A lung nodule analysis method based on morphogenesis steady state evaluation

Abstract

The embodiment of the application provides a lung nodule analysis method based on morphological steady state evaluation, and belongs to the technical field of image recognition, and specifically comprises: labeling the position and boundary of the nodule in the CT image; performing frequency domain morphological spectrum analysis on the CT image, identifying the characteristic wave number corresponding to the nodule; based on the characteristic wave number, constructing a frequency domain ring-shaped mask in the Fourier space, and applying a self-adaptive intensity frequency domain disturbance to generate a disturbed CT image; inputting the CT image and the disturbed CT image into a segmentation model respectively to obtain an original segmentation mask and a disturbed segmentation mask; comparing the difference between the original segmentation mask and the disturbed segmentation mask, and normalizing the average disturbance intensity of the nodule area to calculate the morphological steady state margin of the nodule; according to the morphological steady state margin, the nodule is classified into different development stages, and a development stage label and a clinical interpretation thereof are output. Through the scheme of the application, the analysis accuracy, adaptability and interpretability are improved.

Description

Technical Field

[0001] This invention relates to the field of image recognition technology, and in particular to a method for analyzing lung nodules based on morphogenetic steady-state assessment. Background Technology

[0002] Currently, assessing the developmental stage of pulmonary nodules is a crucial step in the early auxiliary diagnosis of lung cancer. Existing technologies heavily rely on radiologists' empirical judgment based on static morphological features from a single CT image. This method suffers from significant subjective variability and poor consistency, and it is difficult to extract dynamic evolution information of nodules from static images.

[0003] To improve the objectivity and quantifiability of assessments, methods based on radiomics and deep learning have emerged. Existing methods can be divided into two categories: one is radiomics methods based on geometric feature engineering, which use manually designed features combined with classifiers to distinguish between benign and malignant nodules, but it treats nodules as static geometric objects and cannot capture the dynamic properties of morphogenesis; the other is end-to-end methods based on deep learning, which learn implicit feature representations through models such as 3D convolutional neural networks and Vision Transformers, but its decisions lack biological interpretability and are also limited to a static snapshot perspective.

[0004] It is evident that there is an urgent need for a lung nodule analysis method based on morphogenetic homeostasis assessment that offers high accuracy, adaptability, and interpretability. Summary of the Invention

[0005] In view of this, embodiments of the present invention provide a lung nodule analysis method based on morphogenetic steady-state assessment, which at least partially solves the problems of poor analysis accuracy, adaptability and interpretability in the prior art.

[0006] This invention provides a method for lung nodule analysis based on morphogenetic homeostasis assessment, comprising: Step 1: Obtain CT images containing lung nodules and mark the location and boundaries of the nodules in the CT images; Step 2: Perform frequency domain morphological spectrum analysis on CT images to identify the characteristic wavenumbers corresponding to the nodules; Step 3: Based on the characteristic wavenumber, construct a frequency domain annular mask in Fourier space and apply frequency domain perturbation with adaptive intensity to generate perturbed CT images. Step 4: Input the CT images and the perturbed CT images into the segmentation model respectively, and obtain the original segmentation mask and the perturbed segmentation mask through dual-channel morphogenesis field mapping. Step 5: Compare the differences between the original segmentation mask and the segmentation mask after perturbation, and normalize the difference using the average perturbation intensity of the nodule region to calculate the morphological steady-state margin of the nodule. Step 6: Based on morphological homeostasis margin, classify nodules into different developmental stages and output developmental stage labels and their clinical interpretations.

[0007] According to a specific implementation of an embodiment of the present invention, step 2 specifically includes: Perform a two-dimensional discrete Fourier transform on the CT images to convert the images from the spatial domain to the frequency domain, and obtain the spectrum; Transform the spectrum from rectangular coordinates to polar coordinates and calculate the radial energy spectrum; Obtain the reference spectrum of the same type of CT equipment, and calculate the relative deviation spectrum of the radial energy spectrum relative to the reference spectrum; Characteristic wavenumbers are determined using a preset formula and high-pass weighted peak value localization. : ; in, The relative deviation spectrum, This refers to the high-pass filter bandwidth parameter. The radial wavenumber represents the most significant spatial frequency scale at which the nodule morphogenesis process deviates from normal tissue.

[0008] According to a specific implementation of an embodiment of the present invention, step 3 specifically includes: In Fourier space, with the aforementioned characteristic wavenumber Construct a Gaussian ring mask centered on the target: ; in, The standard deviation of the annular mask, This is a relative bandwidth parameter; Calculate adaptive disturbance strength : ; in, As the reference disturbance strength, The energy at the characteristic wavenumber, Maximum energy across the entire frequency band; Perturbed CT images are generated using the following formula. : ; in, Indicates the inverse Fourier transform. The spectrum of CT images, It is complex Gaussian noise.

[0009] According to a specific implementation of an embodiment of the present invention, step 5 specifically includes: Calculate the difference area ratio of the segmented mask before and after the perturbation. : ; in, For the original segmentation mask, This is a segmentation mask after perturbation. This is a pixel-by-pixel XOR operation; Calculate the average disturbance intensity within the nodule region. : ; in, The pixel values ​​of the CT image. The pixel values ​​of the disturbed CT image. and These are the height and width of the image, respectively; The morphological steady-state margin is calculated using the following formula. : ; in, To prevent extremely small constants from being divided by zero, the morphological steady-state margin... Used to characterize the stability of nodule morphology against motion. A larger value indicates a more stable form.

[0010] According to a specific implementation of an embodiment of the present invention, step 6 specifically includes: When the steady-state margin When the value approaches 1, the nodule is classified as terminally differentiated, which is clinically interpreted as the nodule boundary being highly stable and the morphological development having entered the terminal stage. When the steady-state margin When the nodules are within the first preset range, they are classified as steady-state maintenance states. When the steady-state margin When the nodules are within the second preset range, they are classified as morphologically malleable. When the steady-state margin When the value approaches 0, the nodules are classified as active, which is clinically interpreted as the nodule morphology being highly malleable and in the early stage of development or a period of rapid change.

[0011] According to a specific implementation of an embodiment of the present invention, after step 6, the method further includes: When the doctor makes a correction to the nodule boundary, the correction is interpreted as an adjustment to the diffusion coefficient ratio of the morphogenesis field. The inverted coefficient-constrained segmentation network is then used to generate a refined boundary that satisfies the continuity of the morphogenesis field.

[0012] According to a specific implementation of an embodiment of the present invention, the step of interpreting the correction behavior as an adjustment of the diffusion coefficient ratio of the morphogenesis field includes: When doctors enlarge the nodule borders, they interpret it as an upregulation of the activator diffusion coefficient. ; When doctors shrink the nodule boundaries, they interpret it as an upregulation of the suppressor diffusion coefficient. ; Based on the rate of change of nodule area before and after correction, the diffusion coefficient ratio is mapped to... The adjustment amount is used as a constraint input to the conditional inference branch of the segmentation network.

[0013] The lung nodule analysis scheme based on morphogenetic steady-state assessment in this embodiment of the invention includes: Step 1, acquiring CT images containing lung nodules and annotating the location and boundaries of the nodules in the CT images; Step 2, performing frequency domain morphological spectrum analysis on the CT images to identify the characteristic wavenumbers corresponding to the nodules; Step 3, constructing a frequency domain annular mask in Fourier space based on the characteristic wavenumbers and applying adaptive intensity frequency domain perturbation to generate perturbed CT images; Step 4, inputting the CT images and perturbed CT images into a segmentation model respectively, and obtaining the original segmentation mask and the perturbed segmentation mask through dual-channel morphogenetic field mapping; Step 5, comparing the differences between the original segmentation mask and the perturbed segmentation mask, and normalizing by the average perturbation intensity of the nodule region to calculate the morphological steady-state margin of the nodules; Step 6, classifying the nodules into different developmental stages according to the morphological steady-state margin, and outputting developmental stage labels and their clinical interpretations.

[0014] The beneficial effects of the embodiments of the present invention are as follows: Through the solution of the present invention, an analysis process of "frequency domain detection-morphological response-steady-state quantization" is constructed. By mapping the adaptive frequency domain perturbation probe with the dual-channel morphogenesis field, the steady-state margin of nodule morphology is calculated, the abstract concept of developmental stage is transformed into a calculable specific index, and the objective classification of developmental stages is realized based on the index. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a lung nodule analysis method based on morphogenetic steady-state assessment, provided as an embodiment of the present invention. Detailed Implementation

[0017] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0018] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, in the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0019] It should be noted that various aspects of embodiments within the scope of the appended claims are described below. It will be apparent that the aspects described herein can be embodied in a wide variety of forms, and any particular structure and / or function described herein is merely illustrative. Based on this invention, those skilled in the art will understand that one aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number of aspects set forth herein can be used to implement the device and / or practice the method. Additionally, this device and / or method can be implemented using structures and / or functionalities other than one or more of the aspects set forth herein.

[0020] It should also be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. The illustrations only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0021] Furthermore, specific details are provided in the following description to facilitate a thorough understanding of the examples. However, those skilled in the art will understand that the described aspects can be practiced without these specific details.

[0022] Specifically, existing technologies suffer from the following shortcomings: First, current methods struggle to answer the core question of "how stable is nodule morphology," a question involving developmental dynamics, and cannot transform developmental stage concepts such as "terminal differentiation state" and "active developmental stage" into calculable quantitative indicators. Second, frequency domain analysis, exemplified by FPS, applies perturbations to the model training process to improve algorithm generalization; works such as radiomics robustness databases assess the reproducibility of feature extraction algorithms through perturbation evaluation. While the existing AdaDSP framework introduces an adaptive frequency band selection mechanism in the frequency domain, its perturbations also affect the training phase, aiming to enhance data diversity and improve model generalization. The frequency domain perturbations in these methods all serve algorithm evaluation or model training, rather than the detection of the biological attributes of the lesion itself; the potential of frequency domain information in assessing nodule developmental stages is not effectively utilized. Third, when doctors manually correct AI-annotated boundaries, the correction operation is merely used as a label correction signal, lacking a unified biological framework to quantify its significance in inferring nodule morphological parameters.

[0023] Therefore, existing technologies struggle to objectively and quantitatively assess the developmental stages of pulmonary nodules from a morphogenetic dynamics perspective. There is an urgent need for an assessment scheme that can utilize frequency domain perturbations to detect nodule morphological homeostasis, thereby automatically generating quantitative indicators corresponding to developmental stages, and possessing biological interpretability.

[0024] This invention provides a method for analyzing lung nodules based on morphogenetic homeostasis assessment, which can be applied to medical data classification processes.

[0025] See Figure 1 This is a flowchart illustrating a lung nodule analysis method based on morphogenetic homeostasis assessment, provided by an embodiment of the present invention. Figure 1 As shown, the method mainly includes the following steps: Step 1: Obtain CT images containing lung nodules and mark the location and boundaries of the nodules in the CT images; In practice, a dataset of CT images containing lung nodules of different pathological types can be obtained, and the location and boundaries of each nodule can be labeled.

[0026] Step 2: Perform frequency domain morphological spectrum analysis on CT images to identify the characteristic wavenumbers corresponding to the nodules; In practice, the steps aim to convert the input CT images from the spatial domain to the frequency domain, and accurately locate the most active spatial frequency scale of the nodule morphogenesis field by comparing it with the equipment's reference spectrum.

[0027] Unlike traditional methods that only utilize the frequency domain for data augmentation or feature extraction, this approach identifies nodule-specific characteristic wavenumbers from frequency domain energy deviations, providing precise frequency band localization for subsequent adaptive frequency domain perturbations. This step forms the basis for establishing frequency domain detection of morphogenetic fields and can be formally defined as a mapping function from CT images to characteristic wavenumbers.

[0028] ; This is a frequency domain morphological spectrum analysis function, whose input is CT images. (H and W represent the height and width of the image, respectively), the output is the characteristic wavenumber. This characteristic wavenumber represents the most significant spatial frequency scale at which the nodule morphogenesis deviates from normal tissue.

[0029] In the specific implementation of the frequency domain morphological spectrum analysis, a two-dimensional discrete Fourier transform is first performed on the input CT image to convert the image from the spatial domain to the frequency domain, thereby obtaining the spectrum. .

[0030] Subsequently, the spectrum was converted from rectangular coordinates Transform to polar coordinates ,in K is the radial wavenumber, representing the scale of spatial frequency (smaller k corresponds to low-frequency components of coarse texture, and larger k corresponds to high-frequency components of fine texture). Let be the angular coordinates. Calculate the radial energy spectrum E(k), which is to integrate the energy in all directions at a certain frequency scale k and then take the average.

[0031] To eliminate the interference of normal lung tissue spectrum on the localization of nodule characteristic frequency bands, a reference spectrum from the same model of CT equipment was introduced. This baseline was obtained by statistically averaging a large number of normal lung tissue images using the same procedure. The relative deviation spectrum was calculated as follows: ; in, The spectrum anomalies reflected by nodules are indicated by positive values, which show that the energy at that frequency scale is higher than that of normal tissue, while negative values ​​show that it is lower than that of normal tissue.

[0032] Finally, the characteristic wave was determined by high-pass weighted peak localization. In the formula This is the high-pass filter bandwidth parameter, which controls the degree of low-frequency suppression; For high-pass weighting: when k is small, this term is close to 0, which effectively suppresses the low-frequency background formed by normal lung texture; when k is large, this term is close to 1, which preserves the high-frequency nodule features. This indicates the value of k that maximizes the expression within the parentheses.

[0033] For example, in a case of adenocarcinoma nodule with spiculated borders, the high-frequency components had significantly higher energy than normal lung tissue. A significant peak appears in the high wavenumber region, after Qualcomm weighting. The location at this peak corresponds to the characteristic spatial scale of irregular changes in the nodule edge.

[0034] Step 3: Based on the characteristic wavenumber, construct a frequency domain annular mask in Fourier space and apply frequency domain perturbation with adaptive intensity to generate perturbed CT images. In practice, this step aims to apply precise perturbations to the sensitive frequency bands of the nodule morphogenesis field in Fourier space based on the characteristic wavenumbers identified in step 2. Unlike traditional methods that apply uniform perturbations to the entire image or frequency band, this step achieves adaptive positioning of the perturbation frequency band and adaptive adjustment of the perturbation intensity, making the perturbation a precise probe for detecting the steady state of nodule morphology. This step is the core operational step in the steady-state detection of the morphogenesis field. This process can be formally represented as: ; in, For frequency domain perturbation function, For the generated perturbed CT images, Indicates the inverse Fourier transform. To adapt to the disturbance strength, It is a frequency domain annular mask. It is complex Gaussian noise. The noise variance controls the amplitude of random disturbances.

[0035] In practice, the characteristic wavenumber is first used in Fourier space. Construct a Gaussian ring mask around the center This is used to precisely limit the disturbance to the sensitive frequency band where the nodule morphology occurs: ; in, The standard deviation of the annular mask controls the radial width of the mask; This is a relative bandwidth parameter (typically 0.2), indicating that the mask width is 20% of the center frequency. The mask in... The value is set to a maximum of 1 at the specified frequency band and gradually decreases to 0 towards both sides, ensuring that the disturbance is only applied to the neighborhood of the characteristic frequency band and other frequency bands are not affected.

[0036] Secondly, calculate the adaptive disturbance strength. This correlates the intensity of the disturbance with the energy proportion of the characteristic wavenumber—the stronger the energy of the morphogenetic field at that frequency, the stronger the detection of the disturbance. ; in, The baseline disturbance strength is 0.05 (typical value). The energy at the characteristic wavenumber, This represents the maximum energy across the entire frequency band. (Ratio) Measuring the significance of characteristic frequencies in the overall spectrum allows for customized detection capabilities for different nodules.

[0037] Finally, a complex Gaussian noise perturbation is applied within the ring frequency band, multiplied by a mask and intensity factor, and then superimposed on the original spectrum. After inverse Fourier transform and constraint of pixel values ​​to the original range, the perturbed CT image is obtained. The image is visually almost indistinguishable from the original CT scan, but the characteristic frequencies of nodule morphology have been precisely altered, serving as an input probe for subsequent detection of morphological steady state.

[0038] Step 4: Input the CT images and the perturbed CT images into the segmentation model respectively, and obtain the original segmentation mask and the perturbed segmentation mask through dual-channel morphogenesis field mapping. In practice, the original CT image and the perturbed CT image can be input into the segmentation model respectively, and the original segmentation mask and the perturbed segmentation mask can be obtained through dual-channel morphogenesis field mapping.

[0039] Step 5: Compare the differences between the original segmentation mask and the segmentation mask after perturbation, and normalize the difference using the average perturbation intensity of the nodule region to calculate the morphological steady-state margin of the nodule. In practice, this step aims to quantify the morphological response obtained in step 4 into a calculable steady-state margin index. This step defines the core quantitative index of this invention—morphological steady-state margin. It also provides a quantitative calculation method, making the abstract concept of "morphological stability" measurable and comparable. The specific process is described rigorously as follows: First, calculate the difference in area ratio of the segmentation mask before and after the perturbation. Perform a pixel-by-pixel XOR operation on the two masks and statistically analyze the proportion of regions with inconsistent boundaries to the original nodule area: ; in, For pixel-wise XOR operation—if the two masks have different values, the pixel is 1; if they have the same value, the pixel is 0. |·| represents the number of pixels (area) in the region. , =0 indicates that the boundaries before and after the disturbance are completely consistent (highly stable morphology). The larger the value, the more significant the boundary change.

[0040] Secondly, calculate the average disturbance intensity within the nodule region. , as a normalization factor: ; in, The perturbation magnitude (absolute difference) for each pixel; product term To ensure that only perturbations within the nodule are counted, pixels outside the nodule are excluded by multiplying by 0; normalization is performed by dividing by H×W to obtain the average perturbation intensity of each pixel within the nodule region.

[0041] Finally, the morphological steady-state margin is defined. This is the complement of the boundary change caused by a unit disturbance intensity, and its complete calculation formula is defined as follows: ; in, To prevent division by zero by extremely small constants. The physical meaning of this formula is clear: ratio This represents the rate of change of the boundary caused by a unit disturbance intensity, i.e., the "morphological sensitivity" of the nodule; subtracting this sensitivity from 1 yields the steady-state margin. . → 1 indicates that the nodule boundary has almost no response to disturbances and the morphology has entered a deep steady state (similar to the terminal state of a Turing pattern). →0 indicates that a small perturbation can significantly change the boundary, and the shape is still in a highly malleable period (similar to the early stage of Turing pattern formation).

[0042] Step 6: Based on morphological homeostasis margin, classify nodules into different developmental stages and output developmental stage labels and their clinical interpretations.

[0043] In practice, after obtaining the morphological homeostasis margin, nodules can be classified into terminal differentiation state, homeostasis maintenance state, morphological plasticity state, or active development state based on the morphological homeostasis margin, and developmental stage labels and their clinical interpretations can be output.

[0044] Optionally, after step 6, the method further includes: this step aims to incorporate the physician's manual corrections into the unified morphogenetic field framework for interpretation. In morphogenetic field theory, nodule boundaries are determined by the activator diffusion coefficient. and suppressor diffusion coefficient ratio Decision. The doctor widened the nodule's borders, which is equivalent to increasing the dosage. (Activation enhancement); Doctors narrowing the boundaries is equivalent to upregulation. (Suppression of enhancement). The adjustment amount maps the rate of change in area before and after correction to the diffusion coefficient ratio. This is used as a constraint input to the conditional inference branch of the segmentation network, and the output is a refined boundary that simultaneously satisfies the doctor's correction intention and the continuity of the morphogenesis field.

[0045] Steps 2, 3, and 5 together constitute the core quantization closed loop of this invention: Step 2 locates the sensitive frequency of the nodule morphology generation field; Step 3 applies an adaptive perturbation at this frequency as a probe; Step 5 uses the morphological steady-state margin... The morphological response before and after the perturbation is quantified. Based on the numerical range of M (step 6), the nodules are divided into different developmental stages and a clinical interpretation is output. When the doctor modifies the nodule boundary (step 7), the modification magnitude is mapped to the adjustment amount of the morphogenesis field diffusion coefficient ratio, and the segmentation network is constrained to generate a refined boundary.

[0046] The lung nodule analysis method based on morphogenetic steady-state assessment provided in this embodiment constructs an analysis process of "frequency domain detection-morphological response-steady-state quantification". By mapping an adaptive frequency domain perturbation probe with a dual-channel morphogenetic field, it calculates the nodule morphological steady-state margin, transforming the abstract concept of developmental stage into a calculable specific indicator. Based on this indicator, it achieves objective classification of developmental stages. Applying the method of this invention, it is possible to achieve quantitative assessment of developmental stages from a single CT image without the need for a gold standard label, effectively improving the biological consistency between lung nodule assessment and medical description, and providing a quantifiable morphogenetic dynamic interpretation basis for assessment decisions.

[0047] The units described in the embodiments of the present invention can be implemented in software or in hardware.

[0048] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof.

[0049] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for analyzing pulmonary nodules based on morphogenetic homeostasis assessment, characterized in that, include: Step 1: Obtain CT images containing lung nodules and mark the location and boundaries of the nodules in the CT images; Step 2: Perform frequency domain morphological spectrum analysis on CT images to identify the characteristic wavenumbers corresponding to the nodules; Step 3: Based on the characteristic wavenumber, construct a frequency domain annular mask in Fourier space and apply frequency domain perturbation with adaptive intensity to generate perturbed CT images. Step 4: Input the CT images and the perturbed CT images into the segmentation model respectively, and obtain the original segmentation mask and the perturbed segmentation mask through dual-channel morphogenesis field mapping. Step 5: Compare the differences between the original segmentation mask and the segmentation mask after perturbation, and normalize the difference using the average perturbation intensity of the nodule region to calculate the morphological steady-state margin of the nodule. Step 5 specifically includes: Calculate the difference area ratio of the segmented mask before and after the perturbation. : in, For the original segmentation mask, This is a segmentation mask after perturbation. This is a pixel-by-pixel XOR operation; Calculate the average disturbance intensity within the nodule region : in, The pixel values ​​of the CT image. The pixel values ​​of the disturbed CT image. and These are the height and width of the image, respectively; The morphological steady-state margin is calculated using the following formula. : in, To prevent extremely small constants from being divided by zero, the morphological steady-state margin... Used to characterize the stability of nodule morphology against motion. A larger value indicates a more stable shape; Step 6: Based on morphological homeostasis margin, classify nodules into different developmental stages and output developmental stage labels and their clinical interpretations.

2. The method according to claim 1, characterized in that, Step 2 specifically includes: Perform a two-dimensional discrete Fourier transform on the CT images to convert the images from the spatial domain to the frequency domain, and obtain the spectrum; Transform the spectrum from rectangular coordinates to polar coordinates and calculate the radial energy spectrum; Obtain the reference spectrum of the same type of CT equipment, and calculate the relative deviation spectrum of the radial energy spectrum relative to the reference spectrum; Characteristic wavenumbers are determined using a preset formula and high-pass weighted peak value localization. : in, The relative deviation spectrum, This refers to the high-pass filter bandwidth parameter. The radial wavenumber represents the most significant spatial frequency scale at which the nodule morphogenesis process deviates from normal tissue.

3. The method according to claim 2, characterized in that, Step 3 specifically includes: In Fourier space, with the aforementioned characteristic wavenumber Construct a Gaussian ring mask centered on the target: in, The standard deviation of the annular mask. This is a relative bandwidth parameter; Calculate adaptive disturbance strength : in, As the reference disturbance strength, The energy at the characteristic wavenumber, Maximum energy across the entire frequency band; Perturbed CT images are generated using the following formula. : in, Indicates the inverse Fourier transform. The spectrum of CT images, It is complex Gaussian noise.

4. The method according to claim 3, characterized in that, Step 6 specifically includes: When the steady-state margin When the value approaches 1, the nodule is classified as terminally differentiated, which is clinically interpreted as the nodule boundary being highly stable and the morphological development having entered the terminal stage. When the steady-state margin When the nodules are within the first preset range, they are classified as steady-state maintenance states. When the steady-state margin When the nodules are within the second preset range, they are classified as morphologically malleable. When the steady-state margin When the value approaches 0, the nodules are classified as active, which is clinically interpreted as nodule morphology being highly malleable and in the early stage of development or a period of rapid change.

5. The method according to claim 1, characterized in that, After step 6, the method further includes: When the doctor makes a correction to the nodule boundary, the correction is interpreted as an adjustment to the diffusion coefficient ratio of the morphogenesis field. The inverted coefficient constraint segmentation network is then used to generate a refined boundary that satisfies the continuity of the morphogenesis field.

6. The method according to claim 5, characterized in that, The step of interpreting the correction behavior as an adjustment to the diffusion coefficient ratio of the morphogenesis field includes: When doctors enlarge the nodule borders, they interpret it as an upregulation of the activator diffusion coefficient. ; When doctors shrink the nodule boundaries, they interpret it as an upregulation of the suppressor diffusion coefficient. ; Based on the rate of change of nodule area before and after correction, the diffusion coefficient ratio is mapped to... The adjustment amount is used as a constraint input to the conditional inference branch of the segmentation network.

Images