A medical image segmentation method and system based on improved deep learning model

Abstract

The present invention belongs to the field of image segmentation technology and discloses a medical image segmentation method and system based on an improved deep learning model. The method comprises the following steps: using an improved deep learning algorithm to construct an initial medical image segmentation model, and using a swarm intelligence optimization algorithm to optimize the initial medical image segmentation model to obtain an optimized medical image segmentation model; training the optimized medical image segmentation model based on a number of historical medical image data to obtain an optimal medical image segmentation model; collecting real-time medical image data, and inputting the real-time medical image data into the optimized medical image segmentation model to perform medical image segmentation and obtain real-time medical image segmentation data. The present invention solves the problems of insufficient segmentation accuracy, low training efficiency, and poor model generalization in the prior art.

Description

Technical Field

[0001] The present invention belongs to the technical field of image segmentation, and specifically relates to a medical image segmentation method and system based on an improved deep learning model. Background Art

[0002] Medical imaging technologies such as CT and MRI, as the most intuitive non-invasive methods for obtaining structural information about a patient's diseased area, have been widely used in modern clinical diagnostics. Their real-time and non-invasive nature makes them crucial for diagnosing, predicting, and following up on treatment for patients without obvious symptoms. During treatment, medical experts segment and annotate diseased tissues and organs based on acquired medical images, thereby assessing the progression of the disease and proposing targeted treatment plans. The accuracy of segmentation directly impacts the doctor's assessment of the condition and is a crucial step in the diagnostic process. Furthermore, with the continuous advancement of computer technology and medical imaging equipment, the volume of medical image data has increased dramatically, placing higher demands on the accuracy and efficiency of segmentation algorithms. Traditional medical image segmentation methods, which mostly rely on manual labor, suffer from low accuracy and inefficiency, limiting their clinical application.

[0003] With the advancement of artificial intelligence technology, a large number of artificial intelligence algorithms have been used in the field of medical image segmentation. Existing medical image segmentation technologies mainly include traditional image processing methods and deep learning-based methods, which have the following defects:

[0004] 1) Insufficient segmentation accuracy: Traditional image processing methods are prone to mis-segmentation when dealing with complex image structures, and existing deep learning methods also have difficulty maintaining high accuracy when dealing with noise and inhomogeneity.

[0005] 2) Low training efficiency: Deep learning model training requires a lot of computing resources and time, and the cost of obtaining labeled data is high.

[0006] 3) Poor model generalization ability: Existing deep learning models are often optimized for specific datasets and have difficulty maintaining good segmentation performance on different types of medical imaging data. Summary of the Invention

[0007] In order to solve the problems of insufficient segmentation accuracy, low training efficiency and poor model generalization ability in the existing technology, the purpose of the present invention is to provide a medical image segmentation method and system based on an improved deep learning model.

[0008] The technical solution adopted in the present invention is:

[0009] A medical image segmentation method based on an improved deep learning model includes the following steps:

[0010] Use the improved deep learning algorithm to build an initial medical image segmentation model, and use the swarm intelligence optimization algorithm to optimize the initial medical image segmentation model to obtain an optimized medical image segmentation model;

[0011] Based on a number of historical medical image data, the optimized medical image segmentation model is trained to obtain the optimal medical image segmentation model;

[0012] Real-time medical image data is collected and input into an optimal medical image segmentation model to perform medical image segmentation to obtain real-time medical image segmentation data.

[0013] Furthermore, the improved deep learning algorithm is the MSFCNN-Attention-cGAN-Elman algorithm.

[0014] Furthermore, the medical image segmentation model includes an input layer, a main segmentation network built based on the MSFCNN-Attention algorithm, an adversarial training network built based on the cGAN algorithm, a classification layer built based on the Elman algorithm, and an output layer;

[0015] The adversarial training network consists of a generator and a discriminator connected sequentially;

[0016] The main segmentation network consists of a multi-scale feature extraction layer based on the MSFCNN algorithm, an adaptive attention module based on the Attention mechanism, a residual block, a feature fusion layer, and a deconvolution layer.

[0017] The deconvolution layer, classification layer, and output layer of the main segmentation network are connected in sequence, the feature fusion layer is connected to the generator of the adversarial training network, and the discriminator is connected to the classification layer.

[0018] Furthermore, the swarm intelligence optimization algorithm is an ICPO optimization algorithm.

[0019] Furthermore, an improved deep learning algorithm is used to construct an initial medical image segmentation model, and a swarm intelligence optimization algorithm is used to optimize the initial medical image segmentation model to obtain an optimized medical image segmentation model, including the following steps:

[0020] Use the MSFCNN-Attention-cGAN-Elman algorithm to build an initial medical image segmentation model;

[0021] The initial network parameters of the medical image segmentation model are used as the optimization target of the ICPO optimization algorithm, and according to the optimization target, a search is performed to obtain the optimal initial network parameters;

[0022] According to the optimal initial network parameters, the initial medical image segmentation model is optimized to obtain an optimized medical image segmentation model.

[0023] Furthermore, the initial network parameters include initial main segmentation network parameters, initial adversarial training network parameters, and initial classification network parameters.

[0024] Furthermore, the initial main segmentation network parameters include the initial convolution kernel size, the initial number of convolution kernels, the initial step size, the initial activation function, the initial attention ratio, and the first initial neuron weight and threshold of the multi-scale feature extraction layer;

[0025] The initial adversarial training network parameters include the initial generator network structure parameters, the initial discriminator network structure parameters, the first initial learning rate, and the initial noise dimension of the adversarial training network;

[0026] The initial classification network parameters include the initial number of hidden layer neurons of the Elman network, the second initial learning rate, and the second initial neuron weights and thresholds.

[0027] Furthermore, the initial network parameters of the medical image segmentation model are used as the optimization target of the ICPO optimization algorithm, and a search is performed based on the optimization target to obtain the optimal initial network parameters, including the following steps:

[0028] According to the optimization goal, set the encoding vector, fitness function, ICPO population parameters and maximum number of iterations of the ICPO individual;

[0029] According to the coding vector of ICPO individuals and ICPO population parameters, the Circle chaotic mapping sequence is used for initialization to generate several initial ICPO individuals of the initial ICPO population;

[0030] A cyclic population reduction mechanism is introduced to limit the number of individuals in the ICPO population parameters and obtain the updated ICPO population parameters for the next iteration;

[0031] According to the fitness function, the initial fitness value of the initial ICPO individual in the initial ICPO population is calculated;

[0032] According to the initial fitness value and the updated ICPO population parameters, the first defense strategy, the second defense strategy, the third defense strategy, and the fourth defense strategy are used to update the initial ICPO population to obtain an updated ICPO population;

[0033] Use the dynamic reverse learning algorithm to perform dynamic reverse learning on the updated ICPO population to generate a dynamic reverse ICPO population;

[0034] According to the fitness function, the fitness values ​​of all ICPO individuals in the updated ICPO population and the dynamically reversed ICPO population are calculated, and the ICPO individual with the minimum fitness value is regarded as the optimal individual, and the optimal individual is retained;

[0035] If the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements, the optimal solution corresponding to the optimal individual is output to obtain the optimal initial network parameters.

[0036] Furthermore, the optimized medical image segmentation model is trained based on a number of historical medical image data to obtain the optimal medical image segmentation model, including the following steps:

[0037] Collecting a number of historical medical image data and performing preprocessing to obtain a number of preprocessed historical medical image data;

[0038] The pre-processed historical medical imaging data were divided into a model training set and a model test set in a ratio of 7:3;

[0039] According to the model training set, the optimized medical image segmentation model is trained to obtain a trained medical image segmentation model;

[0040] According to the model test set, the trained medical image segmentation model is tested to obtain the model test accuracy;

[0041] If the model test accuracy is greater than the accuracy threshold, or the number of training times reaches the threshold, the optimal medical image segmentation model is output; otherwise, the model training continues.

[0042] A medical image segmentation system based on an improved deep learning model is used to implement a medical image segmentation method. The system includes a model construction unit, a model training unit, and a medical image segmentation unit connected in sequence.

[0043] The beneficial effects of the present invention are:

[0044] The present invention discloses a medical image segmentation method and system based on an improved deep learning model, and proposes an improved deep learning model that combines multi-scale feature extraction, attention mechanism, cGAN network and Elman network for medical image segmentation. The model extracts image features from different scales, ensures fine recognition of complex structures, thereby improving the accuracy of segmentation, and realizes multi-scale feature fusion, which can better handle noise and unevenness problems in images, improve the robustness of segmentation, and enable the model to focus more on key areas in the image through the attention mechanism, reducing the interference of background noise and irrelevant information, further improving the accuracy of segmentation, and using the cGAN network to generate more Diverse and realistic training samples enhance the model's adaptability to different image features, thereby improving the model's generalization ability and improving the ability to segment complex scenes. By utilizing the structural characteristics of the Elman neural network, better information circulation and state memory are achieved during the model training process, which helps to accelerate the convergence of the model and reduce the time and computing resources required for training. The ICPO optimization algorithm is used to optimize the initial network parameters of the improved deep learning model to find the optimal model parameter configuration, so that the model can maintain good performance on different data sets, solve the defect of the model being sensitive to initial values, avoid the model falling into local optimality and being unable to jump out, and the model being premature, and improve the model training efficiency, stability, generalization ability and prediction accuracy.

[0045] Other beneficial effects of the present invention will be further described in the specific embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 This is a flowchart of the medical image segmentation method based on the improved deep learning model in the present invention.

[0047] Figure 2 This is a structural block diagram of the medical image segmentation system based on the improved deep learning model in the present invention. DETAILED DESCRIPTION

[0048] The present invention will be further explained below with reference to the accompanying drawings and specific embodiments.

[0049] Example 1:

[0050] like Figure 1 As shown, this embodiment provides a medical image segmentation method based on an improved deep learning model, comprising the following steps:

[0051] S1: Use the Multi-Scale Feature Convolutional Neural Network (MSFCNN)-Attention-Conditional Generative Adversarial Network (cGAN)-Elman algorithm to build an initial medical image segmentation model. Then use the Improved Crested Porcupine Optimizer (ICPO) optimization algorithm to optimize the initial medical image segmentation model to obtain an optimized medical image segmentation model. The steps include:

[0052] S1-1: Use the MSFCNN-Attention-cGAN-Elman algorithm to build an initial medical image segmentation model;

[0053] The medical image segmentation model includes an input layer, a main segmentation network built based on the MSFCNN-Attention algorithm, an adversarial training network built based on the cGAN algorithm, a classification layer built based on the Elman algorithm, and an output layer;

[0054] The adversarial training network consists of a generator and a discriminator connected sequentially;

[0055] The main segmentation network consists of a multi-scale feature extraction layer based on the MSFCNN algorithm, an adaptive attention module based on the Attention mechanism, a residual block, a feature fusion layer, and a deconvolution layer.

[0056] The input layer is used to receive input medical image data. The multi-scale feature extraction layer is used to extract multi-scale features using parallel convolutional layers with different convolution kernel sizes. The adaptive attention module applies the adaptive attention mechanism immediately after each convolutional layer or convolution block of the multi-scale feature extraction layer, which can help the model focus on more important features while suppressing unimportant features. This can filter out useful information at an early stage and reduce the computational burden of subsequent layers. Introducing residual blocks between the convolutional structures of the multi-scale feature extraction layer is the core idea of ​​the residual network structure, which helps solve the gradient disappearance problem in deep network training and improves the model's ability to extract deep features. The feature fusion layer fuses features of different scales through jump connections to obtain an intermediate feature map of multi-scale fusion features. The deconvolution layer is used to upsample the intermediate feature map and restore the image size.

[0057] The deconvolution layer, classification layer, and output layer of the main segmentation network are connected in sequence, the feature fusion layer is connected to the generator of the adversarial training network, and the discriminator is connected to the classification layer;

[0058] The adversarial training network exists as an independent training stage and is set up in parallel with the main segmentation network. The generator receives the intermediate feature maps from the main network as input and attempts to generate segmentation results. The discriminator compares the generated segmentation results with the true labels. The main segmentation network is responsible for feature extraction and preliminary segmentation, while the adversarial training network further improves the segmentation quality through adversarial training.

[0059] S1-2: The initial network parameters of the medical image segmentation model are used as the optimization target of the ICPO optimization algorithm. Based on the optimization target, a search is performed to obtain the optimal initial network parameters, including the following steps:

[0060] S1-2-1: According to the optimization goal, set the encoding vector, fitness function, ICPO population parameters and maximum number of iterations of the ICPO individual;

[0061] The formula for fitness value is:

[0062]

[0063] Where, is the fitness function; is the mean square error function; For the n The true output value of each pixel; For the n The ideal output value of each pixel; n is the pixel indicator; N is the total number of pixels;

[0064] S1-2-2: Based on the coding vector of the ICPO individual and the ICPO population parameters, the Circle chaotic mapping sequence is used for initialization to generate several initial ICPO individuals of the initial ICPO population;

[0065] The formula is:

[0066]

[0067] Where, is the initial ICPO individual of the Circle chaos map, i.e. the initial solution; is the randomly generated initial ICPO individual; It is the individual indicator of ICPO;

[0068] S1-2-3: Introduce a cyclic population reduction mechanism to limit the number of individuals in the ICPO population parameters and obtain the updated ICPO population parameters for the next iteration;

[0069] The formula is:

[0070]

[0071] Where, For the The number of individuals in the ICPO population parameter of the iteration; For the The number of individuals in the ICPO population parameter of the iteration; is the minimum number of individuals in the ICPO population parameter; Evaluate arguments for functions; Evaluate loop parameters for a function; Evaluate loop parameters for the maximum function; t is the indicator of the number of iterations;

[0072] S1-2-4: Calculate the initial fitness value of the initial ICPO individuals in the initial ICPO population according to the fitness function;

[0073] S1-2-5: Based on the initial fitness value and the updated ICPO population parameters, the first defense strategy, the second defense strategy, the third defense strategy, and the fourth defense strategy are used to update the initial ICPO population to obtain an updated ICPO population;

[0074] The formula for the first defense strategy is:

[0075]

[0076] Where, For the updated ICPO individuals within the first defense range; The initial ICPO individual within the first defense range; is a random number based on normal distribution; is a random value in the interval [0,1]; It is the optimal solution within the first defense range; is the vector generated between the true optimal solution within the first defense range and the optimal solution randomly selected from the ICPO population; It is the individual indicator of ICPO; is the iteration indicator;

[0077] The formula for the second defense strategy is:

[0078]

[0079] Where, For the updated ICPO individuals within the second defense range; The initial ICPO individual within the second defense range; is the search upper limit vector of the second defense range; is a random value in the interval [0,1]; Respectively Initial ICPO individuals; All are [1, ] two random integers between; is the vector generated between the true optimal solution within the second defense range and the optimal solution randomly selected from the ICPO population;

[0080] The formula for the third defense strategy is:

[0081]

[0082] Where, For the updated ICPO individuals within the third defense range; The initial ICPO individual within the third defense range; is the search upper limit vector of the third defense range; Respectively Initial ICPO individuals; is [1, ] a random integer between ; The odor diffusion factor defined for the fitness function; It is a defense factor; Control parameters for search direction;

[0083] The formula for the fourth defense strategy is:

[0084]

[0085] Where, For the updated ICPO individuals within the fourth defense range; For the initial ICPO individuals within the fourth defense range; It is the optimal solution within the fourth defense range; All are random values ​​in the interval [0,1]; It is a defense factor; Control parameters for search direction; is the average force affecting the search direction; is the convergence speed factor;

[0086] S1-2-6: Use the dynamic reverse learning algorithm to perform dynamic reverse learning on the updated ICPO population to generate a dynamic reverse ICPO population;

[0087] The formula is:

[0088]

[0089] Where, It is a dynamically reversed ICPO individual; is the decreasing inertia coefficient; are the maximum and minimum values ​​of the vector space respectively; For the updated ICPO individual;

[0090] S1-2-7: According to the fitness function, calculate the fitness values ​​of all ICPO individuals in the updated ICPO population and the dynamically reversed ICPO population, take the ICPO individual with the minimum fitness value as the optimal individual, and retain the optimal individual;

[0091] S1-2-8: If the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements, the optimal solution corresponding to the optimal individual is output to obtain the optimal initial network parameters;

[0092] The initial network parameters include the initial main segmentation network parameters, the initial adversarial training network parameters, and the initial classification network parameters;

[0093] The initial main segmentation network parameters include the initial convolution kernel size, initial number of convolution kernels, initial step size, initial activation function, initial attention ratio, and the first initial neuron weight and threshold of the multi-scale feature extraction layer;

[0094] The initial adversarial training network parameters include the initial generator network structure parameters, the initial discriminator network structure parameters, the first initial learning rate, and the initial noise dimension of the adversarial training network;

[0095] The initial classification network parameters include the initial number of hidden layer neurons of the Elman network, the second initial learning rate, and the second initial neuron weights and thresholds;

[0096] S1-3: Optimizing the initial medical image segmentation model according to the optimal initial network parameters to obtain an optimized medical image segmentation model;

[0097] This solves the problem of the model being sensitive to initial values, prevents the model from falling into local optimality and being unable to escape, and prevents the model from becoming premature. It also improves the model's training efficiency, stability, generalization ability, and prediction accuracy.

[0098] S2: Based on a number of historical medical image data, the optimized medical image segmentation model is trained to obtain the optimal medical image segmentation model, including the following steps:

[0099] S2-1: Collecting a number of historical medical image data and performing preprocessing to obtain a number of preprocessed historical medical image data;

[0100] Preprocessing includes data screening (removing duplicate or erroneous data), denoising (applying filtering techniques such as Gaussian filtering and median filtering to remove random noise in image data), image enhancement (enhancing the contrast of image data to improve visual effects and make important anatomical structures clearer), standardization (standardizing image intensity values ​​to have a uniform mean and standard deviation to help the generalization ability of the model), cropping and scaling (cropping images to remove irrelevant areas and focus on areas of interest), labeling and annotation (adding labels and annotations to image pixels for supervised learning model training), and data augmentation (applying transformations such as rotation, flipping, scaling, and shearing to increase data diversity and improve the generalization ability of the model).

[0101] S2-2: Divide the pre-processed historical medical imaging data into a model training set and a model test set in a ratio of 7:3;

[0102] S2-3: Training the optimized medical image segmentation model according to the model training set to obtain a trained medical image segmentation model;

[0103] S2-4: Test the trained medical image segmentation model based on the model test set to obtain the model test accuracy;

[0104] S2-5: If the model test accuracy is greater than the accuracy threshold, or the number of training times reaches the number threshold, the optimal medical image segmentation model is output; otherwise, the model training continues;

[0105] S3: Collect real-time medical image data, and input the real-time medical image data into the optimal medical image segmentation model to perform medical image segmentation to obtain real-time medical image segmentation data.

[0106] Example 2:

[0107] like Figure 2 As shown, this embodiment provides a medical image segmentation system based on an improved deep learning model, which is used to implement a medical image segmentation method. The system includes a model building unit, a model training unit, and a medical image segmentation unit connected in sequence;

[0108] A model building unit, configured to use an improved deep learning algorithm to build an initial medical image segmentation model, and to use a swarm intelligence optimization algorithm to optimize the initial medical image segmentation model to obtain an optimized medical image segmentation model;

[0109] A model training unit, used to train the optimized medical image segmentation model based on a number of historical medical image data to obtain the optimal medical image segmentation model;

[0110] The medical image segmentation unit is used to collect real-time medical image data and input the real-time medical image data into the optimal medical image segmentation model to perform medical image segmentation to obtain real-time medical image segmentation data.

[0111] The present invention discloses a medical image segmentation method and system based on an improved deep learning model, and proposes an improved deep learning model that combines multi-scale feature extraction, attention mechanism, cGAN network and Elman network for medical image segmentation. The model extracts image features from different scales, ensures fine recognition of complex structures, thereby improving the accuracy of segmentation, and realizes multi-scale feature fusion, which can better handle noise and unevenness problems in images, improve the robustness of segmentation, and enable the model to focus more on key areas in the image through the attention mechanism, reducing the interference of background noise and irrelevant information, further improving the accuracy of segmentation, and using the cGAN network to generate more Diverse and realistic training samples enhance the model's adaptability to different image features, thereby improving the model's generalization ability and improving the ability to segment complex scenes. By utilizing the structural characteristics of the Elman neural network, better information circulation and state memory are achieved during the model training process, which helps to accelerate the convergence of the model and reduce the time and computing resources required for training. The ICPO optimization algorithm is used to optimize the initial network parameters of the improved deep learning model to find the optimal model parameter configuration, so that the model can maintain good performance on different data sets, solve the defect of the model being sensitive to initial values, avoid the model falling into local optimality and being unable to jump out, and the model being premature, and improve the model training efficiency, stability, generalization ability and prediction accuracy.

[0112] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, apparatuses, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0113] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0114] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0115] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0116] Those skilled in the art will understand that all or part of the steps in implementing the above facts and methods can be completed by instructing relevant hardware through a program, and the program involved or the program can be stored in a computer-readable storage medium. When the program is executed, it includes the following steps: the corresponding method steps are then brought out, and the storage medium can be ROM / RAM, a disk, an optical disk, etc.

[0117] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A medical image segmentation method based on an improved deep learning model, characterized by: The steps include: Use the improved deep learning algorithm to build an initial medical image segmentation model, and use the swarm intelligence optimization algorithm to optimize the initial medical image segmentation model to obtain an optimized medical image segmentation model; The improved deep learning algorithm is the MSFCNN-Attention-cGAN-Elman algorithm; The medical image segmentation model includes an input layer, a main segmentation network constructed based on the MSFCNN-Attention algorithm, an adversarial training network constructed based on the cGAN algorithm, a classification layer constructed based on the Elman algorithm, and an output layer; The adversarial training network includes a generator and a discriminator connected in sequence; The main segmentation network includes a multi-scale feature extraction layer based on the MSFCNN algorithm, an adaptive attention module based on the Attention mechanism, a residual block, a feature fusion layer, and a deconvolution layer, which are connected in sequence; The deconvolution layer, classification layer, and output layer of the main segmentation network are connected in sequence, the feature fusion layer is connected to the generator of the adversarial training network, and the discriminator is connected to the classification layer; The swarm intelligence optimization algorithm is the ICPO optimization algorithm; An improved deep learning algorithm is used to construct an initial medical image segmentation model, and a swarm intelligence optimization algorithm is used to optimize the initial medical image segmentation model to obtain an optimized medical image segmentation model, including the following steps: Use the MSFCNN-Attention-cGAN-Elman algorithm to build an initial medical image segmentation model; The initial network parameters of the medical image segmentation model are used as the optimization target of the ICPO optimization algorithm. Based on the optimization target, a search is performed to obtain the optimal initial network parameters, including the following steps: According to the optimization goal, set the encoding vector, fitness function, ICPO population parameters and maximum number of iterations of the ICPO individual; According to the coding vector of ICPO individuals and ICPO population parameters, the Circle chaotic mapping sequence is used for initialization to generate several initial ICPO individuals of the initial ICPO population; A cyclic population reduction mechanism is introduced to limit the number of individuals in the ICPO population parameters and obtain the updated ICPO population parameters for the next iteration; According to the fitness function, the initial fitness value of the initial ICPO individual in the initial ICPO population is calculated; According to the initial fitness value and the updated ICPO population parameters, the first defense strategy, the second defense strategy, the third defense strategy, and the fourth defense strategy are used to update the initial ICPO population to obtain an updated ICPO population; Use the dynamic reverse learning algorithm to perform dynamic reverse learning on the updated ICPO population to generate a dynamic reverse ICPO population; According to the fitness function, the fitness values ​​of all ICPO individuals in the updated ICPO population and the dynamically reversed ICPO population are calculated, and the ICPO individual with the minimum fitness value is regarded as the optimal individual, and the optimal individual is retained; If the number of iterations of iterative optimization reaches the maximum number of iterations or the fitness value of the optimal individual meets the requirements, the optimal solution corresponding to the optimal individual is output to obtain the optimal initial network parameters; According to the optimal initial network parameters, the initial medical image segmentation model is optimized to obtain an optimized medical image segmentation model; Based on a number of historical medical image data, the optimized medical image segmentation model is trained to obtain the optimal medical image segmentation model; Real-time medical image data is collected and input into an optimal medical image segmentation model to perform medical image segmentation to obtain real-time medical image segmentation data.

2. The medical image segmentation method based on the improved deep learning model according to claim 1, characterized in that: The initial network parameters include initial main segmentation network parameters, initial adversarial training network parameters and initial classification network parameters.

3. The medical image segmentation method based on the improved deep learning model according to claim 2, characterized in that: The initial main segmentation network parameters include the initial convolution kernel size, the initial number of convolution kernels, the initial step size, the initial activation function, the initial attention ratio, and the first initial neuron weight and threshold of the multi-scale feature extraction layer; The initial adversarial training network parameters include the initial generator network structure parameters, the initial discriminator network structure parameters, the first initial learning rate and the initial noise dimension of the adversarial training network; The initial classification network parameters include the initial hidden layer neuron number of the Elman network, the second initial learning rate, and the second initial neuron weight and threshold.

4. The medical image segmentation method based on an improved deep learning model according to claim 1, characterized in that: Based on a number of historical medical image data, the optimized medical image segmentation model is trained to obtain the optimal medical image segmentation model, including the following steps: Collecting a number of historical medical image data and performing preprocessing to obtain a number of preprocessed historical medical image data; The pre-processed historical medical imaging data were divided into a model training set and a model test set in a ratio of 7:3; According to the model training set, the optimized medical image segmentation model is trained to obtain a trained medical image segmentation model; According to the model test set, the trained medical image segmentation model is tested to obtain the model test accuracy; If the model test accuracy is greater than the accuracy threshold, or the number of training times reaches the threshold, the optimal medical image segmentation model is output; otherwise, the model training continues.

5. A medical image segmentation system based on an improved deep learning model, used to implement the medical image segmentation method according to any one of claims 1 to 4, characterized in that: The system includes a model building unit, a model training unit and a medical image segmentation unit which are connected in sequence.

Images