Medical image segmentation method based on Fourier transform and related device

Abstract

The invention provides a medical image segmentation method based on Fourier transform and a related device. The method comprises the following steps: coding and decoding a to-be-segmented medical image to obtain a pre-segmented feature map; fourier transform and inverse Fourier transform are carried out on the medical image to be segmented, convolution features of the inverse transform image and encoded features of the first image feature are spliced to obtain spliced features, and convolution down-sampling processing and pooling processing are carried out on the spliced features to obtain transform features; performing down-sampling on the transformation features according to different resolutions to obtain a plurality of down-sampling features, and fusing the plurality of down-sampling features; generating segmentation head parameters according to the organ semantic features of the fused image features, and configuring segmentation heads corresponding to various organs according to the segmentation head parameters; segmenting the pre-segmented feature map through a segmentation head; edge details and semantic relations between organs are captured and strengthened to be used as priori knowledge, and the medical image is accurately segmented according to tissue structures between different organs.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of image processing, and in particular to a medical image segmentation method and related devices based on Fourier transform. Background Art

[0002] For current medical image segmentation methods, the task conditions only include a priori information of a single organ to be segmented, and the positions and semantic relationships between different organs are not fully utilized, resulting in low model segmentation accuracy.

[0003] For example, using one-hot labels as task encodings provides prior information, thereby guiding the model to process specific tasks and focus on specific areas, but ignoring the semantic relationship between different organs. Given one-hot labels for the liver [1,0,0], pancreas [0,1,0], and spleen [0,0,1], this will result in no semantic difference between the liver, pancreas, and spleen.

[0004] For example, using text embeddings learned from contrastive language-image pre-training provides anatomical prior information, enabling the model to learn structured feature embeddings, but it is difficult to give detailed guidance on edges, textures, and shapes. Summary of the invention

[0005] The following is a summary of the subject matter described in detail herein. This summary is not intended to limit the scope of the claims.

[0006] The purpose of the present application is to solve one of the technical problems existing in the related art to at least a certain extent. The embodiments of the present application provide a medical image segmentation method and related devices based on Fourier transform, which can accurately segment medical images according to the tissue structures between different organs.

[0007] An embodiment of the first aspect of the present application is a medical image segmentation method based on Fourier transform, comprising:

[0008] Acquire a medical image to be segmented;

[0009] Performing encoding and decoding processing on the medical image to be segmented to obtain a pre-segmentation feature map;

[0010] Performing Fourier transform on the medical image to be segmented to obtain frequency domain features, performing inverse Fourier transform on the frequency domain features to obtain an inverse transformed image, splicing features obtained by convolution of the inverse transformed image and features obtained by encoding the first image features to obtain spliced ​​features, and performing convolution downsampling and pooling on the spliced ​​features to obtain transformed features;

[0011] Downsampling the transformation features according to different resolutions to obtain a plurality of downsampled features, and fusing the plurality of downsampled features to obtain a fused image feature;

[0012] Generating segmentation head parameters corresponding to various organs according to the organ semantic features of the fused image features, and configuring segmentation heads corresponding to various organs according to the segmentation head parameters;

[0013] The pre-segmented feature map is segmented by the segmentation head to obtain a segmented image.

[0014] According to certain embodiments of the first aspect of the present application, encoding and decoding the medical image to be segmented to obtain a pre-segmentation feature map includes:

[0015] Performing multiple downsampling processes on the medical image to be segmented to achieve encoding processing and obtain an encoded image;

[0016] The encoded image is subjected to multiple upsampling processes to achieve decoding processing, thereby obtaining a pre-segmentation feature map.

[0017] According to certain embodiments of the first aspect of the present application, performing inverse Fourier transform on the frequency domain features to obtain an inverse transformed image includes:

[0018] The result of multiplying the absolute value of the frequency domain feature by the transformation matrix is ​​subjected to inverse Fourier transformation to obtain an inverse transformation image.

[0019] According to certain embodiments of the first aspect of the present application, the transformation matrix is ​​obtained based on the distance from each pixel point of the medical image to be segmented to a central pixel point.

[0020] According to some embodiments of the first aspect of the present application, downsampling the transformed features to obtain downsampled features includes:

[0021] Flattening the transformed feature to obtain a flattened feature;

[0022] Performing convolution processing, reshaping processing and layer normalization processing on the flattened feature to obtain a first sub-feature;

[0023] Performing linear transformation processing, convolution processing, activation function processing and feature extraction processing on the first sub-feature to obtain a second sub-feature;

[0024] Performing linear transformation processing and activation function processing on the first sub-feature to obtain a third sub-feature;

[0025] A fusion result of the second sub-feature and the third sub-feature is subjected to linear transformation to obtain a down-sampled feature.

[0026] According to certain embodiments of the first aspect of the present application, the medical image segmentation method based on Fourier transform further includes: adjusting parameters of a model corresponding to the medical image segmentation method according to the first loss function;

[0027] The first loss function is: L = l dice +l CE , Where n v is the number of voxels in the data set of the nth medical image to be segmented, p iL is the probability vector of the i-th voxel label L, y iL is the one-hot encoding vector of the i-th voxel label L, TP L Indicates that the label of the medical image to be segmented is a true positive, FP L Indicates that the label of the medical image to be segmented is a false positive, FN L Indicates that the label of the medical image to be segmented is a false negative, ε is the preset parameter, L is the first loss function, l dice is the Dice loss function, l CE is the cross loss function.

[0028] According to certain embodiments of the first aspect of the present application, the medical image segmentation method based on Fourier transform further includes: adjusting parameters of a model corresponding to the medical image segmentation method according to a second loss function;

[0029] The second loss function is: target=label☉softmax(mask), In the formula, Loss is the second loss function, is the category of the dataset of medical images to be segmented, τ is the number of categories annotated by experts in the dataset of medical images to be segmented, and D n is the dataset of the nth medical image to be segmented, m is the total number of organ categories of the medical image to be segmented, and λ n is the adjustment parameter, label is the real label, target is the pseudo label after the prediction result of the unannotated voxel of the medical image to be segmented is removed from the background, mask is the unannotated organ segmentation result in the data set of the medical image to be segmented, and output is L is the segmentation result of the Lth organoid, and Φm is the annotated category in the background.

[0030] An embodiment of the second aspect of the present application is a medical image segmentation device based on Fourier transform, comprising:

[0031] An input unit, used for acquiring a medical image to be segmented;

[0032] A coding and decoding unit, used for performing coding and decoding processing on the medical image to be segmented to obtain a pre-segmentation feature map;

[0033] A Fourier transform unit, configured to perform Fourier transform on the medical image to be segmented to obtain frequency domain features, perform inverse Fourier transform on the frequency domain features to obtain an inverse transformed image, concatenate features obtained by convolution of the inverse transformed image and features obtained by encoding of the first image features to obtain concatenated features, and perform convolution downsampling and pooling on the concatenated features to obtain transformed features;

[0034] A feature fusion unit, configured to downsample the transformation feature according to different resolutions to obtain a plurality of downsampled features, and fuse the plurality of downsampled features to obtain a fused image feature;

[0035] A segmentation head parameter generating unit, used to generate segmentation head parameters corresponding to various organs according to the organ semantic features of the fused image features, and configure segmentation heads corresponding to various organs according to the segmentation head parameters;

[0036] The segmentation head unit is used to segment the pre-segmentation feature map through the segmentation head to obtain a segmented image.

[0037] An embodiment of the third aspect of the present application is an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the medical image segmentation method based on Fourier transform described in the embodiment of the first aspect of the present application is implemented.

[0038] An embodiment of the fourth aspect of the present application is a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the Fourier transform-based medical image segmentation method described in the embodiment of the first aspect of the present application.

[0039] The present application has the following beneficial effects: by encoding and decoding the segmented medical image, the key image feature representation of the semantic information of the organ is extracted; by performing Fourier transform, inverse Fourier transform and feature splicing on the segmented medical image, the edge details and the semantic relationship between organs are captured and enhanced as prior knowledge to guide the subsequent segmentation process; the prior knowledge is integrated with the deep features extracted by the encoder to generate a dynamic segmentation head, thereby improving the accuracy of medical image segmentation. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings are used to provide further understanding of the technical solution of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the technical solution of the present application and do not constitute a limitation on the technical solution of the present application.

[0041] Figure 1 is a step diagram of a medical image segmentation method based on Fourier transform provided in an embodiment of the present application;

[0042] Figure 2It is a sub-step diagram of step S200 provided in an embodiment of the present application;

[0043] Figure 3 It is a sub-step diagram of step S400 provided in an embodiment of the present application;

[0044] Figure 4 is a data processing schematic diagram of step S300 provided in an embodiment of the present application;

[0045] Figure 5 It is a data processing diagram of step S400 provided in an embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0047] It should be noted that, although the functional modules are divided in the device schematic diagram and the logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first", "second", etc. in the specification, claims or the above drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0048] The embodiments of the present application are further described below in conjunction with the accompanying drawings.

[0049] An embodiment of the present application provides a medical image segmentation method based on Fourier transform.

[0050] Reference Figure 1 , a medical image segmentation method based on Fourier transform, comprising the following steps:

[0051] Step S100, obtaining a medical image to be segmented;

[0052] Step S200, encoding and decoding the medical image to be segmented to obtain a pre-segmentation feature map;

[0053] Step S300, performing Fourier transform on the medical image to be segmented to obtain frequency domain features, performing inverse Fourier transform on the frequency domain features to obtain an inverse transform image, concatenating features obtained by convolution of the inverse transform image and features obtained by encoding the first image features to obtain a concatenated feature, and performing convolution downsampling and pooling on the concatenated feature to obtain a transformed feature;

[0054] Step S400, downsampling the transformed features according to different resolutions to obtain multiple downsampled features, and fusing the multiple downsampled features to obtain a fused image feature;

[0055] Step S500, generating segmentation head parameters corresponding to various organs according to the organ semantic features of the fused image features, and configuring the segmentation heads corresponding to the various organs according to the segmentation head parameters;

[0056] Step S600, segmenting the pre-segmented feature map through a segmentation head to obtain a segmented image.

[0057] In step S100, a CT scan is performed by a scanner to obtain a CT image. Exemplarily, the CT image is an image containing the pheochromocytoma and multiple surrounding organs. By performing image segmentation on the CT image containing the pheochromocytoma and multiple surrounding organs, the adjacent relationship between the tumor and the surrounding tissues and the distribution of the tumor blood supply arteries can be more clearly displayed, providing important reference information for formulating a surgical plan before surgery.

[0058] Preprocess the CT images. Adjust the CT images to the RAI direction, resample to 1mm×1mm×1mm, and remove the background using threshold segmentation. Crop the images to 64mm×192mm×192mm and normalize the intensity. The images are cropped to the range [500,500]. Then z-score normalization is performed based on the mean and standard deviation of the intensity values, followed by data augmentation.

[0059] The CT image is input into the image segmentation model as the medical image to be segmented.

[0060] Reference Figure 2 For step S200, encoding and decoding the medical image to be segmented are performed to obtain a pre-segmentation feature map, including the following steps:

[0061] Step S210, performing multiple downsampling processes on the medical image to be segmented to achieve encoding processing, thereby obtaining an encoded image;

[0062] Step S220, performing multiple upsampling processes on the encoded image to achieve decoding processing, and obtaining a pre-segmentation feature map.

[0063] Exemplarily, the medical image to be segmented is encoded and decoded by a U-shaped encoder-decoder based on nnU-Net.

[0064] The encoder performs five downsampling operations. In each downsampling process, the resolution of the input feature map is halved using a convolution with a stride of 2. The number of channels in the first layer of the encoder is set to 32, and the number of channels doubles after each downsampling process, and the number of channels in the last layer is 320. Given an input image X ij, the image features output by the encoder are: Among them, θ e are encoder parameters.

[0065] Accordingly, the decoder fuses the corresponding feature maps from the encoder through 5 upsamplings to obtain high-resolution feature representation. After each upsampling, the number of decoder output channels is reduced by half, and the decoder output features are: Among them, θ de represents the encoder parameters, is the pre-segmentation feature map, and the number of channels C is set to 32.

[0066] The encoder-decoder aims to generate a pre-segmentation feature map, which is rich in image semantic information and contains key visual features of multiple abdominal organs and pheochromocytoma, independent of the specific segmentation task.

[0067] Reference Figure 4 For step S300, the medical image to be segmented is subjected to fast Fourier transform to obtain frequency domain features, which can be expressed as follows: F =FFT(X ij ). Among them, X F is the frequency domain feature.

[0068] The absolute value of the frequency domain feature is multiplied by the transformation matrix and the result is fast Fourier inverse transform to obtain the inverse transform image It is expressed by the following formula: ij ′=IFFT(|FFT(X ij )|T M ).

[0069] The transformation matrix is ​​based on the distance L from each pixel of the medical image to be segmented to the central pixel. d The distance from each pixel to the central pixel is expressed by the following formula: The transformation matrix is ​​expressed as follows: Among them, d and n are hyperparameters to assist in adjusting the degree of noise removal, and are set to 10 and 5 according to experience. M represents the center point, is the pixel point of the medical image to be segmented.

[0070] The main purpose of the transformation matrix is ​​to filter out noise. After filtering, the features of removing various types of noise are obtained.

[0071] Fourier transform frequency domain processing is used to filter out noise and help capture the structural information and texture features in the original image. The original image features contain a lot of image spatial information. It is crucial to fuse the two and extract them into a more effective feature representation. Considering the inherent redundancy between the extracted original image features and the structural texture features, the transformed image is compressed. Features obtained by convolution of the spliced ​​inverse transformed image and the first image feature encoded The splicing features are obtained, which further enhances the spatial receptive field, compresses the number of features, and supplements the spatial information missing from the Fourier transform.

[0072] The concatenated features are processed by convolution downsampling to obtain the feature map Pooling process obtains the transformation features. It is expressed as follows:

[0073] The edge contour features of organs are extracted as prior information through frequency domain processing. Frequency domain processing is mainly for noise reduction and filtering, better capturing the structural information and texture features in the image, improving the perception of semantic boundaries and subtle boundaries, and guiding the model's understanding of complex scenes, thereby improving segmentation accuracy.

[0074] Reference Figure 3 For step S400, downsampling the transformed features to obtain downsampled features includes the following steps:

[0075] Step S410, flattening the transformed features to obtain flattened features;

[0076] Step S420, performing convolution processing, reshaping processing and layer normalization processing on the flattened feature to obtain a first sub-feature;

[0077] Step S430, performing linear transformation processing, convolution processing, activation function processing and feature extraction processing on the first sub-feature to obtain a second sub-feature;

[0078] Step S440, performing linear transformation processing and activation function processing on the first sub-feature to obtain a third sub-feature;

[0079] Step S450: Perform a linear transformation on the fusion result of the second sub-feature and the third sub-feature to obtain a down-sampled feature.

[0080] Exemplarily, the transformed features are instance normalized (IN) to ensure the consistency of each feature channel, and the ReLU activation function is applied to enhance the nonlinear expression of the features.

[0081] Reference Figure 5, the shape of the transformed feature becomes (B, C, D, H, W), and the transformed feature is flattened to (B, C, L), where L = D × H × W. Multi-resolution downsampling is performed through a one-dimensional convolution operation, for example, using three different resolutions of 1, 2, and 4. Processing inputs of three different resolutions enhances the model's perception and understanding of local details and global semantics. Finally, these fused features are mapped to a one-dimensional sequence, and features at different levels are fused and optimized to obtain downsampled features with different organ semantic relationships.

[0082] The first sub-feature is subjected to linear transformation processing, ID convolution processing, SiLU activation function processing and feature extraction processing of the SSM framework to obtain the second sub-feature. The first sub-feature is subjected to linear transformation processing and SiLU activation function processing to obtain the third sub-feature.

[0083] For step S500, segmentation head parameters corresponding to various organs are generated according to the organ semantic features of the fused image features, and segmentation heads corresponding to various organs are configured according to the segmentation head parameters. Exemplarily, the features containing semantic relations of different organs are reduced to 153 by a multilayer perceptron (MLP) to obtain the segmentation head parameters.

[0084] For step S600, the pre-segmentation feature map is segmented by a segmentation head to obtain a segmented image. Exemplarily, a segmentation head consisting of three consecutive convolutional layers with a convolution kernel size of 1×1×1 is used for segmentation. The first two layers of each type of segmentation head have 8 channels, the last layer has 1 channel, and the total number of parameters is 153. 14 independent segmentation heads are configured in the model, each for identifying and locating a specific organ. The prediction result of the segmented image is expressed by the following formula: Among them, * represents convolution, θ k ={θ k1 ,θ k2 ,θ k3} represents the convolutional segmentation head parameters.

[0085] In the different parts of the annotated data, only the anatomical structure of interest is annotated, and the rest is used as the background (the default value is 0). Since the frequency domain information cannot guide the model to process specific tasks, for given annotated voxels, their labels are certain, providing clear and accurate supervision signals, while the rest of the background is ambiguous, causing model segmentation uncertainty. In order to solve this problem, this paper improves the loss function, which can annotate voxels for partial annotated data sets, consider label ambiguity, remove voxels predicted not to belong to the background from the background, dynamically adjust the loss calculation, effectively alleviate the ambiguity of the background area, and further optimize the segmentation accuracy.

[0086] Specifically, the annotated voxels use supervised loss for loss calculation, using Dice and cross entropy loss functions, dynamically adjusted according to the partial annotation data set, and adjusted the parameters of the model corresponding to the medical image segmentation method according to the first loss function; the first loss function is: L = l dice +l CE , Where n v is the number of voxels in the data set of the nth medical image to be segmented, p iL is the probability vector of the i-th voxel label L, y iL is the one-hot encoding vector of the i-th voxel label L, TP L Indicates that the label of the medical image to be segmented is a true positive, FP L Indicates that the label of the medical image to be segmented is a false positive, FN L Indicates that the label of the medical image to be segmented is a false negative, ε is a preset parameter, set to 1, L is the first loss function, l dice is the Dice loss function, l CE is the cross loss function.

[0087] Unannotated voxels are annotated as the real background or one of the m organs. The prediction results of unannotated voxels are removed from the background and used as pseudo labels for loss calculation. The number of annotated organs in different data sets is inconsistent, which causes large fluctuations in the loss function. In order to make the training process more stable, a data set weight coefficient is added. The parameters of the model corresponding to the medical image segmentation method are adjusted according to the second loss function; the second loss function is: target=label☉softmax(mask), In the formula, Loss is the second loss function, is the category of the dataset of medical images to be segmented, τ is the number of categories annotated by experts in the dataset of medical images to be segmented, and D n is the dataset of the nth medical image to be segmented, m is the total number of organ categories of the medical image to be segmented, and λ n is the adjustment parameter, label is the real label, target is the pseudo label after the prediction result of the unannotated voxel of the medical image to be segmented is removed from the background, mask is the unannotated organ segmentation result in the data set of the medical image to be segmented, and output is L is the segmentation result of the Lth organoid, and Φm is the annotated category in the background.

[0088] The dynamic loss function excludes voxels that are predicted as non-background but annotated as background, and dynamically adjusts the loss calculation for the annotated voxels, which effectively alleviates the ambiguity of the background area. It plays a key guiding role in the training process and further improves the quality of tumor and surrounding organ segmentation in medical images.

[0089] An embodiment of the present application provides a medical image segmentation device based on Fourier transform.

[0090] A medical image segmentation device based on Fourier transform comprises: an input unit, a coding and decoding unit, a Fourier transform unit, a feature fusion unit, a segmentation head parameter generation unit and a segmentation head unit.

[0091] Among them, the input unit is used to obtain the medical image to be segmented; the encoding and decoding unit is used to encode and decode the medical image to be segmented to obtain a pre-segmentation feature map; the Fourier transform unit is used to perform Fourier transform on the medical image to be segmented to obtain frequency domain features, perform inverse Fourier transform on the frequency domain features to obtain an inverse transform image, splice the features obtained by convolution of the inverse transform image and the features obtained by encoding the first image features to obtain spliced ​​features, and perform convolution downsampling and pooling on the spliced ​​features to obtain transformation features; the feature fusion unit is used to downsample the transformation features according to different resolutions to obtain multiple downsampled features, and fuse the multiple downsampled features to obtain fused image features; the segmentation head parameter generation unit is used to generate segmentation head parameters corresponding to various organs according to the organ semantic features of the fused image features, and configure segmentation heads corresponding to various organs according to the segmentation head parameters; the segmentation head unit is used to segment the pre-segmentation feature map through the segmentation head to obtain a segmented image.

[0092] It can be understood that the various units of the medical image segmentation device provided in the embodiments of the present application correspond one-to-one to the various steps of the medical image segmentation method. The medical image segmentation device and the medical image segmentation method have the same technical solution, solve the same technical problems, and bring the same technical effects, which will not be described in detail here.

[0093] An embodiment of the present application provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the medical image segmentation method based on Fourier transform as described above when executing the computer program.

[0094] The electronic device may be any intelligent terminal including a computer.

[0095] In general, for the hardware structure of the electronic device, the processor can be implemented using a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, etc., to execute related programs to implement the technical solutions provided in the embodiments of the present application.

[0096] The memory can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory can store an operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program codes are stored in the memory, and the processor is called to execute the methods of the embodiments of this application.

[0097] The input / output interface is used to realize information input and output.

[0098] The communication interface is used to realize the communication interaction between this device and other devices. Communication can be achieved through wired methods (such as USB, network cable, etc.) or wireless methods (such as mobile network, WIFI, Bluetooth, etc.).

[0099] The bus transmits information between the various components of the device (such as the processor, memory, input / output interface, and communication interface). The processor, memory, input / output interface, and communication interface communicate with each other within the device through the bus.

[0100] An embodiment of the present application provides a computer storage medium, wherein the computer storage medium stores computer executable instructions, and the computer executable instructions are used to execute the medical image segmentation method based on Fourier transform as described above.

[0101] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. In addition, it is well known to those of ordinary skill in the art that communication media typically contain computer-readable instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium. In the above description of this specification, the description of reference terms "one embodiment / example", "another embodiment / example" or "certain embodiments / examples" and the like means that the specific features, structures, materials or characteristics described in conjunction with the embodiments or examples are included in at least one embodiment or example of the present application. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0102] Those skilled in the art will appreciate that all or some of the steps in the methods disclosed above, and the functional modules / units in the systems and devices may be implemented as software, firmware, hardware, or a suitable combination thereof.

[0103] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0104] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0105] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including multiple instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of various embodiments of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory, referred to as ROM), random access memory (Random Access Memory, referred to as RAM), disk or optical disk and other media that can store programs.

[0106] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the above-mentioned units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms. Although the embodiments of the present application have been shown and described, it can be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and purposes of the present application, and the scope of the present application is defined by the claims and their equivalents.

[0107] The above is a specific description of the preferred implementation of the present application, but the present application is not limited to the embodiments. Technical personnel familiar with the field can make various equivalent modifications or substitutions without violating the spirit of the present application. These equivalent modifications or substitutions are all included in the scope defined by the claims of the present application.

Claims

1. A medical image segmentation method based on Fourier transform, characterized in that: include: Acquire a medical image to be segmented; Performing encoding and decoding processing on the medical image to be segmented to obtain a pre-segmentation feature map; Performing Fourier transform on the medical image to be segmented to obtain frequency domain features, performing inverse Fourier transform on the frequency domain features to obtain an inverse transformed image, splicing features obtained by convolution of the inverse transformed image and features obtained by encoding the first image features to obtain spliced ​​features, and performing convolution downsampling and pooling on the spliced ​​features to obtain transformed features; Downsampling the transformation features according to different resolutions to obtain a plurality of downsampled features, and fusing the plurality of downsampled features to obtain a fused image feature; Generating segmentation head parameters corresponding to various organs according to the organ semantic features of the fused image features, and configuring segmentation heads corresponding to various organs according to the segmentation head parameters; The pre-segmented feature map is segmented by the segmentation head to obtain a segmented image.

2. The medical image segmentation method based on Fourier transform according to claim 1, characterized in that: The encoding and decoding of the medical image to be segmented to obtain a pre-segmentation feature map includes: Performing multiple downsampling processes on the medical image to be segmented to achieve encoding processing and obtain an encoded image; The encoded image is subjected to multiple upsampling processes to achieve decoding processing, thereby obtaining a pre-segmentation feature map.

3. The medical image segmentation method based on Fourier transform according to claim 1, characterized in that: The performing inverse Fourier transform on the frequency domain features to obtain an inverse transform image comprises: The result of multiplying the absolute value of the frequency domain feature by the transformation matrix is ​​subjected to inverse Fourier transformation to obtain an inverse transformation image.

4. The medical image segmentation method based on Fourier transform according to claim 3, characterized in that: The transformation matrix is ​​obtained according to the distance from each pixel point of the medical image to be segmented to the central pixel point.

5. The medical image segmentation method based on Fourier transform according to claim 1, characterized in that: Downsampling the transformed features to obtain downsampled features includes: Flattening the transformed feature to obtain a flattened feature; Performing convolution processing, reshaping processing and layer normalization processing on the flattened feature to obtain a first sub-feature; Performing linear transformation processing, convolution processing, activation function processing and feature extraction processing on the first sub-feature to obtain a second sub-feature; Performing linear transformation processing and activation function processing on the first sub-feature to obtain a third sub-feature; A fusion result of the second sub-feature and the third sub-feature is subjected to linear transformation to obtain a down-sampled feature.

6. The medical image segmentation method based on Fourier transform according to claim 1, characterized in that: Also includes: Adjusting parameters of a model corresponding to the medical image segmentation method according to the first loss function; The first loss function is: L = l dice +l CE , Where n v is the number of voxels in the data set of the nth medical image to be segmented, p iL is the probability vector of the i-th voxel label L, y iL is the one-hot encoding vector of the i-th voxel label L, TP L Indicates that the label of the medical image to be segmented is a true positive, FP L Indicates that the label of the medical image to be segmented is a false positive, FN L Indicates that the label of the medical image to be segmented is a false negative, ε is the preset parameter, L is the first loss function, l dice is the Dice loss function, l CE is the cross loss function.

7. The medical image segmentation method based on Fourier transform according to claim 1, characterized in that: Also includes: Adjusting the parameters of the model corresponding to the medical image segmentation method according to the second loss function; The second loss function is: target=label⊙softmax(mask), In the formula, Loss is the second loss function, is the category of the dataset of medical images to be segmented, τ is the number of categories annotated by experts in the dataset of medical images to be segmented, and D n is the dataset of the nth medical image to be segmented, m is the total number of organ categories of the medical image to be segmented, and λ n is the adjustment parameter, label is the real label, target is the pseudo label after the prediction result of the unannotated voxel of the medical image to be segmented is removed from the background, mask is the unannotated organ segmentation result in the data set of the medical image to be segmented, and output is L is the segmentation result of the Lth organoid, and Φm is the annotated category in the background.

8. A medical image segmentation device based on Fourier transform, characterized in that: include: An input unit, used for acquiring a medical image to be segmented; A coding and decoding unit, used for performing coding and decoding processing on the medical image to be segmented to obtain a pre-segmentation feature map; A Fourier transform unit, configured to perform Fourier transform on the medical image to be segmented to obtain frequency domain features, perform inverse Fourier transform on the frequency domain features to obtain an inverse transformed image, concatenate features obtained by convolution of the inverse transformed image and features obtained by encoding of the first image features to obtain concatenated features, and perform convolution downsampling and pooling on the concatenated features to obtain transformed features; A feature fusion unit, configured to downsample the transformation feature according to different resolutions to obtain a plurality of downsampled features, and fuse the plurality of downsampled features to obtain a fused image feature; A segmentation head parameter generating unit, used to generate segmentation head parameters corresponding to various organs according to the organ semantic features of the fused image features, and configure segmentation heads corresponding to various organs according to the segmentation head parameters; The segmentation head unit is used to segment the pre-segmentation feature map through the segmentation head to obtain a segmented image.

9. An electronic device, characterized in that: The electronic device includes a memory and a processor, the memory stores a computer program, and the processor implements the Fourier transform-based medical image segmentation method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the Fourier transform-based medical image segmentation method according to any one of claims 1 to 7 is implemented.

Images