Medical image denoising system and method based on 3D chaos and attraction and rejection algorithm

Optimizing filter parameters through 3D chaos and improved attraction and rejection algorithms, the problem of poor denoising effect of medical images is solved, efficient noise suppression and detail retention is achieved, and image quality is improved.

CN120471795AActive Publication Date: 2025-08-12CHONGQING NORMAL UNIVERSITY
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202510556437.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-04-29
Publication Date
2025-08-12
Estimated Expiration
2045-04-29

AI Technical Summary

Technical Problem

The traditional Chinese medicine image denoising effect is poor, and it is difficult to dynamically weigh the noise suppression and detail retention. Traditional random initialization leads to uneven parameter space coverage, which is easy to fall into local optimality, and has poor noise denoising effect.

Method used

The hybrid filter parameter group optimization strategy based on 3D chaos and improved attraction and repulsion algorithm is adopted. The population is initialized through the 3D-LY chaos mapping strategy, combined with the improved attraction and repulsion optimization strategy, the individual population position is dynamically adjusted to achieve the improvement of global optimization ability.

Benefits of technology

It significantly improves the effect and efficiency of medical image denoising, can effectively suppress noise and retain image details, and improves image quality and edge retention capabilities.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120471795A_ABST
    Figure CN120471795A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of image processing, in particular to a medical image denoising system and method based on a 3D chaos and attraction and rejection algorithm, and the method comprises the steps: optimizing a strategy based on a preset hybrid filter parameter group according to a historical original medical image; determining an optimal parameter group of each filter corresponding to a hybrid filter used for carrying out denoising processing on the medical image, and configuring the optimal parameter group into the corresponding filter; the preset hybrid filter parameter group optimization strategy comprises a 3D-LY chaotic mapping strategy used for initializing a population corresponding to a hybrid filter parameter group and an improved attraction and rejection optimization strategy used for optimizing the initialized population; obtaining a current medical image to be processed at the current moment, and carrying out noise addition processing on the obtained current medical image to generate a corresponding current image with noise; and taking the current image with noise as input data, inputting the input data into the hybrid filter, and outputting a corresponding denoised medical image.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a medical image denoising system and method based on 3D chaos and attraction-repulsion algorithm. Background Art

[0002] Medical images play a critical role in disease diagnosis, but they are susceptible to motion artifacts, device electronic noise, and other artifacts, generating interference such as speckle and salt-and-pepper noise. This noise not only degrades image quality but can also obscure lesion features, leading to misdiagnosis or missed diagnosis, and even compromising the accuracy of subsequent automated analyses (such as lesion segmentation and volume measurement). Therefore, efficient denoising technology is a core requirement for medical image processing.

[0003] Existing technologies usually use a single filter for denoising or perform denoising through optimization algorithms. Among them, single filter denoising has inherent defects. For example, although the use of linear filters can effectively suppress Gaussian noise, it will cause edge blurring and texture loss, and it is difficult to cope with complex noise patterns, resulting in poor denoising effect.

[0004] The existing optimization algorithms have the following problems:

[0005] Traditional random initialization leads to uneven coverage of the parameter space and is prone to falling into local optima. For example, one-dimensional chaotic maps (such as logistic maps) lack fractal dimension and have a clustered population distribution. This lack of dynamic balance between global exploration and local exploitation makes it difficult to dynamically strike a balance between noise suppression and detail preservation, resulting in poor denoising results.

[0006] Based on this, there is an urgent need for a medical image denoising system and method based on 3D chaos and attraction-repulsion algorithm, which can solve the problem of poor medical image denoising effect in the existing technology and greatly improve the effect and efficiency of medical image denoising. Summary of the Invention

[0007] One of the purposes of the present invention is to provide a medical image denoising system and method based on 3D chaos and attraction-repulsion algorithm, which can solve the problem of poor medical image denoising effect in the existing technology and greatly improve the effect and efficiency of medical image denoising.

[0008] In order to achieve the above object, a medical image denoising method based on 3D chaos and attraction-repulsion algorithm is provided, which includes the following steps:

[0009] S1. Based on a preset hybrid filter parameter group optimization strategy, determining the optimal parameter group of each filter corresponding to the hybrid filter used for denoising the medical image based on the historical original medical image, and configuring the optimal parameter group to the corresponding filter; the preset hybrid filter parameter group optimization strategy includes a 3D-LY chaotic mapping strategy for initializing the population corresponding to the hybrid filter parameter group and an improved attraction-repulsion optimization strategy for optimizing the initialized population;

[0010] S2. Obtaining a current medical image to be processed at the current moment, and performing noise processing on the obtained current medical image to generate a corresponding current noisy image;

[0011] S3. The current noisy image is used as input data and input into the hybrid filter, and the corresponding denoised medical image is output.

[0012] The technical principle and effect of this scheme: In this scheme, the optimal participating group of each filter of the hybrid filter is first determined through the historical original medical images combined with the hybrid filter parameter group optimization strategy. At the same time, the corresponding hybrid filter parameter group optimization strategy includes a 3D-LY chaotic mapping strategy for initializing the population corresponding to the hybrid filter parameter group and an improved attraction-repulsion optimization strategy for optimizing the initialized population. The 3D-LY chaotic mapping strategy is used to initialize the population corresponding to the parameter group. The chaotic system is highly sensitive to the initial values and parameters and can generate non-periodic and ergodic chaotic sequences in three-dimensional space, avoiding the problem of uneven population distribution caused by traditional random initialization. Through the randomness and ergodicity of the chaotic sequence, the initial population is evenly distributed in the parameter search space, covering more potential optimal solution areas and providing high-quality initial samples for subsequent optimization.

[0013] Then, through the improved attraction-repulsion optimization strategy, the attraction (cooperation) and repulsion (competition) behaviors between individuals in the biological population are simulated, so as to adjust the positions corresponding to the individuals in the population, and then adjust the parameters corresponding to the individuals themselves, so as to improve the search efficiency and global optimization ability.

[0014] The current medical image to be processed is then acquired and subjected to noise processing to generate the corresponding current noisy image. Artificially adding simulated noise (e.g., Gaussian noise or salt and pepper noise) to the original image simulates the noise environment of real medical images (e.g., imaging device noise and signal transmission interference), making the noise components in the image more prominent and easier to identify.

[0015] Finally, the noisy image is input into the hybrid filter with optimized parameters. Each component processes the image in a preset order or in parallel. The multi-level filtering of noise is achieved through the synergy of parameters, and the denoised image is finally output.

[0016] Compared to existing techniques that rely on manual parameter adjustment and a single optimization algorithm to adjust filter parameters, this solution utilizes a hybrid optimization approach using a 3D-LY chaotic system coupled with an improved attraction-repulsion method. This significantly improves the efficiency and reliability of global optimization, significantly enhancing medical image denoising in terms of noise suppression, detail preservation, and generalization. Its core value lies in combining bio-inspired optimization with image processing technology, providing an adaptive and efficient solution for complex medical noise environments. This approach has strong clinical application potential and algorithmic scalability, addressing the poor performance of existing medical image denoising techniques and significantly improving their effectiveness and efficiency.

[0017] Furthermore, the 3D-LY chaotic mapping strategy is:

[0018] According to the number of filter types in the hybrid filter, a corresponding three-dimensional chaotic model is constructed;

[0019] The three-dimensional chaos model is:

[0020]

[0021] Where, is the system state variable in the corresponding three-dimensional chaos model, a and b are the parameters of the three-dimensional chaos model;

[0022] According to the system state variables in the three-dimensional chaotic model, the filter parameters corresponding to each filter in the hybrid filter are respectively mapped to one of the system state variables;

[0023] Running the three-dimensional chaotic model and iterating multiple times, recording the system state variables corresponding to each time, and forming a corresponding chaotic sequence matrix; each row of chaotic sequence in the chaotic sequence matrix corresponds to a filter parameter group;

[0024] Each row of chaotic training in the formed chaotic sequence matrix is normalized to form individuals of the corresponding initial population, and the individuals are merged to form the corresponding initial population.

[0025] Beneficial effects: In this scheme, in response to the multi-parameter optimization requirements of the hybrid filter (such as n filters corresponding to n parameter groups, each group containing m parameters), a three-dimensional chaotic system (x, y, z) is constructed, and a complex chaotic sequence is generated through the nonlinear coupling of three state variables. Each variable can be mapped to different types of filter parameters (such as x corresponding to the median filter window size, y corresponding to the Gaussian filter standard deviation, and z corresponding to the wavelet threshold). The cosine / sine function introduces periodicity and nonlinearity, making the system state variables highly sensitive to the initial values and parameters (a, b); the cross terms enhance the trajectory complexity in three-dimensional space and avoid falling into simple periodic oscillations.

[0026] Compared with traditional random initialization, in which the randomly generated parameter groups may be concentrated in local areas, leading to local optimality, this scheme uses three-dimensional chaotic trajectories to traverse the parameter space almost uniformly, ensuring that the initial population covers various parameter combinations from low to high, providing rich samples for global optimization.

[0027] At the same time, based on the need for synergistic effects of parameters of different filters, a parameter group with implicit parameter correlation can be generated through the cross-influence of x, y, and z in the three-dimensional chaotic model. Based on the coupling characteristics of the chaotic model, the potential synergistic rules of the parameters can be automatically mined. Moreover, the generation of chaotic sequences only requires multiple iterations of the three-dimensional chaotic model, which greatly improves the efficiency of generating high-quality initial solutions.

[0028] Furthermore, the improved attraction and repulsion optimization strategy is:

[0029] Step 1: Obtain a historical original medical image and perform noise processing on the original medical image to generate a corresponding original noisy medical image;

[0030] Step 2: According to the formed population, the filter parameter group corresponding to the individuals in the population is configured into the hybrid filter;

[0031] Step 3: The original noisy medical image is used as input data and input into each hybrid filter configured with a corresponding filter parameter group, the original noisy medical image is denoised, and the original denoised medical image corresponding to each population individual is output;

[0032] Step 4: Calculate the parameter fitness corresponding to each population individual based on the original denoised medical image corresponding to each population individual;

[0033] Step 5: According to the parameter fitness corresponding to each population individual, the parameters are ranked in descending order. The individuals with the first preset percentage of parameter fitness ranking are selected as candidates for the global optimal solution, and the attraction mechanism is triggered to drive other individuals in the population to move towards the optimal solution candidate; the individuals with the second preset percentage of parameter fitness ranking are selected as candidates for inferior solutions, and the repulsion mechanism is triggered to drive other individuals in the population away from the inferior solution candidates;

[0034] Step six, based on the current number of iterations, determine the local search strategy corresponding to the number of iterations, and based on the determined local search strategy, search each individual in the population to form a new individual after optimization of each individual in the population, and re-execute step two until the iteration requirements are met, and output the current global optimal solution as the optimal filter parameter group corresponding to the hybrid filter.

[0035] Beneficial effects: In this scheme, the global optimal solution candidate and inferior solution candidate corresponding to each iteration number are determined by calculating the parameter fitness, and then the individuals in the population are driven to diffuse and move by the two-way pressure corresponding to the attraction mechanism and the repulsion mechanism, so that the attraction pushes the population to gather in the current optimal area, and the repulsion prevents the population from gathering in the inferior area, forming a dynamic balance of "approaching the best and avoiding the worst", that is, ensuring that the optimal solution characteristics of each generation can be quickly spread to the entire population, avoiding the dilution of high-quality solutions due to random mutations, and at the same time, by identifying inferior solutions, forcing other individuals to stay away, avoiding the contamination of the population by low-quality parameter groups.

[0036] Of course, the search mode is dynamically switched according to the iterative progress, the exploration scope is expanded in the early stage of optimization, and the optimal solution neighborhood is focused in the later stage, to achieve an efficient search of "breadth first and depth later", to further optimize the parameter group, and further improve the reliability and accuracy of the search for the filter parameter group.

[0037] Furthermore, the parameter fitness corresponding to each population individual is calculated as follows:

[0038]

[0039] Where F is the parameter fitness of the individual in the population, σ, φ is the weight coefficient, MSE is the mean square error corresponding to the individual, PSNR is the peak signal-to-noise ratio corresponding to the individual, and SSIM is the structural similarity index between individuals; I o (i, j) is the pixel value of the original medical image without noise at the coordinate (i, j), I n (i, j) is the pixel value of the denoised medical image at coordinate (i, j), I, J are the number of rows and columns of the image, N is the maximum grayscale value of the image, x is the original medical image, y is the denoised medical image, u xis the pixel mean of the original medical image, u y is the pixel mean of the denoised medical image, is the pixel variance of the original medical image, is the pixel variance of the denoised medical image, σ xy is the covariance between the original medical image and the denoised medical image, c1 and c2 are stability constants.

[0040] Beneficial Effects: Compared to existing single-metric evaluation methods, this approach, through the negative feedback mechanism of PSNR and SSIM, forces the parameter set to find a balance between denoising strength and structure preservation. The gradient of the fitness function with respect to the parameters can be decomposed into a weighted sum of the gradients of each metric. This gradient guides the attraction-repulsion algorithm to search for simultaneous improvements in multiple metrics, avoiding local optimality in a single metric.

[0041] Furthermore, the attraction mechanism in step five is:

[0042] Each individual x in the population i Subject to the global optimal solution candidate x best The corresponding update formula is driven by the attraction of

[0043]

[0044] Where c is the attraction intensity constant, m is the dynamic balance factor, which decreases linearly to 0.1 as the number of iterations t increases, a1 is a random vector from 0 to 1, r1 is a random number from 0 to 1, p1 is the attraction switching threshold, and b i The attractive displacement corresponding to the attractive force.

[0045] Beneficial effect: In this scheme, the individual moves to the global optimal solution based on the global optimal solution candidate, and the adaptive adjustment of the individual to the global optimal solution is achieved, and the a1⊙x best Generate random points in the neighborhood of the optimal solution, so that individuals can maintain exploratory movement when moving towards the global optimal solution, and avoid sudden changes in the intensity of attraction by setting a dynamic balance factor, so that the population can smoothly transition from rapid aggregation to local exploration, preventing the population from spreading out of control due to a sudden drop in attraction.

[0046] Furthermore, the local search strategy in step six includes a Brownian motion strategy, a trigonometric function strategy, and a random jump strategy.

[0047] Beneficial effects: In this solution, a full-cycle adaptive search of the parameter group optimization process is achieved by setting a variety of local search strategies, thereby achieving high reliability of local search.

[0048] The present invention also provides a medical image denoising system based on 3D chaos and attraction-repulsion algorithm, which uses the above-mentioned medical image denoising method based on 3D chaos and attraction-repulsion algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 This is a flow chart of a medical image denoising method based on 3D chaos and attraction-repulsion algorithm in Example 1 of the present invention;

[0050] Figure 2 : Lyapunov exponent spectrum under the change of parameter b in Example 1 of the present invention.

[0051] Figure 3 This is the bifurcation diagram when the parameter b changes in Example 1 of the present invention.

[0052] Figure 4 This is the phase diagram of the chaotic mapping in the first embodiment of the present invention.

[0053] Figure 5 This is a diagram showing the Shannon entropy projected into the phase space of a chaotic system in Example 1 of the present invention.

[0054] Figure 6 This is a diagram showing the phase space projection Shannon entropy of the 3D-LY chaotic system in Example 1 of the present invention.

[0055] Figure 7 This is a comparison experiment diagram of denoising under speckle noise between Example 1 of the present invention and a certain prior art.

[0056] Figure 8 This is a comparison experiment diagram of denoising under salt and pepper noise between Example 1 of the present invention and a certain prior art.

[0057] Figure 9 This is a comparison experiment diagram of denoising under speckle noise between Example 1 of the present invention and some prior arts.

[0058] Figure 10 This is a comparison experiment diagram of denoising under salt and pepper noise in Example 1 of the present invention and some existing technologies. DETAILED DESCRIPTION

[0059] The following is further described in detail through specific implementation methods:

[0060] Example 1

[0061] Medical image denoising method based on 3D chaos and attraction-repulsion algorithm is basically as follows Figure 1 As shown, the following steps are included:

[0062] S1. Based on a preset hybrid filter parameter group optimization strategy, determining the optimal parameter group of each filter corresponding to the hybrid filter used for denoising the medical image based on the historical original medical image, and configuring the optimal parameter group to the corresponding filter; the preset hybrid filter parameter group optimization strategy includes a 3D-LY chaotic mapping strategy for initializing the population corresponding to the hybrid filter parameter group and an improved attraction-repulsion optimization strategy for optimizing the initialized population;

[0063] The 3D-LY chaotic mapping strategy is:

[0064] According to the number of filter types in the hybrid filter, a corresponding three-dimensional chaotic model is constructed;

[0065] The three-dimensional chaos model is:

[0066]

[0067] Where, is the system state variable in the corresponding three-dimensional chaotic model, a and b are the parameters of the three-dimensional chaotic model; in this embodiment, the initial conditions of the system are set to (0, 0.0001, 0.001), and the system parameter a = 7.72 is fixed, and the value of parameter b is changed. When parameter b changes at (1, 6), the system's Lyapunov exponent spectrum and bifurcation diagram with respect to parameter b are as follows: Figure 2 and Figure 3 As shown. Figure 2 and Figure 3 It can be seen that when the parameter b is in the range of [1,6], the system is always in a chaotic state.

[0068] certainly Figure 4 For system parameters a=7.72,b=4,determine the phase diagram corresponding to the system initial value (0,0.0001,0.001). Figure 4 -a is 3d, Figure 4 -b is the xy plane, Figure 4 -c is the xz plane, Figure 4 -d is the yz plane.

[0069] According to the system state variables in the three-dimensional chaotic model, the filter parameters corresponding to each filter in the hybrid filter are respectively mapped to one of the system state variables;

[0070] Running the three-dimensional chaotic model and iterating multiple times, recording the system state variables corresponding to each time, and forming a corresponding chaotic sequence matrix; each row of chaotic sequence in the chaotic sequence matrix corresponds to a filter parameter group;

[0071] Each row of chaotic training in the formed chaotic sequence matrix is normalized to form individuals of the corresponding initial population, and the individuals are merged to form the corresponding initial population.

[0072] The improved attraction-repulsion optimization strategy is:

[0073] Step 1: Obtain historical original medical images and perform noise processing on them to generate corresponding original noisy medical images. In this example, the data used to implement the proposed algorithm and analyze the performance of image preprocessing was collected from brain MRI images publicly available on Kaggle. The dataset contains approximately 15,000 2D brain MRI slices. This dataset has dimensions of 496×248, a bit depth of 24, and a horizontal and vertical resolution of 100 dpi.

[0074] Step 2: According to the formed population, the filter parameter group corresponding to the individuals in the population is configured into the hybrid filter;

[0075] Step 3: The original noisy medical image is used as input data and input into each hybrid filter configured with a corresponding filter parameter group, the original noisy medical image is denoised, and the original denoised medical image corresponding to each population individual is output;

[0076] Step 4: Calculate the parameter fitness corresponding to each population individual based on the original denoised medical image corresponding to each population individual;

[0077] The parameter fitness corresponding to each population individual is calculated as follows:

[0078]

[0079] Where F is the parameter fitness of the individual in the population, σ, φ is the weight coefficient, MSE is the mean square error corresponding to the individual, PSNR is the peak signal-to-noise ratio corresponding to the individual, and SSIM is the structural similarity index between individuals; I o (i, j) is the pixel value of the original medical image without noise at the coordinate (i, j), I n (i, j) is the pixel value of the denoised medical image at coordinate (i, j), I, J are the number of rows and columns of the image, N is the maximum grayscale value of the image, x is the original medical image, y is the denoised medical image, u x is the pixel mean of the original medical image, u y is the pixel mean of the denoised medical image, is the pixel variance of the original medical image, is the pixel variance of the denoised medical image, σ xyis the covariance between the original medical image and the denoised medical image, c1 and c2 are stability constants.

[0080] Step 5: According to the parameter fitness corresponding to each population individual, the parameters are ranked in descending order. The individuals with the first preset percentage of parameter fitness ranking are selected as candidates for the global optimal solution, and the attraction mechanism is triggered to drive other individuals in the population to move towards the optimal solution candidate; the individuals with the second preset percentage of parameter fitness ranking are selected as candidates for inferior solutions, and the repulsion mechanism is triggered to drive other individuals in the population away from the inferior solution candidates;

[0081] The attraction mechanism in step 5 is:

[0082] Each individual x in the population i Subject to the global optimal solution candidate x best The corresponding update formula is driven by the attraction of

[0083]

[0084] Where c is the attraction intensity constant, m is the dynamic balance factor, which decreases linearly to 0.1 as the number of iterations t increases, a1 is a random vector from 0 to 1, r1 is a random number from 0 to 1, p1 is the attraction switching threshold, and b i The displacement corresponding to the attraction is attracted. In this embodiment, the present application discovered a chaotic system in the prior art. This three-dimensional chaotic system, by introducing a hybrid driving mechanism of cross-product terms and linear terms, combined with asymmetric parameter design (such as the constant term 6(a+b) in z'), while maintaining a relatively simple equation structure, exhibits strong nonlinear dynamic characteristics, significant parameter sensitivity and stable chaotic state generation capabilities, and can simulate complex phenomena such as bifurcation behavior and multistability. It is suitable for research and application in the fields of secure communication encryption, biomedical signal modeling and complex network synchronization control. Compared with this document, this scheme introduces dynamic singularities and positive feedback loops by nesting cosine fractional nonlinear terms, resulting in the appearance of multi-level self-similar structures in the bifurcation diagram, significantly improving the ergodicity of the solution space. Combined with Lyapunov exponent spectrum analysis, the new system exhibits double positive Lyapunov exponents, verifying its hyperchaotic characteristics, while the comparative document only has a single positive exponent. The hyperchaotic state expands the ergodicity of the solution space through high-dimensional attractors, effectively avoiding the population initialization aggregation phenomenon caused by the periodic window of the traditional chaotic mapping.

[0085] To demonstrate the advantages of our chaotic system, we calculated the Shannon entropy of each chaotic system under optimal parameters. We used the same parameters for the number of iterations and population size to ensure fairness in the experiment.

[0086] like Figure 5 and Figure 6 Based on the quantitative evaluation results of the two-group control experiment, the new chaotic system proposed in this study showed significant advantages in the Shannon Entropy (SE) index, specifically: Shannon Entropy improvement: Under the same initial conditions and parameter configuration, the Shannon Entropy value of the new system reached 8.999 bits, which was significantly improved compared with the comparison file 1 (benchmark chaotic system, SE = 4.928 bits), indicating that the uniformity of its phase space distribution and the randomness of information are significantly enhanced.

[0087] In order to verify the corresponding denoising effect under the same noise conditions, the following comparative experiments are conducted:

[0088] The corresponding index comparison table under the same speckle noise is shown in Table 1, and the corresponding experimental figure under the same speckle noise is shown in Figure 7 shown.

[0089] index PSNR MSE SSIM Comparison 1 7.92 100.09 0.7935 LY-ARO 29.53 76.00 0.8379

[0090] Table 1 Comparison of indicators under speckle noise

[0091] The corresponding index comparison table under the same salt and pepper noise is shown in Table 2, and the corresponding experimental figure under the same salt and pepper noise is shown in Figure 8 shown.

[0092] index PSNR MSE SSIM Comparison 1 28.75 69.45 0.8002 LY-ARO 31.92 52.49 0.8756

[0093] Table 2 Comparison of indicators under salt and pepper noise

[0094] Based on comparative experiments under the same speckle noise and salt and pepper noise conditions, this model demonstrates significant advantages in edge preservation performance. The above table and specific experimental figures clearly demonstrate the significant effect of LY-ARO.

[0095] In step 6, based on the current number of iterations, a local search strategy corresponding to that number of iterations is determined. Based on the determined local search strategy, each individual in the population is searched to form a new individual after optimization of each individual in the population. Step 2 is then re-executed until the iteration requirements are met, and the current global optimal solution is output as the optimal filter parameter set corresponding to the hybrid filter. The local search strategies in step 6 include Brownian motion strategy, trigonometric function strategy, and random jump strategy. In this embodiment, the random jump strategy is mainly used before the first preset number of iterations, with the Brownian motion strategy or trigonometric function strategy occasionally inserted to explore potential areas outside the attraction guidance direction.

[0096] When the number of iterations is between the first preset number and the second preset number, the Brownian motion strategy and the trigonometric function strategy are used as the main forces to start fine-tuning the parameters. When the number of iterations is greater than the second preset number, only the Brownian motion strategy is retained.

[0097] Among them, the update of the candidate solution corresponding to the Brownian motion strategy follows the normal distribution, but the standard deviation is dynamically adjusted to:

[0098] r B =u1⊙N(0,σ)

[0099]

[0100] u1 is a binary vector, randomly selecting the perturbation dimension; fr1 = 0.1: perturbation amplitude coefficient, later ((t→t max ))(σ) approaches 0 and turns to fine adjustment.

[0101] The trigonometric function strategy uses the periodic changes of sine and cosine functions to generate displacement. The formula is as follows:

[0102]

[0103] Where x w are other solutions selected by roulette, a2 is a random vector, r4 and r5 are random numbers, fr2 is the control amplitude, and u2 is a binary vector similar to u1.

[0104] The jumping formula corresponding to the random jumping strategy is:

[0105] r R =u3⊙(2a3-1)⊙(x max -x min )

[0106] Where, (a3): [0,1] random vector, (u3) is a binary vector that controls the jump dimension;

[0107] In terms of strategy selection, we rely on the probability switch p2 to select the Brownian motion or trigonometric function strategy, otherwise we use the random jump strategy:

[0108] In this embodiment, the last position update is:

[0109] x i (t) = x i (t-1)+n i +b i +r i

[0110] Compared with the existing technologies and BM3D, NLM, MDF, BF, MF, WF, and GF algorithms, this embodiment conducts comparative experiments on speckle noise and salt and pepper noise respectively:

[0111] Comparison of speckle noise images,Table 3 shows the comparison of denoising effects of LY-ARO and existing filters under the influence of speckle noise.,The experimental data includes 300 brain MRI images with speckle noise added, and the results are shown in, Figure 9 As shown in the figure, the proposed improved algorithm can effectively preserve image edges and details while suppressing speckle noise, demonstrating excellent visual effects. In contrast, existing filtering algorithms are relatively rough in contour processing, often resulting in over-smoothing of noisy areas, which in turn affects the preservation of edge structures.

[0112] Metrics LY-ARO BM3D NLM MDF BF MF WF GF MSE 118 127.22 927.22 319.5 1292.15 188.15 152.56 134.58 PSNR 27.6 26.510 18.67 23.170 17.01 25.31 26.23 26.500 SSIM 0.8586 0.7470 0.7877 0.7972 0.6414 0.7997 0.7792 0.8751

[0113] Table 3 Performance comparison of various filtering methods under speckle noise

[0114] Table 3 shows that our algorithm has advantages over BM3D, NLM, MDF, BF, MF, WF, and GF algorithms in removing image speckle noise. In contrast, existing filtering models perform poorly due to image quality degradation, complex processing, improper edge processing, ineffective minimization of intensity variations, and inadequate noise removal. Through data comparison, our algorithm achieved a 1.28 improvement in PSNR, a 0.1416 improvement in SSIM, and a 330 reduction in MSE. This demonstrates superior image quality, edge information, and structural similarity compared to other algorithms.

[0115] Table 4 shows the comparison of the denoising effects of LY-ARO and existing filters on the dataset under the influence of salt and pepper noise. According to Table 4, the algorithm proposed in this study shows certain advantages over BM3D, NLM, MDF, BF, MF, WF and GF algorithms in removing image salt and pepper noise. Through data comparison, the PSNR increased by 8.46 on average and the MSE decreased by 435 on average. The results of brain MRI images with speckle noise are shown in Figure 4. Figure 10 shown.

[0116]

[0117]

[0118] Table 4 Performance comparison of various filtering methods under salt and pepper noise

[0119] S2. Obtaining a current medical image to be processed at the current moment, and performing noise processing on the obtained current medical image to generate a corresponding current noisy image;

[0120] S3. The current noisy image is used as input data and input into the hybrid filter, and the corresponding denoised medical image is output.

[0121] The above is only an embodiment of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme is excessively described here. Ordinary technicians in the relevant field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the relevant field can improve and implement this scheme in combination with their own abilities under the enlightenment given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the relevant field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A medical image denoising method based on 3D chaos and attraction-repulsion algorithm, characterized by: The following steps are involved: S1. Based on a preset hybrid filter parameter group optimization strategy, determining the optimal parameter group of each filter corresponding to the hybrid filter used for denoising the medical image based on the historical original medical image, and configuring the optimal parameter group to the corresponding filter; the preset hybrid filter parameter group optimization strategy includes a 3D-LY chaotic mapping strategy for initializing the population corresponding to the hybrid filter parameter group and an improved attraction-repulsion optimization strategy for optimizing the initialized population; S2. Obtaining a current medical image to be processed at the current moment, and performing noise processing on the obtained current medical image to generate a corresponding current noisy image; S3. The current noisy image is used as input data and input into the hybrid filter, and the corresponding denoised medical image is output.

2. The medical image denoising method based on 3D chaos and attraction-repulsion algorithm according to claim 1, characterized in that: The 3D-LY chaotic mapping strategy is: According to the number of filter types in the hybrid filter, a corresponding three-dimensional chaotic model is constructed; The three-dimensional chaos model is: Where, is the system state variable in the corresponding three-dimensional chaos model, a and b are the parameters of the three-dimensional chaos model; According to the system state variables in the three-dimensional chaotic model, the filter parameters corresponding to each filter in the hybrid filter are respectively mapped to one of the system state variables; Running the three-dimensional chaotic model and iterating multiple times, recording the system state variables corresponding to each time, and forming a corresponding chaotic sequence matrix; each row of chaotic sequence in the chaotic sequence matrix corresponds to a filter parameter group; Each row of chaotic training in the formed chaotic sequence matrix is normalized to form individuals of the corresponding initial population, and the individuals are merged to form the corresponding initial population.

3. The medical image denoising method based on 3D chaos and attraction-repulsion algorithm according to claim 2, characterized in that: The improved attraction-repulsion optimization strategy is: Step 1: Obtain a historical original medical image and perform noise processing on the original medical image to generate a corresponding original noisy medical image; Step 2: According to the formed population, the filter parameter group corresponding to the individuals in the population is configured into the hybrid filter; Step 3: The original noisy medical image is used as input data and input into each hybrid filter configured with a corresponding filter parameter group, the original noisy medical image is denoised, and the original denoised medical image corresponding to each population individual is output; Step 4: Calculate the parameter fitness corresponding to each population individual based on the original denoised medical image corresponding to each population individual; Step 5: According to the parameter fitness corresponding to each population individual, the parameters are ranked in descending order. The individuals with the first preset percentage of parameter fitness ranking are selected as candidates for the global optimal solution, and the attraction mechanism is triggered to drive other individuals in the population to move towards the optimal solution candidate; the individuals with the second preset percentage of parameter fitness ranking are selected as candidates for inferior solutions, and the repulsion mechanism is triggered to drive other individuals in the population away from the inferior solution candidates; Step six, based on the current number of iterations, determine the local search strategy corresponding to the number of iterations, and based on the determined local search strategy, search each individual in the population to form a new individual after optimization of each individual in the population, and re-execute step two until the iteration requirements are met, and output the current global optimal solution as the optimal filter parameter group corresponding to the hybrid filter.

4. The medical image denoising method based on 3D chaos and attraction-repulsion algorithm according to claim 3, characterized in that: The parameter fitness corresponding to each population individual is calculated as follows: Where F is the parameter fitness of the individual in the population, σ, φ is the weight coefficient, MSE is the mean square error corresponding to the individual, PSNR is the peak signal-to-noise ratio corresponding to the individual, and SSIM is the structural similarity index between individuals; I o (i, j) is the pixel value of the original medical image without noise at the coordinate (i, j), I n (i, j) is the pixel value of the denoised medical image at coordinate (i, j), I, J are the number of rows and columns of the image, N is the maximum grayscale value of the image, x is the original medical image, y is the denoised medical image, u x is the pixel mean of the original medical image, u y is the pixel mean of the denoised medical image, is the pixel variance of the original medical image, is the pixel variance of the denoised medical image, σ xy is the covariance between the original medical image and the denoised medical image, c1 and c2 are stability constants.

5. The medical image denoising method based on 3D chaos and attraction-repulsion algorithm according to claim 4, characterized in that: The attraction mechanism in step 5 is: Each individual x in the population i Subject to the global optimal solution candidate x best The corresponding update formula is driven by the attraction of Where c is the attraction intensity constant, m is the dynamic balance factor, which decreases linearly to 0.1 as the number of iterations t increases, a1 is a random vector from 0 to 1, r1 is a random number from 0 to 1, p1 is the attraction switching threshold, and b i The attractive displacement corresponding to the attractive force.

6. The medical image denoising method based on 3D chaos and attraction-repulsion algorithm according to claim 5, characterized in that: The local search strategies in step six include Brownian motion strategy, trigonometric function strategy and random jump strategy.

7. A medical image denoising system based on 3D chaos and attraction-repulsion algorithm, characterized by: A medical image denoising method based on 3D chaos and attraction-repulsion algorithm according to any one of claims 1 to 6.

Citation Information

Patent Citations

  • WSN target coverage method and device based on chaos adaptive moth algorithm

    CN117939420A

  • Medical patient similarity clustering analysis method and system and application

    CN118919006A

  • Medical image denoising method based on 3D chaotic mapping dung beetle algorithm

    CN119477731A

  • Self-tuning method for control parameters of single-phase alternating-current input low-ripple adjustable direct-current stabilized power supply

    CN119891793A

  • Apparatus and method for using analog circuits to embody non-lipschitz mathematics and properties using attractor and repulsion modes

    US8266085B1