A method and system for MRI tumor boundary extraction based on complex amplitude frequency domain enhancement and complementary differential interference.

Abstract

This application relates to the field of image processing technology, specifically to a method and system for MRI tumor boundary extraction based on complex amplitude frequency domain enhancement and complementary differential interference. The method includes: performing Fourier transform on an MRI image to obtain a frequency domain signal; performing frequency domain modulation on the frequency domain signal using a complex amplitude frequency domain operator, followed by inverse Fourier transform to obtain a spatial domain enhanced image; generating a boundary spectrum evaluation value based on the spatial domain enhanced image; if the boundary spectrum evaluation value does not meet a preset convergence condition, adjusting the adjustable parameters of the complex amplitude frequency domain operator based on the boundary spectrum evaluation value to update the frequency domain modulation signal; performing Fourier complementary differential interference enhancement processing on the spatial domain enhanced image that meets the preset convergence condition to obtain a differential enhanced image; and extracting the closed boundary of the differential enhanced image to obtain a complete tumor WT closed boundary image. This application achieves clear, continuous, and closed extraction of the WT tumor boundary from MRI images.

Description

Technical Field

[0001] This application relates to the field of image processing technology, specifically to a method and system for MRI tumor boundary extraction based on complex amplitude frequency domain enhancement and complementary differential interference. Background Technology

[0002] Magnetic Resonance Imaging (MRI) is a commonly used medical imaging technique that utilizes strong magnetic fields, radio frequency pulses, and the magnetic resonance signals of hydrogen protons in human tissues to create tomographic images through computer reconstruction. Compared to imaging methods such as CT (Computed Tomography), MRI does not rely on ionizing radiation, has better soft tissue resolution, and can clearly reflect the structural differences in brain tissue, tumor tissue, and surrounding edema areas. It plays an important role in the identification, localization, segmentation, and boundary analysis of brain tumors.

[0003] Brain tumors in MRI images typically exhibit localized variations in grayscale or signal intensity, abnormal tissue morphology, altered boundary contours, and structural differences from surrounding normal tissue. The whole tumor (WT) usually includes the tumor itself and its associated abnormal areas; accurate extraction of its closed boundaries is crucial for lesion extent visualization, auxiliary analysis, treatment planning, and subsequent image annotation. However, in actual MRI images, tumor boundaries often appear as weak, blurred, or irregular closed boundaries, and are easily affected by brain tissue texture, grayscale heterogeneity, noise, and local false edges, making boundary extraction challenging.

[0004] Existing methods for MRI tumor boundary extraction mostly rely on digital image processing or deep learning algorithms. Among them, traditional edge detection, threshold segmentation, region growing, and morphological processing methods are easily affected by noise, uneven gray levels, and blurred boundaries, leading to boundary breaks, false detections, or incomplete closures. While deep learning methods have better segmentation results, they usually require a large amount of labeled data and high computational resources. The model training and inference process is relatively complex, and there may still be insufficient generalization ability under different scanning sequences, different devices, or different lesion morphologies.

[0005] Fourier optics can rapidly transform and modulate image information in the frequency domain, featuring parallel processing, fast response, and suitability for boundary spectral enhancement. However, using only a single high-pass filter to enhance MRI image boundaries can easily amplify high-frequency noise, brain tissue texture, and false edge information simultaneously; conversely, excessive suppression of low-frequency components may weaken the overall contour of the tumor region, leading to instability in subsequent closed boundary extraction. Summary of the Invention

[0006] In view of this, this application proposes an MRI tumor boundary extraction method and system based on complex amplitude frequency domain enhancement and complementary differential interference. This application is an MRI image enhancement scheme that can simultaneously perform low-frequency contour protection, ring boundary main frequency band screening, directional boundary enhancement, and noise suppression in the frequency domain plane. It combines FPGA (Field Programmable Gate Array) feedback evaluation and Fourier complementary differential interference enhancement to improve the clarity, continuity, and closure integrity of the extracted tumor boundary.

[0007] Specifically, this application is implemented through the following technical solution: According to a first aspect of the embodiments of this specification, a method for MRI tumor boundary extraction based on complex amplitude frequency domain enhancement and complementary differential interference is provided, comprising: Step S10: Perform Fourier transform on the MRI image to obtain the frequency domain signal; Step S20: Frequency domain modulation is performed on the frequency domain signal using a complex amplitude frequency domain operator to obtain a frequency domain modulated signal; the frequency domain modulation includes low-frequency profile protection, ring spectrum double-threshold modulation, direction enhancement, and fractional derivative enhancement; Step S30: Perform an inverse Fourier transform on the frequency domain modulation signal to obtain a spatial domain enhanced image; Step S40: Generate a boundary spectrum evaluation value based on the boundary energy, background residual, high-frequency noise ratio, and boundary closure trend of the spatial domain enhanced image; if the boundary spectrum evaluation value does not meet the preset convergence condition, adjust the adjustable parameters of the complex amplitude frequency domain operator based on the boundary spectrum evaluation value, and return to step S20 for iterative execution; Step S50: Perform Fourier complementary differential interferometry enhancement processing on the spatial domain enhanced image that satisfies the preset convergence condition to obtain a differential enhanced image; Step S60: Extract the closed boundary of the differential enhancement image to obtain the whole tumor WT closed boundary image; wherein, the whole tumor WT closed boundary image is used to represent the overall outer contour boundary of the tumor region in the MRI image.

[0008] According to a second aspect of the embodiments of this specification, an MRI tumor boundary extraction system based on complex amplitude frequency domain enhancement and complementary differential interference is provided, comprising: The frequency domain transformation module is used to perform Fourier transform on MRI images to obtain frequency domain signals; The frequency domain enhancement module is used to perform frequency domain modulation on the frequency domain signal using a complex amplitude frequency domain operator to obtain a frequency domain modulated signal; the frequency domain modulation includes low-frequency profile protection, ring spectrum dual-threshold modulation, direction enhancement, and fractional derivative enhancement. The spatial domain enhancement module is used to perform an inverse Fourier transform on the frequency domain modulation signal to obtain a spatial domain enhanced image; The feedback adjustment module is used to generate a boundary spectrum evaluation value based on the boundary energy, background residue, high-frequency noise ratio, and boundary closure trend of the spatial domain enhanced image. If the boundary spectrum evaluation value does not meet the preset convergence condition, the adjustable parameters of the complex amplitude frequency domain operator in the frequency domain enhancement module are adjusted based on the boundary spectrum evaluation value, and the frequency domain enhancement module and the spatial domain enhancement module are controlled to run iteratively until the preset convergence condition is met. The differential enhancement module is used to perform Fourier complementary differential interference enhancement processing on the spatial domain enhanced image that satisfies the preset convergence condition to obtain a differential enhanced image; The contour extraction module is used to extract the closed boundary of the differential enhancement image to obtain the whole tumor WT closed boundary image; wherein, the whole tumor WT closed boundary image is used to represent the overall outer contour boundary of the tumor region in the MRI image.

[0009] The embodiments of this application have at least the following technical effects: This application employs a complex amplitude frequency domain operator to selectively modulate the frequency domain signal of MRI images. Low-frequency contour protection preserves the overall contour of the tumor region, while ring spectrum dual-threshold modulation extracts and enhances the main frequency band of the tumor boundary. Directional enhancement improves the response of irregular and weak boundaries, and fractional-order differential enhancement suppresses high-frequency noise amplification while enhancing the boundary. Simultaneously, a boundary spectrum evaluation value is generated based on boundary energy, background residuals, high-frequency noise ratio, and boundary closure trend, and the adjustable parameters of the complex amplitude frequency domain operator are adjusted accordingly, allowing the system to adapt to different grayscale distributions and boundary strength variations in MRI images. Fractional-order differential enhancement is applied to the boundary spectrum components to improve boundary clarity and suppress excessive high-frequency noise amplification caused by ordinary integer-order differential enhancement.

[0010] This application uses Fourier complementary differential interferometry to divide the spatial domain enhanced image into a low-frequency background branch and a high-frequency boundary branch, and then uses energy matching and interferometric beam combining to suppress background information and enhance tumor boundary information.

[0011] This application extracts closed boundaries from differentially enhanced images to obtain continuous and complete whole tumor WT closed boundaries, reducing boundary breakage and false detection problems. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Some specific embodiments of this application will be described in detail below with reference to the drawings in an exemplary and non-limiting manner. The same reference numerals in the drawings indicate the same or similar parts or components. Those skilled in the art should understand that these drawings are not necessarily drawn to scale. In the drawings: Figure 1 This is a schematic flowchart illustrating an exemplary embodiment of the MRI tumor boundary extraction method based on complex amplitude frequency domain enhancement and complementary differential interference. Figure 2 This is a schematic diagram of an MRI brain tumor image according to one embodiment of this application; Figure 3 yes Figure 2 A schematic diagram of the WT region segmentation result obtained after segmenting the brain tumor region in the brain. Figure 4 This is a schematic diagram of an electronic device illustrated in this specification according to an exemplary embodiment; Figure 5 This is a structural block diagram of an MRI tumor boundary extraction system based on complex amplitude frequency domain enhancement and complementary differential interference, as illustrated in an exemplary embodiment of this application. Figure 6 This is a block diagram illustrating the structure of a frequency domain enhancement module according to an exemplary embodiment of this application; Figure 7 This is a block diagram illustrating the differential enhancement module structure in an exemplary embodiment of this application. Detailed Implementation

[0013] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0014] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The singular forms “a,” “the,” and “the” used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0015] This application proposes an MRI tumor boundary extraction scheme based on complex amplitude frequency domain enhancement and complementary differential interference.

[0016] Figure 1 This is a schematic flowchart illustrating an exemplary embodiment of the MRI tumor boundary extraction method based on complex amplitude frequency domain enhancement and complementary differential interferometry. Figure 1 As shown, this application provides a method for MRI tumor boundary extraction based on complex amplitude frequency domain enhancement and complementary differential interference, including: Step S10: Perform Fourier transform on the MRI image to obtain the frequency domain signal; Step S20: Frequency domain modulation is performed on the frequency domain signal using a complex amplitude frequency domain operator to obtain a frequency domain modulated signal; the frequency domain modulation includes low-frequency profile protection, ring spectrum double threshold modulation, direction enhancement, and fractional derivative enhancement; Step S30: Perform an inverse Fourier transform on the frequency domain modulation signal to obtain a spatial domain enhanced image; Step S40: Generate a boundary spectrum evaluation value based on the boundary energy, background residue, high-frequency noise ratio, and boundary closure trend of the spatial domain enhanced image; if the boundary spectrum evaluation value does not meet the preset convergence condition, adjust the adjustable parameters of the complex amplitude frequency domain operator based on the feedback of the boundary spectrum evaluation value, and return to step S20 for iterative execution. Step S50: Perform Fourier complementary differential interferometry enhancement processing on the spatial domain enhanced image that meets the preset convergence condition to obtain a differential enhanced image; Step S60: Extract the closed boundary of the differential enhancement image to obtain the whole tumor WT closed boundary image; wherein, the whole tumor WT closed boundary image is used to represent the overall outer contour boundary of the tumor region in the MRI image.

[0017] The following details each of the above steps.

[0018] Step S10 receives the MRI image to be processed, forms the image signal to be modulated, performs Fourier transform on the MRI image, and obtains the frequency domain signal.

[0019] Step S20 uses a complex amplitude frequency domain operator to perform complex amplitude frequency domain enhancement on the frequency domain signal in the Fourier frequency domain plane, including low-frequency profile protection, ring spectrum double threshold modulation, direction enhancement and fractional derivative enhancement, and finally obtains the frequency domain modulated signal.

[0020] In some embodiments, step S20, low-frequency profile protection and ring spectrum dual-threshold modulation in the frequency domain specifically includes the following sub-steps: Step 1: Obtain the frequency radius of the frequency domain signal; Step 2: When the frequency radius is less than the first frequency threshold, the corresponding spectral components are subjected to low-frequency contour preservation or weakening based on the low-frequency preservation ratio. Step 3: When the frequency radius is greater than or equal to the first frequency threshold and less than or equal to the second frequency threshold, the corresponding spectral components are used as candidate information for the main frequency band of the tumor boundary, and enhancement processing is performed based on the boundary enhancement gain. Step 4: When the frequency radius is greater than the second frequency threshold, perform high-frequency noise suppression processing on the corresponding frequency components based on the noise suppression intensity.

[0021] The complex amplitude frequency domain operator in this embodiment performs low-frequency contour protection on the frequency domain signal, preserving the overall contour and low-frequency structure of the tumor region in the MRI image, and avoiding the breakage of the overall tumor contour during the boundary enhancement process; the ring spectrum dual threshold modulation forms a ring boundary main frequency band according to the first frequency threshold r1 and the second frequency threshold r2, and enhances the spectral components located in the ring boundary main frequency band based on the boundary enhancement gain.

[0022] Wherein, the first frequency threshold r1 is used as the inner frequency threshold r1, and the second frequency threshold r2 is used as the outer frequency threshold, r1 < r2; when the frequency radius is less than r1, the corresponding spectral component is subjected to low-frequency contour preservation or weakening processing based on the low-frequency preservation ratio; when the frequency radius is between r1 and r2, the corresponding spectral component is used as the candidate information of the main frequency band of the tumor boundary, and enhancement processing is performed based on the boundary enhancement gain; when the frequency radius is greater than r2, the corresponding spectral component is subjected to high-frequency noise suppression processing based on the noise suppression intensity.

[0023] Low-frequency contour protection preserves the overall structural information of the tumor region by retaining low-frequency components; ring-spectrum dual-threshold modulation constructs a ring passband in the frequency domain, selectively enhancing the boundary main frequency band components while suppressing excessively high-frequency noise components, thus achieving a balance between boundary enhancement and noise suppression.

[0024] In some embodiments, step S20, the direction enhancement of frequency domain modulation includes the following sub-steps: Step 1: Obtain the frequency domain coordinates of each spectral component in the frequency domain signal, and calculate the corresponding spectral direction angle based on the frequency domain coordinates; Step 2: Based on the directional distribution of the boundary to be enhanced in the MRI image, set at least one directional enhancement center angle; Step 3: Calculate the angle difference between the spectral direction angle and the direction enhancement center angle; Step 4: When the angle difference is less than or equal to a preset angle threshold, apply a first direction enhancement weight to the corresponding spectral component; Step 5: When the angle difference is greater than the preset angle threshold, apply a second direction enhancement weight to the corresponding spectral component; Among them, the weight of the first direction enhancement is greater than the weight of the second direction enhancement; the direction enhancement center angle, the preset angle threshold, the weight of the first direction enhancement and the weight of the second direction enhancement are all adjustable parameters of the complex amplitude frequency domain operator.

[0025] In this embodiment, the spectral components corresponding to the tumor boundary direction are weighted and enhanced to improve the response of irregular and weak boundaries. Specifically, in the frequency domain plane, the orientation angle of each spectral component reflects the spatial change direction corresponding to that frequency component. The spectral orientation angle can be obtained by calculating the orientation angle of each spectral component's frequency domain coordinates relative to the spectral center. Based on the prior orientation distribution of the tumor boundary in the MRI image or the preliminary detection results, one or more directional enhancement center angles are set to focus the enhancement operation on the spectral components corresponding to the main direction of the boundary. The angle difference between each spectral orientation angle and each directional enhancement center angle is calculated. When the angle difference is less than or equal to a preset angle threshold, a larger first directional enhancement weight is applied to the spectral component; when the angle difference is greater than the preset angle threshold, a smaller second directional enhancement weight is applied. Thus, spectral components consistent with the boundary direction are selectively enhanced, while spectral components perpendicular to or significantly deviating from the boundary direction are relatively suppressed, thereby improving the clarity of the target boundary while reducing interference from non-boundary directional textures and false edges.

[0026] The directional enhancement center angle, preset angle threshold, and the two directional enhancement weights can all be used as adjustable parameters of the complex amplitude frequency domain operator, and can be adjusted based on evaluation indicators such as boundary closure trend. For example, when the boundary closure trend is lower than the preset threshold, the directional enhancement weight of the corresponding direction can be increased to promote the connection of the broken boundary and the formation of the closed contour. Multiple directional enhancement center angles can cover different orientation segments of the tumor boundary, adapting to the anisotropic characteristics of irregular tumor contours.

[0027] In some embodiments, step S20, the fractional derivative enhancement of frequency domain modulation includes the following sub-steps: Step 1: Set the fractional derivative order, which is greater than 0 and less than or equal to 2. Step 2: Construct a fractional-order differential gain function based on the frequency radius of the spectral components in the frequency domain signal; Step 3: Use the fractional-order differential gain function to perform fractional-order differential enhancement on the spectral components within the candidate information of the main frequency band of the tumor boundary; Step four involves gain limiting for spectral components above the second frequency threshold to suppress excessive amplification of high-frequency noise while improving boundary clarity.

[0028] This embodiment performs fractional-order differential enhancement on the boundary spectral components to improve boundary clarity and suppress excessive amplification of high-frequency noise caused by ordinary integer-order differentials. Specifically, a fractional-order differential order (greater than 0 and less than or equal to 2) is set, and a fractional-order differential gain function is constructed based on the frequency radius of the spectral components. This gain function is used to perform fractional-order differential enhancement on the spectral components located within the main frequency band of the ring boundary (i.e., candidate information of the main frequency band of the tumor boundary), and gain is limited for spectral components above the second frequency threshold (outer frequency threshold r2), thereby improving boundary clarity while avoiding excessive amplification of high-frequency noise.

[0029] Specifically, when the frequency radius is less than the first frequency threshold r1, the corresponding spectral component undergoes low-frequency contour preservation or weakening based on the low-frequency preservation ratio. When the frequency radius is between the first and second frequency thresholds r1 and r2, the corresponding spectral component is used as candidate information for the main frequency band of the tumor boundary and enhanced. When the frequency radius is greater than the second frequency threshold r2, the corresponding spectral component undergoes high-frequency noise suppression based on the noise suppression intensity. Therefore, the complex amplitude frequency domain operator does not perform a single high-pass filter on the MRI image, but rather selectively enhances the main frequency band of the tumor boundary while preserving the low-frequency contour. The parameters of the complex amplitude frequency domain operator can be adaptively adjusted.

[0030] Fractional derivatives have nonlinear amplitude-frequency response characteristics in the frequency domain. They can enhance the boundary main frequency band while handling the high-frequency band more gently than integer derivatives. Combined with gain limits above r2, they can effectively suppress noise.

[0031] In some embodiments, step S40, the process of generating boundary spectrum evaluation values ​​includes: Step 1: Perform edge candidate point detection on the spatial domain enhanced image to obtain edge candidate regions and background candidate regions; Step 2: Calculate the boundary energy based on the average edge response or the dominant frequency band energy of the edge candidate region; Step 3: Calculate the background residual based on the residual response in the background candidate region; Step 4: Calculate the high-frequency noise ratio based on the ratio of high-frequency spectral energy above the second frequency threshold to the total spectral energy; Step 5: Calculate the boundary closure trend based on the ratio of the closed contour length to the total edge length, or based on the ratio of the number of edge endpoints to the number of edge pixels. Step 6: Generate a boundary spectrum evaluation value based on a weighted combination of boundary energy, background residual, high-frequency noise ratio, and boundary closure trend.

[0032] The preset convergence conditions are met, specifically: The boundary spectrum evaluation value has reached the preset evaluation threshold; Or at least two of the following conditions must be met: The boundary energy reaches the preset boundary energy threshold. Background residue is lower than the preset background residue threshold; The proportion of high-frequency noise is lower than the preset noise proportion threshold. The boundary closure trend has reached the preset closure trend threshold; The change in the boundary spectrum evaluation value obtained in two consecutive iterations is less than the preset change threshold.

[0033] This embodiment acquires spatial domain enhanced images through a multi-channel photosensitive array, an image acquisition interface, a digital image buffer unit, or a combination thereof, and divides edge candidate regions and background candidate regions based on edge candidate point detection. From these regions, quantitative indicators of four dimensions—boundary energy, background residue, high-frequency noise ratio, and boundary closure trend—are extracted and weighted to generate boundary spectrum evaluation values, thereby comprehensively evaluating the enhancement quality and noise control level of the tumor boundary under the current frequency domain modulation.

[0034] The preset convergence condition adopts a dual-mode judgment mechanism, which combines single threshold achievement or multiple index joint achievement, to balance convergence efficiency and stability. When the convergence condition is not met, the corresponding parameters of the complex amplitude frequency domain operator are adjusted according to the deviation of each index: if the boundary energy is insufficient, the ring main frequency band is widened; if the background residual is too high, the low frequency retention ratio is reduced; if the noise ratio is too high, the external frequency threshold r2 and the fractional derivative order are lowered; if the closure trend is insufficient, the directional enhancement weight in the corresponding direction is increased, thereby achieving adaptive optimization of parameters.

[0035] In some embodiments, the adjustable parameters include at least one of the following: a first frequency threshold, a second frequency threshold, a low-frequency retention ratio, a first directional enhancement weight, a second directional enhancement weight, a directional enhancement center angle, a preset angle threshold, a boundary enhancement gain, a noise suppression strength, and a fractional derivative order. Feedback adjustment of the adjustable parameters of the complex amplitude frequency domain operator includes: When the boundary energy is lower than the preset boundary energy threshold, the difference between the second frequency threshold and the first frequency threshold is increased to expand the main frequency band range of the ring boundary. When the background residue is higher than the preset background residue threshold, reduce the low-frequency retention ratio; When the proportion of high-frequency noise is higher than the preset noise proportion threshold, the second frequency threshold and the fractional derivative order are reduced. When the boundary closure trend is lower than the preset closure trend threshold, increase the direction enhancement weight in the corresponding direction; Based on the adjusted parameters, the boundary spectrum evaluation value is recalculated, and it is determined whether the boundary spectrum evaluation value meets the preset convergence condition. If it does, the feedback adjustment is stopped, and the current spatial domain augmented image is taken as the spatial domain augmented image that meets the preset convergence condition. Otherwise, the feedback adjustment is repeated.

[0036] In this embodiment, the adjustable parameters of the complex amplitude frequency domain operator are adjusted based on the boundary spectrum evaluation value. The feedback adjustment strategy is as follows: when the boundary energy is lower than a preset threshold, r2 is increased and / or r1 is decreased to widen the main frequency band of the ring boundary; when the background residual is too high, the low-frequency retention ratio is reduced; when the high-frequency noise ratio is too high, r2 and the fractional derivative order are reduced; when the boundary closure trend is insufficient, the directional enhancement weight in the corresponding direction is increased. After adjustment, the boundary spectrum evaluation value is recalculated, and the process is iteratively executed until the preset convergence condition is met.

[0037] By employing feedback adjustment, the complex amplitude frequency domain operator can adapt to the grayscale distribution, boundary strength variations, and noise levels of different MRI images, avoiding contour loss or high-frequency noise amplification issues caused by a single high-pass filter. A closed-loop mapping is established between the multidimensional parameters of frequency domain modulation and the quantitative evaluation of boundary enhancement effects, automatically optimizing along the gradient direction to achieve adaptive parameter configuration.

[0038] In some embodiments, step S50 specifically includes the following sub-steps: Step 1: Perform a Fourier transform on the spatial domain enhanced image to obtain the frequency domain enhanced signal; Step 2: Divide the frequency domain enhancement signal into low-frequency background branch signal and high-frequency boundary branch signal; Step 3: Perform energy matching and inverse Fourier transform processing on the low-frequency background branch signal to obtain the spatial domain background signal; perform boundary band gain processing and inverse Fourier transform processing on the high-frequency boundary branch signal to obtain the spatial domain boundary signal; wherein, the phase difference between the background component in the spatial domain background signal and the corresponding background projection component in the spatial domain boundary signal is π. Step four: Interferometrically combine the spatial domain background signal and the spatial domain boundary signal to cancel each other out and make the boundary components constructive, thus obtaining a differential enhancement image.

[0039] The closed boundary extraction in step S60 includes: edge candidate point detection, broken edge connection, morphological closure, connected component screening, and contour tracing.

[0040] In this embodiment, a spatial domain enhanced image that meets a preset convergence condition is Fourier transformed to the frequency domain. The frequency domain enhanced signal is then divided into a low-frequency background branch signal and a high-frequency boundary branch signal through complementary frequency domain split modulation. The background component in the low-frequency background branch undergoes energy matching to generate a π-phase inversion, and is then converted into a spatial domain background signal via an inverse Fourier transform. The tumor boundary spectral component in the high-frequency boundary branch is enhanced by boundary band gain and converted into a spatial domain boundary signal via an inverse Fourier transform. The two spatial domain signals are input to an interferometric beam combiner. Since the phase difference between the background component in the spatial domain background signal and the corresponding background projection component in the spatial domain boundary signal is π, the background component cancels out during beam combining, while the tumor boundary information is constructively enhanced, thus obtaining a differential enhanced image, which is then acquired by a differential enhancement acquisition array and output.

[0041] After acquisition by the differential enhancement acquisition array, the differential enhancement images output are sequentially subjected to edge candidate point detection, broken edge connection, morphological closure, connected component screening, and contour tracking to obtain the whole tumor WT closed boundary image and output it for display.

[0042] Figure 2 This is a schematic diagram of an MRI brain tumor image according to one embodiment of this application, wherein the white highlighted area is the brain tumor area; Figure 3 yes Figure 2 A schematic diagram of the WT region segmentation result obtained after segmenting the brain tumor region. (See diagram below.) Figure 2 , Figure 3 As shown, the MRI tumors extracted using the technical solution of this application have clear, continuous, and complete boundaries.

[0043] Based on the above embodiments of this application, the MRI tumor boundary extraction scheme of this application has at least the following advantages: This application selectively enhances MRI images in the frequency plane through complex amplitude frequency domain enhancement. Low-frequency contour protection can preserve the overall contour of the tumor region. Circular spectrum dual-threshold modulation can extract and enhance the main frequency band of the tumor boundary. Directional enhancement can improve the response of irregular and weak boundaries. Fractional-order differential enhancement can enhance the boundary while suppressing the high-frequency noise amplification problem caused by ordinary integer-order differentials.

[0044] This application can adjust the frequency domain modulation parameters based on feedback from boundary energy, background residue, high-frequency noise ratio, and boundary closure trend, so that the scheme can adapt to the gray-scale distribution and boundary strength variations of different MRI images.

[0045] This application uses Fourier complementary differential interferometry to divide the initially enhanced image into low-frequency background branches and high-frequency boundary branches. Through energy matching and interferometric beam combining, background information is suppressed and tumor boundary information is enhanced.

[0046] This application obtains continuous and complete WT closed boundaries by performing edge candidate point detection, broken edge connection, morphological closure, connected component screening and contour tracking on differentially enhanced images, thereby reducing boundary breakage and false detection problems.

[0047] Figure 4 This is a schematic diagram of an electronic device illustrated in this specification according to an exemplary embodiment. Please refer to... Figure 4 At the hardware level, the device includes a processor 402, an internal bus 404, a network interface 406, memory 408, a hardware acceleration device 410, and non-volatile memory 412, and may also include other hardware required for its functions. One or more embodiments of this application can be implemented in software, for example, the processor 402 reads the corresponding computer program from the non-volatile memory 412 into memory 408 and then runs it. Of course, in addition to software implementation, one or more embodiments of this application do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the above processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0048] Figure 5 This is a structural block diagram of an MRI tumor boundary extraction system based on complex amplitude frequency domain enhancement and complementary differential interference, as illustrated in an exemplary embodiment of this application. Figure 5 As shown, the MRI tumor boundary extraction system can be applied to, for example... Figure 5 The electronic device shown is used to implement the technical solution of this application.

[0049] Among them, the MRI tumor boundary extraction system based on complex amplitude frequency domain enhancement and complementary differential interference includes: The frequency domain transformation module is used to perform Fourier transform on MRI images to obtain frequency domain signals; The frequency domain enhancement module is used to perform frequency domain modulation on the frequency domain signal using a complex amplitude frequency domain operator to obtain a frequency domain modulated signal; the frequency domain modulation includes low-frequency profile protection, ring spectrum double threshold modulation, direction enhancement, and fractional derivative enhancement. The spatial domain enhancement module is used to perform an inverse Fourier transform on the frequency domain modulated signal to obtain a spatial domain enhanced image. The feedback adjustment module is used to generate a boundary spectrum evaluation value based on the boundary energy, background residue, high-frequency noise ratio, and boundary closure trend of the spatial domain enhanced image. If the boundary spectrum evaluation value does not meet the preset convergence condition, the adjustable parameters of the complex amplitude frequency domain operator in the frequency domain enhancement module are adjusted based on the boundary spectrum evaluation value, and the frequency domain enhancement module and the spatial domain enhancement module are controlled to run iteratively until the preset convergence condition is met. The differential enhancement module is used to perform Fourier complementary differential interferometry enhancement processing on the spatial domain enhanced image that meets the preset convergence conditions to obtain the differential enhanced image; The contour extraction module is used to extract closed boundaries from the differential enhancement image to obtain the whole tumor WT closed boundary image; wherein, the whole tumor WT closed boundary image is used to represent the overall outer contour boundary of the tumor region in the MRI image.

[0050] Figure 6 This is a block diagram illustrating the structure of a frequency domain enhancement module according to an exemplary embodiment of this application. Figure 6 As shown, in some embodiments, the frequency domain enhancement module includes a low-frequency profile protection unit, a ring spectrum dual-threshold modulation unit, a direction enhancement unit, and a fractional-order differential enhancement unit.

[0051] The low-frequency profile protection unit is used to obtain the frequency radius of the frequency domain signal; When the frequency radius is less than the first frequency threshold, the corresponding spectral components are either preserved in low frequency contours or weakened based on the low frequency preservation ratio.

[0052] The ring spectrum dual-threshold modulation unit is used to take the corresponding spectral components as candidate information of the main frequency band of the tumor boundary when the frequency radius is greater than or equal to the first frequency threshold and less than or equal to the second frequency threshold, and to perform enhancement processing based on the boundary enhancement gain. When the frequency radius is greater than the second frequency threshold, high-frequency noise suppression processing is performed on the corresponding frequency components based on the noise suppression intensity.

[0053] The directional enhancement unit is used to obtain the frequency domain coordinates of each spectral component in the frequency domain signal and calculate the corresponding spectral direction angle based on the frequency domain coordinates; Based on the directional distribution of the boundary to be enhanced in the MRI image, at least one directional enhancement center angle is set; Calculate the angle difference between the spectral direction angle and the direction enhancement center angle; When the angle difference is less than or equal to a preset angle threshold, a first direction enhancement weight is applied to the corresponding spectral component; When the angle difference is greater than the preset angle threshold, a second direction enhancement weight is applied to the corresponding spectral component; Among them, the weight of the first direction enhancement is greater than the weight of the second direction enhancement, and the direction enhancement center angle, the preset angle threshold, the weight of the first direction enhancement and the weight of the second direction enhancement are all adjustable parameters of the complex amplitude frequency domain operator.

[0054] The fractional derivative enhancement unit is used to set the fractional derivative order, which is greater than 0 and less than or equal to 2. Construct a fractional-order differential gain function based on the frequency radius of the spectral components in the frequency domain signal; Fractional differential gain function is used to enhance the spectral components within the candidate information of the main frequency band of the tumor boundary using fractional differential gain function; Gain limiting is applied to spectral components above a second frequency threshold to suppress excessive amplification of high-frequency noise while improving boundary clarity.

[0055] In some embodiments, the feedback adjustment module includes an FPGA boundary spectrum evaluation unit, which is used for: Edge candidate point detection is performed on the spatial domain enhanced image to obtain edge candidate regions and background candidate regions; Calculate the boundary energy based on the average edge response or the dominant frequency band energy of the edge candidate region. Calculate background residuals based on residual responses in background candidate regions; The proportion of high-frequency noise is calculated based on the ratio of high-frequency spectral energy above the second frequency threshold to the total spectral energy. The boundary closure trend is calculated based on the ratio of the closed contour length to the total edge length, or based on the ratio of the number of edge endpoints to the number of edge pixels. The boundary spectrum evaluation value is generated by a weighted combination of boundary energy, background residual, high-frequency noise ratio, and boundary closure trend.

[0056] In some embodiments, the spatial domain enhancement module is used to: perform a Fourier transform on the spatial domain enhanced image to obtain a frequency domain enhanced signal; The frequency domain enhancement signal is divided into low-frequency background branch signal and high-frequency boundary branch signal; The low-frequency background branch signal is sequentially subjected to energy matching processing and inverse Fourier transform processing to obtain the spatial domain background signal; the high-frequency boundary branch signal is sequentially subjected to boundary band gain processing and inverse Fourier transform processing to obtain the spatial domain boundary signal; wherein, the phase difference between the background component in the spatial domain background signal and the corresponding background projection component in the spatial domain boundary signal is π. By interferometrically combining the spatial domain background signal and the spatial domain boundary signal, the background component cancels out and the boundary component constructs, thus obtaining a differentially enhanced image.

[0057] Figure 7 This is a block diagram illustrating the differential enhancement module structure in an exemplary embodiment of this application. Figure 7 As shown, in some embodiments, the differential enhancement module includes: a Fourier transform unit, a complementary spectrum splitter adjustment unit, an energy matching unit, a first inverse Fourier transform unit, a boundary band gain unit, a second inverse Fourier transform unit, an interferometric beam combiner, and a differential enhancement acquisition array. The Fourier transform unit is used to perform Fourier transform on the spatial domain enhanced image to obtain the frequency domain enhanced signal; The complementary spectrum splitter adjustment unit is used to split the frequency domain enhancement signal into a low-frequency background branch signal and a high-frequency boundary branch signal; wherein, the phase difference between the background component in the spatial domain background signal and the corresponding background projection component in the spatial domain boundary signal is π. An energy matching unit is used to sequentially perform energy matching processing on the low-frequency background branch signal; The first inverse Fourier transform unit is used to perform inverse Fourier transform processing on the output of the energy matching unit to obtain the spatial domain background signal. The boundary band gain unit is used to sequentially perform boundary band gain processing on the high-frequency boundary branch signals. The second inverse Fourier transform unit is used to perform inverse Fourier transform processing on the output of the boundary frequency band gain unit to obtain the spatial domain boundary signal. An interferometric beam combiner is used to interferometrically combine spatial domain background signals and spatial domain boundary signals. The differential enhancement acquisition array is used to differentially enhance the output of the interferometric beam combiner, so that the background components cancel each other out and the boundary components construct each other, thus obtaining a differentially enhanced image.

[0058] For the system embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this application according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0059] Accordingly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the methods of any of the above embodiments.

[0060] Accordingly, embodiments of this application also provide a computer program product configured to perform the methods of any of the above embodiments.

[0061] The systems, devices, modules, or units described in the above embodiments can be implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer, which can take the form of a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email sending and receiving device, game console, tablet computer, wearable device, or any combination of these devices.

[0062] In a typical configuration, a computer includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0063] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0064] Computer-readable media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random-access memory (SRAM), dynamic random-access memory (DRAM), other types of random-access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0065] While this specification contains numerous specific implementation details, these should not be construed as limiting the scope of any invention or the scope of the claims, but rather are primarily intended to describe features of specific embodiments of a particular invention. Certain features described in the various embodiments herein may also be implemented in combination in a single embodiment. Conversely, various features described in a single embodiment may also be implemented separately in various embodiments or in any suitable sub-combination. Furthermore, while features may function in certain combinations as described above and even initially claimed in this way, one or more features from a claimed combination may be removed from that combination in some cases, and a claimed combination may refer to a sub-combination or a variation thereof.

[0066] Similarly, although the operations are depicted in a specific order in the accompanying drawings, this should not be construed as requiring these operations to be performed in the specific order shown or sequentially, or requiring all illustrated operations to be performed to achieve the desired result. In some cases, multitasking and parallel processing may be advantageous. Furthermore, the separation of various system modules and components in the above embodiments should not be construed as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

[0067] Thus, specific embodiments of the subject matter have been described. Other embodiments are within the scope of the appended claims. In some cases, the actions recited in the claims may be performed in a different order and still achieve the desired result. Furthermore, the processes depicted in the drawings are not necessarily shown in a specific order or sequence to achieve the desired result. In some implementations, multitasking and parallel processing may be advantageous.

[0068] It should be noted that 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 said element.

[0069] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for extracting MRI tumor boundaries based on complex amplitude frequency domain enhancement and complementary differential interference, characterized in that, include: Step S10: Perform Fourier transform on the MRI image to obtain the frequency domain signal; Step S20: Frequency domain modulation is performed on the frequency domain signal using a complex amplitude frequency domain operator to obtain a frequency domain modulated signal; the frequency domain modulation includes low-frequency profile protection, ring spectrum double-threshold modulation, direction enhancement, and fractional derivative enhancement; Step S30: Perform an inverse Fourier transform on the frequency domain modulation signal to obtain a spatial domain enhanced image; Step S40: Generate a boundary spectrum evaluation value based on the boundary energy, background residual, high-frequency noise ratio, and boundary closure trend of the spatial domain enhanced image; if the boundary spectrum evaluation value does not meet the preset convergence condition, adjust the adjustable parameters of the complex amplitude frequency domain operator based on the boundary spectrum evaluation value, and return to step S20 for iterative execution; Step S50: Perform Fourier complementary differential interferometry enhancement processing on the spatial domain enhanced image that satisfies the preset convergence condition to obtain a differential enhanced image; Step S60: Extract the closed boundary of the differential enhancement image to obtain the whole tumor WT closed boundary image; wherein, the whole tumor WT closed boundary image is used to represent the overall outer contour boundary of the tumor region in the MRI image.

2. The method according to claim 1, characterized in that, In step S20, the low-frequency profile protection and ring spectrum dual-threshold modulation of the frequency domain modulation include the following sub-steps: Obtain the frequency radius of the frequency domain signal; When the frequency radius is less than the first frequency threshold, the corresponding spectral components are either preserved in low frequency contours or weakened based on the low frequency preservation ratio. When the frequency radius is greater than or equal to the first frequency threshold and less than or equal to the second frequency threshold, the corresponding spectral component is used as candidate information for the main frequency band of the tumor boundary, and enhancement processing is performed based on the boundary enhancement gain. When the frequency radius is greater than the second frequency threshold, high-frequency noise suppression processing is performed on the corresponding frequency components based on the noise suppression intensity.

3. The method according to claim 1, characterized in that, In step S20, the directional enhancement of the frequency domain modulation includes the following sub-steps: Obtain the frequency domain coordinates of each spectral component in the frequency domain signal, and calculate the corresponding spectral direction angle based on the frequency domain coordinates; Based on the directional distribution of the boundary to be enhanced in the MRI image, at least one directional enhancement center angle is set; Calculate the angle difference between the spectral direction angle and the direction enhancement center angle; When the angle difference is less than or equal to a preset angle threshold, a first direction enhancement weight is applied to the corresponding spectral component; When the angle difference is greater than the preset angle threshold, a second direction enhancement weight is applied to the corresponding spectral component; Wherein, the first direction enhancement weight is greater than the second direction enhancement weight, and the direction enhancement center angle, the preset angle threshold, the first direction enhancement weight and the second direction enhancement weight are all adjustable parameters of the complex amplitude frequency domain operator.

4. The method according to claim 3, characterized in that, In step S20, the fractional-order differential enhancement of the frequency domain modulation includes the following sub-steps: Set a fractional derivative order, wherein the fractional derivative order is greater than 0 and less than or equal to 2; Construct a fractional-order differential gain function based on the frequency radius of the spectral components in the frequency domain signal; The fractional differential gain function is used to perform fractional differential enhancement on the spectral components within the candidate information of the main frequency band of the tumor boundary. Gain limiting is applied to spectral components above the second frequency threshold to suppress excessive amplification of high-frequency noise while improving boundary clarity.

5. The method according to claim 1, characterized in that, In step S40, the process of generating boundary spectrum evaluation values ​​includes: Edge candidate point detection is performed on the spatial domain enhanced image to obtain edge candidate regions and background candidate regions; The boundary energy is calculated based on the average edge response or the dominant frequency band energy of the edge candidate region. Calculate background residue based on the residual response in the background candidate region; The proportion of high-frequency noise is calculated based on the ratio of high-frequency spectral energy above the second frequency threshold to the total spectral energy. The boundary closure trend is calculated based on the ratio of the closed contour length to the total edge length, or based on the ratio of the number of edge endpoints to the number of edge pixels. A boundary spectrum evaluation value is generated based on a weighted combination of the boundary energy, background residual, high-frequency noise ratio, and boundary closure trend.

6. The method according to claim 5, characterized in that, In step S40, the preset convergence condition is met, specifically: The boundary spectrum evaluation value reaches the preset evaluation threshold; Or at least two of the following conditions must be met: The boundary energy reaches the preset boundary energy threshold. Background residue is lower than the preset background residue threshold; The proportion of high-frequency noise is lower than the preset noise proportion threshold. The boundary closure trend has reached the preset closure trend threshold; The change in the boundary spectrum evaluation value obtained in two consecutive iterations is less than the preset change threshold.

7. The method according to claim 6, characterized in that, In step S40, The adjustable parameters include at least one of the following: a first frequency threshold, a second frequency threshold, a low-frequency retention ratio, a first directional enhancement weight, a second directional enhancement weight, a directional enhancement center angle, a preset angle threshold, a boundary enhancement gain, a noise suppression strength, and a fractional derivative order. Feedback adjustment of the adjustable parameters of the complex amplitude frequency domain operator includes: When the boundary energy is lower than the preset boundary energy threshold, the difference between the second frequency threshold and the first frequency threshold is increased to expand the main frequency band range of the ring boundary. When the background residue is higher than the preset background residue threshold, the low-frequency retention ratio is reduced. When the high-frequency noise ratio is higher than the preset noise ratio threshold, the second frequency threshold and the fractional derivative order are reduced. When the boundary closure trend is lower than the preset closure trend threshold, the direction enhancement weight in the corresponding direction is increased; The boundary spectrum evaluation value is recalculated based on the adjusted parameters, and it is determined whether the boundary spectrum evaluation value meets the preset convergence condition. If it does, the feedback adjustment is stopped, and the current spatial domain enhanced image is taken as the spatial domain enhanced image that meets the preset convergence condition. Otherwise, the feedback adjustment is repeated.

8. The method according to claim 1, characterized in that, Step S50 specifically includes the following sub-steps: Perform a Fourier transform on the spatial domain enhanced image to obtain the frequency domain enhanced signal; The frequency domain enhancement signal is divided into low-frequency background branch signal and high-frequency boundary branch signal; The low-frequency background branch signal is subjected to energy matching processing and inverse Fourier transform processing in sequence to obtain the spatial domain background signal; the high-frequency boundary branch signal is subjected to boundary band gain processing and inverse Fourier transform processing in sequence to obtain the spatial domain boundary signal; wherein, the phase difference between the background component in the spatial domain background signal and the corresponding background projection component in the spatial domain boundary signal is π. The spatial domain background signal and the spatial domain boundary signal are subjected to interference and beam combining, so that the background component cancels out and the boundary component is constructive, to obtain the differential enhancement image.

9. The method according to claim 1, characterized in that, The closed boundary extraction in step S60 includes: edge candidate point detection, broken edge connection, morphological closure, connected component screening, and contour tracing.

10. An MRI tumor boundary extraction system based on complex amplitude frequency domain enhancement and complementary differential interference, characterized in that, include: The frequency domain transformation module is used to perform Fourier transform on MRI images to obtain frequency domain signals; The frequency domain enhancement module is used to perform frequency domain modulation on the frequency domain signal using a complex amplitude frequency domain operator to obtain a frequency domain modulated signal; the frequency domain modulation includes low-frequency profile protection, ring spectrum dual-threshold modulation, direction enhancement, and fractional derivative enhancement. The spatial domain enhancement module is used to perform an inverse Fourier transform on the frequency domain modulation signal to obtain a spatial domain enhanced image; The feedback adjustment module is used to generate a boundary spectrum evaluation value based on the boundary energy, background residue, high-frequency noise ratio, and boundary closure trend of the spatial domain enhanced image. If the boundary spectrum evaluation value does not meet the preset convergence condition, the adjustable parameters of the complex amplitude frequency domain operator in the frequency domain enhancement module are adjusted based on the boundary spectrum evaluation value, and the frequency domain enhancement module and the spatial domain enhancement module are controlled to run iteratively until the preset convergence condition is met. The differential enhancement module is used to perform Fourier complementary differential interference enhancement processing on the spatial domain enhanced image that satisfies the preset convergence condition to obtain a differential enhanced image; The contour extraction module is used to extract the closed boundary of the differential enhancement image to obtain the whole tumor WT closed boundary image; wherein, the whole tumor WT closed boundary image is used to represent the overall outer contour boundary of the tumor region in the MRI image.

Images