Ischemic stroke image segmentation method based on wavelet pooling neural network

Abstract

The invention belongs to the technical field of medical imaging, and particularly relates to an ischemic stroke image segmentation method based on a wavelet pooling neural network, and the method comprises the following steps: firstly, employing a U-Net convolutional neural network as a basic network architecture, and adding a convolution residual module and a convolution residual in each convolution block of a coding path; inputting the preprocessed feature map into a convolution block of a coding path; then obtaining a first preliminary effective feature layer, and carrying out lossless sampling through a wavelet pooling module; a channel attention SE module is connected behind the wavelet pooling layer, and channel feature redirection is carried out through extrusion, excitation and re-calibration; after each down-sampling operation, inputting the feature map into a pyramid expansion coordinate attention module, and realizing multi-scale feature fusion through cavity convolution and coordinate attention; and then in a decoding path, after each up-sampling operation, inputting the feature map into a space attention module, and finally outputting a segmentation result. The segmentation result output by the method is more accurate.

Description

Technical Field

[0001] The present invention belongs to the technical field of medical imaging, and in particular relates to an ischemic stroke image segmentation method based on a wavelet pooling neural network. Background Art

[0002] Ischemic stroke is a highly dangerous brain disease that causes disability or death in adults. In the face of this disease, rapid and accurate diagnosis and treatment methods are particularly critical to reduce brain damage. In the diagnosis process, accurately segmenting ischemic lesions from stroke medical images is the key to timely diagnosis and treatment. Doctors need to assess the condition and formulate treatment plans based on the morphology of the segmented lesions. At present, the traditional method is for doctors to manually segment lesions, but it is time-consuming and labor-intensive, and is prone to missed diagnosis and misdiagnosis due to personal subjective factors, which will delay the best time for treatment. In view of this, the development of a neural network image segmentation method that can quickly and automatically complete accurate lesion segmentation is of great significance for improving the efficiency and accuracy of diagnosis and treatment, and assisting doctors to implement timely treatment within the golden treatment time. Summary of the invention

[0003] The purpose of the present invention is to provide an ischemic stroke image segmentation method based on a wavelet pooling neural network. Aiming at the problems of downsampling information loss and insufficient multi-scale capture capability in ischemic stroke segmentation in current deep learning methods, a ischemic stroke image segmentation method based on a wavelet pooling neural network is proposed. The method achieves lossless wavelet decomposition to retain image details by designing a new downsampling module and a multi-scale feature fusion module, and achieves fine edge segmentation of large lesions, correct regional positioning of small lesions, and correct detection and differentiation of multi-region lesions.

[0004] The technical solution adopted by the present invention is as follows:

[0005] A method for ischemic stroke image segmentation based on wavelet pooling neural network includes the following steps:

[0006] Step 1: Using the U-Net convolutional neural network as the basic network architecture, add the convolutional residual module Identity Block and the convolutional residual block Conv Block to each convolutional block in the encoding path;

[0007] Step 2: Input the preprocessed 256×256×4 feature map into the convolution block of the encoding path. After the feature map passes through the first double-layer convolution residual block Conv Block, a 256×256×64 feature map with inconsistent dimensions is obtained; then pass through the convolution residual module Identity Block once more to obtain the preliminary effective feature map of the first layer, at which time the dimensions before and after the convolution are consistent;

[0008] Step 3: Increase the number of feature map channels to 64 and obtain the first preliminary effective feature layer, which is then losslessly sampled through the wavelet pooling module;

[0009] Step 4: Connect the channel attention SE module after the wavelet pooling layer to redirect channel features through squeezing, excitation and recalibration;

[0010] Step 5: After each downsampling operation, the feature map is input into the pyramid expansion coordinate attention module, and multi-scale feature fusion is achieved through dilated convolution and coordinate attention to improve the network's receptive field and feature expression ability;

[0011] Step 6: In the decoding path, the feature map is input to the spatial attention module after each upsampling operation;

[0012] Step 7: A full-scale skip connection module is introduced between each encoder layer and decoder layer. The full-scale skip connection module improves the efficiency of feature transmission and reuse by fusing features of different scales, so as to capture fine-grained and coarse-grained semantics at all scales and obtain a more accurate segmentation result.

[0013] Preferably, in step 1, the residual module algorithm is described as follows: the residual module consists of two consecutive convolutional layers and a skip connection; the input feature passes through the first convolutional layer, then passes through the ReLU activation function, and the output result is passed to the second convolutional layer; the output of the second convolutional layer is added to the input feature to obtain the output of the residual module, which can avoid the gradient vanishing problem and improve the training efficiency and performance of the model.

[0014] Preferably, the step 3 specifically includes the following steps:

[0015] Step 301: In the wavelet pooling module, the discrete wavelet decomposition process is described as follows: the input image is first subjected to a one-dimensional wavelet transform in the horizontal direction, that is, it is simultaneously subjected to high-pass and low-pass filters to extract high- and low-frequency information in the horizontal direction; the high- and low-frequency components obtained in the first step are subjected to a one-dimensional wavelet transform in the vertical direction respectively, that is, four image components after two-dimensional wavelet decomposition are obtained;

[0016] Step 302: Input the four half-resolution image components obtained by two-dimensional discrete wavelet transform into the pooling layer for processing; while realizing the role of the pooling layer in expanding the receptive field, it also makes up for the information loss problem caused by traditional downsampling, so that the network can use the high-frequency components in the image, such as edges, contours and other areas with large changes. The resolution of the sub-image obtained by decomposition is halved, and the number of channels of the feature map becomes four times the original, that is, by allocating the information encoding of the spatial dimension to the channel dimension, the retention of all information is also guaranteed;

[0017] Step 303: The sub-images obtained by the two down-sampling methods of maximum pooling and discrete wavelet decomposition are stacked in the channel dimension and used together as the input of the next layer of network.

[0018] Preferably, in step 301, the four image components are: the component HH filtered by horizontal high frequency and vertical high frequency represents the details of the diagonal direction of the image, the component HL filtered by horizontal high frequency and vertical low frequency represents the horizontal edge information of the image, the component LH filtered by horizontal low frequency and vertical high frequency represents the vertical edge information of the image, and the component LL filtered by horizontal low frequency and vertical low frequency represents the approximate information of the image.

[0019] Preferably, step 4 comprises the following steps:

[0020] Step 401: Given a feature map x with an input size of H×W×C, where H is the height of a single feature map, W is the width, and C is the number of channels; the squeezing operation compresses the input feature map along the spatial length and width dimensions, that is, performs a global average pooling operation, compresses each two-dimensional feature map into a real number and arranges them along the channel direction, that is, obtains a global angle feature vector 1×1×C with a spatial direction, which represents the description of each feature map on the channel, and its formula for the channel is as follows:

[0021]

[0022] Among them, x c represents the cth channel of the input feature x; z c is the global average value of the cth channel;

[0023] Step 402: In the excitation operation, the 1×1×C global feature vector obtained by the squeezing operation is first passed through a fully connected layer with a small number of neurons for inter-channel information interaction learning, and the dimension of the channel is reduced to C / r to reduce the computational complexity and the number of parameters; then it is activated by a ReLU layer; then it is further passed through a fully connected layer with a length of C for channel information fusion and weight learning, and activated by a sigmoid function to reduce the weight range to between 0 and 1. The size of the weight represents the importance of the corresponding channel feature map; the excitation formula is as follows:

[0024] s=F ex (z,W)=σ(g(z,W))=σ(W2δ(W1z)),

[0025] Where W1 is a fully connected layer with dimension C / r, r is the drop ratio, W2 is a fully connected layer with dimension C, δ is the ReLU activation function, σ is the sigmoid activation function; s is the obtained attention weight;

[0026] Step 403: In the channel feature recalibration operation, the attention weight s obtained through the fully connected layer and nonlinear learning is multiplied by the feature map x before the squeeze operation. The formula is:

[0027]

[0028] In the formula, This is the final feature map with channel attention applied.

[0029] Preferably, step 5 comprises the following steps:

[0030] Step 501: performing a plurality of convolution operations with different dilation rates on the preliminary effective feature map, where the dilation rates selected by the module are 1, 2, 3 and 4, respectively, to obtain a multi-scale feature map;

[0031] Step 502: The multi-scale feature maps obtained by the four types of dilated convolutions are concatenated according to the channel dimension, and then a 1×1 convolution is performed to interactively fuse the multi-scale information of each channel;

[0032] Step 503: The above feature maps are subjected to coordinate attention again to redistribute the attention of channel information and spatial information, and finally an effective feature layer with multi-scale robustness and coordinate sensitivity is obtained.

[0033] Preferably, in step 6: first, perform global feature extraction on the input feature map of size H×W×C in the channel dimension, that is, average and extreme values ​​on all channels corresponding to each feature point to obtain two H×W×1 feature maps; stack the two feature maps representing global information in the channel direction, and then reduce the number of channels to 1 through a 7×7 convolution operation, and then use the sigmoid function for activation to transform the value of the feature map to the range of 0 to 1, that is, obtain the attention weight map in the spatial direction; its formula is:

[0034]

[0035] Among them, f 7×7 represents a convolution kernel of size 7×7, σ represents the sigmoid activation function; Represents the two-dimensional feature map obtained by taking the average value of the channel dimension, Represents the two-dimensional feature map obtained by taking the maximum value of the channel dimension;

[0036] By performing matrix multiplication of the spatial attention weight map and the input feature map, we can obtain the output feature map recalibrated in the spatial direction; then a Conv Block is used to deepen the feature extraction, and then the above process is repeated to obtain the final effective feature layer, and a layer of convolution and activation is performed to obtain the preliminary segmentation result.

[0037] Preferably, in step 7, the resolution of the feature maps of different scales in the encoding path is first unified; the feature maps with more underlying information are and After 4 times non-overlapping maximum pooling downsampling and 2 times downsampling respectively, the feature maps after pooling are further extracted by 3×3 convolution; the feature maps of the same scale are After only 3×3 convolution, the decoding path feature map with small resolution but relatively high-level semantic information is transformed into and The images are upsampled by four times and two times through bilinear interpolation, and each of them is also convolved by 3×3. The five scale feature maps with unified resolution are stacked in the channel dimension and uniformly passed through a 3×3 convolution layer for information fusion and channel interaction, that is, the detailed information such as the target boundary contained in the bottom feature map is aggregated with the target category and positioning information contained in the high-level feature map.

[0038] In the encoding stage, steps 1 to 5 are repeated until five preliminary valid feature layers with resolutions of 256×256, 128×128, 64×64, 32×32 and 16×16 are obtained; in the decoding stage, step 6 is repeated, and step 7 runs through all steps to obtain the final valid feature layer, and the final segmentation result is obtained after a layer of convolution and activation.

[0039] The technical effects achieved by the present invention are:

[0040] In the present invention, in view of the problem that downsampling may cause information loss, a method of lossless two-dimensional discrete wavelet pooling of feature maps is adopted, and the feature maps are reduced in dimension in the spatial dimension, thereby achieving the effect of increasing the network receptive field of the traditional pooling layer, while retaining all feature map information by increasing the number of feature map channels. In addition, in view of the problem that stroke images have individual differences according to different patients, resulting in different lesion locations, shapes, and sizes, the present invention effectively improves the adaptability of the network to lesion inputs with different sizes, and can make full use of the context information between pixels from a global perspective to make more accurate judgments on the segmentation edges, taking into account the segmentation of lesions of different sizes, and can distinguish lesions well for inputs with multiple lesion areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is an overall flow chart of an ischemic stroke image segmentation method based on wavelet pooling neural network proposed in the present invention.

[0042] Figure 2The invention provides an image segmentation network model for an ischemic stroke image segmentation method based on a wavelet pooling neural network, wherein the input of the model is a source image and the output is a segmentation result.

[0043] Figure 3 This is an example of comparing the stroke lesion area segmentation effect diagram of an ischemic stroke image segmentation method based on wavelet pooling neural network proposed in the present invention with the segmentation results of other segmentation algorithms. DETAILED DESCRIPTION

[0044] In order to make the purpose and advantages of the present invention more clearly understood, the present invention is specifically described below in conjunction with embodiments. It should be understood that the following text is only used to describe one or several specific embodiments of the present invention, and does not strictly limit the scope of protection of the specific claims of the present invention.

[0045] like Figure 1 As shown, a method for ischemic stroke image segmentation based on wavelet pooling neural network includes the following steps:

[0046] Step 1: Using the U-Net convolutional neural network as the basic network architecture, add the convolution residual module Identity Block and the convolution residual block Conv Block to each convolution block in the encoding path; the residual module algorithm is described as follows: the residual module consists of two consecutive convolutional layers and a jump connection; the input feature passes through the first convolutional layer, then passes through the ReLU activation function, and the output result is passed to the second convolutional layer; the output of the second convolutional layer is added to the input feature to obtain the output of the residual module, which can avoid the gradient vanishing problem and improve the training efficiency and performance of the model.

[0047] Step 2: Input the preprocessed 256×256×4 feature map into the convolution block of the encoding path. After the feature map passes through the first double-layer convolution residual block Conv Block, a 256×256×64 feature map with inconsistent dimensions is obtained; then pass through the convolution residual module Identity Block once more; obtain the preliminary effective feature map of the first layer, at which time the dimensions before and after the convolution are consistent; Identity Block is an identity mapping module used when the input and output dimensions are the same, and Conv Block is a convolution residual module used when the input and output dimensions are different;

[0048] In this embodiment, each convolution block of the encoding path: In the UNet architecture, the encoder part is usually composed of multiple convolution blocks. Each convolution block usually contains two consecutive 3x3 convolution layers, each followed by a ReLU activation function and a 2x2 maximum pooling layer for downsampling. The main purpose of the convolution block is to extract features and reduce the spatial dimension so that the model can capture different levels of abstraction of the input image.

[0049] Residual Module: The residual module is designed to solve the gradient vanishing / exploding problem in deep networks.

[0050] Convolutional Residual Module Identity Block: Identity Block is a special residual module in which the size and depth of the input feature map are the same as the output feature map, so the input can be directly added to the output. This means that if the data dimension on the main path (i.e. the path through the convolutional layer) does not change, then the identity mapping can be used to simply add the input to the output to form a residual connection.

[0051] Convolutional residual block Conv Block: Conv Block is also a type of residual module, but it involves a change in the size or depth of the input feature map. In this case, the input cannot be directly added to the output because the dimensions do not match. Therefore, Conv Block contains an additional convolutional layer to adjust the dimension of the input to match the output on the main path, and then perform the addition operation.

[0052] In this embodiment, when the input and output dimensions of the residual block are the same, the input and the residual map are directly added at the output end, that is, the identity mapping module Identity Block used when the input and output dimensions are the same is used; when the input and output dimensions are different, the input needs to be dimensionally adjusted through a 1×1 convolution to make the two dimensions the same, and then added to the residual map at the output end, that is, the convolution residual module Conv Block used when the input and output dimensions are different is used; the above two residual modules are used in the network of the present invention to deepen the network depth and feature extraction level.

[0053] Step 3: Increase the number of feature map channels to 64 and obtain the first preliminary effective feature layer, and then perform lossless sampling through the wavelet pooling module; specifically, the following steps are included:

[0054] Step 301: In the wavelet pooling module, the discrete wavelet decomposition process is described as follows: the input image is first subjected to a one-dimensional wavelet transform in the horizontal direction, that is, it is simultaneously subjected to high-pass and low-pass filters to extract high- and low-frequency information in the horizontal direction; the high- and low-frequency components obtained in the first step are subjected to a one-dimensional wavelet transform in the vertical direction respectively, that is, four image components after two-dimensional wavelet decomposition are obtained; the four image components are: the component HH subjected to horizontal high-frequency and vertical high-frequency filtering represents the details in the diagonal direction of the image, the component HL subjected to horizontal high-frequency and vertical low-frequency filtering represents the horizontal edge information of the image, the component LH subjected to horizontal low-frequency and vertical high-frequency filtering represents the vertical edge information of the image, and the component LL subjected to horizontal low-frequency and vertical low-frequency filtering represents the approximate information of the image.

[0055] Step 302: Input the four half-resolution image components obtained by two-dimensional discrete wavelet transform into the pooling layer for processing; while realizing the role of the pooling layer in expanding the receptive field, it also makes up for the information loss problem caused by traditional downsampling, so that the network can use the high-frequency components in the image, such as edges, contours and other areas with large changes. The resolution of the sub-image obtained by decomposition is halved, and the number of channels of the feature map becomes four times the original, that is, by allocating the information encoding of the spatial dimension to the channel dimension, the retention of all information is also guaranteed;

[0056] Step 303: stack the sub-images obtained by the two downsampling methods of maximum pooling and discrete wavelet decomposition in the channel dimension and use them together as the input of the next layer of network;

[0057] Step 4: Connect the channel attention SE module after the wavelet pooling layer to redirect channel features through squeezing, excitation and recalibration;

[0058] Step 401: Given a feature map x with an input size of H×W×C, where H is the height of a single feature map, W is the width, and C is the number of channels; the squeezing operation compresses the input feature map along the spatial length and width dimensions, that is, performs a global average pooling operation, compresses each two-dimensional feature map into a real number and arranges them along the channel direction, that is, obtains a global angle feature vector 1×1×C with a spatial direction, which represents the description of each feature map on the channel, and its formula for the channel is as follows:

[0059]

[0060] Among them, x c represents the cth channel of the input feature x; z c is the global average value of the cth channel;

[0061] Step 402: In the excitation operation, the 1×1×C global feature vector obtained by the squeezing operation is first passed through a fully connected layer with a small number of neurons for inter-channel information interaction learning, and the dimension of the channel is reduced to C / r to reduce the computational complexity and the number of parameters; then it is activated by a ReLU layer; then it is further passed through a fully connected layer with a length of C for channel information fusion and weight learning, and activated by a sigmoid function to reduce the weight range to between 0 and 1. The size of the weight represents the importance of the corresponding channel feature map; the excitation formula is as follows:

[0062] s=F ex (z,W)=σ(g(z,W))=σ(W2δ(W1z)),

[0063] Where W1 is a fully connected layer with dimension C / r, r is the drop ratio, W2 is a fully connected layer with dimension C, δ is the ReLU activation function, σ is the sigmoid activation function; s is the obtained attention weight;

[0064] Step 403: In the channel feature recalibration operation, the attention weight s obtained through the fully connected layer and nonlinear learning is multiplied by the feature map x before the squeeze operation. The formula is:

[0065]

[0066] In the formula, This is the final feature map with channel attention applied.

[0067] Step 5: After each downsampling operation, the feature map is input into the pyramid expansion coordinate attention module, and multi-scale feature fusion is achieved through dilated convolution and coordinate attention to improve the network's receptive field and feature expression ability;

[0068] Step 501: performing a plurality of convolution operations with different dilation rates on the preliminary effective feature map, where the dilation rates selected by the module are 1, 2, 3 and 4, respectively, to obtain a multi-scale feature map;

[0069] Step 502: The multi-scale feature maps obtained by the four types of dilated convolutions are concatenated according to the channel dimension, and then a 1×1 convolution is performed to interactively fuse the multi-scale information of each channel;

[0070] Step 503: The above feature maps are subjected to coordinate attention again to redistribute the attention of channel information and spatial information, and finally obtain an effective feature layer with multi-scale robustness and coordinate sensitivity;

[0071] Step 6: In the decoding path, the feature map is input to the spatial attention module after each upsampling operation;

[0072] In step 6: first, perform global feature extraction on the input feature map of size H×W×C in the channel dimension, that is, average and extreme values ​​of all channels corresponding to each feature point to obtain two H×W×1 feature maps; stack the two feature maps representing global information in the channel direction, and then reduce the number of channels to 1 through a 7×7 convolution operation, and then use the sigmoid function for activation to transform the value of the feature map to the range of 0 to 1, that is, obtain the attention weight map in the spatial direction; its formula is:

[0073]

[0074] Among them, f 7×7 represents a convolution kernel of size 7×7, σ represents the sigmoid activation function; Represents the two-dimensional feature map obtained by taking the average value of the channel dimension, Represents the two-dimensional feature map obtained by taking the maximum value of the channel dimension;

[0075] Perform matrix multiplication of the spatial attention weight map and the input feature map to obtain the output feature map recalibrated in the spatial direction; then go through a Conv Block to deepen the feature extraction, and then repeat the above process to obtain the final effective feature layer, and then go through a layer of convolution and activation to obtain the preliminary segmentation result;

[0076] Step 7: Introduce a full-scale skip connection module between each encoder layer and decoder layer. The full-scale skip connection module improves the efficiency of feature transmission and reuse by fusing features of different scales, so as to achieve the effect of capturing fine-grained and coarse-grained semantics at all scales and obtain a more accurate segmentation result.

[0077] In step 7, since the stacking of channel dimensions requires the resolution of feature maps to be consistent, the resolution of feature maps of different scales in the encoding path is first unified; the feature maps with more underlying information are and After 4 times non-overlapping maximum pooling downsampling and 2 times downsampling respectively, the feature maps after pooling are further extracted by 3×3 convolution; the feature maps of the same scale are After only 3×3 convolution, the decoding path feature map with small resolution but relatively high-level semantic information is transformed into and The images are upsampled by four times and two times through bilinear interpolation, and each of them is also convolved by 3×3. The five scale feature maps with unified resolution are stacked in the channel dimension and uniformly passed through a 3×3 convolution layer for information fusion and channel interaction, that is, the detailed information such as the target boundary contained in the bottom feature map is aggregated with the target category and positioning information contained in the high-level feature map.

[0078] In the encoding stage, steps 1 to 5 are repeated until five preliminary valid feature layers with resolutions of 256×256, 128×128, 64×64, 32×32 and 16×16 are obtained; in the decoding stage, step 6 is repeated, and step 7 runs through all steps to obtain the final valid feature layer, and the final segmentation result is obtained after a layer of convolution and activation.

[0079] The technology of the present invention is implemented in the data set ISLES2022, which provides a total of 250 annotated multimodal MR image training sets, including three modalities: fluid attenuated inversion recovery FLAIR, magnetic resonance diffusion weighted imaging DWI, and apparent diffusion coefficient ADC. Among them, DWI utilizes the difference in the diffusivity of intracellular and extracellular water, and can detect cerebral ischemia earlier than other modalities. It is the most sensitive sequence for detecting acute ischemic stroke, also known as the "stroke sequence". It is highly sensitive to ischemic areas within 15 to 30 minutes after the onset of symptoms, clearly showing high-intensity signal areas, and the image lesions have a high contrast with normal tissues, and the lesions are clear at a glance. This embodiment uses DWI modality images in the case for segmentation, and each case contains more than 22 to 70 two-dimensional slices. Regarding the model process, the number of network iterations used for training is set to 300 times, the batch size is 24 times, Adam is used as the optimizer, and the initial learning rate is set to 0.0001.

[0080] In order to verify the effectiveness of the technology of the present invention, Figure 3 The segmentation results of the technology of the present invention on the ISLES2022 dataset are shown, and compared with the methods of U-Net, U-Net++, and U-Net3++. The red arrows indicate the incorrect segmentation areas. The technology of the present invention can take into account the segmentation of lesions of different sizes, and can distinguish lesions well for inputs with multiple lesion areas, which has a certain performance improvement compared to the benchmark network method. The above results prove the effectiveness of the technology of the present invention.

[0081] In order to verify the effectiveness of the technology of the present invention, Table 1 uses the indicators Dice (twice the intersection of the predicted area and the true area divided by the sum of pixels of the predicted area and the true area), Precision (accuracy), Recall (recall rate, measuring the ratio of the area correctly predicted by the model to the sum of pixels of the true area), and IoU (Intersection over Union, measuring the ratio of the area correctly predicted to the area of ​​the union of the predicted value and the true value) to compare the technology of the present invention with the existing technology. As shown in Table 1, the technology of the present invention has achieved better results in multiple indicators, thus proving the effectiveness of the technology of the present invention.

[0082] Table 1 Quantitative comparison results between the invention technology and the prior art

[0083] Method Dice IoU Recall PRE SegNet 0.7863 0.6522 0.7614 0.7712 PSPNet 0.8213 0.7004 0.8006 0.8080 DeepLabv3+ 0.8301 0.7124 0.7346 0.8333 U-Net 0.8366 0.7250 0.7393 0.8646 U-Net++ 0.8553 0.7511 0.8374 0.8448 U-Net3+ 0.8618 0.7614 0.8306 0.8655 The technology of the present invention 0.8863 0.8005 0.8781 0.9154

[0084] In view of the problem that downsampling may cause information loss, the present invention adopts a method of lossless two-dimensional discrete wavelet pooling of feature maps, reduces the dimension of feature maps in spatial dimension, thereby achieving the effect of increasing the network receptive field of the traditional pooling layer, and retains all feature map information by increasing the number of feature map channels. In addition, in view of the problem that stroke images have individual differences according to different patients, resulting in different lesion locations, shapes, and sizes, the present invention effectively improves the adaptability of the network to lesion inputs with different sizes, and can make full use of the context information between pixels from a global perspective to make more accurate judgments on the segmentation edges, taking into account the segmentation of lesions of different sizes, and can perform good lesion differentiation for inputs with multiple lesion areas.

[0085] The above is only a preferred embodiment of the present invention. It should be noted that, for those skilled in the art, several improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications should also be considered as the protection scope of the present invention. The structures, devices and operating methods not specifically described and explained in the present invention shall be implemented according to the conventional means in the art unless otherwise specified and limited.

Claims

1. A method for ischemic stroke image segmentation based on wavelet pooling neural network, characterized in that: The following steps are involved: Step 1: Using the U-Net convolutional neural network as the basic network architecture, add the convolutional residual module Identity Block and the convolutional residual block Conv Block to each convolutional block in the encoding path; Step 2: Input the preprocessed 256×256×4 feature map into the convolution block of the encoding path. After the feature map passes through the first double-layer convolution residual block Conv Block, a 256×256×64 feature map with inconsistent dimensions is obtained; and then passes through the convolution residual module Identity Block once more; Get the preliminary effective feature map of the first layer, at this time the dimensions before and after convolution are consistent; Step 3: Increase the number of feature map channels to 64 and obtain the first preliminary effective feature layer, which is then losslessly sampled through the wavelet pooling module; Step 4: Connect the channel attention SE module after the wavelet pooling layer to redirect channel features through squeezing, excitation and recalibration; Step 5: After each downsampling operation, the feature map is input into the pyramid dilation coordinate attention module to achieve multi-scale feature fusion through dilated convolution and coordinate attention; Step 6: In the decoding path, the feature map is input to the spatial attention module after each upsampling operation; Step 7: A full-scale skip connection module is introduced between each encoder layer and decoder layer. The full-scale skip connection module fuses features of different scales to capture the fine-grained and coarse-grained semantics at all scales and obtain the final segmentation result.

2. The ischemic stroke image segmentation method based on wavelet pooling neural network according to claim 1 is characterized in that: In step 1, the residual module algorithm is described as follows: the residual module consists of two consecutive convolutional layers and a skip connection; the input feature passes through the first convolutional layer, then passes through the ReLU activation function, and the output result is passed to the second convolutional layer; the output of the second convolutional layer is added to the input feature to obtain the output of the residual module.

3. The ischemic stroke image segmentation method based on wavelet pooling neural network according to claim 1 is characterized in that: The step 3 specifically comprises the following steps: Step 301: In the wavelet pooling module, the discrete wavelet decomposition process is described as follows: the input image is first subjected to a one-dimensional wavelet transform in the horizontal direction, that is, it is simultaneously subjected to high-pass and low-pass filters to extract high- and low-frequency information in the horizontal direction; the high- and low-frequency components obtained in the first step are subjected to a one-dimensional wavelet transform in the vertical direction respectively, that is, four image components after two-dimensional wavelet decomposition are obtained; Step 302: input the four image components with half the resolution obtained by two-dimensional discrete wavelet transform into the pooling layer for processing; Step 303: The sub-images obtained by the two down-sampling methods of maximum pooling and discrete wavelet decomposition are stacked in the channel dimension and used together as the input of the next layer of the network.

4. The ischemic stroke image segmentation method based on wavelet pooling neural network according to claim 3 is characterized in that: In step 301, the four image components are: the component HH filtered by horizontal high frequency and vertical high frequency represents the details of the diagonal direction of the image, the component HL filtered by horizontal high frequency and vertical low frequency represents the horizontal edge information of the image, the component LH filtered by horizontal low frequency and vertical high frequency represents the vertical edge information of the image, and the component LL filtered by horizontal low frequency and vertical low frequency represents the approximate information of the image.

5. The ischemic stroke image segmentation method based on wavelet pooling neural network according to claim 4 is characterized in that: The step 4 comprises the following steps: Step 401: Given a feature map x with an input size of H×W×C, where H is the height of a single feature map, W is the width, and C is the number of channels; the squeezing operation compresses the input feature map along the spatial length and width dimensions, that is, performs a global average pooling operation, compresses each two-dimensional feature map into a real number and arranges them along the channel direction, that is, obtains a global angle feature vector 1×1×C with a spatial direction, which represents the description of each feature map on the channel, and its formula for the channel is as follows: Among them, x c represents the cth channel of the input feature x; z c is the global average value of the cth channel; Step 402: In the excitation operation, the 1×1×C global feature vector obtained by the squeezing operation is first passed through a fully connected layer with a small number of neurons for channel information interaction learning, while reducing the channel dimension to C / r; then it is activated by a ReLU layer; then it is further passed through a fully connected layer with a length of C for channel information fusion and weight learning, and activated by a sigmoid function to reduce the weight range to between 0 and 1. The size of the weight represents the importance of the corresponding channel feature map; the excitation formula is as follows: s=F ex (z,W)=σ(g(z,W))=σ(W2δ(W1z)), Where W1 is a fully connected layer with dimension C / r, r is the drop ratio, W2 is a fully connected layer with dimension C, δ is the ReLU activation function, σ is the sigmoid activation function; s is the obtained attention weight; Step 403: In the channel feature recalibration operation, the attention weight s obtained through the fully connected layer and nonlinear learning is multiplied by the feature map x before the squeeze operation. The formula is: In the formula, This is the final feature map with channel attention applied.

6. The ischemic stroke image segmentation method based on wavelet pooling neural network according to claim 5 is characterized in that: The step 5 comprises the following steps: Step 501: performing a plurality of convolution operations with different dilation rates on the preliminary effective feature map, where the dilation rates selected by the module are 1, 2, 3 and 4, respectively, to obtain a multi-scale feature map; Step 502: The multi-scale feature maps obtained by the four types of dilated convolutions are concatenated according to the channel dimension, and then a 1×1 convolution is performed to interactively fuse the multi-scale information of each channel; Step 503: The above feature maps are subjected to coordinate attention again to redistribute the attention of channel information and spatial information, and finally an effective feature layer with multi-scale robustness and coordinate sensitivity is obtained.

7. The ischemic stroke image segmentation method based on wavelet pooling neural network according to claim 6 is characterized in that: In step 6: first, global feature extraction is performed on the input feature map of size H×W×C in the channel dimension, that is, the values ​​on all channels corresponding to each feature point are averaged and extreme values ​​are calculated to obtain two H×W×1 feature maps; the two feature maps representing global information are stacked in the channel direction, and then a 7×7 convolution operation is performed to reduce the number of channels to 1, and then a sigmoid function is used for activation to transform the value of the feature map to the range of 0 to 1, that is, the attention weight map in the spatial direction is obtained; its formula is: Among them, f 7×7 represents a convolution kernel of size 7×7, σ represents the sigmoid activation function; Represents the two-dimensional feature map obtained by taking the average value of the channel dimension, Represents the two-dimensional feature map obtained by taking the maximum value of the channel dimension; By performing matrix multiplication of the spatial attention weight map and the input feature map, we can obtain the output feature map recalibrated in the spatial direction; then a ConvBlock is used to deepen the feature extraction, and then the above process is repeated to obtain the final effective feature layer, and a layer of convolution and activation is performed to obtain the preliminary segmentation result.

8. The ischemic stroke image segmentation method based on wavelet pooling neural network according to claim 7 is characterized in that: In step 7, the resolution of the feature maps of different scales in the encoding path is first unified; the feature maps with more underlying information are and After 4 times non-overlapping maximum pooling downsampling and 2 times downsampling, the feature maps after pooling are further extracted by 3×3 convolution. After only 3×3 convolution, the decoding path feature map with small resolution but relatively high-level semantic information is transformed into and The images are upsampled by four times and two times through bilinear interpolation, and each of them is also convolved by 3×3. The five scale feature maps with unified resolution are stacked in the channel dimension and uniformly passed through a 3×3 convolution layer for information fusion and channel interaction, that is, the detailed information such as the target boundary contained in the bottom feature map is aggregated with the target category and positioning information contained in the high-level feature map. In the encoding stage, steps 1 to 5 are repeated until five preliminary valid feature layers with resolutions of 256×256, 128×128, 64×64, 32×32, and 16×16 are obtained; In the decoding stage, step 6 is repeated, and step 7 runs through all steps to obtain the final effective feature layer, and the final segmentation result is obtained after a layer of convolution and activation.

Images