photon number interval; the image corresponding to the photon number interval includes an initial denoised image and / or a filtered image.

8. The method according to claim 7, characterized in that, The weighted processing of the target pixel values ​​in the filtered image corresponding to the photon number interval to obtain the weighted and denoised pixel values ​​of the pixel to be processed includes: The weight of the target pixel value is determined based on the photon count of the pixel to be processed and the endpoints of the photon count range. The target pixel value is weighted based on the weights to obtain the weighted denoising pixel value of the pixel to be processed.

9. The method according to claim 6, characterized in that, The pixel value replacement process includes any of the following: The photon count of the pixels in the second pixel region is replaced with the extremely low photon count threshold, and then a logarithmic transformation is performed on the replaced pixels to obtain the transformed pixels. The original pixel values ​​in the second pixel region are replaced with the pixel values ​​of the transformed pixels. The pixel values ​​of the pixels in the second pixel region are filled and replaced using an interpolation method.

10. A radiotherapy system, characterized in that, The radiotherapy system includes: a radiation source, a detector, a rotating gantry, and processing equipment; The X-ray source is used to emit X-rays to irradiate the object to be scanned; The detector is positioned opposite the radiation source to receive X-rays passing through the object to be scanned and generate a raw projection image, and then sends the raw projection image to the processing device. The rotating frame is used to drive the X-ray source and the detector to rotate synchronously, and to collect multi-angle projection data of the object to be scanned. The processing device is communicatively connected to the detector, and the processing device is configured to acquire the raw projected image data collected by the detector and perform the image noise reduction method as described in any one of claims 1-9.

Images