Detail enhancement method, device and equipment based on CT (Computed Tomography) metal artifact removal and medium

Abstract

The invention provides a detail enhancement method and device based on CT metal artifact removal, equipment and a medium, and the method comprises the steps: carrying out the CT metal artifact removal of an original CT image through a preset metal artifact removal algorithm, and obtaining an artifact-removed image; generating a metal artifact mask based on an image difference between the original CT image and the artifact-removed image; performing positive and negative data separation on the metal artifact mask to obtain a binary mask of a bright artifact area and a binary mask of a dark artifact area; fusing the binary mask of the bright artifact area and the binary mask of the dark artifact area in the original CT image to obtain a feature enhanced image; carrying out artifact region feature enhancement on the feature enhancement image to obtain an image domain feature image; performing wavelet transform enhancement on the original CT image to obtain a wavelet domain feature image; and fusing the image domain feature image and the wavelet domain feature image to obtain a detail enhanced image. Therefore, the metal artifact removal quality is improved.

Description

Technical Field

[0001] Embodiments of this disclosure relate to the field of image processing technology, and more specifically, to a detail enhancement method, apparatus, device, and medium suitable for CT metal artifact removal. Background Technology

[0002] Metal artifacts primarily arise from the strong attenuation of X-rays by metallic implants (such as steel plates). X-rays at the metal edges are scattered and hardened, resulting in streak-like, star-shaped, or speckled artifacts in the projected data. Metal artifacts not only obscure the structural details of the surrounding normal tissue, reducing image contrast and spatial resolution, but can also lead to misdiagnosis or missed diagnosis, affecting clinicians' accurate judgment of the condition.

[0003] In related technologies, interpolation repair based on the metal projection trajectory region of the projection domain is mainly used to remove CT metal artifacts. Although most of the metal artifacts can be removed, the filtering back projection will cause the loss of output image details. Furthermore, due to the imperfect interpolation effect, some artifacts will not be completely removed and secondary artifacts will be introduced, resulting in poor metal artifact removal quality. Summary of the Invention

[0004] The embodiments described herein provide a detail enhancement method, apparatus, device, and medium based on CT metal artifact removal, overcoming the aforementioned problems.

[0005] Firstly, according to the content of this disclosure, a detail enhancement method based on CT metal artifact removal is provided, including: The original CT image is processed by a preset metal artifact removal algorithm to remove CT metal artifacts, resulting in an artifact-removed image. Calculate the image difference between the original CT image and the artifact-removed image; and generate a metal artifact mask based on the image difference between the original CT image and the artifact-removed image; The metal artifact mask is subjected to positive and negative data separation to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region; The binary mask of the bright artifact region and the binary mask of the dark artifact region are fused in the original CT image to obtain a feature-enhanced image; The artifact region feature enhancement is performed on the enhanced image to obtain an image domain feature image; and the original CT image is enhanced by wavelet transform to obtain a wavelet domain feature image. By fusing the image domain feature image and the wavelet domain feature image, a detail-enhanced image is obtained.

[0006] Secondly, according to the present disclosure, a detail enhancement device based on CT metal artifact removal is provided, comprising: The removal module is used to remove CT metal artifacts from the original CT image using a preset metal artifact removal algorithm, resulting in an artifact-removed image. The generation module is used to calculate the image difference between the original CT image and the artifact-removed image; and to generate a metal artifact mask based on the image difference between the original CT image and the artifact-removed image. The separation module is used to perform positive and negative data separation on the metal artifact mask to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region. The first fusion module is used to fuse the binary mask of the bright artifact region and the binary mask of the dark artifact region in the original CT image to obtain a feature-enhanced image; The enhancement module is used to enhance the artifact region features of the feature enhancement image to obtain an image domain feature image; and to enhance the original CT image using wavelet transform to obtain a wavelet domain feature image. The second fusion module is used to fuse the image domain feature image and the wavelet domain feature image to obtain a detail-enhanced image.

[0007] Thirdly, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the detail enhancement method based on CT metal artifact removal as described in any of the above embodiments.

[0008] Fourthly, a computer-readable storage medium is provided, on which a computer program is stored, and when executed by a processor, the computer program implements the steps of the detail enhancement method based on CT metal artifact removal as described in any of the above embodiments.

[0009] The detail enhancement method based on CT metal artifact removal provided in this application involves: removing CT metal artifacts from the original CT image using a preset metal artifact removal algorithm to obtain an artifact-removed image; calculating the image difference between the original CT image and the artifact-removed image; generating a metal artifact mask based on the image difference; performing positive and negative data separation on the metal artifact mask to obtain a binary mask for bright artifact regions and a binary mask for dark artifact regions; fusing the binary masks for bright and dark artifact regions in the original CT image to obtain a feature enhancement image; performing artifact region feature enhancement on the feature enhancement image to obtain an image domain feature image; performing wavelet transform enhancement on the original CT image to obtain a wavelet domain feature image; and fusing the image domain feature image and the wavelet domain feature image to obtain a detail enhancement image. Thus, after removing metal artifacts from the original CT image, a metal artifact mask is created to specifically enhance the information of the metal artifact region and the non-metal artifact region of the original CT image. Furthermore, feature fusion is used to enhance the data features of the image after metal artifact removal, effectively improving the ability to restore image detail information in the removed artifacts, thereby improving the quality of metal artifact removal.

[0010] The above description is merely an overview of the technical solutions of the embodiments of this application. In order to better understand the technical means of the embodiments of this application and to implement them in accordance with the contents of the specification, and to make the above and other objects, features and advantages of the embodiments of this application more obvious and understandable, specific implementation methods of this application are described below. Attached Figure Description

[0011] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings of the embodiments will be briefly described below. It should be understood that the drawings described below only relate to some embodiments of this disclosure and are not intended to limit this disclosure, wherein: Figure 1 This is a flowchart illustrating a detail enhancement method based on CT metal artifact removal provided in this disclosure.

[0012] Figure 2 This is a schematic diagram of group fusion provided in this publication.

[0013] Figure 3 This is a schematic diagram of a detail enhancement device based on CT metal artifact removal provided in this disclosure.

[0014] Figure 4 This is a schematic diagram of the structure of a computer device provided in this disclosure.

[0015] It should be noted that the elements in the attached diagram are schematic and not drawn to scale. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the described embodiments of this disclosure without creative effort are also within the scope of protection of this disclosure.

[0017] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this subject matter pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having the meaning consistent with their meaning in the context of the specification and in the relevant art, and shall not be interpreted in an idealized or overly formal form unless otherwise explicitly defined herein. As used herein, the statement of “connecting” or “coupling” two or more parts together shall mean that these parts are directly joined together or joined through one or more intermediate components.

[0018] The term "embodiment" as used herein means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of the phrase "embodiment" in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0019] In this document, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists, A and B exist simultaneously, or B exists. Additionally, the character " / " generally indicates that the preceding and following related objects have an "or" relationship. Terms such as "first" and "second" are only used to distinguish one component (or part of a component) from another component (or another part of a component).

[0020] In the description of this application, unless otherwise stated, "multiple" means two or more (including two), and similarly, "multiple groups" means two or more (including two groups).

[0021] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0022] Figure 1 This is a flowchart illustrating a detail enhancement method based on CT metal artifact removal provided in this disclosure embodiment, as shown below. Figure 1As shown, the specific process of the detail enhancement method based on CT metal artifact removal includes: S110. Perform CT metal artifact removal on the original CT image using a preset metal artifact removal algorithm to obtain an artifact-removed image.

[0023] The preset metal artifact removal algorithm is a traditional metal artifact removal algorithm, such as MAR (Metal Artifact Reduction). For example, threshold segmentation / region growing techniques are used to determine the metal pixels in the original CT image and construct a metal mask. Based on the metal mask, linear interpolation / polynomial interpolation is used in the projection domain or image domain to repair the projection data or image data of the metal-occluded area, generating an artifact-removed image.

[0024] S120. Calculate the image difference between the original CT image and the artifact-removed image; and generate a metal artifact mask based on the image difference between the original CT image and the artifact-removed image.

[0025] The image difference between the original CT image and the artifact-removed image can be obtained by subtracting the pixel values ​​of the corresponding locations in the original CT image from the pixel values ​​of the original CT image. This image difference reflects the abnormal pixel value areas in the original CT image caused by metal artifacts. By setting a threshold, the image difference between the original CT image and the artifact-removed image can be binarized, and the areas with pixel value differences greater than the set threshold are identified as metal artifact areas, thereby generating a metal artifact mask.

[0026] S130. Perform positive and negative data separation on the metal artifact mask to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region.

[0027] Since dark artifacts and bright artifacts have different characteristics in metal artifacts, the metal artifact mask needs to be split into different feature masks. For the metal artifact regions in the metal artifact mask, the regions with pixel values ​​greater than 0 are classified as bright artifact regions, and the regions with pixel values ​​less than or equal to 0 are classified as dark artifact regions.

[0028] In some embodiments, data positive and negative separation is performed on the metal artifact mask to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region. This includes: performing data positive and negative separation on the metal artifact mask to divide the metal artifact mask into a bright artifact mask and a dark artifact mask; performing image binarization processing on the bright artifact mask to obtain a binary mask for the bright artifact region; and performing image binarization processing on the dark artifact mask to obtain a binary mask for the dark artifact region.

[0029] In some embodiments, before performing image binarization processing on the bright artifact mask to obtain a binary mask of the bright artifact region, and before performing image binarization processing on the dark artifact mask to obtain a binary mask of the dark artifact region, the method further includes: performing normalization processing on the bright artifact mask; and performing normalization processing on the dark artifact mask.

[0030] For example, the mask is separated into positive and negative data, and bright artifact masks m_pos= np.maximum(mask1, 0) and dark artifact masks m_neg= np.maximum(-mask1, 0) are created respectively.

[0031] Global max-min normalization is performed based on the entire training set to avoid the stretching of differences in the same physical meaning to different ranges on different images due to the different complexity of artifacts during single-image normalization. The normalized result of the bright artifact mask is m_pos_norm = (m_pos - m_pos_min) / (m_pos_max - m_pos_min + 1e-8); the normalized result of the dark artifact mask is m_neg_norm = (m_neg - m_neg_min) / (m_neg_max - m_neg_min + 1e-8); where m_pos_max, m_pos_min, m_neg_max, and m_neg_min are data obtained from statistics based on the entire dataset.

[0032] The binary mask for the bright artifact region is mask_pos = m_pos_norm>alpha1; the binary mask for the dark artifact region is mask_neg = m_neg_norm>alpha; where alpha1 and alpha2 are learnable threshold parameters.

[0033] S140. In the original CT image, the binary mask of the bright artifact region and the binary mask of the dark artifact region are fused to obtain the feature-enhanced image.

[0034] By fusing a binary mask of the bright artifact region and a binary mask of the dark artifact region in the original CT image, it is possible to highlight the feature differences between the bright artifact region and the dark artifact region while preserving the information of the non-artifact region in the original CT image.

[0035] In some embodiments, a feature-enhanced image is obtained by fusing a binary mask of a bright artifact region and a binary mask of a dark artifact region in the original CT image. This includes: generating a feature image of the bright artifact region based on the binary mask of the bright artifact region; generating a feature image of the dark artifact region based on the binary mask of the dark artifact region; and performing channel fusion of the original CT image, the feature image of the bright artifact region, and the feature image of the dark artifact region to obtain the feature-enhanced image.

[0036] Specifically, feature images of bright artifact regions can be generated by fusing the original CT image with a binary mask of the bright artifact region, such as feature_p = Input * mask_pos + Input * (1-mask_pos) * theta; feature images of dark artifact regions can be generated by fusing the original CT image with a binary mask of the dark artifact region, such as feature_neg = Input * mask_neg + Input * (1-mask_neg) * theta; where theta is a learnable threshold parameter; Input * mask applies the strongest mask feature attention to the original image, and Input * (1-mask) * theta is used to adjust for errors caused by inaccurate masks and can alleviate the information bottleneck caused by hard mask boundary truncation. The feature images of the original CT image, the bright artifact region, and the dark artifact region are then channel-fused to obtain the feature-enhanced image Output = concat(original image, feature_p, feature_neg).

[0037] Therefore, two different masks, one for light and one for dark, were created based on the different data characteristics of artifacts of different brightness and darkness. Based on these masks, the original CT images were enhanced with regional features. The information processing bottleneck problem that the hard mask boundary may cause was specifically addressed by using a soft mask application method and a cascade method that preserves the original image information.

[0038] S150. Perform artifact region feature enhancement on the feature enhancement image to obtain the image domain feature image; and perform wavelet transform enhancement on the original CT image to obtain the wavelet domain feature image.

[0039] Among these methods, artifact region feature enhancement primarily enhances data features through image domain processing, resulting in enhanced data with artifact feature annotations. Wavelet transform processes information in the wavelet domain, expanding the dimensions of information acquisition and fully utilizing the advantages of the wavelet domain, such as local geometry, global frequency, and multi-scale orientation, thereby achieving better data feature enhancement.

[0040] In some embodiments, artifact region feature enhancement is performed on the feature-enhanced image to obtain an image domain feature image, including: inputting the feature-enhanced image into a preset image domain branch network, extracting features of different scales layer by layer through multiple convolutional layers and activation function layers contained in the image domain branch network, and fusing feature information of different scales to obtain an initial image domain image; performing a 1*1 convolution on the initial image domain image, and dividing the initial image domain image after 1*1 convolution into four groups of channel data; processing each group of channel data accordingly to obtain the output of each channel data; concatenating the outputs of each channel data in the channel direction, and performing a 1*1 convolution on the concatenated result to obtain the image domain feature image.

[0041] The image domain branch network employs a progressively deepening network structure across its multiple convolutional layers. For example, the first convolutional layer uses a 3×3 kernel to perform initial low-level feature extraction on the input feature-enhanced image, capturing basic information such as edges and textures. Subsequent convolutional layers gradually extract higher-level semantic features by increasing the number of kernels or using kernels with different receptive fields. ReLU activation functions can be used, introducing a non-linear transformation after each convolutional operation to enhance the image domain branch network's ability to express complex features, effectively mitigating the gradient vanishing problem and ensuring efficient information transfer at different scales during feature extraction. By using skip connections or attention mechanisms, shallow detail features are selectively fused with deep semantic features, enabling the image domain branch network to focus more on key feature information in artifact regions when generating image domain feature images, while suppressing background noise interference. This significantly improves the image domain feature image's ability to represent artifact region features.

[0042] like Figure 2 As shown, the input data Input1 (initial image in the image domain) is subjected to a 1*1 convolution to expand the channels. The output X is then divided into four groups according to the channel direction: x1, x2, x3, and x4. Each group is processed separately: x1 is output directly; x2 is processed by a 3*3 convolution to obtain Y2, which is then divided into two groups, y21 and y22, according to the channel direction; x3 and y22 are concatenated in the channel direction and processed by a 3*3 convolution to obtain Y3, which is then divided into two groups, y31 and y32, according to the channel direction; x4 is concatenated in the channel direction with y22 and y32, and then processed by a 3*3 convolution to obtain Y4. Finally, x1, y21, y31, and Y4 are concatenated in the channel direction and then processed by a 1*1 convolution to obtain Output1, the image domain feature image.

[0043] In some embodiments, wavelet transform enhancement is performed on the original CT image to obtain a wavelet domain feature image, including: performing wavelet transform processing on the original CT image to obtain a wavelet transform image; performing 1*1 convolution on the wavelet transform image and dividing the 1*1 convolutioned wavelet transform image into four groups of channel data; performing corresponding processing on each group of channel data to obtain the output of each channel data; concatenating the outputs of each channel data in the channel direction, and performing 1*1 convolution on the concatenated result to obtain a wavelet domain feature image.

[0044] In the process of performing wavelet transform on the original CT image, four different frequency bands can be obtained by wavelet transform. The four frequency bands are then stitched together according to the channel dimension to obtain the input data of the wavelet transform processing branch, which is the wavelet transform image.

[0045] like Figure 2 As shown, the input data Input2 (wavelet transform image) undergoes a 1x1 convolution to expand the channels. The output X is then divided into four groups according to the channel direction: x1, x2, x3, and x4. Each group is processed separately: x1 is output directly; x2 undergoes a 3x3 convolution to obtain Y2, which is then divided into two groups, y21 and y22, according to the channel direction; x3 and y22 are concatenated along the channel direction and then concatenated again to obtain Y3, which is then divided into two groups, y31 and y32, according to the channel direction; x4 is concatenated with y22 and y32 along the channel direction and then concatenated again to obtain Y4. Finally, x1, y21, y31, and Y4 are concatenated along the channel direction and then concatenated again with a 1x1 convolution to obtain Output2, the wavelet domain feature image.

[0046] In this embodiment, the group fusion process is almost equivalent to a single-layer 3x3 convolution, yielding image information with multiple receptive fields such as 1×1, 3×3, 5×5, and 7×7. This is primarily achieved by leveraging the similarity of features in the output feature maps of adjacent convolutional layers. Therefore, without needing to stack convolutional layers, a single convolution can obtain data feature maps with larger receptive fields by concatenating the feature output results of other groups with the feature maps of the original groups. Furthermore, by reusing feature information through grouping, the number of parameters is almost negligible compared to the original single-layer 3x3 convolution.

[0047] S160. By fusing the image domain feature image and the wavelet domain feature image, a detail-enhanced image is obtained.

[0048] Among them, the detail-enhanced image obtained by fusing image domain feature images and wavelet domain feature images can significantly increase the ability to remove artifacts in image artifact regions and restore overall image detail information.

[0049] In some embodiments, fusing image domain feature images and wavelet domain feature images to obtain a detail-enhanced image includes: upsampling the wavelet domain feature image to restore the image size to the original CT image size to obtain an upsampled feature image; and concatenating the image domain feature image and the upsampled feature image according to the channel direction to obtain a detail-enhanced image.

[0050] Specifically, the output of the wavelet transform branch, output2 (i.e., the wavelet domain feature image), can be upsampled to restore the feature map to the original image size, resulting in output3 (i.e., the upsampled feature image). Output1 (i.e., the image domain feature image) and output3 are concatenated along the channel direction and then subjected to a 1*1 convolution to obtain the final input, out, which is the detail-enhanced image.

[0051] In this embodiment, a preset metal artifact removal algorithm is used to remove CT metal artifacts from the original CT image to obtain an artifact-removed image. The image difference between the original CT image and the artifact-removed image is calculated. Based on the image difference, a metal artifact mask is generated. The metal artifact mask is subjected to positive and negative data separation to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region. The binary masks for the bright artifact region and the dark artifact region are fused in the original CT image to obtain a feature-enhanced image. The feature-enhanced image is then subjected to artifact region feature enhancement to obtain an image domain feature image. The original CT image is then subjected to wavelet transform enhancement to obtain a wavelet domain feature image. The image domain feature image and the wavelet domain feature image are fused to obtain a detail-enhanced image. Thus, after removing metal artifacts from the original CT image, a metal artifact mask is created to specifically enhance the information of the metal artifact region and the non-metal artifact region of the original CT image. Furthermore, feature fusion is used to enhance the data features of the image after metal artifact removal, effectively improving the ability to restore image detail information in the removed artifacts, thereby improving the quality of metal artifact removal.

[0052] Figure 3 This is a schematic diagram of a detail enhancement device based on CT metal artifact removal provided in this embodiment. The detail enhancement device based on CT metal artifact removal may include: The removal module 310 is used to remove CT metal artifacts from the original CT image using a preset metal artifact removal algorithm to obtain an artifact-removed image.

[0053] The generation module 320 is used to calculate the image difference between the original CT image and the artifact-removed image; and to generate a metal artifact mask based on the image difference between the original CT image and the artifact-removed image.

[0054] The separation module 330 is used to perform positive and negative data separation on the metal artifact mask to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region.

[0055] The first fusion module 340 is used to fuse the binary mask of the bright artifact region and the binary mask of the dark artifact region in the original CT image to obtain a feature-enhanced image.

[0056] The enhancement module 350 is used to enhance the artifact region features of the feature enhancement image to obtain the image domain feature image; and to enhance the original CT image by wavelet transform to obtain the wavelet domain feature image.

[0057] The second fusion module 360 ​​is used to fuse the image domain feature image and the wavelet domain feature image to obtain a detail-enhanced image.

[0058] In this embodiment, optionally, the separation module 330 is specifically used for: The metal artifact mask is separated into positive and negative data to divide it into a bright artifact mask and a dark artifact mask. The bright artifact mask is then binarized to obtain a binary mask for the bright artifact region. The dark artifact mask is then binarized to obtain a binary mask for the dark artifact region.

[0059] In this embodiment, optionally, the first fusion module 340 is specifically used for: Feature images of bright artifact regions are generated based on binary masks of bright artifact regions; feature images of dark artifact regions are generated based on binary masks of dark artifact regions; and the original CT image, the feature images of bright artifact regions, and the feature images of dark artifact regions are fused by channel fusion to obtain a feature-enhanced image.

[0060] In this embodiment, optionally, the enhancement module 350 is specifically used for: The feature-enhanced image is input into a predefined image domain branch network. Through multiple convolutional layers and activation function layers contained in the image domain branch network, features at different scales are extracted layer by layer from the feature-enhanced image, and feature information at different scales is fused to obtain the initial image in the image domain. The initial image in the image domain is then subjected to a 1*1 convolution, and the initial image in the image domain after 1*1 convolution is divided into four groups of channel data. Each group of channel data is processed accordingly to obtain the output of each channel data. The outputs of each channel data are concatenated in the channel direction, and the concatenated result is subjected to a 1*1 convolution to obtain the image domain feature image.

[0061] In this embodiment, optionally, the enhancement module 350 is specifically used for: The original CT image is processed by wavelet transform to obtain a wavelet transform image; the wavelet transform image is convolved by 1*1 and then divided into four groups of channel data; each group of channel data is processed accordingly to obtain the output of each channel data; the outputs of each channel data are concatenated in the channel direction and the concatenated result is convolved by 1*1 to obtain the wavelet domain feature image.

[0062] In this embodiment, optionally, the second fusion module 360 ​​is specifically used for: The wavelet domain feature image is upsampled to restore the image size to that of the original CT image, resulting in an upsampled feature image. The image domain feature image and the upsampled feature image are then concatenated along the channel direction to obtain a detail-enhanced image.

[0063] In this embodiment, optionally, a processing module may also be included.

[0064] The processing module is used to normalize the bright artifact mask and the dark artifact mask.

[0065] The detail enhancement device based on CT metal artifact removal provided in this disclosure can perform the above-described method embodiments. Its specific implementation principle and technical effects can be found in the above-described method embodiments, and will not be repeated here.

[0066] This application also provides a computer device. Please refer to the following for details. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0067] The computer device includes a memory 410 and a processor 420 that are interconnected via a system bus. It should be noted that only a computer device with memory 410 and processor 420 is shown in the figure; however, it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented alternatively. Those skilled in the art will understand that the computer device described herein is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0068] Computer devices can include desktop computers, laptops, handheld computers, and cloud servers. These devices allow for human-computer interaction with users through keyboards, mice, remote controls, touchpads, or voice-activated devices.

[0069] The memory 410 includes at least one type of readable storage medium, including non-volatile memory or volatile memory, such as flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. RAM may include static RAM or dynamic RAM. In some embodiments, the memory 410 may be an internal storage unit of a computer device, such as the hard disk or memory of the computer device. In other embodiments, the memory 410 may also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, or flash card equipped on the computer device. Of course, the memory 410 may include both internal storage units and external storage devices of the computer device. In this embodiment, the memory 410 is typically used to store the operating system and various application software installed on the computer device, such as the program code of the method described above. In addition, the memory 410 may also be used to temporarily store various types of data that have been output or will be output.

[0070] Processor 420 is typically used to perform overall operations of a computer device. In this embodiment, memory 410 is used to store program code or instructions, including computer operation instructions, and processor 420 is used to execute the program code or instructions stored in memory 410 or process data, such as program code that runs the methods described above.

[0071] In this article, the bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus system can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0072] Another embodiment of this application also provides a computer-readable medium, which may be a computer-readable signal medium or a computer-readable medium. A processor in a computer reads computer-readable program code stored in the computer-readable medium, enabling the processor to execute the functional actions specified in each step or combination of steps in the above method; and to generate means for implementing the functional actions specified in each block or combination of blocks in the block diagram.

[0073] Computer-readable media include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared memory or semiconductor systems, devices or apparatuses, or any suitable combination thereof, wherein the memory is used to store program code or instructions, the program code including computer operation instructions, and the processor is used to execute the program code or instructions of the above-described methods stored in the memory.

[0074] The definitions of memory and processor can be found in the description of the foregoing computer device embodiments, and will not be repeated here.

[0075] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, 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 system, or some features may be ignored or not executed. Furthermore, the 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.

[0076] In the various embodiments of this application, the functional units or modules 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.

[0077] 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 computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, 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, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0078] In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" as described in this application does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. This application can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims listing several means, several units of these means may be embodied by the same item of hardware. The use of "first," "second," and "third," etc., does not indicate any order and these words should be interpreted as names. Unless otherwise specified, the steps in the above embodiments should not be construed as limiting the order of execution.

[0079] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A detail enhancement method based on CT metal artifact removal, characterized in that, include: The original CT image is processed by a preset metal artifact removal algorithm to remove CT metal artifacts, resulting in an artifact-removed image. Calculate the image difference between the original CT image and the artifact-removed image; A metal artifact mask is generated based on the image difference between the original CT image and the artifact-removed image; The metal artifact mask is subjected to positive and negative data separation to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region; The binary mask of the bright artifact region and the binary mask of the dark artifact region are fused in the original CT image to obtain a feature-enhanced image; The artifact region feature enhancement is performed on the enhanced image to obtain an image domain feature image; and the original CT image is enhanced by wavelet transform to obtain a wavelet domain feature image. By fusing the image domain feature image and the wavelet domain feature image, a detail-enhanced image is obtained.

2. The method according to claim 1, characterized in that, The metal artifact mask is subjected to positive and negative data separation to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region, including: The metal artifact mask is subjected to positive and negative data separation to divide it into a bright artifact mask and a dark artifact mask. The bright artifact mask is subjected to image binarization processing to obtain a binary mask of the bright artifact region; and the dark artifact mask is subjected to image binarization processing to obtain a binary mask of the dark artifact region.

3. The method according to claim 1, characterized in that, The binary mask of the bright artifact region and the binary mask of the dark artifact region are fused in the original CT image to obtain a feature-enhanced image, including: A feature image of the bright artifact region is generated based on a binary mask of the bright artifact region; and a feature image of the dark artifact region is generated based on a binary mask of the dark artifact region. The original CT image, the feature image of the bright artifact region, and the feature image of the dark artifact region are fused by channel fusion to obtain a feature-enhanced image.

4. The method according to claim 1, characterized in that, The feature-enhanced image is subjected to artifact region feature enhancement to obtain an image domain feature image, including: The feature-enhanced image is input into a preset image domain branch network, and features at different scales are extracted layer by layer from the feature-enhanced image through multiple convolutional layers and activation function layers contained in the image domain branch network. The feature information at different scales is then fused to obtain the initial image in the image domain. The initial image of the image domain is convolved with a 1*1 convolution, and the initial image of the image domain after the 1*1 convolution is divided into four groups of channel data; each group of channel data is processed accordingly to obtain the output of each channel data; the outputs of each channel data are concatenated in the channel direction, and the concatenated result is convolved with a 1*1 convolution to obtain the image domain feature image.

5. The method according to claim 1, characterized in that, The original CT image is enhanced by wavelet transform to obtain a wavelet domain feature image, including: The original CT image is processed by wavelet transform to obtain a wavelet transform image; The wavelet transform image is convolved with a 1*1 convolution, and the wavelet transform image after 1*1 convolution is divided into four groups of channel data; each group of channel data is processed accordingly to obtain the output of each channel data; the outputs of each channel data are concatenated in the channel direction, and the concatenated result is convolved with a 1*1 convolution to obtain the wavelet domain feature image.

6. The method according to claim 1, characterized in that, By fusing the image domain feature image and the wavelet domain feature image, a detail-enhanced image is obtained, including: The wavelet domain feature image is upsampled to restore the image size to that of the original CT image, resulting in an upsampled feature image; The image domain feature image and the upsampled feature image are concatenated along the channel direction to obtain a detail-enhanced image.

7. The method according to claim 2, characterized in that, The bright artifact mask is subjected to image binarization processing to obtain a binary mask of the bright artifact region; Before performing image binarization processing on the dark artifact mask to obtain a binary mask of the dark artifact region, the process further includes: The bright artifact mask is normalized. The dark artifact mask is then normalized.

8. A detail enhancement device based on CT metal artifact removal, characterized in that, include: The removal module is used to remove CT metal artifacts from the original CT image using a preset metal artifact removal algorithm, resulting in an artifact-removed image. The generation module is used to calculate the image difference between the original CT image and the artifact-removed image; A metal artifact mask is generated based on the image difference between the original CT image and the artifact-removed image; The separation module is used to perform positive and negative data separation on the metal artifact mask to obtain a binary mask for the bright artifact region and a binary mask for the dark artifact region. The first fusion module is used to fuse the binary mask of the bright artifact region and the binary mask of the dark artifact region in the original CT image to obtain a feature-enhanced image; The enhancement module is used to enhance the artifact region features of the feature enhancement image to obtain an image domain feature image; and to enhance the original CT image using wavelet transform to obtain a wavelet domain feature image. The second fusion module is used to fuse the image domain feature image and the wavelet domain feature image to obtain a detail-enhanced image.

9. A computer device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the detail enhancement method based on CT metal artifact removal as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the detail enhancement method based on CT metal artifact removal as described in any one of claims 1 to 7.

Images